VIEWS: 0 PAGES: 233 POSTED ON: 7/27/2012
1 FOOD AND DRUG ADMINISTRATION CENTER FOR DEVICES AND RADIOLOGICAL HEALTH OPHTHALMIC DEVICES PANEL Friday, August 2, 2002 Salons A-C Hilton Hotel Gaithersburg 620 Perry Parkway Gaithersburg, Maryland 2 IN ATTENDANCE: Jayne S. Weiss, M.D., Chair Arthur Bradley, Ph.D., Voting Member Michael R. Grimmett, M.D., Voting Member Alice Y. Matoba, M.D., Voting Member Karen Bandeen-Roche, Ph.D., Consultant, deputized to vote Stephen A. Burns, Ph.D., Consultant Mark A. Bullimore, MCOptom, Ph.D., Consultant, deputized to vote Andrew J. Huang, M.D., Consultant, deputized to vote William D. Mathers, M.D., Consultant Cynthia Owsley, Ph.D., Consultant, deputized to vote William H. Swanson, Ph.D., Consultant, deputized to vote Glenda V. Such, M.Ed., Consumer Representative Ronald E. McCarley, Industry Representative 3 C O N T E N T S PAGE Call to Order Jayne S. Weiss, M.D. Chair 5 Introductory Remarks and Conflict of Interest Statement Sara M. Thornton Executive Secretary 5 Open Public Hearing John A. Vukich, M.D. University of Wisconsin 9 Questions for Public Speaker 17 Open Committee Session Jayne S. Weiss, M.D. 25 General Issues Discussion on Design Elements for Clinical Studies of Phakic Intraocular Lenses (IOLs) FDA Presentation Donna R. Lochner Chief, Intraocular and Corneal Implants Branch 25 Guest Speaker Presentations Specular Microscopy (Endothelial Cell Counts) Bernard E. McCarey, Ph.D. Emory University 28 Henry F. Edelhauser, Ph.D. Emory University 42 Questions for the Speakers 55 4 C O N T E N T S PAGE Guest Speaker Presentation Lens Opacity Liliana Werner, M.D., Ph.D. Storm Eye Institute 84 Questions for the Speaker 104 Panel Deliberations and Questions for Panel Discussion Introduction of Questions Donna R. Lochner 111, 149, 161 Question 1 Michael R. Grimmett, M.D. Discussion Leader 115 Discussion 137 Question 2 William D. Mathers, M.D. Discussion Leader 151 Discussion 156 Question 3 Mark A. Bullimore, MCOptom, Ph.D. Discussion Leader 163, 185, 186, 188 Discussion 165, 186, 187, 189 Discussion on Remaining Issues 190 Closing Comments Jayne S. Weiss, M.D. 231 Sara M. Thornton 232 5 1 P R O C E E D I N G S (8:30 a.m.) 2 DR. WEISS: I'd like to call this meeting of 3 the Ophthalmic Devices Panel to order, and we will have 4 introductory remarks from Sara Thornton. 5 MS. THORNTON: Good morning and welcome to the 6 104th meeting of the Ophthalmic Devices Panel. Before we 7 proceed with the agenda, I do have a few short 8 announcements to make. 9 I'd like to remind everyone who is new to the 10 sessions today to sign in on the attendance sheet in the 11 registration area. I'd actually like everyone to sign in, 12 but I know some of you have heard this announcement before. 13 But please do sign in right outside on the registration 14 table. 15 All handouts for the meeting are available out 16 there on the table, and we'd like, if you have messages for 17 the panel members, the FDA participants, if you need 18 information or have special needs, they should be directed 19 to Ms. Annmarie Williams, who's over here by the door, and 20 Ms. Jennifer Weber. They're both available also at the 21 registration area. 22 The phone number for calls to the meeting area 23 is (301) 977-8900, and in consideration of the panel, the 24 sponsor, and the agency, we ask that those of you with cell 25 phones and pagers please turn them off and put them on 6 1 vibration mode while you're in this room. 2 Lastly, will all meeting participants please 3 into the microphone, directly into the microphone, less 4 than four inches away, according to my instructions, and 5 give your name clearly, so that the transcriber will have 6 an accurate recording of your comments and others will be 7 able to hear you. 8 Now, at this time, I would like to announce to 9 those new to the session today the confirmation of the new 10 Ophthalmic Devices Panel Chair, Dr. Jayne Weiss, to my 11 left. We also have three newly appointed voting members, 12 Dr. Anne Coleman, Allen Ho, and Timothy McMahon, who are 13 regrettably unable to be with us today. However, we look 14 forward to their attendance at future meetings. 15 I'd also like to extend a special welcome and 16 introduce to the public, the panel, and the FDA staff three 17 panel consultants who today are with us for the first time. 18 Dr. Stephen Burns, on my left, who comes to us from 19 Boston, Massachusetts, where he is a senior scientist at 20 the Schepens Eye Research Institute and an associate 21 professor at Harvard University. Dr. Cynthia Owsley, to my 22 right, is from Birmingham, Alabama, where she is a 23 professor of ophthalmology in the School of Medicine and 24 co-director of the Center for Research on Applied 25 Gerontology at the University of Alabama. Lastly, Dr. 7 1 William Swanson, on my right, is a senior research 2 scientist in the Department of Clinical Sciences at the 3 State University of New York College of Optometry in New 4 York, New York. 5 I'd like now to read the conflict of interest 6 statement for this session of the 104th meeting. "The 7 following announcement addresses conflict of interest 8 issues associated with this meeting and is made part of the 9 record to preclude even the appearance of an impropriety. 10 "To determine if any conflict existed, the 11 agency reviewed the submitted agenda for this meeting and 12 all financial interests reported by the committee 13 participants. The conflict of interest statutes prohibit 14 special government employees from participating in matters 15 that could affect their or their employers' financial 16 interests. However, the agency has determined that 17 participation of certain members and consultants, the need 18 for whose services outweighs the potential conflict of 19 interest involved, is in the best interests of the 20 government. 21 "Therefore, waivers have been granted to Drs. 22 Mark Bullimore and Stephen Burns for their interests in 23 firms that could potentially be affected by the panel's 24 recommendations. The waivers allow these individuals to 25 participate fully in today's deliberations. Copies of 8 1 these waivers may be obtained from the agency's Freedom of 2 Information Office, Room 12A-15 of the Parklawn Building. 3 "We would like to note for the record that the 4 agency took into consideration other matters regarding Drs. 5 Arthur Bradley, Michael Grimmett, and Jayne Weiss, who 6 reported interests in firms at issue, but in matters not 7 related to today's agenda. The agency has determined, 8 therefore, that they may participate fully in all 9 discussions. 10 "Lastly, we would like to note for the record 11 that Drs. Henry Edelhauser, Bernard McCarey, and Liliana 12 Werner, all invited guest speakers today, reported 13 interests with firms at issue. Dr. Edelhauser reported a 14 personal financial interest, a consulting relationship, and 15 a professional relationship in the form of contracts and 16 research grants. Drs. McCarey and Werner reported 17 professional relationships in the form of contracts, 18 grants, or research. 19 "In the event that the discussions involve any 20 other products or firms not already on the agenda for which 21 an FDA participant has a financial interest, the 22 participant should excuse him or herself from such 23 involvement and the exclusion will be noted for the record. 24 "With respect to all participants, we ask in 25 the interest of fairness that all persons making statements 9 1 or presentations disclose any current or previous financial 2 involvement with any firm whose products they may wish to 3 comment upon." 4 Thank you, Dr. Weiss. 5 DR. WEISS: Thank you, Sally. 6 We're going to have some comments at this point 7 by Mr. David Whipple. 8 MR. WHIPPLE: Thank you. 9 I only had one comment that I wanted to make 10 this morning. I wanted to thank this panel for yesterday's 11 discussion of the labeling for the wavefront technology 12 LASIK device. I know it was long and difficult, but a very 13 important discussion for us in the agency. Not only will 14 we use your comments and recommendations as guidance in a 15 framework for building labeling for devices of this type, 16 but we will also use it for monitoring the promotion and 17 advertising when they go to market their products as well. 18 So thank you for that discussion yesterday. 19 DR. WEISS: Thank you very much. 20 We're going to proceed to the open committee 21 session and start with the FDA presentation. Excuse me. 22 I'm out of order. We're going to start with the open 23 public hearing session and Dr. John Vukich of the 24 University of Wisconsin is going to make his presentation. 25 DR. VUKICH: Thank you and good morning. 10 1 DR. WEISS: Would you be able to start by -- 2 just start by identifying yourself and your conflicts, if 3 any. Thank you. 4 DR. VUKICH: Okay. Thank you. 5 My name is John Vukich. I am an associate 6 professor at the University of Wisconsin in the Department 7 of Ophthalmology. I'm an investigator and medical monitor 8 for Staar Surgical, and it is from this experience that I 9 draw the information from which I form my opinions that I 10 will be presenting this morning. 11 My testimony today, however, is as an private 12 citizen. I have not received support or reimbursement for 13 my visit today, but I'm here to speak on behalf of phakic 14 IOLs as a segment of the refractive industry and as an 15 option for the correction of myopia. 16 Right now, LASIK is gold standard by which we 17 need to compare all future refractive technologies. We're 18 clearly trying to improve the outcomes of LASIK, and I 19 believe the decision yesterday to allow custom ablations is 20 a step in that direction. Any new technology that comes 21 along certainly is going to be compared to LASIK, and what 22 I'd like to do is present some information on the 23 comparison of LASIK and phakic IOLs. 24 We have looked at this in my practice. I am 25 primary a refractive surgeon. Most of my practice revolves 11 1 around this. We looked at 198 phakic IOLs with a mean 2 myopia of -10. We compared this to a similar number, 219 3 LASIK patients similarly high myopia of between -9.5 to 12. 4 The mean was -9.5. Predominately female and younger 5 patients in their mid-30s. 6 When we looked at the percent of patients who 7 achieved 20/20 or greater, again, in this relatively highly 8 myopic group of patients, we found that about 32 percent of 9 patients on their treatment achieved 20/20 or better with 10 LASIK -- again, with a mean correction of close to -10 -- 11 and close to 50, or 48.5 percent, were able to achieve 12 20/20 or better with phakic IOLs, and this was 13 statistically significant. The curve did fall off towards 14 the end and this was due to an early in the clinical trial 15 difference in the nomograms, and this curve has in fact 16 stayed the same all the way out and has remained consistent 17 for this difference. 18 So we believe that phakic IOLs, at least as a 19 single procedure, offer an alternative to LASIK in the 20 quality of vision that a patient might expect in the higher 21 ranges. 22 We know that recovery of visual acuity is an 23 important issue. Ultimately, patients need to retain their 24 visual acuity, but the rapidity of recovery is also an 25 issue, and the length of disability is an important issue, 12 1 of course. 2 When we look at lines lost or gained early in 3 the recovery at one week, we see that with LASIK, again in 4 the higher ranges, it is not uncommon -- almost 28 percent 5 of patients lost two lines or more of acuity early on, and 6 we can explain this because of epithelial irregularities, 7 surface irregularities, or edema in the early postoperative 8 period. We contrast that with phakic IOLs, in which not a 9 single patient in this clinical trial lost two lines or 10 more of visual acuity at week. 11 When we look at six months out, of course, we 12 would expect the epithelial changes to have recovered, and 13 in fact that is the case. However, there still were 6 14 percent of LASIK patients who had lost two lines or more. 15 This predominately fell from 20/15 to 20/25. Nevertheless, 16 this is a demonstrable loss of quality of vision that has 17 persisted through six months. Again, contrasting with 18 phakic IOLs, in which there was not a single patient who 19 had lost two lines of visual acuity. So in terms of 20 preservation of acuity, we believe that phakic IOLs offer a 21 good alternative, and perhaps superior, to corneal ablative 22 procedures. 23 We have anecdotal reports that patients prefer 24 the quality of vision with phakic IOLs. There have been a 25 limited number of patients who have had a phakic IOL in one 13 1 eye and LASIK in the other, and we have heard in fact that 2 at least one clinical trial from a prominent researcher in 3 Greece was discontinued because of the strong preference of 4 the phakic IOL eye and it was felt that continuing on that 5 did not make sense for them. 6 So based on this, we felt that perhaps maybe 7 there was something we could do to demonstrate a 8 difference, and what we have done is we have looked at 9 wavefront analysis as an objective assessment of optical 10 quality, and have now looked at comparison of the induced 11 aberrations in patients who have either received a phakic 12 IOL or LASIK. 13 These were 10 patients, 20 eyes, two eyes in 14 each patient. The mean myopia in the phakic IOL group was 15 -12 ranging to -15. The LASIK group ranged up to -10.5 16 with a mean myopia of 8.75 or a few diopters less. 17 When we looked at coma group means square 18 values, for phakic IOLs the average value was .22 or less 19 than half of the amount of coma observed in LASIK patients 20 postoperatively, and this was highly significant at the 21 .001 level. 22 We can do image convolutions to demonstrate 23 this difference looking at the standard Snellen chart. 24 This is what a patient might expect to see in simulation 25 with this much induced coma in LASIK, and this is what they 14 1 might expect to see with this much induced coma from a 2 phakic IOL. 3 We can do the same image convolution with a 4 photograph. Again, with LASIK and with the phakic IOL, and 5 we believe that there's a demonstrable difference in the 6 quality of the images that the patients see and what we can 7 demonstrate mathematically. 8 We looked at spherical aberration as well as an 9 isolated fourth-order higher term, and we see that there 10 were three times as much induced, or at least three times 11 as much observed, RMS of spherical aberration in the LASIK 12 compared to the phakic IOL. Again, significant at the .001 13 level. 14 When we look at the image convolutions of this, 15 we see the LASIK image compared to the phakic IOL image, 16 and again we can look at photographic convolutions with 17 LASIK and with the phakic IOL. 18 All of these images again demonstrate what we 19 have heard anecdotally, and that is that the patients with 20 phakic IOLs seem to be very pleased with the quality of the 21 image that they receive. 22 These images or these RMS values do combine and 23 it would probably make more sense to look at the 24 combination of terms. When we look at spherical and coma 25 RMS combined, we see LASIK versus phakic IOL, and once 15 1 again the image with phakic IOL. 2 Well, custom corneal ablation is not ideal 3 option for high myopia. The approval up to -7 yesterday I 4 believe is a step forward in our ability to correct myopia, 5 but one of the issues that I think will limit this 6 application is the fact that it can remove up to 20 microns 7 of tissue per diopter with larger ablation zones and with a 8 custom application. I believe this will ultimately limit 9 custom ablations to ranges that are already approved, at 10 least from one manufacturer, but in fact the simple physics 11 and the simple anatomy may eliminate this as a possibility 12 for higher corrections. So it would certainly be 13 beneficial to have a noncorneal alternative. 14 There is in fact a limit to how much corneal 15 tissue can be removed. This is a macroscopic view of a 16 cadaver eye that has had corneal tissue removed down to a 17 level of 100 microns. This is clearly thinner than what we 18 would do clinically, but it does demonstrate grossly the 19 elastic character of the posterior surface of the cornea 20 and again is consistent with what we can observe with 21 advanced imaging technologies. 22 Corneal ablation is certainly not appropriate 23 for some patients no matter what the correction achieved. 24 Patients with keratoconus, as demonstrated here, clearly 25 would not be suitable for corneal reshaping, but there are 16 1 certainly many more patients who have subtle changes that 2 come to our attention when we screen them for refractive 3 surgery, changes of mild elevation, changes on the 4 posterior surface elevation, or keratometric changes that 5 are subtle variants of what we would consider an abnormal 6 corneal topography or corneal anatomy, and in fact a 7 noncorneal alternative may be a superior alternative for 8 these patients as well. 9 The fact remains that there are few options 10 available to patients who have high myopia. This has led 11 to some options being employed that are not approved and in 12 fact may pose dangerous situations for patients. We have 13 certainly seen clear lens extraction as an unapproved use 14 of an approved IOL for cataract surgery, but again used in 15 a refractive manner. This is controversial. However, it 16 is being done. 17 With one anecdotal report from a clinical trial 18 center in the United States of a refractive-based practice, 19 we looked at the incidence of clear lens extraction before, 20 during, and after the availability of phakic IOLs in this 21 individual practice. The white bar beneath represents the 22 time during which enrollment was available for phakic IOLs, 23 and we can see that there was a significant decrease in the 24 total number of clear lens extractions performed as a 25 refractive procedure. At the conclusion of enrollment, 17 1 there was an over doubling of the number of clear lens 2 extractions. 3 Again, this is consistent with the researcher's 4 or with the investigator's observation that given an 5 alternative, this particular researcher would shy away from 6 clear lens extractions, and we believe that this is a 7 better alternative and perhaps something that I think would 8 offer our patients perhaps a better, or we hope safer, 9 alternative. 10 In conclusion, I would like to suggest that 11 corneal refractive surgery is an excellent opportunity for 12 patients. Many of them enjoy -- most all enjoy -- the 13 benefits of this, but I believe that a noncorneal 14 alternative is an important step forward. I believe that 15 the safety and efficacy of phakic IOLs needs to be 16 demonstrated, but certainly the opportunity to provide 17 quality of vision seems to be quite high. 18 DR. WEISS: Thank you. 19 Do any of the members of the panel have any 20 questions for Dr. Vukich? I would actually start off with 21 one question myself, which is how would you weigh the 22 potential risks of intraocular surgery -- namely, 23 endophthalmitis, albeit rare, and corneal edema -- against 24 the benefits of the visual recovery and quality of vision? 25 DR. VUKICH: Well, certainly, any time we go 18 1 inside the eye, we have to hold a different standard than 2 we would on the surface. We know that infection is an 3 option, or at least is a problem, with LASIK, albeit rare, 4 but it's somewhat devastating when it does occur. It is 5 something that needs to be looked at very carefully in 6 terms of the real incidence. 7 I think the bigger issue in my opinion is the 8 potential for removability of these devices, and that is 9 that there is an alternative to restore the eye perhaps not 10 exactly to what it was preoperatively, but to remove 11 whatever the patient may not have liked about the quality 12 of vision. If there is edge glare or halos or night vision 13 problems or decentrations with LASIK, the remedies are 14 typically not satisfactory, and in fact with the phakic 15 IOL, at least the opportunity to reverse that or remove the 16 offending treatment certainly I believe offers a 17 significant advantage. 18 DR. WEISS: Depending on if the offense is 19 irreversible endothelial cell loss or infectious organisms 20 or cataract formation. 21 DR. VUKICH: Very clearly, those are things 22 that have to be looked at. Endothelial cell counts are a 23 critical issue, as are the potential for infection. 24 DR. WEISS: I think Dr. Bradley had a question, 25 and Dr. Bullimore as well. Let's start with Dr. Bradley. 19 1 Dr. Bradley, Dr. Bullimore, and then Dr. Mathers. 2 DR. BRADLEY: I was just looking at your slide 3 where you showed visual acuity as a function of time after 4 the procedure, and I didn't quite follow your explanation 5 of why the phakic IOL percent of patients achieving 20/20 6 fell off at 12 months. You sort of ran through some sort 7 of excuse it sounded like. 8 DR. VUKICH: I'd like to think of it as the 9 reason. The early nomograms for calculation of power will 10 be represented at the last data point collected as time 11 goes on. So the first several phakic IOLs that were 12 implanted in this clinical trial systematically 13 undercorrected all the patients, and mid-course adjustment 14 or early-course adjustment and the attempted correction 15 versus achieved correction became substantially better. So 16 that dip between six months and 12 months, which was the 17 two data points at six months and 12 months and there was 18 no in-between visit, remains something that between 12 19 months and two years we have seen that seem dip, and now 20 between two years and three years we see that same dip. It 21 is just simply the leading edge representing the earliest 22 patients who were enrolled, but the remainder of the line 23 stays as it has been with the improved nomogram. 24 DR. BRADLEY: Thank you. 25 DR. WEISS: Dr. Bullimore? 20 1 DR. BULLIMORE: Yes, this is Mark Bullimore. 2 If my memory serves me correct, one of the issues discussed 3 by the panel at previous visits to this phakic IOL guidance 4 document was whether these devices should be held to the 5 same standard as LASIK. What's your impression? 6 DR. VUKICH: To the same standard in what 7 regard? 8 DR. BULLIMORE: In terms of, say, vision. 9 DR. VUKICH: In terms of quality of vision? 10 DR. BULLIMORE: Yes. 11 DR. VUKICH: I believe that that is both 12 appropriate -- I don't know that we would want to see a 13 step backward in the evolution of any technology, and so 14 holding to the same standard I believe makes sense. 15 DR. BULLIMORE: So in the absence of any other 16 information, you would argue that we should use the same 17 criteria for uncorrected visual acuity and corrected visual 18 acuity and loss of visual acuity that is currently used for 19 refractive lasers? 20 DR. VUKICH: I think that makes sense. I think 21 that it would also make sense to stratify the data into the 22 various ranges of power that you're looking at, knowing 23 that the standard for LASIK at -12 diopters should be 24 different than the standard at -1 or -2, and that the 25 outcome at that level could be a different expectation. 21 1 DR. BULLIMORE: You mean in terms of safety or 2 efficacy or both? 3 DR. VUKICH: Both. 4 DR. BULLIMORE: So you would expect your LASIK 5 patients in your -10 group to be not doing as well as 6 patients with lower degrees of myopia? 7 DR. VUKICH: I would expect higher enhancement 8 rates. I would expect potentially higher levels of 9 reported edge effect, glare, and those sort of symptoms in 10 the higher ranges. We might also anticipate that the 11 higher ranges of LASIK may in fact become lower as we 12 implement custom corneal ablations limited by the tissue 13 effect that needs to be removed. We simply don't do -12 14 LASIKs anymore. At least, I don't, and many reputable 15 surgeons or high-volume surgeons have lowered the upper 16 limit at which they will perform LASIK, and that number I 17 believe is still going down. 18 DR. BULLIMORE: But clearly there are some less 19 than reputable people doing a lot of clear lens exchange. 20 DR. VUKICH: Again, I can't speak to the 21 decisions that other surgeons make. 22 DR. BULLIMORE: Okay. That's fine. 23 DR. VUKICH: In the face of not having an 24 alternative, it seems to be happening. 25 DR. BULLIMORE: Yes. And one final question. 22 1 I mean, you presented data on 200 patients who had phakic 2 IOLs and LASIK. Were they all from your practice? 3 DR. VUKICH: Yes. 4 DR. BULLIMORE: And where these people who had 5 the phakic IOLs, were they single device? 6 DR. VUKICH: Excuse me. The phakic IOLs were 7 part of the multicenter trial. Excuse me. All the LASIK 8 patients were from my practice. 9 DR. BULLIMORE: Okay. Thanks very much. 10 DR. WEISS: Dr. Mathers? 11 DR. MATHERS: Do you have any information about 12 observation of cataract formation that might have occurred 13 later? I mean, these lenses have been implanted for some 14 time now, but we don't see any data on three years, four 15 years, or whatever, and certainly something was implanted a 16 little bit longer than we have seen data for. 17 DR. VUKICH: Yes, and we are collecting data 18 and do have a substantial amount. In fact, all of the 19 patients in at least one of the clinical trials is 20 submitted. Not submitted, but is through the two-year 21 point, and we're about halfway through the three-year 22 collection of data. So yes, that data does exist on the 23 formation of cataracts in all of the safety and efficacy 24 parameters that were approved in the protocol that's been 25 undertaken. Again, I am not prepared to do a thorough 23 1 disclosure or presentation of that information, other than 2 to say it is going to be submitted and we believe 3 represents a standard that we believe is acceptable. 4 DR. WEISS: Dr. Huang? 5 DR. HUANG: Given the known potential 6 complications of cataract, glaucoma, and retinal 7 detachment, are you implying that this phakic IOL device 8 should be limited to the higher myopias? 9 DR. VUKICH: Ultimately, the complication rate 10 is something that is going to determine whether or not 11 phakic IOLs will be appropriate. The quality of the optics 12 and the ability to correct a refractive error I believe is 13 intuitive and has been demonstrated and will be 14 demonstrated. Ultimately, how safe they are is going to be 15 the issue as to where they should be used. 16 If a product can be demonstrated to be safe at 17 any range, I see no reason that it should be limited only 18 to the higher myopic patients. I believe initially it 19 would make sense to offer this as an alternative for higher 20 myopic patients in which we know LASIK has limitations or 21 may not even be appropriate. 22 DR. WEISS: Dr. Grimmett? 23 DR. GRIMMETT: Dr. Michael Grimmett. Is your 24 experience with phakic IOLs mostly of the posterior chamber 25 type? You don't have any other experience or data 24 1 otherwise regarding anterior chamber, either angle- 2 supported or iris clip? Is that correct? 3 DR. VUKICH: All of my personal experience has 4 been with posterior chamber phakic IOLs. 5 DR. GRIMMETT: Okay. Thank you. 6 DR. WEISS: Just one follow-up question in 7 terms of the standards that the IOL should held to 8 visually. In a LASIK patient who has higher myopic error, 9 we can easily lift up the flap and enhance, but with a 10 phakic IOL, the risk of entering the eye again is higher 11 than a flap relift. With that in mind, would you still 12 hold them to the same visual results postoperatively? 13 DR. VUKICH: I think comparisons would need to 14 be made based on a one- or two-procedure comparison. I 15 believe that all of the trials for LASIK have been as a 16 single procedure without enhancement, and I think the 17 ability to enhance we understand is real and people can do 18 that, but I believe that all the submissions have been on 19 primary treatment, not enhanced data. 20 Now, the ability to do a minor -- or not minor. 21 To do a corneal treatment on top of a phakic IOL certainly 22 exists, although the answer to question is yes. Going back 23 in for a small refractive error probably could easily be 24 done on the corneal level, perhaps more appropriately so 25 than exchanging the implant. 25 1 DR. WEISS: Any other questions from the panel? 2 (No response.) 3 DR. WEISS: If not, thank you very much for 4 your presentation. 5 DR. VUKICH: Thank you. 6 DR. WEISS: Are there any other comments from 7 anyone else for the open public hearing session? 8 (No response.) 9 DR. WEISS: If not, we will now -- obviously, I 10 was anxious for the FDA presentation. So now, we will have 11 it. 12 MS. LOCHNER: I'm just going to make a few 13 introductory comments before we actually present the 14 questions to the panel. 15 Today we plan to discuss with the panel 16 clinical study design for phakic intraocular lenses. We 17 have prepared for your review a document entitled "Phakic 18 Intraocular Lenses: Clinical Guidance for Ophthalmic 19 Devices Panel Discussion, August 2, 2002," which is a 20 compilation of several activities in which the FDA 21 participates. It generally represents a composite of the 22 American National Standards Institute standard, the 23 International Organization for Standardization standard, 24 and the FDA's guidance document for phakic IOLs. 25 We last received the panel's recommendations 26 1 for phakic IOL studies in 1998 and so we thought it 2 important to receive updated recommendations from the 3 panel. We will then compile your recommendations and 4 present to the ANSI and ISO Standards Committees and update 5 FDA's guidance document accordingly. By having this 6 discussion today, we believe sponsors of these studies will 7 gain valuable information to successfully prepare 8 investigational device exemption and premarket approval 9 applications for their phakic IOLs. 10 We will begin this morning with presentations 11 from our invited speakers on two topics. First, Drs. Henry 12 Edelhauser and Bernard McCarey from Emory University will 13 discuss methodology and analysis for endothelial cell 14 density specular microscopy measurements. Next, Dr. 15 Liliana Werner from Storm Eye Institute will provide 16 background for the measurement and analysis of lens 17 opacity. 18 Following the invited speakers' presentation, 19 we will focus the panel's discussion on three areas. 20 First, the endothelial cell density study with Dr. Michael 21 Grimmett as the primary reviewer; second, measurement of 22 lens opacity with Dr. William Mathers as the primary 23 reviewer; and third, contrast sensitivity with Dr. Mark 24 Bullimore as the primary reviewer. 25 Questions have been provided to each of these 27 1 panel members for these topics to help to generate 2 discussion. However, we hope the panel will allow the 3 discussion to move to any area of significance to them. We 4 hope to step through each of the three areas -- endothelial 5 cell counts, lens opacity, and contrast sensitivity -- one 6 by one, opening each topic to full panel deliberations 7 after each of the primary reviewers' comments. After these 8 three primary areas have been discussed, Dr. Weiss will 9 open the discussion to comments on any section of the 10 clinical study guidance. 11 Unless there are any questions about the 12 agenda, I would like to move on to the invited speakers. 13 First, I'd like to express my gratitude to Drs. Henry 14 Edelhauser, Bernard McCarey, and Liliana Werner for taking 15 time from their schedules to present to us today. We are 16 honored to have people of their caliber providing their 17 insights to these important topics. 18 I would like to introduce the first two invited 19 speakers. Dr. Henry Edelhauser is professor of 20 ophthalmology and director of ophthalmic research at Emory 21 University University School of Medicine in Atlanta. He 22 received his Ph.D. from Michigan State University and 23 joined the faculty of the Medical College of Wisconsin. In 24 1989, he became the Ferst Professor of Ophthalmology and 25 director of ophthalmic research at Emory. He has served as 28 1 president of ARVO and has received the Honor and Senior 2 Achievement Award from the American Academy of 3 Ophthalmology. He also has received the Castroviejo Medal 4 and the Alcon Research Award. He presented a keynote 5 lecture at the 55th Congress of Clinical Ophthalmology in 6 Japan entitled "Cataract and Refractive Surgery: The 7 Effect on the Corneal Endothelium." His research interests 8 include surgical pharmacology, corneal physiology, drug 9 delivery, and ocular toxicology. 10 And Dr. Bernard McCarey is professor of 11 ophthalmology at Emory University School of Medicine and 12 affiliate scientist at Yerkes Regional Primate Center at 13 Emory. He received a Ph.D. from Marquette University and 14 joined the Department of Ophthalmology at the University of 15 Florida College of Medicine. He joined Emory in 1979 and 16 has served as chairman of their Institutional Animal Care 17 and Use Committee. He has received the American Academy of 18 Ophthalmology Section Honor Award, the Barraquer Award, the 19 Alcon Research Award, the CIBA Vision Research Excellence 20 Award, and Everett Kinsey Lecture Award at CLAO. His 21 research interests include corneal physiology, refractive 22 surgery procedures, ocular toxicology, and contact lenses. 23 And so, without further ado, I'll turn it over 24 to Drs. McCarey and Edelhauser. 25 DR. McCAREY: Thank you. 29 1 My name is Bernard McCarey. As you've been 2 told, I'm at Emory University. I have been interested in 3 specular microscopy as a laboratory science and also as a 4 clinical science. At present, I am a reading center for 5 Medennium. 6 What I'd like to discuss today with you is -- 7 whoops. We're not moving forward. It's hooked up, but it 8 doesn't move. 9 DR. WEISS: I'd just mention at this point, 10 after all the speakers give their presentation, they can 11 actually sit at this table over here with your names there, 12 so the panel has an opportunity to ask you questions. 13 DR. McCAREY: Now we're moving. Thank you. 14 There are several specular microscopy units on 15 the market presently and they break into two categories, 16 contact and non-contact. I present this as a list for your 17 handout. You can look at it later, but what I would really 18 like to do is to express to you the major differences 19 between the contact and non-contact. 20 Obviously, contact means you have to use an 21 anesthetic. You have to applanate the surface of the eye, 22 but you also are flattening the surface of the eye, and 23 when you do this, you generally can look at a larger field. 24 So generally, the contact units are considered large-field 25 specular microscopy. The non-contact has a smaller field. 30 1 You can see on the very bottom of your slide 2 we're talking about 700 or 800 cells can be visualized with 3 a contact unit, whereas only 160 or so for the non-contact, 4 and this has to do with the height of the slit. If we had 5 time, we'd go into specular glare, but basically the slit 6 in both of these instruments is the same width. It's just 7 a different height. 8 We are going to be collecting data about the 9 cell morphology, cell area, cell density, polymegethism, 10 which is an order of variation in size, and pleomorphism, 11 which tells you about how many sides there are on a cell. 12 I add this slide just for your notes. It 13 expresses the calculation for cell density and coefficient 14 of variation. 15 I also add this for groundwork. It gives you a 16 feeling as to what people would say at middle-age the 17 number of cells would be on a corneal endothelium, and it 18 varies with age. This is well-known in the literature and 19 we can find many references in the literature towards these 20 numbers. 21 But what I'd really like to show you is that 22 these numbers are from linear regression lines. They are 23 not a number. A person doesn't have a 2,700 cell density 24 because they're the age of 50. Rather, there's quite a 25 wide spread, as illustrated from this data from Dr. 31 1 Edelhauser in 1985. You can see a person of age 50 can 2 have anything from 2,200 on up to 3,300. So there is quite 3 a spread. 4 We also have a convention, an issue, that I'd 5 like to mention. That is, polymegethism is often referred 6 to in the literature as a value like .27 for a normal young 7 adult, but you can also see 27. Don't be confused. It's 8 just a literature convention. 9 The spread in coefficient of variation is 10 sometimes even larger in the normal population, as 11 illustrated here. So please don't expect to find one 12 number. 13 The major question that we're going to have 14 here is if you do a surgical trauma or something else to 15 the eye, how representative is a central endothelial cell 16 density to the information of what happened to that tissue? 17 If you cause local damage in one area of the cornea, how 18 fast does it affect another area, what does that time 19 duration spell, and can you really look at central 20 endothelium and get a feeling for what trauma occurred? 21 I reached back in the literature back to '79, 22 and I use this not as an example of what a surgical 23 technique may do to the tissue, but rather as an example of 24 how the tissue responds to a surgical event. 25 In this case, there was phacoemulsification and 32 1 extracataract extraction, and what the person did was 2 they're obviously making an incision in the superior area, 3 going into the eye, potentially damaging endothelium in the 4 superior zone, and if we look at the control tissue, we can 5 inferior, central, and superior clustered together. After 6 the surgery, we see superior has dropped considerably, 7 central has less, and inferior less. 8 So the question is will this spread of damage 9 rapidly congeal to one point again? And if you look at 10 this data, for 24 months there was only slight difference, 11 and it took on up to five years or more before all three 12 zones expressed themselves with the same value. So these 13 things go slowly. 14 DR. BULLIMORE: Before we move on, am I correct 15 in assuming there's only three people in that last data 16 point? 17 DR. McCAREY: I'm glad you mention it. No. 18 There are three people, but what this data did was he 19 collected the data at one time point. So he had 28 20 patients and some of them were out five years and some were 21 out four years or whatever, and he just collected the data 22 in that manner. Yes, there are very few data points in 23 each one of these, but I show this as an illustration as to 24 what can occur. 25 As another example, we're looking at data from 33 1 intracapsular cataract extraction, and if I point just to 2 mean cell area, you look at rapid changes going on for the 3 first four months and then some kind of a linear response. 4 Now, this is characteristic, if we looked at keratoplasty, 5 where you'll see much the same kind of event, rapid changes 6 -- in this case, six months -- and then a progressive 7 change. So this gives you a feeling as to when you might 8 want to collect data because of the kind of tissue 9 response. 10 If we reach back into the literature from Dr. 11 Bourne, we can see that he chose to follow patients for 10 12 years after cataract extraction. It doesn't matter if it's 13 with or without an implant. The point is that he followed 14 the patients for 10 years and he saw a rather in form cell 15 loss of 2.5 percent per year. 16 We often are referring to what a normal patient 17 would lose as far as endothelial cell density over time 18 simply because of aging, and we can refer to a paper by 19 Bourne. He says he followed them up for 10 years and he 20 has .6 percent, plus or minus 5 percent. I feel that's a 21 rather conservative number, but this is a number that's in 22 the literature. 23 One of the things we must realize is that we're 24 often using patients in clinical trials, such as refractive 25 surgery clinical trials, and these patients have a history, 34 1 and the history may often be contact lenses. So we're 2 entering patients in that are not "pure normal" patients, 3 but somebody that has a history of some kind of possible 4 trauma to the endothelium, and we can definitely say in the 5 literature that there are both transient problems with 6 contact lens wear, but also chronic problems in which we 7 have pleomorphism and polymegethism shifts in the tissue 8 over time. 9 So these patients will often look like this, 10 and I point these three examples out because you'll notice 11 that the cell densities are all fairly high and all fairly 12 much the same. Cell density is not going to give you the 13 full answer as to what that history of the patient was. We 14 have to look at the coefficient of variations, and we'll 15 see that they vary from 45 on up to 76. They can be quite 16 high, and that's expressing the fact that we have some 17 large cells interdispersed among the smaller cells. 18 The way we're going to analyze this tissue is 19 by multiple methods out there in the literature. The first 20 one is a comparison method. Look at a picture, look at a 21 honey comb, and how many cells do we have? It just tells 22 you cell density. 23 The frame method is another method. Just cell 24 density. 25 The next two methods, corner method and center 35 1 method, are going to give you actual cell area, and from 2 that we can calculate coefficient of variation and other 3 values. 4 Let's take a moment to look at the frame 5 method. The technique is, on the screen or from a 6 photograph, you put a box, and then you simply count the 7 cells within the box. The box is a small portion of a 8 millimeter. You adjust it up to a full millimeter, and 9 that's going to be the answer. So you just simply get cell 10 density. 11 One of the really easy pitfalls in this is that 12 in this example I've made a blue box, and the data over 13 here is in blue, and a yellow box. The blue box is twice 14 the surface area of the yellow box, even though it may not 15 look it, and if you give somebody an opportunity and tell 16 them to put a box on a field, they're going to make a small 17 box. What this means is that if it's twice the surface 18 area, if you counted 90 cells within the blue box, you have 19 to multiply it by 27 in order to get it up to a full square 20 millimeter. If you had the yellow box, you'd count 45 -- 21 it was half the size -- and you'd have to multiply it by 22 55. So right off the bat, you have a two-fold error 23 magnifier in your calculations simply by the size of the 24 box, and most people make small boxes. 25 The other method is the center method. In this 36 1 one, we put dots in the center and calculate what the 2 nearest neighbor is, and from this you calculate a polygon, 3 which is the cell, and so forth. Dr. Edelhauser will 4 discuss the patterns of this method in a few moments. 5 One of the things we must worry about is if you 6 put the dot offcenter, does it louse up the calculation? 7 And with the Konan software, you have a very nice 8 opportunity of simply a dot and then asking for this 9 analysis. Come back, move another dot, ask for the 10 analysis. 11 So I did this for 10 different cells, and you 12 can see that the error is fairly small. It's less than 1 13 percent. 14 I then dropped a cell. That is, took a dot 15 away, and tried it with various sizes of cells, and it 16 didn't seem to matter. I even used a hexagon pattern, 17 which was a perfectly uniform pattern, and it looked like 18 each cell that you missed the dot on, you lost about a 19 percent in the accuracy of the answer. 20 Another question is how many cells do we need 21 to count? This is the classic one you see in the 22 literature. If you count more cells, you get a more 23 accurate answer. 24 So I got a large field like this, divided it up 25 into multiple boxes, and then counted 10 cells, 20 cells, 37 1 30 cells, and so forth to create a series of lines, as 2 illustrated on this. So each one of these is an effort of 3 increasing the number of cells in the count. 4 You'll notice if you have a uniform, low 5 coefficient of variation cell pattern, you can get a fairly 6 good answer right off the bat. It improves when you get to 7 about 50 and it's a slight improvement beyond that. So you 8 don't need to count an awful lot of cells. 9 Coefficient of variation? It's a little bit 10 noisier. You certainly want to be over 50 cells counted in 11 order to get a reasonably uniform answer, but there's 12 always a spread in the answer. 13 This is more real life. This is a patient that 14 may have had a contact lens or some history of something or 15 simply an older patient with a higher coefficient of 16 variation, 45. Do the same kind of analysis. Now look at 17 the spread. It's tremendous. If you counted 25 cells, you 18 could have anything from 2,000 to 3,200 for cell density. 19 It gets better over 50 or 100 cells, but it never gets 20 really tight. 21 Coefficient of variation is even worse. This 22 is just summarizing. You can look at this on your handout, 23 but basically it says that if you have a large coefficient 24 of variation, you're going to get a larger spread and you 25 can't get away from that. 38 1 Also I want to mention some of things that are 2 pitfalls in the analysis. That is, when you ask the 3 clinical site to count 100 cells, they may tap 100 cells on 4 the cell pattern, as illustrated here on your right, but 5 the analysis using nearest neighbor -- for instance, if 6 this was seven dots and you asked for the calculation to be 7 performed, you'd only see that you actually counted one 8 cell, the one in the center. The others were just nearest 9 neighbor in the analysis. 10 So if you want somebody to count 100 cells, 11 coming down we have to count actually about 140 cells in 12 the analysis. Now, that sounds like a small point, but 13 when you have a limited field to look at, you may not be 14 able to achieve that because there simply aren't enough 15 cells on the field. 16 This shows data from a Medennium clinical 17 trial. These are strictly the control eyes, 123 good 18 images plotted out after the counting, and we had 19 everything from 900 or so on up to 3,600. 20 You'll notice that if you had asked for 100 21 cells in the analysis, we'd have to have a field of about 22 2,400 cell density. Less than 2,400 cell density, there 23 simply aren't the cells to look at in the specular 24 microscope field if using the Konan specular microscope. 25 Then every once in awhile we'll get poor 39 1 images, which we are unfortunately forced to used because 2 maybe the control data wasn't better than that, and you're 3 really using very few cells in the analysis. 4 Another issue is how uniform is the surface of 5 the eye. This happens to be my eye, and what I did was I 6 looked at the central target, I looked at a little bit 7 further out -- 1 millimeter, 2 millimeters, and 4 8 millimeter zones -- and then I took pictures as I looked in 9 various spots on the field, as you can see here. 10 Then I asked the question, statistically, is 11 this dot the same as this dot and so forth? And it came 12 out to no. So my surface, even though I have no history of 13 contact lenses and so forth, has a lot of variability, and 14 if you took answers from all over the place, you are going 15 to look like multiple patients. It's not going to look 16 like one patient. 17 Narrow the field down, still the same problem. 18 Narrow it down, still the same problem. Get down to about 19 a millimeter out and it's certainly better, but it's really 20 good if you look at the dead center. If I looked at that 21 green target very carefully and took 10 pictures in a row, 22 they would really look like the exact same patient. 23 There is a little trick involved in this, and 24 that is I happen to be an emmetrope -- these are reading 25 glasses -- so as I look off at the target inside the 40 1 machine, I see what you see on that screen, a red circle 2 with a green dot. If a person is a myope with 2 or 3 3 diopters off, he sees a blur, and so asking him to 4 cooperate to look at dead center becomes an increasing 5 challenge. 6 One more piece of data. What I did was I 7 looked at the control data from the Medennium group that I 8 have and there's a single reading group, which is me, but 9 there are 58 subjects at seven clinical sites. So this is 10 real life. The patients had a real-life coefficient of 11 variation, 36, not the nice normal of 25 or so, and what we 12 did was we had data collected at baseline and three months, 13 and I asked the question on that patient is the baseline 14 the same as the three-month data? And what it showed me in 15 a paired T-test was a .7, which is pretty good. 16 But there's a little more to the story. If I 17 then looked at the data and massaged it a little bit 18 further and graphed out the percent difference between 19 these two time periods for each of the patients and then 20 made a cumulative graph on your Y axis, I can then ask if I 21 have a spread of 2.5 percent, how many patients are going 22 to fall in that group? And this said 50 percent. So you'd 23 have to have less than 2.5 percent difference between these 24 two numbers to have 50 percent of your patients in your 25 group. If we wanted all of the patients, we'd have to go 41 1 up to 9 percent to get 100 percent. 2 So it's really quite surprising, and there are 3 references in the literature that support these kinds of 4 numbers. We're not looking at 1 or 2 percent spread in the 5 data between these two time periods. In this group, we 6 were looking at 9 percent. So we'd have to have a 10 7 percent change in the event in order to be 100 percent sure 8 that it was caused by the event, rather than just the 9 spread. 10 Some guidance for setting up clinical trials. 11 You certainly want to have careful control of the criteria 12 of your study, which specular microscope you're using, your 13 experience, and the data capture. Who's doing the capture 14 and how often does he do it and how much experience? How 15 many sites are located? Each time you add a site, you're 16 increasing the variabilities. 17 So let's go right down to the final slide and 18 what I would suggest. First of all, I would suggest we all 19 have the same microscope, and I like the non-contact 20 specular microscope simply because it means you're asking 21 for less skill from the technician. 22 I prefer one technician, I prefer to train that 23 technician, and I'd like to check on the training of that 24 technician to see if they really are doing it frequently 25 enough that they have kept their skills up. Most of these 42 1 people have lots of other things to do and they kind of get 2 soft on their skills. 3 I think a central reading center to limit some 4 variability is also a very good idea. 5 Thank you very much. 6 DR. WEISS: Thank you very much. You can take 7 a seat when you're done, Dr. McCarey, at the table. 8 DR. McCAREY: I'm rebooting for the next one. 9 DR. WEISS: Oh, you're rebooting. 10 DR. EDELHAUSER: While Dr. McCarey's doing 11 that, I'm Dr. Henry Edelhauser, and I'm at Emory. My 12 laboratory has been a reading center for KeraVision and 13 Staar, and I do a number of research contracts with Alcon 14 on intraocular irrigating solutions. 15 What I'd like to summarize for everybody today 16 is a very practical summary now of what Bernie talked about 17 and a little bit of the theory, and Dr. Ramzy Azar was the 18 one that helped out and was one of our major reading center 19 individuals, though he's presently in the Navy right now 20 down in Bethesda. 21 The purpose of what I'd like to summarize today 22 is using the robo non-con specular microscope, which seems 23 to be about the best specular microscope to run a clinical 24 trial, and particularly when you're thinking about 25 refractive surgery, because you're not applanating onto the 43 1 cornea. So we want to understand the variable issues that 2 may be found in specular microscopes. So our objectives in 3 this 10-minute presentation are to provide good examples in 4 good and poor photography, illustrate the variability, and 5 illustrate the variability within a single image that data 6 has to be obtained from. 7 What is a good image? I think Dr. McCarey 8 showed you. This is typical panel that one would receive 9 from the robo non-con. You notice on here you can find 10 distinct cells. You can identify at least 150 cells. The 11 cells can be grouped in a very uniform manner. 12 Then you may have to say, well, what may be 13 good for clinical purposes may not be good for research. 14 For example, in many clinical times, they'll only count 15, 15 20 cells, but for a research study, particularly where 16 you're quantitating endothelial cells over a long time, you 17 want to try and put a dot in every one of these corneal 18 endothelial cells. 19 Things to consider that may affect the optical 20 image. Dry eye, contact lens use, wrong specular 21 microscope settings -- you can go into the manual mode with 22 the non-contact -- and patients with keratoconus are very 23 difficult to get endothelial specularscopes. 24 Patient compliance. This is a real issue if 25 the individuals can't see that little green circulating dot 44 1 and you have to work with the patient. So training becomes 2 very important. 3 Age of the patient, a little bit more 4 difficult, and then training and experienced photographer, 5 and as I emphasize over and over again for individuals and 6 companies that are trying to under specular microscopy, you 7 have to train the individual out in the field. 8 Poor quality images are something that are an 9 issue, and particularly with preoperative because if your 10 preoperative photos are poor quality, this is going to 11 carry through the whole study. 12 Here is an example where you have just a panel 13 over here, and here's another panel here. So really what 14 was happening is that the patient was moving his eye when 15 this picture was captured. 16 Another type of poor quality image. Very 17 difficult. Here you might be able to obtain 30 to 40 cells 18 in this particular panel. If this was a preop, this 19 patient is lost because there's identifiable cells that 20 could be measured. 21 Again, poor quality images where it's very, 22 very difficult, and I can tell you from being involved in a 23 reading center and looking at over 10,000 of these with the 24 laboratory, you get photos like this that are preoperative 25 and when you want this as a preoperative photo, how do you 45 1 start a baseline for this particular individual? This is 2 where training of the photographer or the individual 3 running the specular microscope out in the field is very 4 important because that image will come up on the screen, 5 and if it's this one, they should sit the patient down 6 again and retake the photograph. 7 Conditions that potentially increase 8 variability. Patients that have Fuch's, polymegethism, 9 pleomorphism, injury, and low cell density. Particularly, 10 there are some patients that do have a low cell density. 11 Here's an example of a patient that has guttata 12 or Fuch's, and notice you see these black spots here. 13 Actually, these black spots are covered by a very, very 14 thin part of the corneal endothelial cell and the 15 refractive index is different here. Well, how do you 16 analyze this photograph? Well, you'd have to group here or 17 you'd have to group here. 18 So capturing the best image is very important. 19 You have to make sure the patient is comfortable. You 20 have to instruct the patient to blink. You have to 21 instruct the patient not to move and to open his eyes wide. 22 You have to instruct the patient to focus on the green 23 light, and as Dr. McCarey said, it's difficult for somebody 24 who has type of disease or is extremely myopic because you 25 can't see the green light as well as somebody who's an 46 1 emmetrope. You have to be patient. You have to work with 2 that patient and use of the manual setting to improve the 3 quality of the cornea is -- sometimes the corneas are 4 thicker than the normal setting, so you have go to the 5 manual. 6 Things to consider when you analyze images. 7 You have to locate the best and most representative area, 8 the number of cells, you have to look at the quality of 9 cells, and you have to use the area with the fewest 10 distortions, as I'd shown in the very early aspect. 11 Here is an example of the best image. Well, 12 the best image on this specular here would have to be here, 13 and this one you'd have a very hard time finding the best 14 image. It may be somewhere along in this area. 15 Dotting cells. You have to dot all the cells 16 in the center and you have to remain accurate and 17 consistent throughout. We always recommend dotting at 18 least 150 cells if there are 150 or more than that on the 19 photograph because at least you'll get an analysis of 100 20 cells, 110 cells, and as Dr. McCarey showed you on the 21 graph, the more cells you can dot, the better the 22 statistical analysis will be. 23 Where to group the analysis? Now, this is 24 interesting. Well, if you could dot every one of these, 25 this would be the appropriate way to go, but if you dot 47 1 here, if you would look here, you'd have a lot of big cells 2 up in this area and small cells here. So certainly, the 3 diversity in the cell count would be very large, and this 4 is one of the disadvantages of having specular microscopy 5 done out in the field and have the technician because the 6 technician may just pick this area and then that will be 7 the preop, or they may pick this area, and then you come 8 back and your three-month data or six-month, they're going 9 to analyze up here in this area. So having a reading 10 center or having one person do all the analysis is very 11 important. 12 What's wrong with this analysis? Well, here's 13 an example of analysis done in this area, just localized 14 down in this area. Only 71 cells were counted, but it 15 still had an endothelial cell density of 2,639. So the 16 analysis really is not representative. It's introducing 17 bias because you're looking at a population of a lot of 18 small cells here, you're not likely to be able to repeat 19 this analysis, and really we say that you have not counted 20 enough cells. 21 It is very important to group the analysis like 22 this or as illustrated here, and when you look at your 23 grouping analysis, notice the box that we've drawn here. 24 This is an improper grouping that you would see here 25 because you're doing this nearest neighbor analysis as the 48 1 algorithm of the specularscope and may only end up with 50 2 cells or something like this, whereas if you group the 3 whole group, you'd end up at least putting a dot in 150 4 cells. But see, technicians have to be trained if they're 5 going to do this, or your reading center. 6 Patients that have guttata. You may have some 7 of these in a study group, grouping here, here, here, or 8 here. 9 DR. BULLIMORE: This is Mark Bullimore. Excuse 10 me interrupting. You've talked about guttata twice and you 11 seem to infer that you should count around them when 12 estimating cell density. 13 DR. EDELHAUSER: Right. 14 DR. BULLIMORE: So you don't include that area 15 at all in your analysis, even though there are no cells 16 there. 17 DR. EDELHAUSER: Well, there's just usually one 18 large cell that covers that area, but we have found that in 19 doing patients like that, if they're part of the study 20 group, the best analysis is to just use the cells without 21 including that guttae in there. The reason is is that if 22 you put that one large cell, you're multiplying this by 23 such a large factor that your cell number is extremely 24 variable. See, you're wedded to the algorithm of the 25 specular microscope. 49 1 DR. BULLIMORE: Thank you. 2 DR. EDELHAUSER: And hopefully, in doing a 3 study, you would not have many of these patients in there, 4 but I can tell you from the studies that we have done with 5 LASIK and things, you do find some patients that do have 6 guttata. 7 To analyze the cells, you need to be able to 8 visualize the cells. You have to identify a pattern. And 9 would it be appropriate to take this endothelial cell that 10 has shown some damage? 11 Where the image is analyzed can create a great 12 degree of variability, as illustrated here. The old way 13 which we used to do specular microscopy with contact 14 specular microscopy is take a wide field and then we'd have 15 to trace the corneal endothelial cells, put a number in, 16 and then we would digitize each one of the corners. This 17 is very accurate, but again, notice the grouping of the 18 cells. 19 Here are examples of variability. This is one 20 specular image here. If you analyze the endothelial cells 21 in the lower portion, you'd up with a cell density of 22 2,976. If you analyze the endothelial cell density in the 23 upper portion, you'd end up with a cell density of 2,873, a 24 difference of 103 cells and a 4 percent difference. So you 25 can see the variability that can occur and this can very 50 1 easily occur if the training of that individual is not 2 appropriate. 3 Here are examples of variability within 4 readers, and this occurred out at our center when we were 5 training the people out of our Vision Correction Center to 6 do this. Endothelial cell density 2,531 here and 2,358 7 here, a 7 percent difference just on the same pattern. 8 Here are examples of variability between 9 readers, with having different readers. So in this case, 10 we had five different readers put dots in each one of the 11 cells, and you can see they varied from 2,531 up to 2,631. 12 So this is really a degree of variability. So training 13 not only needs to occur with the photographer, but also the 14 person doing the analysis. 15 Just to show you this, the consequences of 16 overcounting, if you skip two cells, you have a significant 17 difference. If you overcount three cells, you have a 18 significant difference. So this is where the training is 19 extremely important for the individuals. 20 Well, just to put this into a little 21 perspective on this, and this is a graph that we've 22 recently put together, the first study that was done that 23 we did back with Richard Yee, et al., this is what happens 24 with contact specular microscopy. Notice, from age groups 25 from 10 up to 89. Notice, this is the distribution of 51 1 endothelial cells here. 2 Okay. We recently went back and did 125 3 patients through the various decades with the non-contact 4 robo, which is illustrated in the yellow line, and notice 5 that the lines overlap. Very early, we published a paper 6 in the AJO and I took the preoperative data from our LASIK 7 patients, which varied from 20 to 50, and notice those 8 lines overlap. 9 There are two areas of outliers, and this was a 10 mixed Asian group of patients that we had in Emory when we 11 looked at this, and notice the Asian patients have a higher 12 endothelial cell density, and a number of years ago we had 13 access to a Japanese population in Osaka when Dr. Matsuda 14 was with us, and in this population, notice that the 15 Japanese population in Osaka had many more endothelial 16 cells than a Caucasian American. 17 So this becomes a very interesting point of 18 view when one wants to look at endothelial cells in grouped 19 patients if you're doing a study, say, in the West Coast 20 compared to, say, in the South or the Midwest. 21 Mike? 22 DR. GRIMMETT: Just a quick question. I hope 23 you don't mind the interruption. 24 DR. EDELHAUSER: No, not at all. 25 DR. GRIMMETT: The non-contact robo data, is 52 1 that published somewhere? Is that an abstract? 2 DR. EDELHAUSER: No, that's published, Mike. 3 That's the original data from Richard Yee's paper we 4 published in Current Eye Research. 5 DR. GRIMMETT: In '85. 6 DR. EDELHAUSER: In 1985, yes. 7 DR. GRIMMETT: Is there an updated one? 8 DR. EDELHAUSER: No, that's not published yet. 9 DR. GRIMMETT: Oh, I see. 10 DR. EDELHAUSER: It's in the process of being 11 written up. We just completed that within about two months 12 ago. 13 DR. GRIMMETT: Thank you. 14 DR. EDELHAUSER: But I thought it was a very 15 interesting comparison because much of the data in the 16 literature is from the contact scope, and this is really 17 the first longitudinal study with the non-contact. 18 Just to give you an example of the variability 19 in the best of hands, this was taken from our LASIK paper, 20 where Ramzy Azar took all of the photographs of the 21 patients, and he then used his own eye throughout a three- 22 month time period where he took the pictures of his right 23 eye and his left eye. So this is 36 different photos, 18 24 of the OD and 18 of the OS. His endothelial cell density 25 is 2,545 and 2,600. 53 1 Notice the standard deviation, 45 cells. So 2 the precision in the best of hands with this, and this is 3 from this AJO paper, is 1.7 percent and 1.5 percent. 4 That's with one person looking and taking all the photos, 5 and I think this is very important. 6 A recent study that we've done and reported at 7 ARVO, which I think is very important as we think about the 8 endothelial cells, we measure central endothelial cells, 9 but peripheral endothelial cells are going to become of 10 interest, too, particularly when you think about phakic 11 IOLs. 12 These are three graphs where lyserine red- 13 stained human corneas, something like 71 human corneas that 14 we looked at. You can see the N listed here. This line is 15 the endothelial cell density in the center, the paracenter, 16 and the far periphery, about 4.5 millimeters from center. 17 You can see there is a progressive decrease of roughly a 18 half a percent per year all the way across the line, but 19 still you do have a higher cell density in the periphery 20 that could aid in the sliding of endothelial cells to the 21 periphery. 22 So just to summarize this, then, what are the 23 sources of variability? It's difficult to return to the 24 same location. When we were a reading center for 25 KeraVision, we did a reproducibility study with 10 patients 54 1 at three different sites where they took three consecutive 2 readings, and we ended up plus or minus 56 cells, very 3 similar to what we measured in the LASIK study. So you 4 have an inherent error, just the reproducibility, of 2 5 percent. 6 Poor image quality. We suggest trying to get 7 at least 100 cells. 8 Training error. Training, and you have to have 9 consistency. Reading analysis. Training and consistency, 10 and equipment calibration and alignment is another very 11 important issue that has to be. 12 So in summary, what the ideal situation might 13 be is it could be a company, it could be an independent 14 reading center, it could be you need your specular sites, 15 and this data then should be sent to a reading center. You 16 should not have the sites do their own analysis. Then the 17 data in a mass fashion, which would be received to the 18 reading center, would be sent to the data processing 19 center, and then for statistical analysis then to the 20 technology company, and obviously then on to the FDA. 21 So I hope this little bit of a summary was 22 important and I was able to show you some of the 23 variability of the technique. It's a good measurement, 24 it's a hard measurement to get, and what is very important 25 is that you do certainly want to see what the state of the 55 1 corneal endothelium is. 2 Thank you for your attention. 3 DR. WEISS: I'd like to thank you both for 4 excellent presentations. Perhaps you could come up to the 5 table and the panel could start their questions. 6 I would just like to ask you a couple of 7 elementary ones. What would you suggest as the minimal 8 time of follow-up in order to detect endothelial cell loss 9 after phakic IOL implantation? So what's the shortest 10 study? 11 DR. EDELHAUSER: Well, I think certainly a 12 three-month time period, that's a reasonable approach to 13 take it. I mean, are you trying to say how soon after -- 14 DR. WEISS: Two years, three years, five years? 15 What would your wish list be? 16 DR. McCAREY: If the literature is any 17 indicator, I would imagine that the most active changes are 18 going to occur probably within the first six months, and 19 then afterwards you're probably going to level out into 20 some kind of a steady effect. So the initial trauma of the 21 procedure, let's say the first six months you need the 22 data. Afterwards, you might need data every six months for 23 maybe two years, and then hopefully you're going to see 24 some kind of a level line that you're dealing with. If you 25 don't, you're just going to have to go further. 56 1 DR. WEISS: So just to clarify if I understand 2 you correctly, you would say two and a half years from 3 implantation? 4 DR. McCAREY: Certainly two years. 5 DR. WEISS: Two years from implantation as a 6 minimal, unless there is anything else that you can tell. 7 DR. McCAREY: Correct. 8 DR. WEISS: If a patient is a contact lens 9 wearer before, that's an evolution of what is occurring 10 with the endothelium as well. Before they have their 11 phakic IOL, how long should they be out of contact lenses 12 so you get a stable specular microscopy before they can be 13 entered into a study? 14 DR. McCAREY: Yes. That's almost a 15 presentation on its own as to how the patient is able to 16 return from this polymegethism state from contact lens 17 wear. It looks like it's very, very, very slow, if at all, 18 there is a correction from that polymegethism change. So 19 it means that you could look at a patient one, two, three 20 years, five years out of having not wearing their contact 21 lenses and they would still have the effect of wearing the 22 contact lenses. 23 The next part of the issue is does this mean 24 that the eye is a little susceptible to further trauma? 25 And the literature would indicate that these eyes are more 57 1 susceptible to trauma. They respond more negatively to 2 trauma than a person who had a normal CV and no contact 3 lens history. 4 DR. WEISS: So should that be an exclusion 5 criteria then? 6 DR. EDELHAUSER: No, I wouldn't think so. It 7 shouldn't be an exclusion criteria if somebody has 8 polymegethism. 9 DR. WEISS: No, I mean contact lens wear, 10 because it sounds like it's such a confounding variable 11 that you -- 12 DR. EDELHAUSER: It is, but I think there is 13 variability within the degree of polymegethism because if 14 you're thinking of phakic IOLs and everything, basically 15 all these people have worn contact lenses. 16 DR. McCAREY: I agree with Dr. Edelhauser. 17 That's the problem. You're going to lose an awful lot of 18 your patients. 19 The second part of the story about 20 polymegethism is it's a stress from oxygen. If you are an 21 old-fashioned PMMA lens wearer, you have the most stress. 22 If you have the more modern, let's say the silicone 23 materials or a high-water content hydrogel, you'll probably 24 have less stress. So it's a sliding scale as to how much 25 stress they have had from their contact lenses. 58 1 I think what I want to point out is that the 2 endothelial cell density is not alone an indicator as to 3 the history of that patient. You'd also want to know 4 what's going on with their CV, the spread in the cell 5 sizes, and I think good data collection is probably more 6 important a statement than eliminating these patients. 7 DR. WEISS: If I had to put you on a spot and 8 ask you to give a bare minimum -- not the range, but the 9 bare minimum -- you think that someone would have to be out 10 of contact lenses, would you be able to give me a number? 11 DR. McCAREY: It would just be a wild number 12 and I would certainly -- 13 DR. WEISS: A wild number might still be 14 helpful. 15 DR. McCAREY: Certainly six months, but I don't 16 really know for sure. 17 DR. WEISS: Okay. Six months. 18 Would Asian corneas then have to be grouped 19 differently because they're starting out with more 20 endothelial cells? 21 DR. EDELHAUSER: I would think that would be a 22 subset that should be analyzed separately. From past 23 experience with that, all Asians that we've looked at have 24 a very higher cell density, and it depends upon where the 25 study is going to be conducted, but I wouldn't use an 59 1 exclusion. I'd just use it as an added subset. 2 DR. WEISS: A subset. Thank you very much. 3 DR. McCAREY: Could I add one more thing on 4 this? 5 DR. WEISS: Yes. 6 DR. McCAREY: I think that the contralateral 7 eye is a goldmine, that a lot of these issues that you're 8 referring to can be lessened as far as you're concerned if 9 you know the history of the contralateral eye. Follow both 10 eyes and make the data relevant within the patient 11 themselves, rather than some kind of a standard regression 12 line from a group. I think will solve a lot of problems. 13 DR. WEISS: So in this case, you would suggest, 14 at least from the endothelial cell portion, that it would 15 be very helpful only to do unilateral phakic IOL surgery. 16 DR. McCAREY: Yes. Unfortunately, that's what 17 I would state. 18 DR. WEISS: Thank you. 19 We're going to go around. Dr. McCarley, then 20 Dr. Burns, then Dr. Bullimore, Dr. Matoba, Dr. Grimmett, 21 and then we'll go on down the line. 22 MR. McCARLEY: This is Rick McCarley. I knew 23 if I waited around long enough, I'd get a degree, too. So 24 thank you. 25 (Laughter.) 60 1 MR. McCARLEY: It's just Rick. 2 DR. WEISS: I think I'm going to start with 3 first names and go back to the first names. 4 (Laughter.) 5 MR. McCARLEY: There you go. Thank you. 6 You'll see my eyebrows going up several times 7 today because obviously I'm involved in a clinical study on 8 phakic intraocular lenses in the U.S. that's been going for 9 about five years now, and in fact 15 years in Europe. So I 10 have a lot of -- I'll call it practical experience, and 11 boy, I wish I knew then what I know now. We do have a lot 12 of data and I wanted to share some of this because it 13 applies to all of us. These discussions have happened in 14 the ANSI standards, so I'm pretty well up to date on what 15 has happened in the industry and what data we've collected. 16 But a couple of the comments, one is the 17 patients not only wear contacts. Most of these patients 18 that we've seen have polymegethism, but many of them, we're 19 talking about -15 to -20. 20 DR. WEISS: Actually, I'm going to interrupt 21 you for a moment. Because of the time constraints, what 22 I'd like to do is use the benefits of our experts' 23 expertise, and I would prefer the comments get directed to 24 the comment portion. 25 MR. McCARLEY: That's okay. 61 1 DR. WEISS: And if you have any questions to 2 direct to Dr. McCarey or Dr. Edelhauser, use this time for 3 that. 4 MR. McCARLEY: Okay. Then I will ask the 5 question directly. Have you ever studied a population of 6 high myopes individually and compared that to the normal 7 population? 8 DR. EDELHAUSER: No, we haven't. The only high 9 myopes that we've really looked at were part of our study 10 that we published on LASIK, where had 100 consecutive 11 patients that we looked at and there were only, I think, 12 eight to 10 at the most that were high myopes. 13 MR. McCARLEY: I see. And when you're 14 analyzing the endothelial cell data over a large 15 population, the comment about the subgroup of Asian eyes, 16 are we looking at percent changes, so it really doesn't 17 matter what their beginning or ending is? So looking at a 18 subgroup really doesn't matter. If they start off with 19 more, we'd expect them to end up with more. 20 DR. EDELHAUSER: I think that's how the study 21 design is set up and how the statistical analysis is done. 22 MR. McCARLEY: Right. 23 DR. EDELHAUSER: I mean, as an independent 24 reading center, we want to be masked in that. So the 25 criteria would be in the early study design and how you're 62 1 going to do that. 2 MR. McCARLEY: Thank you. 3 DR. WEISS: Dr. Bandeen-Roche? 4 DR. BANDEEN-ROCHE: Yes, I have maybe three 5 related questions. The first one is both of you alluded to 6 cell density not being uniform over the surface of the eye. 7 Would you think it would be worth some sort of a dynamic 8 sampling strategy to try to isolate the maximum cell loss 9 or is that relatively reliable in the center? I'd just 10 appreciate your thoughts on that. 11 DR. McCAREY: I think you'd be opening up a can 12 of worms if you approached it that way. I think you're 13 probably best trying to get central readings, and with 14 instruments like the Konan, hopefully the patient can 15 cooperate and look at the target and you're getting the 16 same field. 17 On my own eye, I could take 10 pictures, and 18 every one of us having slightly different patterns and you 19 can see an odd cell here and there, and within the 10 20 pictures, I could see the same little couple of cells. I 21 can come back two or three months later and do that again 22 and once again see those same little odd cells. 23 So I think that the key here is training and 24 cooperation from the patient and the central cells, and 25 that gets rid of some other cans of worms. 63 1 DR. BANDEEN-ROCHE: Thank you. 2 DR. EDELHAUSER: Let me just add one other 3 thing. I think in order to get a representative, it is 4 recommended, and we certainly have recommended this, that 5 if you're going to take specular photographs, at least try 6 and get three photographs, and then at least have the 7 reading center or whoever is doing the analysis try and 8 then analyze the best one of this, or in the early training 9 of your individual, if you have them take three, and then 10 take the average to see where they are. So again, it comes 11 back to teaching the photographer to get a representative 12 photo. 13 DR. BANDEEN-ROCHE: Thank you. 14 The second question goes to variability over 15 repeated readings, say, over a three-month time period, and 16 you alluded to in the best case percentage variability of 17 1.7 and 1.5 percent. I'm curious. How much of that is 18 variability in the reader versus natural variability in 19 cell density over three months? And when you refer to 20 those rates being the best, I tend to say, well, suppose 21 somebody just counted more cells? Couldn't it get better? 22 If you could respond to that. 23 DR. EDELHAUSER: In this particular case, I say 24 it was the best because we had one person taking the 25 pictures of his eye and counting as many cells as possible 64 1 to come up to do this, and I think that what you're dealing 2 with in this particular case is the variability of the 3 machine and the algorithm. Because don't forget, in this 4 particular case, you're, say, dotting 150 cells. At best, 5 160 maybe is all you can put on there. Then you push the 6 button and the algorithm pops up here, and what you're 7 dealing with is that's what you're left with, the analysis 8 of those particular cells that you're dealing with. 9 DR. McCAREY: I kind of take a little bit 10 different approach to that. I think the math is the math. 11 It's not changing. The computer's not changing. The 12 machine is hopefully focusing the same and keeping the same 13 magnification. So your variation isn't in the equipment. 14 It's in either where the picture was taken -- well, that's 15 probably the heart of it, where the picture was taken, 16 because there is a variation across the surface. 17 Now, you mentioned the best answer and the 18 worst. I showed you 9 percent of a spread in order to get 19 all the patients in the group. That's with seven clinical 20 sites, 58 patients, and no expectation that I was going to 21 do that analysis. That's the hardcore reality. 22 Whenever you look at a paper that comes from a 23 given site -- whoever it may be, our lab or Bill Bourne's 24 or someone else's -- they are doing the entire study 25 themselves. They are giving you their best shot. They 65 1 know they're going to do the study, and it always comes out 2 cleaner that way. So my 9 percent is probably reality and 3 the equipment probably is as good as 1 percent. 4 DR. BANDEEN-ROCHE: Thank you, and a very brief 5 final question that goes to guttata. So the point of not 6 counting that, I take it that induced guttata is not a 7 primary problem in phakic IOLs? 8 DR. EDELHAUSER: Provided it doesn't cover the 9 whole surface of the corneal endothelium. I mean, we've 10 all seen patients that have guttata that go from limbus to 11 limbus, and then occasionally you'll have a person who will 12 sit down and you'll get one guttae in the area. I mean, 13 you still can use that photograph. So I think there are 14 various degrees. 15 DR. WEISS: Thank you. 16 Dr. Burns, did you have some questions? 17 DR. BURNS: Yes, two. Steve Burns. Two simple 18 questions. 19 First, is it true that if you take, say, three 20 sets of images that the one with the highest count is best? 21 I thought I heard you say that, but I might have 22 misinterpreted it. 23 DR. EDELHAUSER: No, I think you the best image 24 that I was implying is the one where you have a good 25 distribution of cells over the whole screen that you can 66 1 see, and not one where a patient may have moved their eye a 2 little bit and you only get a strip of the endothelial 3 cells. So the best image that I would suggest would be one 4 that has as many cells uniform across the full screen. 5 DR. BURNS: So given that you've got three good 6 images, do you just average them? Do you recommend 7 averaging those counts for that data point? 8 DR. EDELHAUSER: Yes, yes. 9 DR. BURNS: The second question is when you 10 were talking about how long you'd follow up, was that sort 11 of in laboratory-type data or were you taking into account 12 the realities of variance from a trial? So the two and a 13 half years you were suggesting, do you think you'd have to 14 lengthen that given the variability you get from a 15 multisite study? 16 DR. EDELHAUSER: You might. The best 17 illustration I'd give you is our LASIK data, and we looked 18 at this very carefully for 100 patients and we followed 19 them up to three years. Of the 100 patients, we able to 20 get 63 of them back -- again, this is all volunteer at this 21 particular stage -- and we really found no change over a 22 three-year time frame. But I think with something that has 23 the potential where endothelial cell populations may be 24 decreasing, going out to two or three years might be 25 important. 67 1 DR. WEISS: Dr. Bullimore? 2 DR. BULLIMORE: Mark Bullimore. I'd like to 3 thank both of you for coming. It's a fabulous presentation 4 and it's very encouraging for those of us who do clinical 5 research in general to see the level of rigor with the 6 reading center. I mean, that was very refreshing. 7 We took some data on the Konan probably under 8 the worst-case scenario, and our precision was probably 9 closer to 10 percent than 2 percent, but we weren't 10 counting the number of cells and we weren't doing all the 11 other sophisticated things that you gentlemen do. 12 I have hopefully a couple of quick questions. 13 First of all, when you talk about coefficient of variation 14 that's what other people, or maybe yourselves, would also 15 refer to as polymegethism? Are those two terms equivalent 16 or interchangeable? 17 The other thing is, and this is a particular 18 concern given the fact that many of the patients having 19 phakic IOLs will be long-term contact lens wearer, you 20 inferred or implied that if you take a patient out of 21 contact lenses, there would ultimately be a very slow or 22 potentially no long-term recovery from that insult. Did I 23 hear you correct? 24 DR. McCAREY: That's strictly from the 25 literature. It's not from my own laboratory experience, 68 1 but Brian Holden has done lots of work in this area, and it 2 repeatedly shows up that that's true. 3 DR. BULLIMORE: So taking that to the next 4 step, were we at some future date to be looking at these 5 data, you would not worry about that contact lens or prior 6 contact lens wear as a confounding factor in any changes in 7 endothelial count that occurred after the patient had a 8 surgery? 9 DR. McCAREY: There is literature out there 10 that tells you that large polymegethism values often lead 11 to a patient being more susceptible to the trauma of a 12 given surgery as compared to patients with a lower 13 coefficient of variation. So they do have the potential to 14 be more susceptible to damage. 15 DR. BULLIMORE: I see, but that's a short-term 16 effect of the surgery, rather than their sort of five-year 17 history, say. 18 DR. McCAREY: I don't really know if it would 19 mean that the five- and 10-year periods would still be at a 20 higher rate of loss or not. I don't know. 21 DR. BULLIMORE: Did you want to go ahead? 22 DR. EDELHAUSER: There may actually be a 23 benefit of removing the contact lens because I say from our 24 experience, in our three-year data from the LASIK patients, 25 we actually did see an improvement in the coefficient of 69 1 variation, and all these patients were contact lens 2 wearers. 3 DR. BULLIMORE: And a final question. There 4 has been a great deal of emphasis on your slides and in 5 other materials I've looked at on the topic on endothelial 6 cell density. Do you think that should be the primary 7 outcome measure or should we be looking at coefficient of 8 variation as the primary outcome measure in a long-term 9 study or would you consider both to have equal weight? 10 DR. EDELHAUSER: I think cell density would be 11 the number one aspect of it. The difficulty with the 12 coefficient of variation that we have if you use the robo, 13 to some extent that data is a little soft, and the reason I 14 say that is soft is because you're using the center dot 15 method. If you were tracing cells and using the corner 16 method, that data is much stronger. 17 One of the things that I have found over the 18 years of using the robo, if you have a patient or a subject 19 that has a high coefficient of variation -- say, .6, or 20 like many of the diabetic patients -- it shows up, but the 21 difference between .27 and .35 is not really a significant 22 change. 23 DR. BULLIMORE: So without wanting to put words 24 in your mouth, the precision of the coefficient of 25 variation method is not particularly -- let me rephrase 70 1 that. The precision of the coefficient of variation 2 measure is not particularly impressive. 3 DR. McCAREY: Well, the coefficient of 4 variation, I passed the slide very quickly, but the 5 calculation is that you're dividing the mean cell area into 6 the standard deviation or the spread in that mean cell 7 area. So I feel that it is a piece of informative 8 information, but you do get a lot of noise, though, in it 9 from the fact that there is a spread in the data. I'd 10 personally like to see that information carried forth in 11 the study. 12 DR. BULLIMORE: I lied, Madam Chairman. I do 13 have another question. 14 I'm very familiar with a paper by Scott McCray 15 on long-term PMMA contact lens wear, where he looked at a 16 cohort of patients who'd worn PMMA lenses for 20 years and 17 reported their outcomes, and one of the things I found 18 compelling about the paper is actually, and it's not 19 emphasized in the paper, but the cell density in the 20 contact lens wearers centrally was actually higher than the 21 controls. 22 Okay? The take-home message was that there was 23 this subset of patients who fell below a given value of 24 cell density. I think it was nine out of the 81 contact 25 lens wearers compared to two of the controls, which was 71 1 statistically significant. 2 So in his analysis, and using that as, if you 3 like, an analogy to what we're doing here with phakic IOLs, 4 would it not be more appropriate, rather than emphasizing 5 the mean endothelial cell density, to look at sort of, for 6 want of a better phrase, incident cases of significant or 7 substantial reductions in cell density? Have you explored 8 that parameter yourselves or is it in the literature? 9 DR. EDELHAUSER: It's not in the literature, 10 but in the database that KeraVision submitted to the FDA, I 11 know it's there because they went back and looked very 12 significantly at patients who may have lost greater than 10 13 percent of their cells, and they were reported. 14 The other interesting thing in that paper 15 you're referring to by Dr. McCray, of that subset of 16 patients, there were a group of contact lens wearers who 17 actually lost cells. 18 DR. BULLIMORE: Yes. That was my point, and it 19 was nine out of the 81 were way below what you might 20 consider the normal range, but when you actually averaged 21 the cell density -- 22 DR. EDELHAUSER: It was lost. 23 DR. BULLIMORE: It was lost. So I think we 24 should maybe keep that in mind. 25 DR. McCAREY: When you get polymegethism, you 72 1 actually appear to be getting smaller cells, not just 2 losing cells and big ones reappear. That would be a 3 misconception. There's actually an appearance of what 4 appear to be smaller cells. Recently, and I unfortunately 5 don't know the author, there was a description of how this 6 occurs. 7 DR. BULLIMORE: That was Michael Delaty, 8 probably. 9 DR. McCAREY: I don't remember, but he has 10 certainly has a lot of articles. 11 You can actually see a triangular-like pattern 12 occurring and that triangle shows a smaller top to the 13 aqueous, what looks like a smaller cell. The volume of the 14 cell may be the same. So there is a shift in the 15 dimensions of cells, rather than a loss of cells. 16 DR. BULLIMORE: Sort of from a cylindrical 17 profile to a trapezial. 18 DR. McCAREY: Right, and so it appears that 19 they could have a very high cell density when that's really 20 true if you counted the whole number of cells across the 21 surface. 22 DR. BULLIMORE: Thank you both again. This has 23 been very, very helpful. 24 DR. WEISS: Thank you. 25 Dr. Matoba, did you have questions? 73 1 And I will again reiterate, because we're 2 already going to be off time and running over. So if the 3 members of the panel could make their questions short and 4 sweet and directed to the issue at hand, I'd appreciate it. 5 DR. MATOBA: Do you want us to identify 6 ourselves? 7 DR. WEISS: Yes. I mean, we can identify 8 ourselves for the first time around, and then I think the 9 transcriptionist won't have a problem. 10 DR. MATOBA: Okay. Alice Matoba. My question 11 is your presentation indicated that there are significant 12 differences between races in terms of endothelial cell 13 density, and I wonder if you have any sense whether there 14 can also be differences between races in terms of the 15 minimum endothelial cell count you would need before one 16 starts to develop clinical edema? 17 DR. EDELHAUSER: I don't think that's in the 18 literature at all. I mean, just as a little bit of a 19 sidelight, if you go back and looked at the original radial 20 keratotomy, it was done by Sado, and the reason he was 21 somewhat successful I think is that the population of 22 individuals he did the study on had more endothelial cells. 23 So no, we don't know the minimum, and the best data would 24 probably come from Japan, Dr. Matsuda's laboratory. 25 DR. WEISS: Dr. Grimmett? 74 1 DR. GRIMMETT: Dr. Grimmett. My question was 2 asked by Dr. Bandeen-Roche and answered, so I'll pass at 3 this time. 4 DR. WEISS: I'm just going to ask you a quick 5 follow-up question, then. I understand that you would want 6 CV as part of specular microscopic studies. Is there a 7 number that you would assign beyond which this is of 8 concern? I'm being a very concrete person. 9 DR. EDELHAUSER: Are you talking about an upper 10 level? 11 DR. WEISS: Upper level, yes. 12 DR. EDELHAUSER: Well, the thing that you see 13 in your upper levels are CVs up around, say, 45 and above. 14 You know, that's a very high number. 15 DR. WEISS: That's of concern. 16 DR. EDELHAUSER: Yes. 17 DR. WEISS: Thank you. 18 Dr. Bradley? 19 DR. BRADLEY: Dr. Bradley. A couple of 20 statistical questions, really. You seem to be suggesting 21 you should be avoiding contact lens wearers in the sense 22 that I got the impression you were treating that as perhaps 23 a confounding source of an independent variability, but if 24 the contact lens wear interacts with the effects of phakic 25 IOLs, are you then more susceptible to this? And these are 75 1 the sorts of patients that might be getting phakic IOLs. 2 Surely, they shouldn't be excluded, but should they be 3 specifically included in such a study? Could you comment 4 on that? 5 DR. McCAREY: I agree completely with what 6 you've said. I just want to make sure that you're aware of 7 which patients has had a contact lens history. 8 DR. BRADLEY: Second question. Again, I think 9 statistically you did quite a thorough job of demonstrating 10 different sources of variance, and I lost track a little 11 bit of the individual sources of the variance, but it 12 seemed you put that all together in a graph and you 13 suggested that in order to detect a change in cell density 14 in an individual eye, it would have to change by 9 percent. 15 DR. McCAREY: With that set of data that I 16 presented to you, yes. 17 DR. BRADLEY: So the question, and I wondered 18 if you knew the answer, is what would be the sample size 19 required to detect a clinically significant change in the 20 sample of eyes that might be used, for example, in the 21 study of phakic IOLs? 22 DR. McCAREY: I'd have to go back to the 23 computer. I don't know the answer. It's a statistical 24 question that I didn't ask. 25 DR. GRIMMETT: Michael Grimmett. I'll present 76 1 that during my talk. 2 DR. WEISS: Dr. Huang? 3 DR. HUANG: Andrew Huang. My question is that 4 so far we have emphasized the physical characteristics of 5 the endothelium, but we all know corneal thickness is a 6 function of the corneal endothelial functions. Do the 7 speakers have any thought about what's the value of corneal 8 thickness in terms of evaluating the cornea's general 9 health? 10 DR. McCAREY: Well, as part of an answer, and 11 I'm sure Dr. Edelhauser can expand on this, but you can get 12 endothelial cell counts down below 1,000 to 900 and 500 and 13 still have normal corneal endothelial thickness. 14 DR. HUANG: Exactly, yes. 15 DR. McCAREY: So it seems to be not as 16 sensitive an indicator as to what trauma may have happened 17 to that tissue. 18 DR. HUANG: But by the same token, the flipside 19 of the coin is that if you have a decrease of cell density 20 from your graph, from the aging population, from 3,000 or 21 4,000 at birth to 2,000, but the patient did not really 22 have any functional visual disturbance, is that a bad thing 23 to have a decrease in those endothelial cells or is that a 24 good thing to have a healthy corneal thickness? 25 DR. EDELHAUSER: Well, I think there's a limit 77 1 obviously. I mean, a corneal endothelial cell population 2 between, say, 1,000 and 2,000 will survive very nicely, 3 because we see this in many cases with patients with 4 guttata and we see it with postkeratoplasty with patients 5 and it's fine. 6 I think that obviously when you set up a study 7 that you want to do this where there's a potential loss of 8 endothelial cells or you want to see it, you don't want to 9 start out with patients that have 1,400 cells, for example. 10 So you'd really want to have -- a "normal" endothelial 11 cell population with some polymegethism would be fine, up 12 around 2,500 cells per se. 13 DR. HUANG: But by the same token, you may now 14 have started with the patient in, say, the Asian population 15 with an endothelial cell count of 2,500, but with a corneal 16 thickness of thicker than 600 microns. 17 DR. EDELHAUSER: Occasionally, yes, you do see. 18 Some people do have thicker corneas. 19 DR. HUANG: But that itself may be indicative 20 of the corneal endothelium is compromising. 21 DR. EDELHAUSER: I understand. Sure. That's 22 true, very much so. 23 DR. HUANG: Yes. Seeing the cell number does 24 not necessarily mean the cell is alive. 25 DR. EDELHAUSER: Right, but I think you bring 78 1 up, if corneal thickness is going to be an issue, again, 2 that's another training point, and, one, the corneal 3 thickness measurements off of some of the specular 4 microscopes are not that accurate compared to ultrasound. 5 DR. WEISS: Dr. Mathers? 6 DR. MATHERS: Presumably, you would recommend 7 that we do not include patients with Fuch's dystrophy and 8 significant guttata because that would confound this 9 measurement considerably, correct? 10 DR. EDELHAUSER: Yes. 11 DR. MATHERS: But you see an occasional 12 guttata, maybe one or something like this, in a fairly 13 large percentage of the population. Could you hazard a 14 guess as to how many in single field could be there and you 15 would qualify them to be excluded? 16 DR. McCAREY: Well, as a reader of the images, 17 it isn't so much if they're present. It's can I get a 18 large enough area remaining that's contiguous for counting 19 of the cells. So I do not count around the cells, and I 20 don't want to count a narrow sheet between guttata because 21 that will louse up the algorithm. 22 DR. MATHERS: But to do this study, you want to 23 have as clean a group as possible. So presumably, we 24 wouldn't want to have patients where we're fighting the 25 guttata. 79 1 DR. McCAREY: Yes. This is probably one of the 2 reasons I like to see more than one picture taken because 3 if there is a random guttata, I try to select the picture 4 with the least problem. 5 DR. MATHERS: So you're not going to -- 6 DR. McCAREY: Rather than averaging the three. 7 That's not so important to me because I think that when 8 you consider the surface area, it's .003 percent of the 9 surface for one picture. What's one more picture out of 10 the whole surface? Very little. 11 DR. MATHERS: Yes, right. 12 DR. McCAREY: But you want to get a good 13 picture, and if it means getting a good contiguous field of 14 cells, then that's what a good picture is. 15 DR. MATHERS: All right. 16 DR. WEISS: I get the impression that Dr. 17 Mathers is trying to quantify it because we're doing a 18 guidance document. 19 DR. MATHERS: Yes, right, because if you take 20 enough pictures, you're going to get a spot where you can 21 count 150 cells and in a single field you might still have 22 15 guttata. We may not want to include this in a study 23 where we're looking at this because those guttata are going 24 to indicate a confounding population. We need some kind of 25 measure as to say this person has too many guttata to 80 1 include in a study like this. 2 DR. EDELHAUSER: In a preops study, yes. I 3 mean, I think that this could be an exclusion criteria. 4 DR. MATHERS: Right. 5 DR. EDELHAUSER: And that would be put in 6 there. You know, obviously, there are patients that have 7 few guttata, but there are a lot of patients that have a 8 lot of guttata and they should be excluded. 9 DR. MATHERS: Okay. We'll call it a lot. 10 And presumably, when you suggested a two-year 11 follow-up, you're speaking of an insult and then a process 12 of evaluating the endothelium after that time, but if you 13 have an ongoing insult, ongoing inflammation, presumably 14 then would you modify your two-year recommendation if 15 you're going to assess that? 16 DR. EDELHAUSER: Indeed, I would. You know, if 17 you're going to follow these patients, three years 18 certainly would be reasonable to follow these patients out. 19 DR. McCAREY: Are you implying a chronic 20 inflammation? 21 DR. MATHERS: Chronic inflammation. 22 DR. McCAREY: If it was a chronic inflammation, 23 wouldn't you expect a chronic loss? 24 DR. MATHERS: Correct. 25 DR. McCAREY: So you would expect to see 81 1 eventually a linear line occurring, and that's when I'm 2 telling you that you've followed them long enough. You 3 don't have to follow them infinitum. You need to know 4 what's going on at a steady state. 5 DR. MATHERS: Right, but that would presumably 6 be longer than an initial insult period study. So you were 7 asked to give an estimate and when you said two years, 8 certainly that would be longer if you have a chronic 9 process. Can you give a time? 10 DR. McCAREY: I still kind of flip the response 11 back to you by saying that I'm looking for a linear 12 response. 13 DR. MATHERS: Yes. 14 DR. McCAREY: And if it's still not linear at 15 the end of two years, I have to keep going. I have to go 16 for a longer follow-up time. 17 DR. MATHERS: Right, but we need -- 18 DR. McCAREY: And I would not know how long it 19 would be. 20 DR. MATHERS: You don't know. 21 DR. McCAREY: Yes. 22 DR. MATHERS: Okay. Fine. 23 DR. EDELHAUSER: I think, Dr. Mathers, the 24 longest study that we've followed through are the LASIK 25 patients, and that's been three years out. 82 1 DR. MATHERS: Right. 2 DR. EDELHAUSER: So I haven't really 3 participated and I think with the KeraVision, their three- 4 year data is just now coming in and being analyzed. 5 DR. WEISS: Dr. Owsley? 6 DR. OWSLEY: Cynthia Owsley. Dr. McCarey, you 7 mentioned that when people stop wearing their contact 8 lenses, the cell density counts, if we look at the 9 literature, suggest that it's pretty stable or there's not 10 this miraculous going back to the norm, whatever that is. 11 From the pragmatics of doing a clinical study on patients, 12 most of whom will be contact lens wearers and who have very 13 severe myopia, do you feel it's too burdensome on patients 14 to have them be without the contacts for six months and is 15 that maybe a little inflated? I mean, I know they can do 16 the spectacles, but being a myope myself, I know that a lot 17 of contact lens wearers, they like to wear their contacts 18 and not the spectacles. Just in terms of patient 19 enrollment issues. 20 DR. McCAREY: I have a great answer for that. 21 I'm a Ph.D. I don't have to worry about the clinical part. 22 I really don't know how to answer you because -- 23 (Laughter.) 24 DR. OWSLEY: That's a good answer. 25 DR. WEISS: Dr. Swanson, did you have any 83 1 questions? 2 DR. SWANSON: No questions at this time. 3 DR. WEISS: Great. 4 I want to thank you both very much. Those were 5 excellent presentations and extremely helpful to us, and I 6 thank you for your good humor with putting up with our 7 questions as well. 8 So you can move back from the table if you 9 would like, and we next have Dr. Liliana Werner from the 10 Storm Eye Institute, who will be speaking to us on lens 11 opacity. 12 Donna, do you have something to say first? 13 MS. LOCHNER: I'd just like to introduce Dr. 14 Werner a little more formally. She is an assistant 15 professor of ophthalmology at the Storm Eye Institute at 16 the Medical University of South Carolina in Charleston. 17 She is the senior scientist of the Center for Research on 18 Ocular Therapeutics and Biodevices. 19 She received her doctor of medicine from the 20 faculty of medicine of the Federal University of Minas 21 Gerais in Brazil in 1989, her residency in ophthalmology at 22 the Felicio Rocho Hospital in Brazil, and two postresidency 23 programs at the University of Paris and the Hotel-Dieu 24 Hospital in Paris. In 1999, she received a Ph.D. from the 25 University of Paris and began her work at the Storm Eye 84 1 Institute. 2 She is editor, together with David J. Apple, of 3 the summer 2001 issue of the International Ophthalmology 4 Clinics and is currently serving as a scientific referee 5 for many ophthalmology journals. She was recently selected 6 to joint the International Intraocular Implant Club and 7 starting in September of 2002, she will be the director of 8 research of the new David J. Apple Laboratories for 9 Ophthalmic Devices Research at the John Moran Eye Center at 10 the University of Utah in Salt Lake City. 11 Thank you. 12 DR. WERNER: Good morning. I would like first 13 of all to thank the FDA members for the opportunity to be 14 here and participate in this meeting, and I will be 15 discussing the issue of cataract formation after 16 implantation of phakic posterior chamber intraocular 17 lenses. This presentation is based on a review of the 18 literature, but also on some studies we performed in our 19 center and, as mentioned, soon we'll be moving back to Salt 20 Lake City, where the center was in fact founded in the '80s 21 by Dr. Apple and Dr. Olson. 22 So in fact what we did is an update of the 23 report we prepared for the ANSI meeting in Newport Beach 24 last year, and I would like to start with a brief overview 25 of the cell types involved in the problem of crystalline 85 1 lens and capsular bag opacification. 2 So if you look at this picture, we can in fact 3 divide the crystalline lens epithelium into two different 4 biological zones, and in this zone we have the A-cells, and 5 these cells are attached to the inner surface of the 6 anterior capsule, and when the cells are disturbed, they 7 have the tendency to remain in place and undergo a process 8 of fibrous metaplasia. These cells are in continuity with 9 the cells in the equatorial region, or the E-cells, and 10 these cells, on the contrary, when they are disturbed, they 11 have the tendency to migrate and proliferate, forming 12 bloated cells. 13 So both cell types are involved in the 14 different forms of capsular bag opacification. For 15 example, anterior capsular opacification after implantation 16 of different intraocular lenses. Also, different forms of 17 secondary anterior to capsular cataracts. Also, 18 interlenticular opacification between piggyback lenses, and 19 finally posterior capsular opacification after implantation 20 of different intraocular lenses. 21 So here, for example, you have the A-cells 22 being the most important cell type involved in the process 23 of anterior capsular opacification after implantation of 24 different intraocular lenses for cataract surgery, and you 25 can see here a beautiful example of the anterior capsule 86 1 being opacified only where it's in contact with the IOL 2 material. 3 So here you have another example, and this is a 4 case of capsule contraction syndrome, and because you have 5 asymmetric fibrotic formation and asymmetric contraction, 6 the lens is also decentered. 7 Whenever you prepare those specimens for 8 histopathological evaluation, that's what you're going to 9 find at the level of the capsule axis edge. So there are 10 always these fibrocellular tissue attached to the inner 11 surface of the anterior capsule, and this in fact 12 corresponds to the opacification in this border. 13 And of course, the A-cells are also the cell 14 type most involved in the anterior capsular cataract 15 associated with phakic IOL implantation, and here I have a 16 bilateral case provided by Dr. Koch. So we have here the 17 opacity in the right eye and in the left eye. I'll be 18 talking a lot today about the ICL because the available 19 literature is related to the ICL and also these are the 20 specimens we have available to us in our center. 21 So what happens in this case is that the 22 surgeon has to explant the lens and then he is going to 23 perform the cataract procedure. So we are recommending 24 them to save for us the capsule excess fragment so we can 25 perform histopathological analysis of this anterior capsule 87 1 fragment, and in fact it's been very interesting to notice 2 that there are many similarities between anterior capsule 3 specimens obtained in different situations. 4 So for example, here you have the capsule 5 excess edge of a case of anterior capsular opacification 6 after implantation of a silicone lens, here you have the 7 specimen, the capsule excess specimen obtained during the 8 surgery of an anterior subcapsular cataract secondary to 9 uveitis, and finally here is the specimen we received in 10 our center. It's the capsule excess obtained during the 11 cataract procedure for a case of anterior subcapsular 12 cataract after phakic IOL implantation. So in fact, if you 13 see these three examples, you will always find this 14 fibrocellular tissue attached to the inner surface of the 15 anterior capsule, corresponding to the opacification. 16 Finally, the E-cells are the most important 17 cells involved in the process of posterior capsule 18 opacification and mostly in the pearl form of posterior 19 capsule opacification. Both cells, and mostly the E-cells, 20 are involved in the process of interlenticular 21 opacification between piggyback lenses, and when you 22 analyze such specimens in histopathology, you are going to 23 find residual corneal material and pearls very similarly to 24 what is observed with posterior capsular opacification. 25 I'd like now just to summarize the evolution of 88 1 the designs of the phakic posterior chamber lenses. 2 So, as you know, these lenses were introduced 3 by Fyodorov in the '80s and the first designs were pupil- 4 fixated lenses. So this lens, for example, was supposed to 5 stay in the sulcus, but the optic component would protrude 6 through the anterior chamber through the pupil. 7 The second generation was represented by the 8 Chiron Adatomed silicone lens, and in fact this lens was 9 withdrawn from the market because of cataract formation. 10 This was really a very thick lens. 11 Finally, the third and current generation is 12 represented by the Staar ICL manufactured from the collagen 13 material. This is a much thinner lens, and in fact there 14 are different models of this design and each model has 15 different vaulting characteristics. 16 The third generation is also represented by the 17 Medennium phakic refractive lens, or PRL, manufactured from 18 silicone, and this is also a very thin lens, and you have 19 here the myopic model and the hyperopic model. 20 So let's talk about some relevant aspects of 21 fixation and sizing. We had the opportunity to analyze 22 some Chiron Adatomed silicone lenses which were explanted 23 because of the problem of cataract formation. Then, after 24 analysis of the lenses, we reimplanted the lenses in eyes 25 of different sizes, and in fact we could observe from the 89 1 posterior view or a side view that the lens was really too 2 big and too thick. The lens was really sitting on the 3 zonulas. We could not fixate the lens in the sulcus, and 4 because it was very thick, it was in large contact with the 5 posterior surface of the iris and anterior surface of the 6 crystalline lens. 7 With respect to the ICL, consecutive V models 8 of this design had different vaulting characteristics, and 9 this was done in order to reduce the possibility of 10 cataractogenesis. Apparently, the sizing is important for 11 this design. So a lens that's too large will be followed 12 by excessive vaulting, but a lens that's too small for the 13 eye will be unstable and eventually become decentered. 14 Dr. Ferdinand Trinidad from Brazil very nicely 15 summarized different situations with incorrect sizing of 16 the ICL. So for example, if you have a lens that's 17 oversized, the vaulting will be excessive and there will be 18 a large area of contact between the lens and the posterior 19 surface of the iris. 20 Here, for example, there is a central vault, 21 but a large mid-peripheral contact, and you have a pool of 22 aqueous humor that's stagnated between the lens and the 23 crystalline lens, and eventually, as we're going to discuss 24 further, there is the possibility of some metabolic 25 disturbances. 90 1 Finally, if the lens is clearly undersized, the 2 lens will be unstable and there will be a large area of 3 contact between the lens and the anterior surface of the 4 crystalline lens. 5 So the sizing issue is eventually a very 6 important issue for phakic IOL implantation in general, and 7 in general surgeons are using the measurement of the white 8 to white to finally choose the size of the lens that's 9 going to be implanted in the eye of the patient. 10 For example, for the ICL, if you review the 11 literature, surgeons would measure the white to white and 12 then add 0.5 millimeters for a myopic eye or they subtract 13 0.5 millimeters for a hyperopic eye, but this measurement 14 can be so inaccurate, and sometimes we receive some cadaver 15 eyes in our lab and we don't even know exactly where to 16 measure. 17 This is a very recently published paper by this 18 group. They analyzed 43 eyes of 24 patients. They 19 patients were aged at around 34 years and they were highly 20 myopic or hyperopic. They performed measurements of the 21 white to white with surgical calipers, and they tried to 22 look for a correlation between the white to white and the 23 sulcus diameter measured with composites of UBM 24 photographs. 25 They concluded that the traditional estimation 91 1 of the sulcus size through the limbal measurement is 2 inadequate. So the limbus size alone would not be able to 3 predict the sulcus size. 4 We are also trying to do different studies 5 using cadaver eyes regarding the sizing. For example, we 6 are actually working on this protocol where we get 7 different cadaver eyes, we measure the anterior/posterior 8 length, then we mark the 12 o'clock position, and we 9 localize the horizontal meridian and the vertical meridian. 10 Also, we perform measurements using a plastic sizer of the 11 anterior chamber diameter, and after that we try to fixate 12 the eye with special techniques which allow us to keep the 13 geometry of the whole anterior segment. Then we select the 14 meridian to be studied with the form sections, and then we 15 directly measure the angle to angle and the sulcus to 16 sulcus with surgical calipers. 17 So in fact, we have some eyes where we studied 18 the vertical meridian and other eyes where we studied the 19 horizontal meridian, and this is preliminary data and the 20 study is not finished, but it's already very interesting to 21 notice that, for example, here, for the same measurement of 22 the white to white, which here is 11, we obtained real 23 measurements of the sulcus to sulcus which went from 11 to 24 almost 13. So then if you would choose an 11.5 ICL to 25 implant in these eyes, what would happen with the eyes 92 1 between 12 and 13? 2 So I would like to mention the new technology 3 that's being developed that will help in the issue of the 4 sizing, and I mention this device because this is the 5 device we are having the opportunity to work with right 6 now. 7 So we are working this protocol where again we 8 use cadaver eyes and we measure the anterior/posterior 9 length, then we mark the 12 o'clock position, measure the 10 white to white, and with this prototype of ultrasound we 11 are performing of the anterior chamber diameter and then 12 the sulcus-to-sulcus diameter. Then we prepare the eye 13 with these special techniques for fixation. We perform the 14 sections in the region we choose, and finally we perform 15 the direct measurements of angle to angle and sulcus to 16 sulcus. 17 This is also preliminary data, but so far we 18 analyzed nine phakic cadaver eyes, and if you compare the 19 angle-to-angle measure by calipers and ultrasound, the 20 results are very similar and this is valid also for the 21 sulcus to sulcus, and this is also valid for pseudophakic 22 cadaver eyes we analyzed. So far, we analyzed only six. 23 When you look at the pictures you obtain with 24 the ultrasound, in fact they apparently reflect very nicely 25 the morphology we obtain after the fixation of the 93 1 specimen, which allows us to perform the measurement which 2 appeared to be very accurate. So here you have an example 3 of a phakic cadaver eye, and here the same analysis is 4 performed in the pseudophakic cadaver eye. 5 Of course, this technology is going to be 6 extremely important in the follow-up of patients implanted 7 with different phakic IOLs. It will be important for the 8 measurement of the distance between the edge of the lens 9 and the mid-periphery of the cornea. For example, in the 10 phakic anterior chamber intraocular lenses, and also 11 extremely important to the measurements of the posterior 12 surface of the lens and the anterior surface of the 13 crystalline lens in phakic posterior chamber IOLs. 14 So let us review briefly the surgical 15 implantation. We have to remind you there are lots of 16 opportunities for the surgeon to create the cataract 17 observed in the postoperative period. 18 So the first thing the surgeon has to do is in 19 fact to perform these YAG laser iridotomies. In general, 20 they use two superior iridotomies placed 90 degrees apart, 21 and this is performed one or two weeks before surgery. 22 There are some studies indicating that these have 23 eventually a cataractogenic effect also and that they 24 contribute to pigment deposition which we always observe on 25 the surface of these lenses. Also, in an alternative way, 94 1 the surgeon can perform one single surgical iridectomy. 2 After that, the surgeon has to perform the 3 incision. These are foldable lenses, the incision is very 4 small, and it can be used to correct preexisting 5 astigmatism. The surgeon has to inject viscoelastics, 6 which is extremely important in the protection of 7 intraocular tissues and also to allow the lens to unfold in 8 very controlled manner. 9 Both lens types can be inserted with forceps 10 and also injected within the anterior chamber, and then 11 finally the haptics will be placed behind the iris with 12 spatulas or hooks, and this is a very important step 13 because no pressure should be placed on the crystalline 14 lens at that time. Then the pupil is constricted with 15 miotic agents, the viscoelastic is removed, and the wound 16 is closed. 17 So of course, the crystalline lens should 18 ideally not be touched at all during the whole surgery, but 19 as you can see, there are many opportunities to have 20 accidental contact with the anterior capsule of the 21 crystalline lens not only during the placement of the 22 haptics behind the iris, but also injection of viscoelastic 23 behind the iris, et cetera. So anterior capsule trauma, as 24 we review the literature, you will notice that this may 25 lead to crystalline lens opacities months later after the 95 1 procedure. 2 There are some studies indicating that in many 3 ways high myopic patients are going to have cataract and 4 with earlier onset. For example, I can cite this study 5 indicating that moderate to high myopic patients had an 6 association with age-related cataract. For lower levels of 7 myopia, this relationship has been disputed. They also 8 indicated that early onset of myopia is a strong 9 independent risk factor for cataract. 10 But we can not forget that they are talking 11 here about age-related cataract. They are talking about 12 mostly nuclear cataract, and when you review the different 13 forms of cataract according to the age they appear, you're 14 going to notice that anterior subcapsular cataracts are 15 very rare forms of age-related cataract unless they are 16 caused by inflammation or injury. 17 So I'd like to summarize some of the specimens 18 we are receiving in our center. We had the opportunity, as 19 I mentioned, to analyze some silicone Chiron Adatomed 20 lenses and we have recently received eight ICLs, all 21 explanted because of cataract formation. There are some 22 bilateral cases and the lenses were explanted between one 23 year and four years after implantation. 24 Here you have some examples. This is one of 25 the eyes of the patient, and here the corresponding ICL 96 1 explanted from this eye, and this is the contralateral eye 2 and the corresponding ICL explanted from the eye. So this 3 surgeon not only submitted the ICL he explanted, but also 4 he submitted the fragment of capsular excess while he 5 performed the cataract surgery. 6 So in general, when you analyze the surfaces of 7 these ICLs, you always find some pigment deposition, as you 8 can observe here, and in these specimens, stained with 9 different techniques, you always can observe the 10 fibrocellular tissue attached to the inner surface of the 11 anterior capsule which is corresponding to the opacity. 12 This pigment deposition can be very discrete, 13 as in the previous case, but it can also be very important, 14 as you can see here in this bilateral case. These lenses 15 were also explanted because of cataract. 16 What about the mechanisms of this cataract 17 formation? This study is very interesting because it 18 summarized many of the factors that are eventually 19 important. So these patients were implanted with the ICL 20 and they observed an anterior chamber reduction in 9 to 12 21 percent of the cases. Central endothelial cell density 22 decrease, not progressive, but very interesting, they 23 report an increase of the aqueous flare in 50 percent of 24 the cases with stabilization, but always above preoperative 25 values. They reported progressive decrease of crystalline 97 1 lens transmittance with time, contact ICL iris in all eyes, 2 peripheral contact ICL and crystalline lens in 60 percent 3 of the cases, central contact of the ICL and crystalline 4 lens in 15 percent of the cases, and changes in ICL axis in 5 10 percent of the cases with rotation of the lens in the 6 postoperative period. But they didn't observe any cataract 7 formation after the follow-up. 8 So of course, we mentioned already surgical 9 trauma can cause these cataracts were are observing after 10 implantation of phakic lenses. We cannot forget the 11 possible effect of the YAG laser for the iridotomy. We 12 cannot forget the accidental contact of the anterior 13 capsule is possible during different surgical steps. 14 Intermittent microtraumas can also cause these cataracts. 15 There is an increased crystalline lens curvature during 16 efforts for accommodation. It was demonstrated that the 17 lens can rotate inside the eye in the postoperative period, 18 and of course, there is always an increase in the overall 19 lens size throughout life, so the distance between the 20 phakic lens and the crystalline lens is not always going to 21 be the same. 22 So what about constant trauma? This apparently 23 is extremely important. So here in cases of clearly 24 undersized lenses, there would be a large area of contact 25 between the lens and the anterior surface of the 98 1 crystalline lens. 2 Also, there is the possibility of a continuous 3 disruption of the blood/aqueous barrier with subclinical 4 inflammation, and this is caused by friction between the 5 iris and the phakic lens, and eventually by the ciliary 6 sulcus fixation also. These have effects not only on the 7 crystalline lens transmittance, but eventually on the 8 corneal endothelium. 9 What about crystalline lens metabolic and 10 nutritional disturbances? We already commented on this 11 situation, for example. There is a pool of aqueous humor 12 stagnated between the lens and the anterior surface of the 13 natural crystalline lens. So this could be caused by the 14 previously mentioned subclinical inflammation, but by any 15 cause of blockage of normal circulation of the aqueous 16 humor. 17 When we reviewed the literature, in fact we 18 performed a review from '96 to 2002, and we could only find 19 studies regarding the early Fyodorov lenses, the Chiron 20 Adatomed silicone lenses, and the ICLs. We could not find 21 any studies regarding the PRL. I'd like to comment on some 22 of these studies because they have very interesting points 23 which are eventually very important. 24 So for example, in this study of patients 25 implanted with the Chiron Adatomed silicone lens, there was 99 1 no space between the phakic lens and the natural 2 crystalline lens in all cataract cases, which places 3 eventually this factor as one of the most important 4 factors. 5 In this study by Zaldivar and coworkers and 6 studying patients implanted in the ICL, he reported one eye 7 with a peripheral anterior subcapsular opacity which 8 developed in the region of the peripheral laser iridotomy, 9 showing again that these iridotomies eventually have a 10 cataractogenic effect. 11 So this group in Brazil studying patients 12 implanted with the Staar ICL reported an anterior 13 subcapsular opacity in the central non-contact area. So 14 the contact eventually is very important, but maybe there 15 are other factors or maybe the follow-up was just not 16 enough. 17 This group, also studying patients implanted 18 with the ICL, reported anterior subcapsular opacity which 19 developed 24 hours after surgery. So we may think that 20 this was really caused during the surgery and not by the 21 lens itself. 22 So this group reported one case of nuclear 23 cataract in a 53-year-old male. He had already some degree 24 of nuclear sclerosis, so of course, this is not the kind of 25 cataract we are talking about here. This is maybe just 100 1 age-related cataract and is not related to the procedure 2 and not to the lens also. 3 This is one of the very few studies which 4 really describes the evolution of the opacities. So they 5 describe opacities which appeared superiorly and then 6 progressed involving the optical zone, and in all cases 7 there was in fact a satisfactory central ICL vaulting in 8 the cases of cataract, indicating that in these particular 9 cataract cases, the contact was not the most important 10 factor. I like to cite this study because in general there 11 are many surgeons that would say that some peripheral 12 opacities would never progress. 13 Finally, this is maybe the only study which 14 compares patients implanted with an ICL in one eye and the 15 Chiron Adatomed silicone in the other eye, and they 16 demonstrated that the silicone lens was associated with 17 more cataracts and the cataracts appeared earlier in the 18 postoperative period. In all cases, no space was observed 19 between phakic lens and natural lens, so again, here we 20 believe that the contact is the most important factor. 21 So when you group all these papers together, 22 it's a big problem to really understand what is the real 23 incidence of cataract. With regards to the silicone lens, 24 the rates vary from 0 to 52.9 percent and regarding the 25 ICL, they go from 0 to 25 percent. 101 1 So why is that? First of all, because the 2 definition of cataract and opacity is really not the same, 3 some surgeons will report just the cataracts that are 4 creating visual problems for the patient, and some 5 opacities that are not decreasing the visual acuity would 6 not be reported. 7 Also, there is such a variation in the age of 8 the patients included, and also such a variation in the 9 follow-up considered, and in some studies they would say 10 that the follow-up is from three months to two years, but 11 in fact maybe just two patients had the follow-up of two 12 years and the great majority were followed up for six 13 months only. And in some studies -- for example, for the 14 ICL -- during the same study different models of the same 15 design were implanted, and you know with different vaulting 16 characteristics, so the effect for cataract formation would 17 not be the same. 18 So there is a great need of standardization of 19 these studies evaluating cataract formation. First of all, 20 the parameters used for the YAG iridotomies should be 21 described, because eventually this is cataractogenic. All 22 trauma to the anterior capsule during this surgery should 23 be noted for future reference. A follow-up period should 24 be at least two years because in fact, according to the 25 literature, the majority of cataracts appear between one 102 1 and two years after the procedure. And of course, we have 2 to evaluate very nicely the relationship of the phakic lens 3 with anatomic structures because the contact is one of the 4 most important factors. 5 Also, of course, we need a very accurate method 6 to choose the IOL size. To use just the white to white to 7 choose the overall size of the lens that's going to be 8 implanted is not accurate at all, and that's why we're 9 having so many complications, and this could really be 10 avoided. 11 There is a need of evaluation of subclinical 12 inflammation with laser flare meters because in some cases, 13 there was a very good vaulting lens and a cataract appeared 14 anyway. 15 There is a need of evaluation of explanted 16 phakic lenses with histopathological analysis of adjacent 17 tissues, but if you ask me right now if just with 18 histopathologic studies alone we will be able to 19 differentiate cataracts caused by the surgeon and cataracts 20 caused by the lenses, the answer is that we don't know yet. 21 We have just one specimen and we would like to look for 22 more specimens to have an impression about that. 23 And of course, we need to describe the 24 evolution of the anterior subcapsular opacity. 25 So let's talk about the possibility of a 103 1 classification for cataract formation after phakic IOL 2 implantation. So as I mentioned, some surgeons would say 3 that all the opacities they saw are peripheral and they are 4 non-progressive, but we have this paper indicating that the 5 peripheral superior opacity progressed, involving the 6 optical zone. So if you have a classification, we should 7 maybe classify the opacity in each visit to have an 8 impression about the progression of the problem. 9 So as you know, there are systems for the 10 classification of cataracts and they are all based on 11 standard retroillumination photographs and the total area 12 of the opacity. These are three well-known systems, the 13 LOCS system, Wilmer system, and Oxford system. 14 Here, we have some pictures showing how to 15 classify nuclear, cortical, and posterior subcapsular 16 cataracts with the LOCS system, all based on these standard 17 photographs. This could eventually be applied to cataract 18 formation after phakic IOL implantation. This is how we 19 grade the cortical opacities according to the Wilmer system 20 and this is the way we score anterior subcapsular and 21 posterior subcapsular opacities according to the Oxford 22 system. 23 Also, there are very sophisticated systems 24 combining high-resolution digital retroillumination imaging 25 with image analysis systems, allowing objective and 104 1 quantitative measurement, for example, of posterior 2 capsular opacification, which eventually would be very 3 useful in this clinical situation. 4 So in a way, if there is a way to do a 5 classification, this classification should indicate the 6 location of the opacity. For example, peripheral or 7 paracentral or central opacity. Also, maybe there is a 8 possibility to have an index for the intensity of the 9 opacity, and of course, we would have to score the area of 10 the opacity, and by doing that in each visit, we will have 11 an impression about the evolution and the progression of 12 the opacity and we would really understand better the 13 phenomenon. 14 Thank you very much for your attention. Thank 15 you very much again for the opportunity. 16 DR. WEISS: Thank you very much for an 17 excellent presentation. Would you be able to take a seat 18 at the table, and we'll open up to the panel for some 19 directed questions. 20 Dr. Bandeen-Roche, then Dr. Matoba, and then 21 we'll continue around. Dr. Bandeen-Roche? 22 DR. BANDEEN-ROCHE: Yes, I want to thank you 23 for your presentation -- very clear -- and your careful 24 recommendations about the data to collect involving the 25 surgery I think was great. 105 1 Just one brief question. You talked about the 2 importance of tracking the evolution of the cataract and 3 discussed grading. What implications do you think there 4 are for the frequency of evaluation and do you think that 5 that should be by passive surveillance or by active 6 surveillance? Just elaborate a little bit more on how to 7 track the evolution of cataract. 8 DR. WERNER: I don't have a precise idea about 9 when all these gradings should be done. Of course, 10 immediately after the postoperative period, within one 11 month to anything that's very fast developed, and maybe six 12 months, one year, two years, because this is what we see in 13 the literature. 14 But this is not only to just have the score. 15 It's also for us to understand the phenomenon because still 16 there are many surgeons who believe that the cataract is 17 really not a problem because they have just opacities in 18 the periphery that never progress, and we need to 19 understand if they are not really progressing or he is just 20 not observing. 21 DR. BANDEEN-ROCHE: Thank you. 22 DR. WEISS: Dr. Matoba? 23 DR. MATOBA: In regard to the development of 24 cataracts in areas of contact between the IOL and the 25 crystalline lens, how much do you think the lens material 106 1 or the nature of the lens material contributes or is it 2 mostly, you think, a mechanical effect? 3 DR. WERNER: Well, I don't know if I can answer 4 this question because we only have results about the ICL, 5 which we know the material. There is another lens made of 6 silicone and I'm not aware of their results, so we cannot 7 really compare if there is a material effect. 8 Also, as I mentioned, there are papers showing 9 that cataract is formed only in the area of contact. Other 10 papers would say that it was formed in a different area, 11 but maybe the follow-up was not enough. So there are still 12 many questions about that. 13 DR. WEISS: Dr. Bradley, Dr. Huang, and then 14 Dr. Mathers. 15 DR. BRADLEY: Just a clarification on the 16 peripheral cataracts you described. How peripheral are 17 they and, for example, would they become visually 18 significant under nighttime viewing conditions where the 19 pupil would dilate? 20 DR. WERNER: Well, when you look at the 21 literature, it is really not described, and sometimes, in 22 very early papers, they would describe some peripheral 23 opacities that would cause some glare in light conditions 24 of evening or something like that. But talking to 25 surgeons, they would say that the peripheral opacities 107 1 which are barely visible in the pupil dilation, they would 2 not cause any problem. 3 DR. WEISS: Dr. Huang? 4 DR. HUANG: Two questions. The first question 5 was regarding your earlier presentation, you seem to have 6 implied there were two types of cells that were induced in 7 the two different locations of the cataract. One is A- 8 cells induced in the anterior capsule, and then E-cells 9 induced in the posterior capsule. Is there any vital stain 10 that can help you to distinguish what type of cells are 11 responsible for the evolution of this cataract? 12 DR. WERNER: When you perform histopathological 13 studies, in fact what you see is that both cells are always 14 involved in everything, but there is always a predominate 15 type. For example, even for PESU, you have a fibrotic form 16 of PESU and you have a firm form of PESU, and when you 17 perform normal stains, you can see even morphologically 18 they are very different because E-cells always have the 19 tendency to be bloated, and the other are elongated 20 fibrotic cells, fibrotic-like cells. 21 DR. HUANG: In the LOCS III grading system, 22 it's really a numerical system and there is a highly 23 individual variation. Do you have any suggestion how to 24 standardize if that system were to be chosen for the 25 cataract characterization? And also, I believe the LOCS 108 1 III does not have any geographical information about a 2 cataract, and so do you have any suggestion how to modify 3 the system? 4 DR. WERNER: I believe there should be a 5 geographic thing because it's very important for the areas 6 of contact who are outside the contact of the lens. So in 7 this case, it's very important. 8 But with regards to your first question, I 9 think we should start by collecting many pictures from 10 surgeons to create standard pictures, as they have in the 11 LOCS system, and we have some, but we need more pictures to 12 have really different grading if we use such a system based 13 on standard photographs. 14 DR. WEISS: Dr. Mathers? 15 DR. MATHERS: Do you think that the 16 photographs, the retroillumination of the opacification, 17 can the photographs detect smaller anterior subcapsular 18 cataract formation than the slit lamp can detect it? What 19 do you think is actually the finest, highest resolving 20 method? 21 DR. WERNER: We have experience with these 22 systems based on retroillumination photographs for 23 posterior capsular opacification, and this is the best we 24 can have for that. 25 DR. MATHERS: Do you think that's higher 109 1 resolution than the human eye achieves with the slit lamp? 2 DR. WERNER: Eventually, yes. 3 DR. MATHERS: Yes. And have you any experience 4 using confocal microscopy, which could actually focus onto 5 the anterior capsule and have even higher resolution? Do 6 you think that that's possible? 7 DR. WERNER: Well, my own experience with 8 confocal microscopy regards the cornea, and I know that 9 there are some objectives which you could switch in some 10 devices and have an imaging of the anterior surface of the 11 crystalline lens. I have no experience with that and I 12 don't know if there is any data available published about 13 that. 14 DR. MATHERS: Sizing is clearly an important 15 process here. Do you think that the high-resolution 16 ultrasound will give the best sizing data and can that be 17 used clinically to determine which size to put in? 18 DR. WERNER: We are evaluating this right now 19 with these cadaver eyes. The results have been very 20 interesting. So I don't know exactly the status of 21 development of the technique, when it's going to be 22 available -- maybe this year -- but apparently it's the 23 best we can get for the moment. 24 DR. MATHERS: Do you think that the issue of 25 visual significance could be assessed best with glare 110 1 testing or what would you suggest as the most significant, 2 highest-resolving method to test the small amounts of 3 visual impairment you might get from an early cataract? 4 DR. WERNER: Yes, it is a very good question, 5 because in some cases discussed with the surgeon, the 6 lenses were explanted because of glare, and not really 7 decreasing visual acuity. So glare is very important. 8 DR. MATHERS: You think it's better than 9 contrast sensitivity testing as it's normally performed? 10 DR. WERNER: Maybe both. 11 DR. MATHERS: Just your opinion. 12 DR. WERNER: Maybe both should be associated in 13 this case, yes. 14 DR. WEISS: Mr. McCarley, did you have a 15 question as well? 16 MR. McCARLEY: One question just quickly. In 17 your experience, and maybe one of the clinical 18 ophthalmologists can answer this maybe even better, a cell 19 flare meter is used to determine subclinical inflammation. 20 Is that different in a posterior chamber lens than it 21 would be, for instance, in an anterior chamber lens? 22 DR. WERNER: Well, what we saw in the 23 literature is that there are also increased values of flare 24 cell meter with anterior chamber lenses, and apparently the 25 values are even higher and they also stabilize above the 111 1 preoperative values, contrary to cataract surgery, where 2 you have higher values, but these have a tendency to come 3 back to preoperative values after one year or so. 4 DR. WEISS: If there are no other questions, I 5 want to thank you, Dr. Werner, for your excellent 6 presentation. 7 We can move on to the open public hearing 8 session. Is there anyone who wanted to make any 9 statements? 10 (No response.) 11 DR. WEISS: If not, I'm going to put a question 12 to the panel. If we have perhaps a 15-minute coffee break 13 and skip lunch, we might be able to catch earlier flights, 14 as I know is in the interest of some. Are any of you 15 interested in doing that, taking a 15-minute break, rather 16 than -- so we have two hands up for Dr. Bullimore and Dr. 17 Matoba, and Dr. Mathers for sure. I would say that passes 18 without a formal vote. 19 So we'll take a 15-minute coffee break and 20 we'll see you back here in 15 minutes. 21 (Recess.) 22 DR. WEISS: We will now start the FDA 23 presentation and Donna Lochner will introduce the questions 24 for panel discussion. 25 MS. LOCHNER: Yes. I'm just going to go 112 1 through the questions and give a little bit of background 2 to where we were coming from with each question. 3 We're going to begin today with the endothelial 4 cell density study. After I've stepped through this 5 question, I'll turn the floor over to Dr. Grimmett, and 6 then the panel will discuss the endothelial cell issue 7 before going on to the next questions. 8 First, "Please comment upon the inclusion 9 criteria recommendations found in Table 1." This topic is 10 still being actively discussed, and particularly with the 11 ANSI Standards Committee, and so we believe any comments 12 will be very timely. 13 Table 1, which is just shown right here, 14 provides the recommended minimum endothelial cell 15 densities, and these values for minimum endothelial cell 16 density are generally being used in current U.S. phakic IOL 17 studies. These values were determined over the course of 18 several meetings over the years with input from FDA, 19 industry, and ophthalmologists that attend these standards 20 meetings. Allow me to hopefully clarify how the minimum 21 densities per age category were determined. 22 This slide is included as Attachment B in the 23 handout. First, the approximate initial cell density for a 24 21-year-old, as shown in the second column in this table, 25 was taken from the 1997 Moller-Pedersen article and the 113 1 citation for this article is provided in the handout, but 2 not on this slide. 3 For the 35- and 46-age categories, the cell 4 density at time of implant was approximated by assuming .6 5 percent yearly cell loss due to normal aging, with the .6 6 percent figure taken from the 1997 Bourne article, as 7 referenced earlier by Drs. Edelhauser and McCarey. This 8 was done to provide a check of whether the minimum 9 inclusion criteria per age group were reasonable. 10 The third column, the estimated rate of cell 11 loss per year, represents potential rates of loss due to 12 the phakic IOL. In other words, 1.5 and 2 percent assumed 13 loss from the phakic IOLs were used as examples to then 14 calculate the age when the cell density would be less than 15 1,200 cells per millimeter squared and less than 1,000. 16 These ages, shown in the fourth and fifth columns, assume a 17 surgical loss of 10 percent and compound the 1.5 and 2 18 percent loss annually. 19 Finally, in order to determine the minimum cell 20 density inclusion criteria, we looked at the starting 21 densities that would ensure greater than 1,000 cells at age 22 70 for the 21- to 25-age range, and at 75 for 26 and older. 23 So this table verifies that the minimum 24 inclusion criteria, as shown on Table 1, would be 25 sufficient in a worst case situation to allow for adequate 114 1 cell density for the health of the cornea for roughly the 2 life of the patient, assuming a 2 percent annual loss from 3 the phakic IOL and that patients would have at least 1,000 4 per millimeter squared at ages 70 to 75. 5 As you can see, there are various assumptions 6 inherent in the inclusion criteria and because of the 7 iterations this has undergone because of the committee 8 process, we are very much interested today in the panel's 9 comments on this inclusion criteria. 10 Our statistical calculations suggest that 200 11 subjects should be sufficient to detect a 2 percent loss 12 using measurements at multiple visits in order to establish 13 linearity of the loss. The measurements, as currently 14 proposed, would taken at the three- or six-month visit, the 15 12-, 24, and 36-month visits. 16 Further, although the statistics suggest that 17 200 subjects would be sufficient, we recommend that 18 specular microscopy be performed on all subjects enrolled 19 in the study to ensure that 200 analyzable photographs are 20 obtained. 21 Last, we recommend that multiple images be 22 captured at each visit and the mean endothelial density 23 from those multiple images be used in the analysis. 24 We are asking for panel comments on these 25 criteria as well. 115 1 Now, I'd like to turn the floor over to Dr. 2 Grimmett for his review. 3 DR. GRIMMETT: Thank you, Donna. 4 This is Michael Grimmett. I've prepared some 5 comments in outline form which at least the panel members 6 should have on their table. It's a 10-page outline, and I 7 promise to go very quickly through it. There are four 8 tables as well. 9 Regarding the questions, I think I'll go 10 through my outline first. I was doing the bulk of this 11 thought process and review prior to having the questions. 12 I was doing the work, I'd gone on vacation out in Santa Fe, 13 and I'm relieved and grateful that in the open public 14 hearing session, my wife did not comment on the timing of 15 that review. 16 (Laughter.) 17 DR. GRIMMETT: Notwithstanding that, I think 18 I'll do this review first, and then we'll come back to the 19 questions if some of the issues are not resolved. 20 Just in general, and some of this has been 21 alluded to by the previous speakers, the peer-reviewed 22 literature on phakic IOLs has numerous limitations, and 23 it's important to recognize that when reviewing any data 24 that's reported in the literature. Mostly, the data are 25 retrospective in design, they're non-randomized case 116 1 series, they have extremely low numbers of eyes reported, 2 there is poor accountability for the longer follow-up 3 intervals, and in general morphometric endothelial analyses 4 are not generally reported. So coefficient of variation 5 and percent hexagonality are not there. 6 Additionally, my review is not exhaustive or 7 comprehensive. 8 I think it's instructive to look at phakic IOL 9 types because we can separate them into three different 10 animals that have different implications for the corneal 11 endothelium. I've separated them into anterior chamber 12 type and posterior chamber type. Of the anterior chamber 13 type, there are two, angle-supported and iris-fixated. 14 Looking at the literature on the angle- 15 supported lens, I'll just give you a smattering of some 16 articles to go over what are reported endothelial cell 17 losses and what's known about design parameters that would 18 impact our recommendations regarding future phakic IOL 19 studies. 20 To start with, angle-supported lenses, a first- 21 generation lens, the Baikoff ZB lens, had a distance 22 between the IOL edge and the endothelium of only 1.16 23 millimeters, and that was the key factor. It was 24 determined that there was a high endothelial cell loss 25 secondary to excessive contact between the IOL optic edge 117 1 and the endothelium. One report by Jimenez-Alfaro reported 2 a 16 to 18.8 percent loss at one year and a 20 to 28 3 percent loss at two years. Another report reported up to 4 56 percent loss with a shorter follow-up. Of the second 5 report, 37.5 percent had morphologic or cell density 6 changes. 7 Quotes in these articles include "We have 8 stopped doing this surgical procedure" and "Further 9 implantation of this IOL is unacceptable." 10 The second-generation lens, the Baikoff ZB5M 11 lens, was manufactured until 1997. The major difference 12 here is that they increased the distance between the IOL 13 edge and the endothelium, increasing that to 1.56 14 millimeters. Endothelial cell loss in one study was 15 reported at 4.5 to 5 percent cell loss at one year, 5.6 to 16 6.8 at two years, and 5.5 to 7.5 loss at three years. 17 A separate study by Perez -- forgive me for my 18 pronunciation -- Perez-Santonja found an endothelial cell 19 loss of 12.33 percent at one year and remaining relatively 20 stable at two years. 21 Then the larger study was recently reported in 22 Ophthalmology found that at one year there was a 5.53 23 percent loss, and interestingly, after year 2, while the 24 numbers fell down significantly in terms of total eyes 25 examined, the overall loss approximated normal aging 118 1 losses. That is, preop to year one was at 5.5 percent 2 loss; year 1 to year 2, 1.37 percent loss; year 2 to 3, .72 3 percent loss; year 3 to 4, there was a .28 percent loss; 4 year 4 to 5, there was a .55 percent loss; year 5 to 6, .37 5 percent; and year 6 to 7, .56 percent. 6 So based on this study, and the reason I find 7 this instructive, is that they give us some information 8 regarding what is a reasonable duration to follow these 9 particular lenses before they come to the panel. If we 10 take this data, which is one of the largest subsets that I 11 could find in the published literature, in year 3 they 12 still had 157 eyes, albeit they're mixing lens times. 13 They're either started or stabilized, and they found 14 stabilization from the two- to three-year period, and you 15 can tell three years is the appropriate duration of a 16 study. 17 A fourth-generation lens, ZSAL-4 by Morcher, 18 had a distance from the IOL edge to the peripheral cornea 19 of 1.65 millimeters and they reported an endothelial cell 20 loss of 3.5 percent at one year and 4.2 percent at two 21 years, albeit the numbers are low. There are only 18 eyes. 22 They commented that there is a fifth-generation lens, but 23 I could not locate it in the published data regarding this 24 product. 25 Another anterior chamber angle-supported lens 119 1 in the literature is the Nuvita lens by Bausch & Lomb, and 2 it's a single-piece PMMA IOL. In one study of 21 eyes by 3 Allemann in Ophthalmology, they reported a 10 percent cell 4 loss in the first year, another 4.3 percent in the second 5 year, and, given these large numbers, certainly if that 6 were to come to the panel, I would want to see more data. 7 The next category of anterior chamber lenses 8 are iris-fixation lenses. A Worst-Fechner lens developed 9 in the late '80s, one particular study just reported two 10 eyes of greater than 50 percent cell loss, but they didn't 11 provide any mean cell density analyses or morphometric 12 analyses. A separate study in 1996 by Perez-Santonja 13 reported a 13 percent loss at one year, another 4.6 percent 14 at two years, indicating to me that the central endothelial 15 cell loss did not stabilize over a two-year period. The N 16 was only 30 in this particular study. 17 The Artisan or Iris-Claw lens, the prior 18 nomenclature which is unfortunately called the Worst Iris- 19 Claw lens -- 20 (Laughter.) 21 DR. GRIMMETT: Those are synonymous as far as I 22 understand. There may be some design changes I'm not aware 23 of, but I believe they're in the same family. 24 It had reported some distances from the corneal 25 endothelium in various publications. For example, a -15 120 1 diopter lens with a 3.2 millimeter anterior chamber leaves 2 1.97 millimeters from the corneal endothelium. Because 3 these lens are fixated at the iris itself and not vaulted 4 above the iris, certainly you would expect more distance, 5 and this is what we're seeing. There's more distance from 6 the endothelium. So the angle-supported lenses, in my 7 view, given their proximity to the cornea, have a higher 8 risk to the corneal endothelium than, let's say, an Iris- 9 Claw lens, assuming that the anterior chamber (inaudible) 10 is constant, whatever it happens to be. 11 Looking at some data regarding endothelial cell 12 loss on the Artisan lens, a study by Menezo showed a 6.6 13 percent loss at 12 months for 109 eyes and 9.22 percent at 14 two years, but I think it's instructive just to look at the 15 differences. It's 6.6 percent preop to year 1; 2.63 year 1 16 to 2; 2.5 percent year 2 to 3; 1.74 year 3 to 4. 17 They found that the cell loss correlated to 18 increased power of the lens, the thicker lens -- that is, 19 perhaps closer to the endothelium -- and shallower anterior 20 chamber depth, which would also tend to bring the lens 21 closer to the corneal endothelium. 22 This particular study reported morphometric 23 measurements regarding percent hexagonality and coefficient 24 of variation. 25 They also find some changes that I found 121 1 instructive. They found that there were statistically 2 significant decreases or changes at six months and then 3 return at approximately two years with gradual resolution 4 back to near preop factors in their four-year study. 5 Then the final category of lenses would be 6 posterior chamber lenses, which we've heard a great deal 7 about already here today by the speakers. Dr. Werner's 8 excellent presentation showed us pictures of the Fyodorov 9 posterior chamber lens, which is not in general clinical 10 use, with cell losses of 10 percent at 12 months. 11 Interestingly, the Chiron Adatomed lens that 12 Dr. Werner indicated was pulled because of cataract 13 formation, at least the two studies I pulled, did not 14 report any endothelial cell data whatsoever, reports by 15 Fechner and Marinho. I didn't locate any other studies on 16 those lenses. 17 The Staar Surgical ICL that we heard about 18 today already, one study by Zaldivar of 124 eyes didn't 19 report endothelial cell counts, and in another study by 20 Arne, cell loss was actually remarkably low, 2 percent at 21 12 months and 2 percent at 24 months, and no eye had an 22 endothelial cell loss greater than 3.8 percent at one year 23 in 58 eyes. Since these lenses are in Phase III trials, 24 I'm certain there's some data targeted and reviewed by the 25 FDA in a confidential fashion regarding larger sample 122 1 sizes. 2 Knowing what the literature has to say about 3 endothelial cell loss and the differences between these 4 three types of lenses would help us give guidance regarding 5 future manufacturers' studies. It's important, as Dr. 6 Edelhauser alluded to, to look at normative data, so that 7 we know what the natural rate of loss is for corneal 8 endothelium in order to gauge our comments. 9 Rates of normal endothelial cell loss range 10 between .3 and 1 percent per year, depending on where you 11 look. I've quoted the various studies. The Bourne article 12 in 1997 with the .6 percent rate is frequently quoted, but 13 there are slight differences in the literature in that 14 regard. It's in somewhere in that range. 15 As far as surgical procedures, we all know that 16 operative procedures can create both a direct surgical 17 instantaneous hit to the endothelium as well as possibly 18 change the annualized cell loss rate. Cataract surgery as 19 far as Bourne's article in 1994 reported a mean 2.5 percent 20 cell loss per year over a 10-year period. It's important 21 to realize that these were intracapsular and extracapsular 22 surgeries with iris-sutured lenses, transiridectomy clip 23 lenses, and a few posterior chamber lenses. There are some 24 other articles that indicate that cataract surgery causes a 25 1.1 percent cell loss rate per year, and a Werblin article 123 1 in '93 said that one year is an 8.8 percent loss, and 2 you'll note that in the FDA table previously shown by Ms. 3 Lochner and in some of the assumptions I'll later make, we 4 just round that figure to a 10 percent surgical loss at the 5 time of the procedure, which obviously is different from 6 the future cell loss rate that may be increased over 7 normal. 8 As a point of interest, penetrating 9 keratoplasty has a 7.8 cell loss per year. 10 I went ahead and quoted the age-stratified 11 normal endothelial cell density values, both from the Yee 12 article in 1985 as well as a pathologic analysis. The 13 difference between the studies, one is specular microscopy, 14 the other is pathologic. I found they were similar in 15 terms of mean values. The lower age bracket, age 20 to 29, 16 they start at 2,900 cells and by the time you get up to age 17 80-89, there are 2,300 mean cells. The main difference is 18 in the standard deviations. 19 Dr. Edelhauser earlier indicated that the non- 20 contact robo data agrees with the Yee data, and I quite 21 frankly find the standard deviation values in the path 22 study to be huge, plus or minus 500, plus or minus 690. In 23 talking with Dr. Edelhauser on the break, he's going to 24 look further into the Moller-Pedersen article regarding the 25 standard deviations, but I tend to gravitate toward the Yee 124 1 article with the standard deviations. 2 The peripheral cornea, if measured, is known to 3 have increased cell density. Per some data that Dr. 4 Edelhauser provided, there's a 5.8 percent increase in cell 5 density in the paracentral region and a 9.8 percent 6 increase in the peripheral cornea. 7 The next category that I did, knowing the peer- 8 reviewed literature and now knowing normative data, is I 9 tried to determine what would be some thresholds for 10 unacceptable rates of endothelial cell loss. There are two 11 different questions, I believe. 12 One, as a panel member, when an application 13 comes to panel, we all be concerned about what is an 14 acceptable cell rate when we get that application with a 15 particular observed cell loss rate. The reason it's 16 important to get some judgement about acceptable cell loss 17 rates now is it will help us give guidance regarding what 18 or how low should thresholds be that we're actually trying 19 to screen for to make the determination how big the sample 20 sizes must be. So we have to have a sense for what are the 21 maximum and minimum ranges for thresholds that we're even 22 looking at, so we can give some type of guidance regarding 23 sample sizes. 24 I certainly don't expect anyone right now to 25 define an acceptable cell loss rate. That will certainly 125 1 be a hotly debated topic once an application's received, 2 but I think it's instructive that we go through the 3 exercise to see what our limits are, maximum and minimum 4 unacceptable rates of cell loss. 5 I've approached this argument by first looking 6 at life expectancy data. The RP-2000 mortality table is 7 based on a study of the mortality experience of pension 8 plans conducted by the Society of Actuaries and was in 9 response to pension legislation that directed the Secretary 10 of Treasury to promulgate the use of updated mortality 11 tables for various pension calculation purposes. 12 According to that table, the life expectancy 13 for a 21-year-old male is 58 future years or an age of 14 death of 79. The life expectancy for a 21-year-old female 15 is 62 future years, so an age of death of 83. Those are 16 United States data. 17 Realize that depending on your entry date, 18 you'll have change, obviously, to your age of death. If 19 you enter at age of 80, you don't have an age of death of 20 79. 21 (Laughter.) 22 DR. GRIMMETT: But I used it as a fixed value 23 for this particular analysis, so as to not get confused 24 with multiple iterations of the tables. Suffice it to say 25 that when you enter at 20, 30, or 40, it may only differ by 126 1 a few years in terms of your age of death. 2 The minimal acceptable corneal endothelial cell 3 density value is critical, and Ms. Lochner presented a 4 table regarding a 1,200 threshold and a 1,000 threshold 5 near death. I picked values on either side of the range to 6 show the range of acceptable cell loss for sake of 7 argument. 8 In one of Dr. McCarey's earlier versions of a 9 slide, he quoted a minimal acceptable rate of 1,500 cells. 10 Whether we call that number 1,500 or 1,400 or 1,300 or 11 whatever it's picked as is not really the crucial value, 12 but I think a larger number has the argument that if these 13 patients get into any kind of trouble with their phakic 14 IOL, number one, it allows you to do a surgical procedure 15 to possibly correct that, whether it be explantation of the 16 IOL or manipulation of the IOL. 17 Secondarily, we all know as clinicians that as 18 patients enter their early 70s, they have a much higher 19 likelihood of having us see them with cataracts just from 20 age-related phenomena. Whether or not it's increased with 21 phakic IOLs remains to be determined, but this larger 22 target value near or at the age of death will certainly 23 allow a future intraocular surgical procedure, such as 24 cataract surgery. If we run all of our calculations and 25 run them right down to the wire and leave them the bare 127 1 minimum and leave half of the years to the time of death, 2 you are not giving that patient the opportunity to have any 3 intraocular intervention of any kind. So that's why I 4 picked the number of 1,500, and Dr. McCarey in an earlier 5 version happened to put that number there, so I went with 6 it. 7 The next number I used is a potential corneal 8 edema of 800 cells per square millimeter. Acceptance of 9 this particular target will not allow a future intraocular 10 procedure, in my opinion, and certainly if I had a patient 11 of advanced age with immense nuclear sclerotic cataract 12 requiring higher phaco times, I would not be comfortable 13 performing phacoemulsification with entry cell count of 14 800. I would certainly advise the patient clinically that 15 they would have a higher chance of having postoperative 16 corneal edema. 17 In the literature, there's been a quote of 500 18 cells for imminent corneal decompensation, but I think it 19 depends on the actual function of the remaining cells. So 20 the exact, precise figure is not locked down, but in my 21 opinion, clinically it's somewhere in the area of 800. 22 The assumptions I made for my threshold 23 analysis is that the endothelial cell loss, as calculated, 24 was an instantaneous, exponential endothelial cell loss 25 rate, that they lose 10 percent at the time of the surgical 128 1 procedure, and then they have a continuous stable 2 annualized exponential cell lose rate after that time. I 3 used a round down analysis of any remaining fraction. I 4 used .99 rounded down because partial cells do not survive. 5 I did not alter the life expectancy target 6 values, as I had previously mentioned. 7 Table 1. I'm not going to go through this, 8 obviously, in detail, but all I did is create an Excel 9 spreadsheet with a formula. I set the target at the end at 10 1,500, I set the percent drop per year, and I was back- 11 calculating the cells you would need to enter the study. 12 That's all it is. So I was trying to find out what cell 13 count would you need at age 21 in order to end up with a 14 cell count of 1,500 at the time of death for male or 15 female, and then supplying the percent drop per year. 16 For that particular assumption, 1,500 cells at 17 the time of death, 2 percent cell loss per year, I found, 18 for example, age 21, you need 5,900 cells if you're female 19 and you need 5,400 male. Not possible. We looked at the 20 normative data and those numbers exceed the normative data. 21 So the conclusion from that scenario is that if 22 a 1,500 cell target is desired at death, a 2 percent annual 23 cell loss rate is not acceptable. It's not possible to 24 enter with a high enough cell count for all but the very 25 older age ranges. So everyone will fall below 1,500 129 1 because you can't enter with a high enough cell count. 2 I did that same type of analysis trying to 3 bracket what would be inclusive of all ages in order to 4 allow everyone to enter the study. A 1.5 percent cell lose 5 rate, which is Table 2, they're still entering with some 6 very high cell counts. At age 21, for example, 4,300 for a 7 woman and 4,100 cells for a male. 8 The conclusion from that scenario with 1,500 9 target and 1.5 percent cell loss is that no one could enter 10 with a high enough cell density, except those older than 11 approximately 50. So that cell loss rate is unacceptable, 12 the 1.5, if you're targeting for 1,500, that target, of 13 course, having the advantage of allowing someone in the 14 future an intraocular surgical procedure. 15 It turns out a .9 percent cell loss rate per 16 year allows all ages to enter with a reasonable cell count 17 that could be achieved based on the normative data. The .9 18 percent cell loss per year on Table 3 shows all the entry 19 cell requirements, and they reasonably match or were below 20 the normative data for all ages, telling us that the cell 21 density values will approximate or exceed 1,500 at the time 22 of death for all patients entering the study. 23 So the .9, assuming you'd want 1,500 to be your 24 target, is the maximum allowable rate that's inclusive of 25 all ages, and that rate is approximately 50 percent higher 130 1 than the normal of .6 percent cell loss rate per year. 2 We'll later hear from our statisticians regarding the 3 feasibility of trying to read for that low level of cell 4 loss rate in the face of the variability and precision 5 issues that Dr. Edelhauser brought up, because it's my 6 opinion, based on review of some initial data, that the 7 sample sizes would have to be unreasonably large. 8 Just as an example from the literature, an 9 angle-supported phakic IOL from Alio, 1999, has a cell loss 10 rate of .72 percent from years 2 to 7 after experiencing a 11 6.83 percent loss from preop to year 2. So based at least 12 on something in the literature that the numbers are not 13 huge, it's doable. 14 Looking at it a different, looking at the 15 target of 800, so running right up to the edge of corneal 16 edema, I can give you all the tables, but the number that I 17 am giving you is 1.9 percent cell loss rate per year. That 18 number of 1.9 percent cell loss rate per year would allow 19 all patients to enter based on the normative data. 20 So if you desire an 800-cell target value, a 21 1.9 percent rate of annual endothelial cell loss is the 22 maximum allowable rate of loss that's inclusive of all 23 ages, and that's about three-fold higher than the normal .6 24 percent rate per year, and at least based on the Menezo 25 1998 Artisan lens data, I have an average cell loss rate of 131 1 2.28 percent from years 1 to 4. 2 Based on all these analyses and review of the 3 literature, I went ahead and just prepared a bunch of 4 different issues that I would want to see or issues that I 5 would want considered in phakic IOL studies regarding the 6 endothelium, and then we can back to the specific 7 endothelial questions. Most of the question issues I 8 believe will be covered. 9 General issues that I would have for all phakic 10 IOL studies is that certainly endothelial cell density 11 measurements are mandatory. We saw in some of the 12 published literature it did not report endothelial cell 13 densities whatsoever, which I think is unacceptable for a 14 new product of this design because it's a critical issue 15 for the survival of corneal health. 16 Certainly, a central count is mandatory. A 17 peripheral count will be important, especially if it's an 18 anterior chamber lens. You'd want a cell density in the 19 region near the IOL edge or in the area of minimum distance 20 between the IOL and endothelium. That was learned from 21 early lens design. So especially with an angle-supported 22 lens, I would be highly interested in reviewing the 23 peripheral cell count. 24 It is my belief that morphometric analyses are 25 mandatory. We know that analysis of cell shape and size 132 1 provides a more sensitive indication of endothelial cell 2 damage than cell density alone. I understand the 3 limitations of the issues, as Dr. McCarey and Dr. 4 Edelhauser have outlined, regarding the coefficient of 5 variation and the reliability, especially with the 6 algorithms of the non-contact robo. Notwithstanding those 7 limitations, I still feel it's an important variable to 8 consider and may give us some additional information. 9 Just as a matter of course, corneal pachymetry, 10 corneal functional analysis, most certainly would be 11 suggested. 12 The duration of the study that I would be 13 interested in prior to that study coming to panel would be 14 three years. Based on the literature, it seems like there 15 is an initial larger decrease and subsequent next decrease, 16 and then stabilization after a year or perhaps two to 17 three. Depending on what is seen in panel, it would be my 18 guess that there would be discussion of postmarket 19 surveillance of endothelial cell data for possibly a year 20 or two more. 21 There are some data that it may take four years 22 to see the morphometric data return to baseline levels and 23 ensure stability, but I do not believe, based on what I've 24 seen in the peer-reviewed literature and based on my 25 concern about the corneal endothelium, especially for 133 1 anterior chamber lenses, that I would be comfortable 2 approving a lens with only two years of data. So for me, 3 three years would be the minimum I would favor. 4 I would favor a longer endothelial study 5 duration, perhaps postmarket-mandated, for higher risk 6 factors, such as angle-supported lenses, which have a 7 higher risk to culture the endothelium than iris-fixated 8 lenses, which are closer than posterior chamber lenses. 9 The same issue. Thicker lenses are closer to 10 the endothelium than thinner lenses. So we would want 11 longer follow-up for those. 12 A shallower anterior chamber depth, such as 13 hyperopic patients, would have a higher risk than deeper 14 anterior chamber depth, perhaps if the IOL is closer to the 15 endothelium, and certainly chronic anterior chamber 16 inflammation, as previously mentioned, would be a higher 17 risk factor for endothelial loss than a quiet anterior 18 chamber. 19 Interestingly, in one early, approximately 50- 20 page paper by Drews in 1991, he went over study parameters 21 versus the FDA grid regarding what he would expect for 22 phakic IOLs, and he recommended a five-year study duration. 23 I certainly don't disagree with that, given the importance 24 of corneal endothelium. 25 With respect to anterior chamber phakic IOLs, 134 1 we know that the optic-endothelium distance plays an 2 important role in potential endothelial damage. Therefore, 3 high-resolution ultrasound, as mentioned by Dr. Werner, I 4 believe would be mandatory to disclose the optic- 5 endothelial distance and distances between other eye 6 structures and the IOLs, such as the crystalline lens. 7 The peripheral endothelial cell density and 8 morphometric measurements would be mandatory in my opinion 9 in the region of the IOL optic edge, in addition to the 10 central endothelial analysis, because if the examination is 11 limited to the central cornea, it may fail to detect 12 significant endothelial injuries, and then specular images 13 can show significant morphologic changes over the edge of 14 the IOL in the absence of central cell density changes. 15 I would caution that a preop history of eye 16 rubbing may be a contraindication for entry into the study, 17 especially when we're talking about such low tolerances 18 between the distance between the IOL edge and the corneal 19 endothelium. 20 Depending on the IOL design, such as an angle- 21 supported lens, a manufacturer must specify a minimum 22 anterior chamber depth that contraindicates IOL insertion. 23 Because chronic inflammation is a known factor 24 in endothelial damage and because some studies have 25 disclosed chronic anterior segment inflammation after 135 1 phakic IOLs using a laser flare cell meter, I would 2 recommend laser flare fluorophotometry to evaluate chronic 3 anterior chamber inflammation, and just raise the question 4 would iris fluorescein angiography be of any help if a lens 5 were either rubbing the iris or directly affixed to the 6 iris? So that would be an issue that perhaps in a small 7 subset of patients in early studies may be relevant. 8 Posterior chamber phakic IOLs certainly would 9 have advantages for corneal endothelium. They would have a 10 maximum distance between the IOL and the endothelium. They 11 would avoid optic-endothelial contact, but of course, their 12 location behind the iris would have a higher risk for 13 pigment dispersion and introduction of cataract. If these 14 lens induce chronic anterior segment inflammation, they 15 certainly may have secondary effects on the endothelium, so 16 I do believe that endothelial studies are also important 17 for posterior chamber lenses, irrespective of the fact that 18 they're the furthest lens away from the endothelium. 19 As far as that table -- now getting back to the 20 two questions, I added these last two portions right at the 21 end. Getting back to the table, what would be the 22 recommended minimum endothelial cell density to enter these 23 studies? And we saw that table put forth by the FDA. 24 I see two approaches to attack the problem. 25 One is standard deviations. You could look at accepted 136 1 normative data and say you don't want anyone entering the 2 study outside of, let's say, lower than, let's say, two 3 standard deviations. 4 Let's talk about two standard deviations. 5 Assuming the standard deviations are reasonable and your 6 studies are reproducible, you would want to exclude anyone 7 out in that 2.5 percent tail, and I went ahead and just 8 listed the values as they would range from age 20 at about 9 2,700 going down to 2,400 or so by age 60. That's sort of 10 one way of looking at it. 11 The other way to look at it would be the way 12 that the FDA has approached it. Look at do you have enough 13 cells to make it near the age of death? And that's the 14 other way to look at it. 15 In reviewing the FDA Attachment B, the 16 threshold values selected don't exactly -- they do not 17 likely allow a patient to undergo a secondary ocular 18 procedure such as cataract surgery or further phakic IOL 19 manipulation during the life of the patient. I would have 20 concerns as a clinician if a patient had a cell count of 21 1,000 with a moderately dense cataract doing a phaco 22 procedure. I have to be worried that I might tip them over 23 into clinical corneal edema. 24 So I think we have to understand that the chart 25 values selected, and I selected 1,500 and the FDA selected 137 1 1,200 and 1,000, are variable depending upon what our 2 expectations are for these patients to have elective 3 procedures down the road, and it's impossible to precisely 4 determine the minimum entry requirements without knowing 5 the exact rate of endothelial cell loss per year at the 6 various sites that run these. So when we discuss that 7 particular table, I'm certain all those issues will come 8 into play. 9 The issue that I haven't addressed, and I'll 10 leave it to the statisticians, is regarding sample size 11 analysis. Drs. Edelhauser and Bernie McCarey talked about 12 that in the best case scenario, the best precision they can 13 get on endothelial cell measurements is 2 percent and real- 14 world data is at 9 percent according to Dr. McCarey. I 15 wold like to know how that translates into the standard 16 deviation values that our statisticians have calculated 17 regarding sample sizes, if that's known, to see what kind 18 of numbers we would actually need if we wanted to screen 19 for the lower rates of annualized loss. 20 That would conclude my introductory comments at 21 this time. 22 DR. WEISS: Thank you very much for a very 23 thorough, as usual, presentation. 24 DR. GRIMMETT: Thank you. 25 DR. WEISS: Dr. Matoba has a question. Then 138 1 we'll go around. 2 DR. MATOBA: Mike, I might have missed this, 3 but when you did your calculations, did you assume any 4 amount of cell loss from the actual surgery? 5 DR. GRIMMETT: Yes. I assumed a 10 percent 6 instantaneous loss at the time of the surgery based on the 7 (inaudible) of having 8.8 percent with a phaco procedure. 8 DR. WEISS: Mr. McCarley, did you have a 9 question? 10 MR. McCARLEY: I just had a couple of 11 questions. The 9 percent that you're referring to is what 12 Dr. Edelhauser and Dr. McCarey were talking about? Weren't 13 those the controlled laboratory percentages and not actual 14 real data? 15 DR. GRIMMETT: I believe it was real data from 16 a real study, but we could probably have Dr. McCarey answer 17 it, but Dr. Edelhauser's number of 2 percent precision was 18 sort of in the best of hands what is the best precision. 19 MR. McCARLEY: Right, exactly. 20 DR. GRIMMETT: And Dr. McCarey's number was in 21 order to -- it might be better to have Dr. McCarey answer, 22 but I believe it was real data. I'm not sure how many eyes 23 were in the study. 24 DR. WEISS: You can over there. As I recall, 25 it was different centers, with the variation 9 percent 139 1 between different people reading the same thing. 2 DR. McCAREY: Yes, it was real data from a 3 clinical trial that Medennium's working. It's the control 4 data. There were 58 patients, seven clinical sites. I 5 just started to collect the data and asked the question and 6 came up with that answer. 7 MR. McCARLEY: Okay. I wonder is the FDA able 8 to provide the panel, especially the voting members and 9 reviewers, with data that they currently have on phakic 10 intraocular lenses. In order words, the endothelial cell 11 data that has come already out of the studies? There must 12 be over 2,000 implants so far over -- 13 MR. WHIPPLE: You mean the ones that are under 14 IDE? 15 MR. McCARLEY: Pardon? 16 MR. WHIPPLE: Dave Whipple. You mean the ones 17 that are already under IDE? 18 MR. McCARLEY: Yes, correct. 19 MR. WHIPPLE: We can summarize it. 20 MR. McCARLEY: In other words, we're reviewing 21 the literature and making decisions based on the 22 literature, and a lot of these have been recognized as 23 being small studies. We don't know what methodology was 24 used to validate -- the validity of the methodology and so 25 forth. A lot of reference goes back to original studies 140 1 that are even from the 1970s, from Bourne and group, and of 2 course, he did a later study, but he did longitudinal study 3 showing I think 10 years' follow-up. 4 But I'd be curious, one, is can the FDA provide 5 the panel members with the data that they have now to show 6 them what the standard deviation is right now? 7 MR. WHIPPLE: Yes. They're special government 8 employees. As long as they kept it amongst themselves, 9 they could use it, yes. 10 MR. McCARLEY: Right. I would just say before 11 a recommendation is made on limits and so forth, I would 12 say you have real-time data, real data on real phakic 13 intraocular lenses, that is up to date that you may want to 14 consider before you make a final recommendation. 15 DR. WEISS: Well, actually, the advantage of 16 this sort of meeting is that everyone's entitled to their 17 opinion, there is no final vote, and there has to be 18 consensus. So basically, the FDA will use all the 19 information gathered here today, and including any other 20 information we think would be helpful to obtain in the 21 future, to come up with some final recommendations on their 22 own. 23 Dr. Grimmett? 24 DR. GRIMMETT: Yes. Michael Grimmett. I 25 alluded to it in my introductory comments that while it 141 1 would be wonderful to have the Phase III endothelial cell 2 loss data versus what is published in the literature, we're 3 not trying to set a threshold rate of what is acceptable to 4 this panel or what is not acceptable. That's not even the 5 purpose of this. But I approached the analysis in that 6 fashion to show what the edges of the approach would be, so 7 that we can get some data on how many patients would we 8 need to screen in that range. That was my goal. 9 Understandably, the literature is not giving us 10 that much guidance regarding actual cell loss rates because 11 most of the lens designs have been just started and we're 12 now on the newer lens designs. So we actually don't know 13 what the newer loss rates are. 14 DR. WEISS: Dr. Mathers? 15 DR. MATHERS: Yes, I want to compliment Dr. 16 Grimmett on his excellent analysis, and it may sound like 17 it is rather strict, but I don't think that using the 18 actuarial tables as you have is in any way offbase. In 19 fact, it's probably ultraconservative because, as we have 20 seen, as medical advances continue, it isn't impossible to 21 look at ages beyond what you are saying, and we all know a 22 lot of 80-year-olds who do not feel like dying right now. 23 (Laughter.) 24 DR. MATHERS: And who are probably going to 25 live longer as these issues become more relevant. So 10 142 1 years from now, with one advance in cardiac pathology, this 2 would become not strict enough. 3 I would also like to ask who you think should 4 do the reading and the entrance qualification to get into a 5 study like this, because clearly the endothelium, which is 6 actually quite accessible as the cell to study, can be best 7 read perhaps by a central office, and should the entrance 8 requirements be that a central group does the initial 9 reading to get in? What would you think of that? 10 DR. GRIMMETT: Michael Grimmett. Based on the 11 comments of Dr. Edelhauser, given all the variability that 12 exists if the technician is not trained properly and all 13 the parameters to be analyzed, I obviously am in favor of 14 the highest trained, highest precision measurer because of 15 the critical nature of endothelial cell loss over time as 16 our population ages. 17 DR. MATHERS: It wouldn't be difficult to do 18 this with digital capture. You can transfer these images 19 instantaneously and you can then decide whether you have 20 good images, because it would seem that a lot of this 21 depends upon having good images to start, so that you know 22 how to get good data afterwards, and that's certainly 23 possible on an instantaneous basis. 24 DR. WEISS: The way the document reads at the 25 present time is that "The use of a reading center is 143 1 strongly recommended. If the use of a reading center is 2 not possible, the sponsor should establish a protocol for 3 collection and analysis of images to be used by each 4 participating site." 5 Would you then change it from strongly 6 recommended to required or you'd leave it as strongly 7 recommended? 8 DR. GRIMMETT: This is Michael Grimmett again. 9 I think there are always multiple ways to skin a cat, and 10 if a particular study or a sponsor can demonstrate the 11 reliability that they have internal mechanisms to validate 12 precision and validity and it appears to be equivalent to a 13 standardized reading center, I think there is always 14 flexibility in that regard. 15 DR. WEISS: Dr. Huang? 16 DR. HUANG: My first comment is that regarding 17 Mr. McCarley's comment earlier, using the FDA existing data 18 is almost like using the soccer player to be the referee. 19 You know, that we are judging the safety of the data and 20 then using the data to be the reference for its own safety. 21 I think it's questionable. 22 The second comment I would like to make is also 23 that Dr. Grimmett used a very nice life table actuary to 24 analyze this, but I think the endpoint is a little bit 25 strict because we don't do cataract surgery at the time of 144 1 the death. We do that cataract surgery maybe hopefully 2 five or 10 years before the patient's life expectancy in 3 order to improve their quality of life. 4 So maybe we have two points that we can use. 5 You know, maybe one at one point is 10 years before death 6 and then at the death point, and then to find a reasonable 7 middle ground for the starting point. 8 DR. GRIMMETT: This is Michael Grimmett again. 9 Just a quick response. I first attempted to do the 10 cataract surgical procedure 10 years before the age of 11 death with an increased rate of annual cell loss. The 12 table became so complicated in the formula that I would 13 have to do to change the rate of cell loss midstream and 14 back-calculate to the entry cell data that I couldn't get 15 the spreadsheet to work in that regard. 16 I took took the give-up approach. You know, I 17 was on vacation. 18 (Laughter.) 19 DR. WEISS: Well, maybe in that case, the 20 guidance could be to the FDA to recalculate this with the 21 average age of cataract surgery as the final. 22 Yes, Dr. Huang? 23 DR. HUANG: I'm just joking. I say, you know, 24 maybe he can take another vacation to calculate that. 25 (Laughter.) 145 1 DR. WEISS: Dr. Burns? 2 DR. BURNS: Yes, but I think all things being 3 equal, if you move the cataract surgery earlier, you're 4 going to actually end up with stricter numbers because 5 you'll have the loss from the surgery and then you'll have 6 an increased growth and you're going to come lower on the 7 curve. So I think that will only make things stricter. 8 DR. HUANG: Andrew Huang again. I think this 9 is just a recommendation. Not all the patients eventually 10 are going to need cataract surgery, but I certainly 11 understand that there will be additional loss, and I don't 12 know which one is greater, 2 percent annual loss versus the 13 10 percent initial loss. So statistics will tell us. 14 DR. WEISS: I think it would be up to the FDA 15 to crunch the numbers both ways to see what the differences 16 are and if they're clinically relevant or not. 17 Dr. Grimmett? 18 DR. GRIMMETT: I would just point out on the 19 tables, you realize that it is a 10 percent initial loss 20 for the phakic IOL surgery. Please realize there are some 21 data on some of the more current studies that say that that 22 initial loss may at that rate, 5 to 6 percent. The earlier 23 studies did have a higher rate. 24 So I used a cataract surgery 25 phacoemulsification initial rate that may not exactly be 146 1 true for phakic IOLs. I used the most conservative 2 approach to make sure that we leave people with enough 3 cells at the end. 4 DR. WEISS: While we're on the endothelial cell 5 topic, I just want to pose one other question relating to 6 this, and then perhaps Ms. Lochner could come up and then 7 we can go on with our other presenters and other 8 discussions. 9 The other question that I would pose to the 10 panel, as I already did to our experts, is what would you 11 think would be the reasonable number of months to tell a 12 patient they would have to be out of contact lenses? If 13 you have an opinion. 14 Dr. Bullimore? 15 DR. BULLIMORE: None. I mean, I think keeping 16 it -- 17 DR. WEISS: Not even a week? 18 DR. BULLIMORE: Well, let me say less than a 19 month. 20 DR. WEISS: Less than a month. 21 DR. BULLIMORE: I mean, I think trying to sort 22 of take them out of contact lenses with the expectation 23 that the endothelium's going to change in any meaningful 24 fashion, based on my experience and what I heard from the 25 experts today, is futile. 147 1 DR. WEISS: So you would keep them out long 2 enough to get a proper keratometry or corneal topography? 3 DR. BULLIMORE: I mean, I'd use the same 4 guidelines that exist for corneal refractive surgery. 5 DR. WEISS: Gee, I think we just eliminated two 6 questions with that comment. 7 Dr. Mathers? 8 DR. MATHERS: Yes, I strongly agree with that. 9 Aside from the fact that it would be really difficult to 10 have patients go through that period of time, since the 11 polymegethism afterwards doesn't really evolve very quickly 12 and we don't know exactly the significance of it, I think 13 that it's not terribly relevant to have them out of their 14 lenses a long period of time, except to establish their 15 refractive error issue. 16 DR. WEISS: Dr. Bullimore? 17 DR. BULLIMORE: And I think we need to be 18 generalizable, and these products and these procedures are 19 going to be done on people who, to a large extent, are 20 long-term contact lens wearers. We saw yesterday that 21 something like 80 to 90 percent of the patients enrolled in 22 a study for low myopia or low to moderate myopia were 23 contact lens wearers. So in the high group, it's going to 24 be probably even higher. 25 DR. WEISS: Another question, which I warn you 148 1 in advance I'm going to limit the discussion on because it 2 will come up again, is with this in mind, would you only 3 want one eye done at a time or one eye done and use the 4 other eye as a control for the endothelial cell study? 5 Dr. Bullimore? 6 DR. BULLIMORE: Again, I think you don't want 7 to burden the patients too much, and enrolling them in, 8 say, a three-year study where they can only have one eye 9 done with a device and the other eye has to wait I think is 10 an unreasonable burden to be placed on the patient. We 11 have a lot of historical control data. It seems to me we 12 have some historical data on endothelial cell count as a 13 function of age. We can use that, and what we're looking 14 for, I guess, are the extreme or the worst cases where 15 people really do loss a lot of endothelial cells within a 16 relatively short amount of time, and I don't think we need 17 a control group to necessarily look at those event rates. 18 DR. WEISS: Dr. Grimmett? 19 DR. GRIMMETT: Dr. Grimmett. My concern with a 20 unilateral one-eye study would be for quality of life for 21 the patient. These patients would be typically those that 22 don't qualify for other refractive procedures, and hence 23 they have a higher range of myopia. There is significant 24 (inaudible) of emmetropia. So in a three-year duration 25 study, I think that would be unwieldy and probably not very 149 1 reasonable for the patient. So I'm in agreement with Dr. 2 Bullimore. 3 DR. WEISS: Dr. Mathers? 4 DR. MATHERS: Yes, I'm also highly in 5 agreement. I think we'll need to discuss again the time 6 delay between the first and second operation, but three 7 years is too long. 8 DR. WEISS: Then I would sort of conclude -- 9 Dr. Bandeen-Roche? 10 DR. BANDEEN-ROCHE: Yes, just very briefly. 11 You know, statistically, we'd obviously like to have a 12 contralateral, but I am absolutely swayed by the quality of 13 life considerations as long as we have good quality control 14 data. So that would include both a precise estimate of the 15 rate of loss, but also of the variability in rates. 16 DR. WEISS: Okay. So then I would conclude 17 that the fact that the contact lens issue in terms of the 18 change and shape of the cells and number of the cells might 19 still be evolving after the implant was placed in will be a 20 confounding variable, but not objectionable by the panel. 21 Fine. 22 Ms. Lochner, perhaps we can go on. 23 MS. LOCHNER: For analysis of the crystalline 24 lens for lens opacity, we currently recommend that "The 25 natural lens should be evaluated preoperatively and at each 150 1 of the postoperative intervals. The level of evaluation 2 should be commensurate with the risk of cataractogenesis or 3 lens changes identified by the risk analysis performed by 4 the manufacturer. For phakic IOLs where the design or 5 surgical procedure may lead to lens changes, a grading 6 system or quantitative method should be used to evaluate 7 lens changes over time. For IOLs for which lens changes 8 are not an identified risk, qualitative observation may be 9 adequate." 10 The analyses should include the number of 11 patients with any change in the appearance of the lens 12 stratified by the type of change and the number of patients 13 with clinically significant lens opacities, and the term 14 "clinically significant" is as yet undefined. 15 We are asking for panel comments on whether you 16 believe evaluation of lens changes should be requested of 17 all sponsors or whether this evaluation should only be 18 performed if the sponsor's risk analysis warrants and 19 whether you have any specific recommendations for defining 20 the term "clinically significant" lens opacities. 21 We're also asking for your recommendations 22 about the use of quantitative methods for measurements of 23 lens changes versus the use of grading systems, and finally 24 we'd like your thoughts on the duration of the study and 25 request that you specifically discuss the length of follow- 151 1 up you believe would be adequate for panel review of this 2 cataractogenesis outcome. 3 Now, Dr. Mathers will provide his review. 4 DR. MATHERS: Thank you. 5 Regarding the first question we are asked to 6 address, the evaluation of the lens changes, it is this 7 reviewer's opinion that all phakic IOLs need to be 8 evaluated for any cataract or changes in the lens. 9 Anything that happens essentially inside the anterior 10 chamber in front of the lens is an issue here, and even 11 something that would not touch the lens or is known to be 12 touching the lens would still be a problem. Certainly, 13 central touch is not the only issue nor is just peripheral 14 touch. There's anterior chamber inflammation, and all of 15 this can affect the lens changes over time. So any 16 perturbation would be of interest and all cataract 17 processes need to be assessed. 18 In this direction, I think it's going to be 19 important not just to look at the lens itself, but to look 20 at the source of the possible problem, such as we have 21 heard regarding flare assessment or anterior chamber 22 inflammation, and I think that it would useful to measure 23 flare in these patients with perhaps the laser flare 24 system, and also, regarding the evaluation of the cataract 25 process, that the sizing and the structure of the anterior 152 1 chamber is going to be a key issue. So the use of high- 2 resolution ultrasound I think is going to be extremely 3 important in evaluating this. 4 Regarding the clinical significance of the lens 5 opacities, this is a more subjective area and we don't have 6 very good data on this, but this of course refers to vision 7 changes, and in this regard we need I think to be as 8 precise as possible because when one measures vision under 9 various circumstances, you get very, very different 10 results. 11 So if you have a central opacification of the 12 anterior subcapsular area, it will not show up in vision 13 testing in a dim room. You're going to have to do this 14 with a small pupil under conditions that will induce some 15 glare, and in fact, the more glare, the better. This is 16 not a non-real-life situation. I mean, when people are in 17 a high-light environment, which they often are, this comes 18 into play. 19 So I think glare testing is the most relevant, 20 but the only way that this should be assessed. For this, 21 we need to have careful measurements of the patients as 22 they enter the study and a standardized method of 23 evaluating the glare, and that high-glare settings should 24 be used. 25 Now, everyone will have some decrease in visual 153 1 acuity with a high setting of a glare. So we have to 2 decide whether we think one, two, or more lines of loss 3 compared with the preoperative evaluation would be 4 significant. My recommendation would be two or more lines 5 compared with what they had in the loss before. 6 I think that all lens changes should be 7 reported, not just the anterior subcapsular fibrosis, but 8 we also have anterior cortical changes which may result 9 from these anterior subcapsular processes. 10 The posterior subcapsular cataract is also an 11 issue because the anterior lens cells are not the only ones 12 going to be involved. As you build lenses that do not ride 13 or touch the central lens, they may ride in the periphery 14 of the lens, and they are going to get closer to those 15 cells on the outside which can migrate posteriorally, and 16 certainly may do so. So that becomes an issue, and the 17 possibility of inflammation that occurs with the anterior 18 chamber lens of this design may affect the development of 19 nuclear sclerosis, and this is going to be something that 20 is going to be harder to assess, but I think that we need 21 to at least monitor this. 22 In Part C, we're asked to comment on the use of 23 quantitative measures for the measurement of lens changes 24 versus the more semi-quantitative grading system, and by 25 quantitative here, we mean the assessment or the 154 1 visualization, the optical visualization, of the changes 2 underneath the anterior capsule. 3 There is no standard way to do this and the 4 examination of this is very light-dependent. If the 5 lighting system is a little bit off, you get a little 6 different view of this and it becomes harder to see, and 7 Dr. Werner's presentation on this was excellent, but even 8 she could not really tell us exactly how much better, say, 9 a photographic system is from a slit lamp system, which of 10 course is going to be subjective. But I believe that the 11 backlighting and retroillumination of the lens with high- 12 resolution color photography probably offers us the most 13 objective and reliable way to follow this over time. 14 The development of a scale to do this has 15 already been done. I don't think this needs to be 16 reinvented, but can be perhaps modified slightly. The LSCS 17 system can grade these opacifications and can be used for 18 all of this -- the anterior subcapsular, the posterior 19 subcapsular, and the nuclear -- and I think something like 20 that would be appropriate. 21 I would strongly recommend the use of digital 22 photography to perform this process and I also think it is 23 possible, as technology improves, that we will have a 24 better understanding through other means of visualization, 25 perhaps confocal microscopy of these cells now that this 155 1 becomes an issue and it becomes relevant to look at this. 2 But there are no standards for this now, and that would not 3 necessarily be used in an early study like this. 4 In Part D, we're asked to comment on phakic 5 IOLs and the length of time that it might be useful to 6 evaluate this for, and here, as we've heard with the 7 endothelium, when we're studying the endothelium, we have a 8 fairly clear endpoint. We have endothelial cell loss and 9 we can follow this fairly objectively. 10 With lens changes, it's much less objective, 11 and one study, noted recently, showed that there was some 12 change in light transmittance with these lenses, not 13 necessarily based on cataract, but I think that with the 14 long time span that we're talking about and the possibility 15 of chronic inflammation associated with this that is 16 subclinical, that at least three years would be necessary 17 to evaluate it. I think that the monitoring of this 18 perhaps should go on longer than that, but I think the 19 three years is probably enough to give us an idea of what 20 is happening. 21 The capsular process of cataract formation is 22 tied to a number of different other processes. Not just 23 lens touch, but, as I said, inflammation, and as the lens 24 is redesigned to minimize the cataract process, the other 25 issues, such as iris touch and development of pigment 156 1 dispersion and glaucoma, become more of an issue, and I 2 think that the industry or lens manufacturers will be 3 tempted to avoid the obvious problems of lens touch by then 4 shifting the burden to the back of the iris. I think that 5 this kind of monitoring is also going to be important. I 6 know it's not part of this particular issue, but I think 7 it's relevant because there are tradeoffs here. There is 8 not much space in this area and as you design the lens to 9 perform in one way, you then create other issues of 10 significance. 11 That is my summary. 12 DR. WEISS: Thank you very much. 13 I just have just two questions on things which 14 you've already mentioned, but I just want to clarify. What 15 would you call a clinically significant cataract? 16 DR. MATHERS: Clinically significant cataract 17 refers to a loss of a number of lines, but it's highly 18 dependent upon that is assessed, and I think that needs to 19 be assessed not just with standard -- well, our assessment 20 of vision can be done in the standard way in dimly lit room 21 to optimize vision, but it needs to go beyond that. We 22 need to have glare testing as well because our standard 23 measures of vision will not be adequate to pick up the kind 24 of changes we're going to see with capsular opacification. 25 DR. WEISS: So then to just restate that, with 157 1 glare testing, it depends on sensitive you want to be to a 2 very early cataract, and that's the question I'm asking, is 3 how sensitive do you want to be? So if we said let's it 4 bring it to the glare testing realm and say, okay, we're 5 going to determine that by a loss of X number of lines by 6 glare testing, is there an idea you have? 7 DR. MATHERS: Certainly, it has to be more than 8 one line, so I would say two. 9 DR. WEISS: So would you prefer, if I was going 10 to make you quantify it, would you then say a loss of two 11 lines by glare testing, rather than just a straight loss of 12 two lines without the glare testing or would you want to 13 say something else or it's totally unknown? 14 DR. MATHERS: I think without the glare testing 15 loss of one line would be important, would be significant. 16 With glare testing, it's going to be two, because the 17 glare testing is much more sensitive. 18 DR. WEISS: And would you require contrast 19 sensitivity data or that would be optional? 20 DR. MATHERS: I think contrast sensitivity data 21 also should be included. 22 DR. WEISS: Okay. So you'd like to have both, 23 but at least at this point of the discussion, a loss of two 24 lines at glare testing would be considered significant. 25 Dr. Bradley? 158 1 DR. BRADLEY: Yes, just a comment on the means 2 by which one does glare testing and how that interacts with 3 the underlying spatial distribution of the cataract. We 4 might think of two scenarios, one that was just mentioned 5 of a central cataract that is anterior or posterior. The 6 idea of most glare testing is you employ a bright light 7 source and, in so doing, the pupil constricts, and 8 therefore the cataract fills a larger proportion of the 9 pupil, and therefore the scattered light becomes a larger 10 proportion of the retinal image, and that leads to an 11 increased visual effect. 12 Obviously, the converse is true. If you have a 13 peripheral or marginal cataract, as the pupil is 14 constricted, a smaller and smaller proportion of the pupil 15 is covered by the cataract, and therefore a smaller 16 proportion of the retinal image is scattered light, and 17 thus the visual effects are decreased under those 18 circumstances. 19 So I'm not sure there is a single way one can 20 do a glare test that would sensitize the tester to the 21 visual impact of a cataract, and it may be necessary to 22 employ more than one approach. Just an off-the-cuff 23 suggestion would be to employ the one suggested, which is 24 the standard approach, perhaps, where the pupil constricts 25 in the presence of the bright light and it's highly 159 1 sensitive for picking up the visual impact of a central 2 cataract. I might also suggest performing the same test 3 under cycloplegic pupil dilation to emphasize the impact of 4 a marginal or peripheral cataract that, of course, would be 5 visual manifest for the patient under, for example, night 6 driving circumstances, which arguably are the most 7 important ones. 8 DR. WEISS: The question, Arthur, has anyone 9 done cycloplegic glare testing? Do we have any data to 10 know what the results are in the normals? 11 DR. BRADLEY: I don't know. 12 DR. WEISS: Dr. Mathers? 13 DR. MATHERS: I would strongly agree with what 14 you have suggested. I meant that we should do standard 15 testing, contrast sensitivity testing, and glare testing, 16 and I do not have any experience doing glare testing in a 17 dilated pupil, but it is similar to night driving, but we 18 just don't have any data on that, and I don't think that 19 the visual assessment is going to give us all of the 20 answer. I think we're going to see a lot more with the 21 objective and quantitative than we do simply with the 22 vision changes. 23 DR. WEISS: Well, perhaps if we don't have that 24 data, we could request a subset. If panel thought that was 25 helpful and so did the agency, we could request a subset of 160 1 patients when they have their initial entry dilated exam to 2 be glare tested while dilated and not dilated, so not to be 3 too onerous to any sponsors. 4 Dr. Grimmett, then Dr. Bullimore. 5 DR. GRIMMETT: Dr. Grimmett. Just a quick 6 comment. If my memory serves me correctly, I think I've 7 seen some dilated glare testing data in an ARVO abstract by 8 Arthur Ginsberg out of California, San Ramon. He does 9 functional driving tests and other activities and has some 10 data on that kind of stuff. He's a contrast sensitivity 11 guru. I think I've seen that data before, so I think it 12 could exist. 13 DR. WEISS: Dr. Bullimore? 14 DR. BULLIMORE: My impression is we're going to 15 come to contrast sensitivity in a minute and glare testing, 16 but since it's on the table, this is a real sticky area, 17 and anybody who a few years ago was involved in the Eye 18 Care Technology Forum knows that agreement was quickly 19 reached on some areas, like measurement of intraocular 20 pressure and visual acuity and visual fields, but contrast 21 sensitivity and glare testing became a thorn in the side of 22 the organizers of that meeting, and Morris Waxler was the 23 point person on that and Arthur was involved in the panel 24 as well. It was very difficult. 25 As far as assessment of cataract, I think one 161 1 of our speakers put forward a number of mechanisms by which 2 a standardized grading system could be used. I think the 3 panel should consider whether the FDA should strongly 4 recommend a reading center for cataracts or for lens 5 opacities in the same way they're recommending it for 6 endothelial cell density. 7 It becomes difficult because in the case of 8 endothelial cell density, you have a standardized 9 instrument that can capture the image. Photographing lens 10 opacity, since you want to capture the different features 11 of the lens, is a much more difficult and sophisticated 12 procedure. 13 So I think using standardized grading systems 14 is appropriate, paying attention to the kind of opacities 15 that are likely to occur. Anterior and posterior capsular 16 opacities I think are appropriate, and cortical opacities 17 maybe, but I'm not in favor of requiring a reading center 18 in the same way that it's currently recommended for the 19 endothelium. 20 DR. WEISS: If there are no other comments on 21 this section, Ms. Lochner, if we could proceed to the third 22 and last question. 23 MS. LOCHNER: At the most recent ANSI meetings, 24 a consensus appeared to be reached on the general 25 parameters of the contrast sensitivity substudy that's 162 1 outlined in Section 8.3, and Dr. Bullimore, you might be 2 interested to know that we had Dr. Ginsberg, who's the 3 contrast sensitivity guru, and we had Dr. Jack Holiday, 4 who's advocated contrast acuity testing, actually agreeing 5 on this point. So it was a red letter day. 6 (Laughter.) 7 DR. BULLIMORE: For the record, I never called 8 him a contrast sensitivity guru. That was Dr. Grimmett. I 9 want to strictly go on the record on that. 10 (Laughter.) 11 MS. LOCHNER: And I think that the use of the 12 contrast sensitivity systems, rather than contrast acuity, 13 was recommended because of contrast sensitivity's ability 14 to capture the full range of spatial frequencies and 15 contrasts, and it was felt that the contrast acuity charts 16 would potentially miss significant contrast losses because 17 of the unpredictability of the spatial frequency at which 18 these losses may be seen. Of course, we will have letter 19 recognition performance under low light conditions assessed 20 by the best-corrected visual acuity testing. 21 The contrast sensitivity testing, as proposed, 22 includes mesopic and mesopic with glare conditions. Please 23 comment on the clinically significant decrease being set at 24 .3 log units, and also on whether this decrease should be 25 at one or two or more spatial frequencies to be considered 163 1 significant. 2 Next, should charts with the minimum contrast 3 at each spatial frequency be used to minimize the problem 4 of missing data, and perhaps first of all, are these charts 5 commercially available? 6 Please also comment on the recommended analyses 7 of these data, including how missing data should be 8 handled, and by missing data, we mean when the patient is 9 unable to see the target at a particular spatial frequency 10 at any of the available contrast levels. 11 Last, please provide any additional comments, 12 particularly any recommendations you may have to improve 13 the quality of the data generated from this testing. 14 Dr. Bullimore? 15 DR. BULLIMORE: Can you go back to the first 16 slide? I want to take these questions one at a time with 17 panel input, if that's okay, Madam Chairman. 18 DR. WEISS: Anything you want. 19 DR. BULLIMORE: Wow. I guess lunch is not on 20 the table. 21 (Laughter.) 22 DR. WEISS: That's true. 23 DR. BULLIMORE: Let me paraphrase my comments 24 by saying that I have a long and distinguished record of 25 being a fan of letter charts over grating, so anything I 164 1 say should be taken in that context. 2 That notwithstanding, I think first of all, the 3 statement about .3 log units being a clinically significant 4 decrease in contrast sensitivity is reasonable. Just to 5 put that in context, we've come to accept two lines of 6 visual acuity on a logMAR chart as being a meaningful 7 decrease and representing a complication or an adverse 8 event or, to put it more broadly, an unsatisfactory outcome 9 of a refractive procedure. Here, we're talking in the 10 contrast domain of an equivalent of three lines, and I 11 think this is reasonable and conservative. 12 I think saying that the drop should be at two 13 or more spatial frequencies, again, we get into the martial 14 end of contrast sensitivity testing pretty quickly. One of 15 the limitations and reservations that some people have 16 about these tests is that unlike letter testing, for 17 example, where the patient has to name a letter, the 18 patients is asked either whether they can actually see the 19 grating or not or is asked to say is the grating on the top 20 part of the chart or on the bottom part of the chart? So 21 the opportunity for a bias based on shifts in criteria if 22 you use the first approach or the opportunity to sort of 23 guess correctly when it's just a one in two chance compound 24 the analysis of some of these data. 25 But again, in the interests of the goodwill 165 1 exhibited between Dr. Holiday and Dr. Ginsberg, I think 2 this again is a reasonable, practical approach, and with a 3 letter chart, of course, you might not have the problem 4 with then having to say, well, is it one or two spatial 5 frequencies, but really I think the panel at this stage 6 should be presented with the data when it's available and 7 let the panel decide, so to speak. 8 Does anyone want to comment on that first 9 thing? I'm sure Arthur would. I'd appreciate Dr. Owsley's 10 input anytime she wants to say something. 11 DR. WEISS: Dr. Bradley? Or Owsley. Whoever. 12 DR. OWSLEY: Why don't keep going and then I 13 can probably just make a few comments at the end? 14 DR. BULLIMORE: Okay. Arthur, can I keep going 15 for you as well or do you want to interject? 16 DR. BRADLEY: I really appreciate the 17 opportunity to interject. 18 (Laughter.) 19 DR. WEISS: So why don't you? 20 DR. BRADLEY: Mark raises an interesting sort 21 of comparative analysis to try and decide, well, is .3 log 22 units, obviously a factor of two changing contrast 23 sensitivity, clinically significant? And he draws the 24 parallel between what people decide is clinically 25 significant in terms of logMAR for visual acuity change, 166 1 and two lines, obviously that's .2 log units on a logMAR 2 chart, and I question, Mark, that that is somehow 3 equivalent to .3 log units in contrast sensitivity. Was 4 the equivalence based upon some sort of Z score change, 5 Mark, or -- 6 DR. BULLIMORE: I didn't mean to imply that 7 they were equivalent, but I said it sort of parallels the 8 change. 9 Now, you could say, well, if we're doing .2 log 10 units of visual acuity, we should use .2 log units for 11 contrast sensitivity. Unfortunately, most of the 12 commercially available tests for contrast sensitivity go in 13 steps of .15. 14 Again, we're going to have the data. We're 15 going to be able to look at the number of patients that 16 have lost .3 or more log units at one or two spatial 17 frequencies. We'll have mean contrast sensitivity data for 18 each spatial frequency and for each lighting condition. 19 There will be a colossal amount of data that we'll be able 20 to sort of chew over in depth when the opportunity arises. 21 DR. BRADLEY: Yes, I think we'll have the data, 22 but we'll still be left with the question about what's 23 going to be significant. 24 Just as a suggestion, then, I would make that 25 if there is consensus that a .2 log unit change of visual 167 1 acuity is clinically significant, would you think it 2 reasonable that we convert that into some sort of acuity Z 3 score change -- say, two, three, four standard deviations, 4 whatever it is -- and propose that an equivalent Z score 5 change in contrast sensitivity be considered significant. 6 Does that make any sense at all? 7 DR. BULLIMORE: It makes sense, but I don't 8 think it's an approach that I would advocate at this stage. 9 DR. WEISS: Dr. Owsley? 10 DR. OWSLEY: This is Cynthia Owsley. I think 11 one of the -- 12 (Telephone rings.) 13 DR. OWSLEY: Could somebody get that? 14 DR. BULLIMORE: I think it's Art's. 15 DR. WEISS: It's probably Pizza Hut returning 16 someone's surreptitious call. 17 (Laughter.) 18 DR. OWSLEY: I mean, I think both Mark's 19 approach and Arthur's approach are reasonable approaches. 20 The problem for me, when I think about this, is when you 21 decide how much decline on a visual function test is bad in 22 some sense, clinically significant in a bad sense, you have 23 to ask yourself what is it you're trying to prevent. 24 Answering these questions in vacuo without 25 looking to see how much of a loss you need in contrast 168 1 sensitivity or acuity or glare or whatever the visual field 2 causes a problem in functional performance, without looking 3 at it in that way, I just don't see how you could answer 4 the question. 5 I'm not suggesting that we propose all kinds of 6 -- whatever they're called -- substudies to answer that, 7 but I think it's an important dilemma, because when it's 8 considered on its own, it's abstract. For the patient, 9 it's in terms of their everyday life. What implications 10 will a .2 loss in logMAR acuity mean? What implications 11 for their everyday will be a .3 loss in contrast 12 sensitivity? 13 So I haven't proposed any answers to this, but 14 I see it as a very sticky dilemma that we might just have 15 to kind of go with something that feels like it has some 16 face validity sort of on a clinical level. 17 DR. WEISS: Mr. McCarley, Dr. Mathers, Dr. 18 Bandeen-Roche, and then Dr. Bradley. 19 MR. McCARLEY: Yes, just one question. Is this 20 the first time contrast sensitivity testing has been 21 required by the panel? My understanding is that 22 manufacturers of refractive lasers also collected this 23 data. I mean, the question behind that is is there a 24 standard for contrast sensitivity or are we now making the 25 standard? 169 1 DR. WEISS: Well, we're making a standard for 2 phakic IOLs, so even if it wasn't required for anything 3 else, it really -- 4 MR. McCARLEY: In my understanding, it was. 5 Maybe I'm wrong. 6 DR. WEISS: Well, I'll defer to Mr. Whipple. 7 MR. WHIPPLE: Yes, I believe we have required 8 contrast sensitivity for LASIK studies. 9 DR. BULLIMORE: Yes, and the current guidance 10 document says that either you have to measure it or 11 basically to say that you didn't measure it. 12 MR. WHIPPLE: Right, and I'll defer also if 13 Donna and Malvina -- 14 PARTICIPANT: That sounds definitive. 15 DR. BULLIMORE: I'm paraphrasing a little bit, 16 but that's the general spirit of it. 17 DR. WEISS: Donna? 18 MS. LOCHNER: Well, it's not the first time 19 it's been required. As you point out, the LASIK example, 20 and of course the panel has reviewed extensive contrast 21 sensitivity data in the multifocal IOL example. 22 I think it's also important to point out this 23 question about the clinical significance because I think 24 what Dr. Bullimore was saying about, you know, you're going 25 to have to look at the data and make some judgements when 170 1 it's received, but this question is backing up to the 2 sample size calculations. It's not backing up that if it's 3 .3, the device is approved and if it's .29, it isn't. It's 4 really more is this a reasonable effect to be looking at 5 the sample size calculations, et cetera. 6 I think also, as Dr. Owsley said, the meaning 7 of this is really not going to be clear to many of the 8 panel members, as well as to the patients, unless this were 9 there were this daily living-type testing required, which 10 we really have not -- we're not suggesting that, but I 11 think it's a point well taken. So really, if you think of 12 this in the context of the sample size. 13 DR. WEISS: I think Dr. Bullimore -- 14 DR. BULLIMORE: Yes, I mean, you can look at a 15 .3 loss as being clinically significant two ways, whether 16 it occurs in a subset of individual patients or in one 17 individual patient or whether it occurs in the population 18 as a whole, and in terms of sample size, you're interested 19 in the population as a whole. In terms of the safety of a 20 device, you might say, well, if a two-line loss of visual 21 acuity sets off alarm bells and appears on some summary 22 statistics for safety, then should there be an equivalent 23 here? 24 I'm not comfortable doing the later. You know, 25 we heard from a speaker this morning who has a lot more 171 1 experience with phakic IOLs than the rest of us, and he was 2 advocating that the IOLs be held to the same standard 3 visually as LASIK, and we have some historical precedent 4 here with these tests being done on everybody and I think 5 we should carry on with that. 6 In terms of the sample size for the entire 7 cohort, I think it's much more likely to be driven by 8 endothelial cell density considerations than anything 9 visual. 10 MS. LOCHNER: So it's your recommendation that 11 this testing be performed on all individuals? 12 DR. BULLIMORE: Yes. That would be my 13 recommendation. 14 DR. WEISS: I think we're going to go along 15 this time even if it's out of order I originally said. 16 DR. BULLIMORE: I'm not asking that everybody 17 flies to Iowa for a driving simulation. I mean, this a 18 test that should be reasonably easy to incorporate into a 19 protocol and I would like to see data on as many patients 20 as possible. 21 MR. CALOGERO: Don Calogero. Right now, it's 22 set up as a substudy, and I believe the sample size is 23 somewhere between 60 and 100 patients, and that gives us 24 the ability to detect down to .15 log units. So you're 25 saying that in spite of perhaps having sufficient precision 172 1 to the study, you'd like to see it on the entire 2 population? 3 DR. BULLIMORE: Well, I think where Dr. Mathers 4 was going with this is that he would like to see data on 5 contrast sensitivity with and without glare that would give 6 you clues, maybe not definitive decisions, as to whether 7 significant cataractogenesis had occurred in these 8 patients. 9 MR. CALOGERO: Okay. So you're also using it 10 for that purpose, then. 11 DR. BULLIMORE: Exactly. Now, if my memory 12 serves me correctly, I mean, yesterday's presentation for 13 an intraocular -- 14 DR. WEISS: Wavefront. 15 DR. BULLIMORE: Wavefront. That was it. 16 DR. WEISS: How quickly we forget. 17 DR. BULLIMORE: A good dinner. 18 For that, we had data on all the patients. Am 19 I correct? 20 MR. CALOGERO: I believe so, yes. 21 DR. BULLIMORE: Yes. So I would have thought, 22 with a newer technology, which, let's face it, these phakic 23 IOLs are, and I think increased or elevated safety concerns 24 in terms of lens opacities, I'd want to have that data on 25 as many patients as possible. 173 1 Now, I realize we may be into latter phases of 2 some PMA investigations, and that's fine, but I think it's 3 going to inform at this stage whether the patients have 4 indeed developed anything that may cause concern in the 5 lens opacity department. 6 MR. CALOGERO: Thanks for clarifying that. 7 DR. WEISS: Dr. Mathers, then Dr. Bradley, and 8 we'll move our way around, and then come back to Dr. 9 Owsley. 10 Dr. Mathers? 11 DR. MATHERS: Yes, I agree with Dr. Owsley that 12 it's difficult to assess how this impacts the patient, but 13 what we need to do is have reasonably high resolution of 14 the data that we're picking up. When we're talking about 15 whether this is actually a good thing, you have to contrast 16 that with the struggles the patients have with enormous 17 myopia and their tremendous problem and the alternative of 18 clear lens extraction and other significant issues, but we 19 want to have reasonable resolution and an ample sample size 20 so that we can tell what's going on, and it may be that two 21 or three lines is a reasonable expectation considering the 22 other struggles that they have. But we would determine 23 that later. We need to have the data now, and I think the 24 contrast sensitivity not only gives us something about the 25 visual function of the system, but also the 174 1 cataractogenesis process. 2 DR. WEISS: I think that probably most people 3 here are in agreement that we need the data and whether or 4 not someone postulates that they know what the data means 5 upfront versus whether they don't know what the data means 6 upfront really won't affect the FDA. They can just tell 7 you after the fact whether you were right or not. 8 So if anyone has any comments not related to 9 their opinion on that particular topic, we can proceed with 10 them. 11 Dr. Bradley? 12 DR. BRADLEY: Yes, I think just because this is 13 a new product doesn't necessarily mean we have to measure 14 contrast sensitivity. We really need to have sort of 15 underlying theoretical rationale for why contrast 16 sensitivity measurement might or might not be useful in 17 this particular case. 18 I think the primary concern we have here is 19 degradation of optical quality, whether it be in the cornea 20 due to endothelial problems or primarily in the lens due to 21 cataract development. The likely optical and visual 22 consequences of that are related to scattered light, and 23 they have fairly predictable optical effects. Indeed, they 24 will have some effect on our ability to see fine detail, 25 which ought to be revealed by some visual resolution task. 175 1 In addition to that, they will have impact on 2 the image quality for larger targets, and the primary 3 impact will be on contrast reduction. One method for 4 assessing this is to examine contrast sensitivity. 5 That said, as Mark alluded earlier, there are 6 devotees of letter contrast sensitivity testing and there 7 are devotees of grating contrast sensitivity testing. One 8 of the arguments in favor of grating testing is that one 9 can test at many spatial frequencies, which academically 10 might be quite interesting, but unless one can come up with 11 a reasonable theoretical argument for why a measurement at 12 a single low spatial frequency might not provide the same 13 information, I think one is wasting one's time measuring at 14 lots of different spatial frequencies. 15 Therefore, I wonder about most grating tests 16 for that reason, and I think Dr. Owsley might be able to 17 comment on that. 18 DR. OWSLEY: Yes. I very much agree with the 19 perspective or the question you're raising. I would not 20 describe myself as a devotee or a guru of any test. 21 The reason I favor letter tests in situations 22 like this is that I know of no convincing evidence in the 23 peer-reviewed literature that shows that letter tests are 24 worse than grating tests or grating tests are better than 25 letter tests, however you want to say it. 176 1 I think that there are a lot of us who do 2 clinical studies, clinical intervention evaluation studies, 3 including bodies like the National Eye Institute, who go 4 the way of letter tests not only based on the evidence that 5 they're not worse than the grating tests, but because of 6 two things. One, if you're getting a measure of high- 7 resolution visual acuity, and then you do a contrast 8 sensitivity letter test, say like the Pelly-Robson, which 9 has large letters, you're basically getting information 10 about the entire shape of the function. 11 Then the second reason is that we're talking 12 about -- I don't know which specific test the sponsor would 13 be using or any sponsor would be using, but if you're doing 14 spatial frequency testing in the kinds of things we're 15 talking about in this kind of study, you're doing those 16 spatial frequencies for topically, mesopically, with glare 17 and without glare, before and after surgery. This is a 18 long testing time, okay? 19 And then I think -- what was my last point on 20 this? Well, I'll just leave it like that, but I know that 21 this issue has been visited by this panel before and I've 22 heard about it. I've never been to the panel to hear it 23 argued as a panel member, but I think that it's an 24 important point that sponsors should hear, the public 25 should hear, and the FDA should hear that there really are 177 1 really sound arguments for not, in every case, doing all 2 the spatial frequencies. You need to ask yourself, what 3 are you doing with that information? What's it really 4 providing for you that high-contrast acuity in a single 5 measure on a large letter contrast sensitivity chart would 6 not provide? 7 DR. WEISS: Dr. Swanson? 8 DR. SWANSON: I agree with both sets of 9 comments and wanted to add one other thing. I've used 10 letters and spatial frequencies. It depends on what the 11 task is. 12 The point that Dr. Bullimore raised becomes 13 important for sample size calculations as well as for the 14 amount of time that goes on, particularly if we're looking 15 at this question of glare, but as Dr. Bradley mentioned, in 16 general, these effects should be in a range of spatial 17 frequencies. 18 If you are using a grating and you have a two- 19 alternative forced choice, basic researchers who use those 20 things do lots of trials because that's the only way you 21 can reduce test/retest variability. For letter testing, 22 where there's going to be between 10 and 26 options that 23 the person's guessing amongst, a much smaller number of 24 trials is needed. 25 So in order to have a significant change, you 178 1 need to have a test where the test/retest variability is 2 low. For a small number of trials, which needed to run all 3 these conditions in all these people, a multiple 4 alternative, 10- or 20-alternative forced choice test, is 5 going to be much better than a two-alternative forced 6 choice, and there aren't any commercially available 10- 7 alternative forced choice grating tests and it's going to 8 be very difficult to create on given the response demands 9 on the patients. So a letter type of acuity test is going 10 to be much more suitable in terms of gathering a fair 11 amount of useful data in a short period of time. 12 Then the questions that come up, such as, well, 13 do we have to have two spatial frequencies down or three 14 spatial frequencies or one spatial frequencies down, those 15 become quite complicated by the high test/retest 16 variability of a two-alternative forced choice with just a 17 single endpoint. 18 DR. WEISS: So let me get to a bottom line. 19 What do you want? 20 DR. SWANSON: I'm just trying to hammer home 21 all the points they made about letter charts being superior 22 for this purpose. I understand there was agreement before. 23 DR. WEISS: So you would like a letter chart 24 for this purpose. 25 DR. SWANSON: For the purpose, a letter chart 179 1 is going to allow a much smaller sample size and a much 2 larger -- 3 DR. WEISS: And Dr. Owsley, you agree, and Dr. 4 Bradley? 5 DR. BRADLEY: Yes, I agree, a letter chart. I 6 think Donna asked us if such charts were available. There 7 certainly is a letter contrast sensitivity chart, as I 8 think it's been referred to. Sometimes it's called a 9 Pelly-Robson chart. 10 One thing I would alert the FDA to is if you 11 are to request use of that chart, it doesn't have to be 12 used at the standard distance at which it was originally 13 designed, and there are reasons to pick your distance, 14 depending upon the scale or size of target for which you 15 wish to study. 16 DR. WEISS: Dr. Bullimore, your comment? 17 DR. BULLIMORE: I mean, the spirit in sort of 18 guidance documents before, whether for this issue or 19 others, was that the sponsor or a potential sponsor would 20 be able to speak with the FDA and the FDA, in terms of the 21 guidance document, wouldn't specify a given test, but there 22 would be some scope. I was just a little surprised that 23 this was so specific, even naming spatial frequencies. I 24 mean, you come pretty close to naming a test or naming a 25 couple of tests, and I'd just like to see some statement 180 1 that other tests could be considered if the sponsor in 2 consultation with the agency would be able to agree that 3 these were acceptable. 4 DR. WEISS: Don Calogero? 5 MR. CALOGERO: I can provide a little 6 background. At the last ANSI meeting, we sort of discussed 7 this issue, and there was a presentation and the basic 8 summary of the presentation was that it's really not 9 possible for us to predict where the largest effect is 10 going to be in terms of the spatial frequencies. We went 11 through the literature containing some information -- 12 DR. BULLIMORE: But Don, the panel's telling 13 you that that's not an appropriate -- well, not necessarily 14 a widely held view of the world. 15 MR. CALOGERO: Well, this was from the 16 literature and this is from sort of the internal data we 17 have. Some devices, depending on the type, had the largest 18 effect at 1.5 cycles per degree, others 12 cycles per 19 degree, and the committee felt that if we had recommended 20 the letter charts, based on sort of the spatial frequency 21 domain that they evaluate, you're essentially just simply 22 getting the highest spatial frequencies, maybe 12 cycles 23 and above, whereas we could potentially miss large drops at 24 other spatial frequencies, and with our current to predict 25 where the effect is, we felt it would be judicious to 181 1 recommend that the entire contrast sensitivity function be 2 defined. 3 DR. OWSLEY: Can I make a comment? 4 DR. WEISS: Yes. Make a comment, Cynthia, and 5 then also if we could direct it at, from what I understand 6 from the vision scientists on this panel, they don't agree, 7 and is that the case or is that not the case? 8 DR. OWSLEY: The vision scientists on this 9 panel I think do agree the letters would be better. 10 DR. WEISS: No, you agree that letters would be 11 better, but are you agreeing with what's been put forward 12 by the FDA? 13 DR. OWSLEY: Well, let me put it like this. 14 DR. WEISS: That's what I'm asking. 15 DR. OWSLEY: And that's exactly what my comment 16 is on. I've been reading this literature for 20, 25 years, 17 like Arthur -- Mark's a little younger -- and Steve. 18 DR. WEISS: He's thinking a lot younger. 19 DR. OWSLEY: I would be happy as a consultant 20 to the panel to look at those peer-reviewed papers. Of 21 course, if you have internal data you can or cannot share 22 with me, that's your decision, but I have not seen any 23 convincing evidence that measuring all these different 24 spatial frequencies would matter in any of the decisions 25 that you would be faced with at the FDA regarding these 182 1 types of devices. But I'm openminded. I just haven't seen 2 those papers. 3 DR. WEISS: So I see sort of agreement by Dr. 4 Swanson. 5 Dr. Burns has a comment. 6 DR. BULLIMORE: I mean, my overall view is that 7 these -- 8 DR. WEISS: Dr. Burns has a comment. 9 (Laughter.) 10 DR. BULLIMORE: This -- 11 DR. WEISS: Well, I guess Dr. Bullimore has a 12 comment. 13 (Laughter.) 14 DR. BULLIMORE: Since it's my -- 15 DR. WEISS: Dr. Burns is deferring to you, 16 Mark, so why don't you give your comment? 17 DR. BULLIMORE: I have no problem with gratings 18 being used. It was just having a sentence in there that 19 other avenues could be pursued if -- 20 DR. WEISS: So you don't like the rigidity of 21 it, but you don't have a -- 22 DR. BULLIMORE: It's more the rigidity, and I 23 mean, as I said before, you're almost sort of saying you've 24 got to use this test or this test, and those of us who work 25 in other arenas, and we've already spoken about the 183 1 limitations of certain types of tests, we just find that a 2 little offensive, I think. 3 DR. WEISS: Dr. Burns, and then Karen Bandeen- 4 Roche and Mr. McCarley. 5 DR. BURNS: Yes, I just want to chime in that I 6 also believe that the combination of the high contrast and 7 the low contrast letter chart will give us enough 8 sensitivity both for some of the safety issues, such as 9 contrast degradation from cataracts combined with glare, 10 and remember also, this is a refractive procedure. So we 11 do want to get general decrement information. But I do 12 believe that a contrast letter chart combined with a high 13 contrast letter chart will be acceptable. 14 DR. WEISS: Mr. Whipple? 15 MR. WHIPPLE: Yes, Dave Whipple. I just wanted 16 to address Mark's comment about the rigidity of the 17 guidance. We may recommend certain tests or prefer certain 18 things, but a guidance is exactly that. It's guidance. It 19 always carries with it the option of using other tests, 20 other test methods, and making arguments why the 21 recommended tests aren't appropriate for that particular 22 device. So that's inherently built into the guidance 23 document development. 24 DR. WEISS: We'll go on to the other two 25 comments if they're not related or they don't change what 184 1 my next statement is going to be. It's from why I 2 understand, the members of the panel prefer a letter chart 3 and Dr. Bullimore would also prefer that the wording 4 doesn't sound so restrictive, even if it actually isn't. 5 Having said that, do you have anything else to 6 add to that, Dr. Bandeen-Roche or Mr. McCarley? 7 DR. BANDEEN-ROCHE: My comment was a brief one 8 on a different topic, so should we finish this? 9 DR. WEISS: If it's related to this question -- 10 DR. BANDEEN-ROCHE: It's related to clinical 11 significance in terms of functional performance. 12 DR. WEISS: Well, why don't you just proceed? 13 DR. BANDEEN-ROCHE: I just wanted to mention 14 that there is detailed data from the Salisbury Eye 15 Evaluation Project on glare, contrast, et cetera, with many 16 measures of functional performance. So I don't think we're 17 completely in the dark. I think that there are some data 18 that can inform that question, albeit it is a sample of 19 older adults. 20 DR. WEISS: Good. Thank you. 21 Mr. McCarley? 22 MR. McCARLEY: Yes, just very quickly. I'd 23 like to understand or get clarified for me the intent. My 24 understanding, at least from the ANSI side of it when we've 25 been discussing this for the last several years, was that 185 1 doing contrast sensitivity was to see if the combination of 2 optical components in the eye degraded, similar to what the 3 intent was when they did LASIK studies. Now, it seems to 4 be that's what's being used to determine whether or not the 5 patient's undergoing a degradation as a result of cataract, 6 for instance. So one would be a sample size and one would 7 be all patients, I think. 8 DR. WEISS: Dr. Bullimore, did you want to 9 address that? 10 DR. BULLIMORE: No. 11 DR. WEISS: No one wants to address that. I 12 guess we have no takers? 13 DR. BURNS: If you're measuring a visual 14 performance like this, you're tapping the whole system, and 15 the degradation can come about from optical imperfections 16 or from what are actually just very high-order optical 17 imperfections of scattering and tissue turbidity. So it 18 probes the whole system. 19 DR. WEISS: I think we've beat this one into 20 the ground and it's no longer even gasping. So why don't 21 we go on to the next one? 22 DR. BULLIMORE: The next one I think is very 23 easy to deal with. 24 DR. WEISS: We'll hold you to that. 25 DR. BULLIMORE: Again, we're asking people to 186 1 be very specific about the tests, and I think the FDA have 2 enough savvy and flexibility to deal with this, but if 3 somebody can't read or can't see the highest contrast on 4 the chart, I think their contrast sensitivity has to be 5 scored as zero. One would hope that preoperatively the 6 test was sufficiently intelligently designed such that they 7 could read well above that bottom level. 8 Certainly, recalling some of the data we were 9 presented with yesterday, we were dealing with contrast 10 sensitivities of one or above most spatial frequencies. So 11 as long as we have something down at the 40 percent or 60 12 percent, which of course is .3 or so contrast sensitivity, 13 we should be okay. 14 But if they can't see it, it shouldn't be 15 counted as missing data. We should assume that their 16 contrast sensitivity is zero or some other intelligent 17 value based on the range of the test. 18 DR. WEISS: So can the transcript reflect there 19 are about six heads bobbing up and down? 20 Next question. 21 DR. BULLIMORE: And the next question, please, 22 Donna? 23 I think this is kind of tied in with the last 24 one. If I can answer, I think tests are available and if a 25 patient can't see a targeted spatial frequency at any 187 1 contrast, it should be scored as a contrast sensitivity of 2 zero. 3 MR. WHIPPLE: That is information. 4 DR. WEISS: Dr. Bradley, and then Dr. Swanson. 5 DR. BRADLEY: I'm having a bit of trouble with 6 the question. I mean, we're not anticipating these 7 patients are going to have horrible vision, are we? 8 DR. WEISS: Before or after the implant? 9 (Laughter.) 10 DR. WEISS: Sorry. 11 DR. BRADLEY: I mean, we're talking about some 12 serious methodology here. We're trying to ascertain 13 something fairly subtle, and I don't think in any of these 14 cases we're going to have the problem that people can't see 15 the target at all. 16 MR. CALOGERO: I think why we're asking the 17 question is under the mesopic test conditions. Under the 18 mesopic test conditions, we do in fact have cases where at 19 the higher spatial frequencies, the patients are unable to 20 perform anything, and so you might have 20 or 30 percent of 21 the population that you're testing that essentially has a 22 zero, let's say, 12 or 18 cycles per degree. So we wanted 23 some sort of standardized way of handling all of those zero 24 data points. 25 DR. BRADLEY: Well, I think the panel has given 188 1 you our suggestion on that, which is that you should not do 2 these grating tests. Particularly, don't try to find out 3 if people can see fine detail in the dark. We know they 4 can't. 5 DR. OWSLEY: This is Cynthia Owsley. It's very 6 unusual to find a patient who can't see the first triplet 7 of the Pelly-Robson chart, and I see Karen Bandeen-Roche 8 from the SEE Project nodding. 9 DR. WEISS: But in that case, I think we have 10 the answer to this question. We may not like the question, 11 but I think Dr. Bullimore has already given an answer to 12 it, which I assume you would agree to, but you don't think 13 it's going to even come up. 14 Fine. Next? 15 DR. BULLIMORE: Beaten to death. Beaten to 16 death. 17 I really have nothing to add on this one at the 18 moment. I'll read it again. "Please provide any 19 additional comments on the contrast sensitivity substudy" 20 -- now it's a substudy -- "including any other guidance 21 that could be provided to enhance the quality of the data 22 that are generated from this testing." 23 Use a good test would be my recommendation. 24 (Laughter.) 25 DR. WEISS: That's why we call the experts to 189 1 figure this out. 2 DR. BULLIMORE: And use it well. 3 DR. WEISS: So is that the end of your 4 questions? 5 DR. BULLIMORE: I'd like to hear what Dr. 6 Owsley has to say. 7 DR. WEISS: Dr. Owsley? 8 DR. OWSLEY: I think the test should have the 9 best test/retest reliability as possible and should be 10 brief. 11 DR. WEISS: And brief is always dear to my 12 heart. 13 Yes? 14 DR. BULLIMORE: This is Dr. Bullimore again, 15 not wanting to shut up. I think it would be ideal and 16 preferable and maybe mandatory that for any test that's 17 going to be used that normal data are available on a wide 18 range of age ranges and for the measured conditions. 19 DR. WEISS: Dr. Bradley? 20 DR. BRADLEY: I think just to reiterate what 21 everybody has said here and maybe to put the comment that 22 Bill Swanson made earlier, what we're arguing for regarding 23 the quality of a contrast sensitivity test can be thought 24 of by thinking of an analogy with a visual acuity test. 25 Imagine you had a visual acuity chart, one 190 1 letter per line, and it was either A or B. Would we 2 consider that an appropriate test? And clearly not, and 3 all we're saying is let's hold the contrast sensitivity 4 test to a similar standard psychophysically in terms of 5 rigor and test/retest reliability that we are quite 6 comfortable demanding from our visual acuity tests, and it 7 turns out at the moment the only convenient one out there 8 that we know of is a letter chart. 9 DR. WEISS: Dr. Bradley, we get the idea you 10 like the letter chart. 11 DR. BULLIMORE: He is so proud that he can 12 read. 13 DR. WEISS: He likes that letter chart. 14 DR. BULLIMORE: He is so proud. 15 (Laughter.) 16 DR. WEISS: If there are no other comments on 17 this, what I'd like to do then is move on to panel 18 discussion on remaining issues, and I have three issues, 19 and then we'll introduce if the agency has any additional 20 issues or if any members of the panel want to introduce any 21 additional issues. 22 One issue, which has been alluded to and 23 discussed, but I would like to see if we can give as many 24 opinions as possible, is duration of study before it's 25 presented to the panel. Somewhere between two and three 191 1 years has been mentioned. Perhaps any of those of you who 2 would like to give your opinions can raise your hand and 3 just give me a number and tell me why. 4 Dr. Burns? 5 DR. BURNS: From what I see of the endothelial 6 cell measurements and their likelihood of accuracy in this 7 population, I think three years is a minimum. 8 DR. WEISS: So Dr. Burns feels three years. 9 Dr. Bullimore, had you -- three years. We have 10 a sign. We're going to signs now. Okay. So we have Dr. 11 Matoba at three, Dr. Grimmett at three. Dr. Bradley likes 12 the number chart. 13 (Laughter.) 14 DR. WEISS: Dr. Huang, three. Dr. Mathers is 15 three. Dr. Owsley is three. Dr. Swanson is three. So I 16 think we have actually, even though we're not looking for a 17 vote, essentially it's almost unanimous, if not unanimous, 18 for three years. 19 DR. BANDEEN-ROCHE: Madam Chair? 20 DR. WEISS: Yes? 21 DR. BANDEEN-ROCHE: May I make a very brief 22 comment? 23 DR. WEISS: Yes. 24 DR. BANDEEN-ROCHE: And that is that I believe 25 it's important, based on Dr. Grimmett's comments, that at 192 1 least one, preferably two, evaluations be scheduled between 2 two and three years because he suggested that linearity was 3 not beginning to be achieved until then. 4 DR. WEISS: Okay. So one to two endothelial 5 cell counts should be performed between the two- and three- 6 year mark. 7 Dr. Grimmett? 8 DR. GRIMMETT: Dr. Grimmett. I couldn't agree 9 more with Dr. Bandeen-Roche that the data that is present 10 in the peer-reviewed literature do not show a linear 11 approach. We don't have enough data long enough. So that 12 is correct. In an intermediate point. 13 DR. WEISS: Dr. Mathers? 14 DR. MATHERS: I would not end the data 15 collection necessarily with three years because I think 16 that ongoing data could be important. 17 DR. WEISS: That's an important question in 18 terms of a postmarket study. It could be brought to panel 19 at three years, at least from the recommendations from this 20 panel today. After that, a postmarket study of what 21 duration of time would you like to see, Dr. Mathers? 22 DR. MATHERS: Another two years. 23 DR. WEISS: Another two years. 24 Dr. Grimmett? 25 DR. GRIMMETT: Dr. Grimmett. I would agree 193 1 with Dr. Mathers' sentiments, but that decision would be 2 made more accurately with the data in hand with the PMA at 3 three years obviously, but I would feel certainly 4 comfortable with probably a postmarket study. 5 DR. MATHERS: Yes. 6 DR. WEISS: Dr. Bullimore? 7 DR. BULLIMORE: This is endothelium only? 8 DR. WEISS: It can be whatever you want. 9 DR. MATHERS: Cataract. 10 DR. WEISS: And you don't have to define it 11 now. Basically, what the agency would like is your 12 opinions. 13 DR. BULLIMORE: My opinions will be more well- 14 informed once I have some data, but I have considerable 15 concerns about the long-term safety of these intraocular 16 devices. As the lens continues to grow, as it gets to 17 cohabit with the phakic IOL for a long period of time, I 18 don't think any of us can predict with any accuracy or 19 precision what that's going to do to the crystalline lens 20 on a long-term basis. 21 You know, these devices we can expect to be 22 better than some of the first-generation phakic intraocular 23 lenses, but they haven't been around long enough for us to 24 tell, so I could foresee a substantial -- i.e., five-year 25 -- postmarket study to track these patients and see, for 194 1 example, how many people develop significant lens opacity, 2 whatever that may be, how many people have to have a 3 cataract extraction and conventional IOL implantation, how 4 many of those patients have complications from that 5 procedure, and it can be nothing more than fairly 6 straightforward head counting, but I think tracking these 7 patients long-term, given what we know happened with, for 8 example, extended wear contact lenses in the '80s and 9 anterior chamber intraocular lenses, I think it would be 10 prudent. 11 DR. WEISS: So the impression I get -- and if 12 there's something you would like to add, then we'll add it 13 -- is that the panel members would like to see postmarket 14 studies from two years on up, maybe even to five years, but 15 pending what the data shows. 16 DR. BULLIMORE: Once we have a PMA before us, 17 we can make a more informed decision, but certainly, given 18 the fact that this is an intraocular device that's 19 cohabiting with a lens that's continuing to grow and 20 function, I would -- 21 DR. WEISS: You want to see a postmarket study. 22 DR. BULLIMORE: Well, I would anticipate the 23 possibility of this requirement. 24 DR. WEISS: Dr. Mathers? 25 DR. MATHERS: We have not focused on pigment 195 1 dispersion, but in the literature there are a number of 2 papers that do. Not so much that they've had a lot of 3 glaucoma yet, but they often have pigment dispersion in the 4 range of 30 percent, and therefore gonioscopy and pressure 5 measurement. 6 DR. WEISS: Actually, that's an excellent 7 point. I don't recall if that's in the document, 8 gonioscopy, and then how often -- 9 DR. MATHERS: It is. 10 DR. WEISS: And how often should postop. 11 DR. MATHERS: It's in there now and it hasn't 12 been focused on, but it's in the literature, and I think 13 it's an important issue. 14 DR. WEISS: So we haven't defined here, but 15 there is concern in terms of follow-up as far as not only 16 the cataract and the endothelium, but also gonioscopy and 17 how frequently that would be done and what the postmarket 18 would be would be defined in terms of when some of the data 19 comes in. 20 Another question is what is an independent 21 entity? Is the subject an independent entity or is the eye 22 an independent entity? 23 For example, classically, for intraocular 24 lenses, each patient has been an independent entity. Each 25 subject has been an independent entity. So if the FDA 196 1 statistically said that 300 data points were required, then 2 each patient would be considered separately an entity 3 whether or not they had one or both eyes treated. 4 On the other hand, for LASIK, each eye has been 5 considered an independent entity in the same patient, so 6 that if 300 data points were required, then you might only 7 need 150 patients. 8 So this is a very important point for sponsors 9 to know whether they need 300 patients, and we're going to 10 be getting to sample size as well, but if the sample size 11 is determined to be necessary to be 300, should that be 300 12 patients or can it be 150 patients with both eyes treated? 13 Dr. Bandeen-Roche? 14 DR. BANDEEN-ROCHE: Yes, I mean, statistically 15 speaking, from a strict point of view, sites are the 16 independent entities. Now, obviously, we're not going to 17 base sample size on that, but methods are available to 18 account for correlation of eyes within individuals, and in 19 turn, of individuals within sites if necessary in computing 20 power. 21 So one doesn't necessarily need to go to the 22 extreme of saying, for instance, we're going to count 23 people as entity. One can still account for the 24 information provided by two eyes within an individual, yet 25 accounting for the correlation in that information that two 197 1 eyes shouldn't count as much as two separate people, and 2 indeed that perhaps two people within a site shouldn't 3 count as much as two people in different sites. 4 I actually have a figure that I brought on 5 this, but if you prefer to save it, we can. 6 DR. WEISS: No, sure. 7 DR. BANDEEN-ROCHE: I need the overhead. So 8 can I just quickly set that up? 9 DR. BANDEEN-ROCHE: Sure. Maybe while you're 10 setting that up, we can go on to a comment by Dr. Burns, 11 and then to Dr. Matoba. 12 DR. BURNS: Yes, I just want to support that 13 strongly because we're talking here a lot about biological 14 reactions to a foreign body. So there will be a 15 significant amount of correlation, I assume, between eyes, 16 and so the studies should explain how they're going to 17 handle it statistically in the design, and not in post-hoc 18 analyses. 19 DR. WEISS: So I want to understand. Would you 20 feel that an eye should be an independent entity or a 21 subject? 22 DR. BURNS: I think Karen is going to explain 23 the statistical way to handle it, but it's important that 24 it be handled that way from the outset, that you're 25 accounting for the correlation between eyes. 198 1 DR. WEISS: Now, we're going to go along with 2 other comments, but we also need to look at this not only 3 statistically, but from a safety standpoint because this is 4 a new device. 5 Dr. Matoba? 6 DR. MATOBA: Yes, in regard to the safety 7 standpoint, since we've already decided that because of the 8 issues of quality of life we're going to allow patients to 9 have both their eyes done instead of limiting it to one eye 10 over a three-year period of time, we might as well get data 11 from both eyes, and so fewer people would have to be 12 involved in the initial study. 13 DR. WEISS: Dr. Mathers? 14 DR. MATHERS: Yes, I strongly agree with that 15 comment. I think we should have both eyes, but we should 16 account for the correlation, and that allows us to get over 17 this quality of life issue, which is going to be very, very 18 important to the patient because only doing one eye is not 19 going to work. 20 DR. WEISS: Well, we're not talking about only 21 doing one eye. We're talking about whether you need 300 22 subjects who can both have both eyes done or 150 subjects 23 who can still both have both eyes done. 24 DR. MATHERS: I think we shouldn't throw away 25 the data on the second eye, but we should use it with an 199 1 appropriate correlation. 2 DR. WEISS: So you basically want to count it 3 and not require another 150 patients. 4 DR. MATHERS: We're about to hear how we can 5 account for the correlation. 6 DR. WEISS: Okay. 7 DR. MATHERS: In estimate. 8 DR. BANDEEN-ROCHE: Yes, and so certainly one 9 can just state what is the expected design, including eyes 10 within patients, patients within sites, et cetera, and then 11 do power and all other statistical analyses accounting for 12 that correlation. So what I had diagrammed here was 13 thinking of people within sites, but it could equally well 14 apply to eyes within people. There might be several 15 levels. 16 So here, M is referring to the larger level. 17 So let me just stick to people within sites because that's 18 the way that I did it. So M is the number of sites and N 19 is the number of individuals within sites. 20 So for instance, if it were people and eyes, it 21 would be M people, and for two eyes, N would be two eyes 22 per person, and so what I'm showing you here is the 23 standard deviation of the mean -- say, a rate, say, a 24 safety rate -- or I guess to simplify things, this was 25 meant to be a continuous measure. So it would be something 200 1 more like an acuity measure that's measured on a continuous 2 scale. 3 So you can see that the standard deviation 4 depends both on the overall number of units, either people 5 times eyes or sites times people, and the symbol rho, which 6 I'll point to here, that's the correlation between outcomes 7 within a person -- so two eyes within a people -- or 8 between visual acuities within a site, and then sigma is 9 just the standard deviation of the measurement that's being 10 taken in the population. 11 So one thing to point out is that if N equals 12 1, then this top term goes away, and you just get the usual 13 standard deviation of the mean. 14 So what I'm showing you here is how the 15 standard deviation varies for the -- I assumed the FDA 16 sample size of 300, but what's varying here is the number 17 of clusters, and so here's where this is really more 18 relevant to sites and people within sites, because the FDA 19 guidance for the number of sites says something like at 20 least 20 per site and no more than a quarter being 21 accounted for by any one site, and so if we had four sites, 22 then that would just be meeting no more than a quarter by 23 any one site. Then here, up at 300, you only have one 24 patient per site, so that's the outer limit. 25 So the bottom line here, the straight line, is 201 1 what you get if there's no correlation. That's just the 2 standard deviation of the mean. If there's no correlation, 3 then it doesn't matter how you distribute people by sites. 4 The other three curves, just to save time, the 5 bottom one here assumes a correlation of .05. That's 6 small. That's assuming that 5 percent of the total 7 variation in outcomes is accounted for by variation between 8 sites. 9 So even with that minimal amount of 10 correlation, notice how much bigger the standard deviation 11 is. I apologize that the scale on this isn't better, but 12 that's twice as high. It's approximately twice as high as 13 your analysis ignoring correlation says that it should be. 14 Incidentally, we were about here yesterday. We 15 had five sites yesterday, and I think two of them accounted 16 for more than 50 percent of the data. 17 The one point is that I would encourage going 18 to more sites than no more than allowing any one to be up 19 to a quarter of the data, but the second point is that if 20 we were just talking about eyes within people, it would be 21 straightforward to do this calculation and just account for 22 it in calculating power and doing the analyses and what 23 have you. 24 DR. WEISS: Thank you very much. 25 I'm going to ask Donna Lochner to speak with us 202 1 a little bit more about this issue. 2 MS. LOCHNER: Well, I think this is helpful, 3 but the kind of recommendation we need clinically is what 4 factors do we base the correlation on and what is the 5 correlation that would be plugged prospectively, as you're 6 stating, into this equation? 7 DR. WEISS: And I should also add with 8 information from you already is the precedence to date, is 9 that LASIK is a little unusual because each eye has been 10 considered a separate entity, but the history has basically 11 been for intraocular lenses one patient, whether or not 12 they had both eyes done or one done, was a separate entity. 13 Viscoelastics, whether they had one eye or both eyes done, 14 was a separate entity. So there's been a little bit of a 15 history of using patients as separate entities, as opposed 16 to eyes. 17 MS. LOCHNER: Right, and I mean, we want to get 18 a feel for clinically how you would assign this correlation 19 factor for all the various variables that you're looking at 20 outcomes. 21 Now, we, of course, in our guidances, have not 22 addressed this method of potentially using the second eye, 23 and so all of our calculations and sample sizes and whatnot 24 are based on independent people, but if we were to allow 25 this approach, we would have to have some sense of what 203 1 would the panel consider acceptable for how correlated are 2 the two eyes for the various safety outcome variables and 3 effectiveness outcome variables. 4 DR. WEISS: Dr. Bandeen-Roche? 5 DR. BANDEEN-ROCHE: Yes, so certainly the 6 approach of calculating it for people is conservative. 7 MS. LOCHNER: Right. 8 DR. BANDEEN-ROCHE: That would essentially be 9 it, and so I have a hard time arguing with that, although I 10 know that others would raise we can't afford to waste any 11 data or money or what have you. 12 So in terms of getting a sense of the 13 correlation, I would think that if you have decent pilot 14 data, that the correlations I'm talking about could be 15 estimated in a straightforward fashion. 16 DR. WEISS: Could you perhaps have a subset of 17 the first 20 patients, 50 patients, to draw whatever those 18 correlations are to see whether you could then use separate 19 eyes as separate subjects or the same patient with two eyes 20 as separate? 21 MS. LOCHNER: I mean, my general impression is 22 that for most sponsors, developing this pilot study and 23 determining this correlation -- I mean, I've seen nobody 24 suggest that to us, first of all, and secondly, how would 25 they do it if at the end of the study they want to use both 204 1 eyes? 2 But I've never seen it presented. I've never 3 even seen a suggestion for what this pilot study would be, 4 and in fact probably what we've recommended in terms of the 5 slow phase in to not put eyes at risk for unproven phakic 6 IOLs -- I mean, if they have foreign experience, we do 7 allow them to phase in quicker, but I kind of suspect that 8 given the slow phase in, doing the pilot study, it's going 9 to be complicated. I mean, it's going to take a while for 10 them to gather that information. I'm not sure how many you 11 typically see in a pilot study to establish these 12 correlations. 13 DR. BANDEEN-ROCHE: And so, you know, certainly 14 I would be perfectly happy with the conservative approach, 15 but my question to the panel, and I have no idea what the 16 answer to this would be, would one expect the data to be so 17 different in these particular studies that a good sense of 18 reasonable correlation could not be obtained from either -- 19 I mean, LASIK data, that's probably ridiculous, but aphakic 20 IOL trials or is there historical data that are similar 21 enough in nature that a reasonable estimate of the 22 correlation of outcomes in a subject could be done? 23 DR. WEISS: I mean, I always wonder, if this is 24 new technology, I wonder in terms of we heard about if the 25 IOL is too small, it can induce cataract by sitting on the 205 1 lens. How often is that phenomena happening and is it 2 correlated between the quality of the measurement, a vagary 3 of that individual and their eye, and is it the quality of 4 the surgery? So I think some of these may be unknown 5 factors. 6 Dr. Burns? 7 DR. BURNS: I mean, I'd be happy with the 8 conservative approach, too. I wouldn't be happy with 9 getting the final analysis coming back suddenly treating 10 the eyes as 600 eyes at the end. 11 MS. LOCHNER: Oh, no. No. That maybe needs to 12 be clarified. In the sample size calculations that we've 13 done, we've recommended 300 individuals, their first eye, 14 and we require that they collect data on the second eye. 15 The second eye data is not combined in with the first. 16 It's just a separate analysis that's provided to the panel 17 that's really more of a confirmation check that the 18 outcomes are holding up in the second eye and giving you a 19 little bit more numbers. But no, we don't combine them 20 into a 600 sample size and improve the precision. 21 DR. WEISS: Dr. Mathers? 22 DR. MATHERS: I was just concerned that there 23 was some thought of holding up doing the second eye for the 24 three-year period. 25 MS. LOCHNER: Oh, no. No. 206 1 DR. MATHERS: But if that's not the issue, then 2 you can afford to be conservative in your statistical 3 approach if that's what you want. 4 MS. LOCHNER: And I think that basically, 5 without the detail and the sort of clear way you've just 6 presented it today, Dr. Bandeen-Roche, we've basically 7 given this advice to sponsors that, should they want to use 8 the second eye, they need to determine the correlation 9 between the two eyes and relook at their sample sizes. 10 So I think just getting this out on the table 11 is important, and I think the point that Dr. Weiss made is 12 probably where most of us sit in the FDA of there's some 13 unknown information and how do you determine the 14 correlation? Perhaps a pilot study, but there's just so 15 much unknown, and it's helpful for us to hear you reiterate 16 what we've basically told sponsors is determined on how the 17 two eyes interact. 18 DR. BANDEEN-ROCHE: And the only thing I would 19 add to that is that at the end of the study, the data 20 itself provides an estimate of the correlation, and that 21 should be accounted for in any analyses of two eyes. 22 MS. LOCHNER: Right. 23 DR. WEISS: Dr. Mathers? 24 DR. MATHERS: We are saying, though, that the 25 data on the second eye will be collected. Is that correct? 207 1 MS. LOCHNER: Oh, yes, and will be reported to 2 the panel. 3 DR. MATHERS: Fine. 4 MS. LOCHNER: But it will not be combined with 5 the first eye to get a bigger sample size. It will be 6 reported separately. 7 DR. MATHERS: Yes, that's key. 8 DR. WEISS: Mr. McCarley? 9 MR. McCARLEY: Yes, just quickly. What are the 10 ISO standards? What are the requirements in ISO right now? 11 MS. LOCHNER: Well, again, all the sample sizes 12 were based on a straightforward calculation not taking the 13 second eye into account, so I believe that the ISO is 14 basically asking for 300 individuals as well. 15 You know, it's a little different with a 16 standard in terms of -- I mean, you still can go to the 17 notified bodies and present whatever you want, but 18 certainly with the FDA, we would allow companies to propose 19 alternate proposals, but taking into account what's been 20 discussed. 21 DR. WEISS: Thank you very much. 22 The other question that I wanted to answer, and 23 we would probably need your input on this as well, is the 24 question of sample size. The 300 is what was put forward 25 by the FDA. Of course, it depends very much on the issues 208 1 that have been discussed by the three panel reviewers, but 2 does anyone have any comments on that 300 number? 3 Dr. Bandeen-Roche? 4 DR. BANDEEN-ROCHE: Dr. Grimmett was kind 5 enough to hand me a set of analyses yesterday, and I 6 believe they were done at FDA. Are they going to be 7 presented? 8 MS. LOCHNER: Well, I don't know if now's the 9 right time to bring it up, but we did have a bit of a 10 question about the endothelial cell density discussion that 11 went on earlier, and that is based on some of Dr. 12 Grimmett's questions, we prepared that table, which is a 13 table of different potential rates of cell loss due a 14 phakic IOL in sample size, and also different standard 15 deviations in the measurement. 16 DR. WEISS: Why don't show it now, because that 17 is probably -- 18 DR. GRIMMETT: Why doesn't Don Calogero just go 19 over some of the salient points? 20 MS. LOCHNER: And let me just, before he does 21 that, say that from the earlier discussion, we heard the 22 1,500 cells and we heard that using the actuarial data, and 23 even perhaps backing the age of cataract and calculating 24 different point on. We heard all that, but what we didn't 25 get from the earlier discussion was what rate do you want 209 1 to be able to detect? And that essentially is what Don 2 will present. I mean, just before giving you a clinical 3 impression, of course, you have to see what the sample size 4 is. We're talking about what really kind of seems 5 reasonable. It's a combination of your clinical judgement 6 along with the sample size is that translates into you 7 arrive at a rate that seems reasonable. 8 So Don's passing that out, and I'll let him 9 explain that to you, because we really didn't walk away 10 understanding whether you felt the 2 percent rate that 11 we've set up is reasonable. 12 DR. HUANG: Can I make a comment? Donna, I 13 think the sample size itself is not just for statistical 14 analysis. It also has to be considered for practical 15 matters. You know, as we mentioned earlier, in some of the 16 aniridia patients, you may not be able to get all those 17 patients, and so you cannot mandate that 300 eyes. 18 DR. WEISS: This is very relevant to the slide 19 you're about to see. 20 MS. LOCHNER: And let me also say that the 300 21 sample size actually didn't originate from the endothelial 22 cell density study. I mean, way back when we started some 23 of studies, it originated from the IOL work and being able 24 to detect low rate of complications, and then, as we 25 developed the statistical analysis for the endothelial cell 210 1 density, et cetera, it was a reality check to that sample 2 size and it coincided very nicely, but originally we 3 carried over a lot of the assumptions from aphakic studies 4 in what we wanted to be able to detect in the complication 5 arena. 6 DR. HUANG: But also the current discussion is 7 really limited to the phakic population, and then we are 8 probably targeting towards a higher myopia patient, and 9 those are patients more difficult to come by as indicated 10 from yesterday. For the wavefront technology, they could 11 only recruit 130 eyes with a -7. 12 MS. LOCHNER: That's true. However, I think 13 most of the studies that are ongoing in the U.S. go down as 14 low as a diopter. They aren't necessarily limiting -- 15 beyond the initial stages, when some of the preliminary 16 safety data is being gathered, in later stages of the 17 study, they are going down to 1, 2, 3 diopters. 18 DR. HUANG: But if this were to be limited to 19 the higher myopic patients -- 20 MS. LOCHNER: You wouldn't have the problem 21 you're discussing. Right. 22 DR. HUANG: Yes. 23 MR. CALOGERO: Okay. As Donna said, we need 24 some additional guidance in terms of the sample size. The 25 sample size that we have in the document now of 211 1 approximately 300 subjects is probably powered to detect 2 endothelial cell loss rates of maybe 2 percent and maybe as 3 low as 1.5 percent, but it's not going to get down to the 4 .9 percent, which was the lower extreme that Dr. Grimmett 5 brought up. 6 In terms of this document that I passed around, 7 what it looks at is it looks at sort of a true yearly loss. 8 The left column is the yearly loss from .9 percent up to 2 9 percent, and then the next three columns on the right are 10 the sample size that you need with an observed, allowable 11 rate per year, which is the fourth column, to give you 90 12 percent confidence that the true rate is up to the yearly 13 loss rate in the lefthand column. 14 Like in the first one, if you want the true 15 yearly loss rate to .9 percent, if you in your data you 16 have a standard deviation of 10 percent, you would need a 17 sample of 296 subjects, and your allowable observed rate 18 could be as high or as low as .63 to meet that level. 19 So it becomes important to first get a sense of 20 what the true standard deviation is in this data, and then 21 secondly, to have a sense of what the panel members would 22 like to see in terms of defining a true rate associated 23 with endothelial cell loss for these devices. 24 I simply generated the numbers, and we'd like 25 some feedback. 212 1 DR. WEISS: I think Dr. Grimmett has a 2 question. 3 DR. GRIMMETT: Drs. McCarey and Edelhauser put 4 forth precision numbers for those, 2 percent and 9 percent, 5 and I want to know how those correlate over the standard 6 deviation numbers. 7 PARTICIPANT: Put up the slide. Are they the 8 same? 9 DR. BANDEEN-ROCHE: No, they're not quite. 10 They're not quite, and so as I understand it, the numbers 11 that were presented this morning were essentially the 12 absolute difference in two measurements over the maximum of 13 those two measurements. I read that off of the handout 14 from this morning. 15 So what that means is the numerator is the 16 difference in two measurements, and so essentially what you 17 need to do with that, each one of them has a variance. 18 Each one of them contributes one of those standard 19 deviations from your table, except it has to be done in 20 terms of variance. 21 So to make a long story short, the conversion 22 is that the FDA percent standard deviation is the percent 23 variation that was reported this morning divided by the 24 square root of 2, and that division being that in the 25 numerator of the statistic this morning, there were two 213 1 measurements being subtracted. So if you approximate no 2 correlation between them, they each contribute a variance, 3 and then take the square root of that because it's a 4 standard deviation, rather than a variance. 5 So that's where that square root of 2 comes 6 from, and so if you do that, then I just sketched out on a 7 thumbnail sketch what the 5 percent, 10 percent, and 15 8 percent on the table in front of you corresponds to in 9 terms of what we hearing this morning. So respectively, 10 that would be 7 percent, 14 percent, and 21 percent. 11 That's just multiply by 1.4, the approximation of square 12 root of 2. 13 One more number would be that if you went down 14 to 3.5 percent on the FDA table, that would correspond to 5 15 percent in terms of the figures that we were hearing this 16 morning. 17 DR. GRIMMETT: This is Dr. Grimmett. If I 18 interpret that correctly, that's actually good news, 19 because that means the numbers that were quoted this 20 morning may be achievable because they actually translate 21 into lower standard deviations. 22 DR. BANDEEN-ROCHE: I interpret it the same 23 way. Yes, and it seems important to me that -- I mean, 24 measurement, good quality of measurement, is where we stand 25 to gain precision and power, and however that can be 214 1 absolutely pushed for people to up their standards, it's 2 important. 3 DR. GRIMMETT: Dr. Grimmett again. So in Dr. 4 Edelhauser's best case scenario, though, the 2 percent 5 precision factor could be achieved. Assuming that could be 6 done, then we're talking a standard deviation of just 7 slightly higher than that. For example, 3 percent or 8 something like that, whatever the number is. 9 DR. BANDEEN-ROCHE: Well, even lower, right? 10 Yes. 11 DR. BURNS: Excuse me. Just a clarification. 12 That's the precision of the measurement, but not the 13 standard deviation of the population, is it? 14 DR. BANDEEN-ROCHE: Right, but that's what I 15 understood this to be. We were talking about the precision 16 of the measurement, right? Yes. So that is what appears 17 in the FDA Table 2. Those standard deviations refer to 18 standard deviation of measurement in a single person. 19 DR. WEISS: So with this information before us 20 and your extra analysis, I'd like some opinions as far as 21 what numbers of subjects we're looking at and what yearly 22 loss. 23 MR. CALOGERO: Don Calogero. Can I just 24 mention one thing? I believe that data was on the 25 KeraVision rings, and those were sort of low myopia 215 1 patients. I believe they went up to 3 or 4 diopters. The 2 population that we're looking for with these devices is 3 going to be higher. 4 I've actually had my endothelial cell counts 5 taken and I'm 4 diopters without my glasses. It's correct 6 what they're saying. It's very difficult to focus on that 7 green light. I can imagine if you're 8 diopters or 12 8 diopters. So I suspect in the populations we're actually 9 looking at, that's a very conservative estimate. 10 DR. BANDEEN-ROCHE: There was a number being 11 cited of 9 percent this morning. I mean, again, just 12 purely interpolating that would put us at about 7.5. I 13 mean, it's in-between the 5 and 10 percent on this table. 14 MR. CALOGERO: Okay. That's the KeraVision 15 number. 16 MS. LOCHNER: So I think what Don is saying is 17 that 9 percent figure came from the KeraVision, which puts 18 you in-between the 5 percent standard deviation and the 10 19 percent. Maybe it would be prudent to go up at least to 20 the 10 percent standard deviation because the population 21 these will be used in will be a much more difficult 22 population than the KeraVision. 23 DR. BANDEEN-ROCHE: If I could ask one more 24 question, these calculations, were they based on just a 25 three-year minus three-month difference? That's what was 216 1 being analyzed? 2 MR. CALOGERO: Yes, yes. 3 DR. BANDEEN-ROCHE: Because -- 4 MS. LOCHNER: No, they were repeated measures. 5 DR. BANDEEN-ROCHE: All four measurements? 6 MS. LOCHNER: Yes, repeated measures, not -- 7 MR. CALOGERO: Okay. As Ashley said, we 8 established linearity with the four measurements, but in 9 terms of this particular calculation -- 10 MS. THORNTON: Don, please speak into the 11 microphone. 12 MR. CALOGERO: In terms of this particular 13 calculation, it's I believe the three-month value, the 36- 14 month minus the three-month, and then you simply divide by 15 2.75. The actual method and equation is right in the 16 information that we provided to you. I simply used the 17 equation that's in that document. 18 DR. BANDEEN-ROCHE: Right, and so certainly I 19 would expect that some precision could be gained by using 20 all four measurements, rather than just the difference 21 between the last and the first, and so that would impact 22 this table. 23 Go ahead. Interject, interject. 24 DR. BRADLEY: This is Dr. Bradley. Could 25 somebody clear up for me, the 9 percent that we're talking 217 1 about from this morning, if I recall the presentation, was 2 the difference that would have to occur in a single eye to 3 confirm with 100 percent certainty that in fact a change 4 had occurred. Therefore, that was an estimate of the 5 overall range, not the standard deviation in that 6 distribution. Perhaps the speaker from this morning can 7 clarify that. 8 PARTICIPANT: I agree with what you just said. 9 DR. BRADLEY: But I think it's being treated 10 here as a standard deviation. 11 DR. BANDEEN-ROCHE: Well, let me just clarify. 12 So the overall range -- now, let me see if I read the 13 wrong thing off of your handout, but the way that I 14 understood it was two measurements, maximum minus minimum 15 over maximum? 16 DR. McCAREY: If you're referring to the 17 graph -- 18 DR. WEISS: Can you identify yourself first for 19 the transcript? 20 DR. McCAREY: My name is McCarey. If you're 21 referring to the 9 percent one, that was simply a 22 subtraction of baseline and three months for each 23 individual. 24 DR. BANDEEN-ROCHE: Right, but that's an 25 absolute difference. 218 1 DR. McCAREY: Yes. 2 DR. BANDEEN-ROCHE: Yes, and so you can 3 approximate an absolute difference by the square root of 4 the squared difference, and so in turn -- I admit there are 5 multiple approximations here, but it's not a bad 6 approximation. The square root of the square, then 7 expectation of the square is a variance, and that's how 8 that enters in. 9 Yes, but I agree. It's worth doing this more 10 carefully than on my thumbnail. 11 DR. WEISS: So I think we could actually -- I 12 think you've given us the data to look at and try to 13 balance what we're willing to detect as a yearly loss 14 versus what we're willing to balance against as a maximal 15 amount of endothelial cell loss, and then we can choose the 16 numbers we want. 17 I would ask Dr. Grimmett if this is basically 18 and opinion-type thing at this point, but that's basically 19 all you want right now. So do you have an opinion as far 20 as what you would wish for a yearly loss and an allowable 21 rate? 22 DR. GRIMMETT: Sure, but I'm taking into 23 account that some of these numbers have largely varied. 24 For example, in the .9 category of the study of 669 25 patients, to have good accountability over three years is 219 1 pretty incredible, and which I don't think is really 2 achievable or reasonable. 3 Keeping that in mind, the higher numbers I 4 guess at this point are 1.9, 2 percent loss. I would be 5 extremely disappointed and worried if a phakic IOL actually 6 achieved that rate. I think it would indicate that 7 patients would actually develop corneal edema during their 8 lifetime, especially if they need cataract surgery. I 9 would hope that they'd have a lower rate of cell loss. 10 What would I like to detect versus what is 11 reasonable? Based on the data here, I suppose if we could 12 be at the worst, assure it's not higher than 1.5. I'd 13 still hope it's a little lower than that. I think a 2 14 percent threshold is too high based on the actuarial tables 15 that I ran. 16 Even for some of these lower numbers, even the 17 1 percent, if they have a 250 sample size -- 244 in this 18 example with a 10 percent standard deviation -- you know, 19 they would be allowed to see a rate of .7 to be sure with 20 90 percent confidence is not higher than 1. You could 21 still determine other factors, just not with this much 22 precision. It's going to be much harder at the lower 23 rates, and then we admit the normal endothelial cell loss 24 rate is .6 percent per year or so. So we have to account 25 for that factor, and then there will be zero differential 220 1 between what he phakic IOL is actually doing versus normal 2 cell loss rate. The 1.5 is what I'm looking at. 3 DR. WEISS: So I think perhaps you could say 4 the 1.5 percent and allowable rate being -- 5 DR. GRIMMETT: Yes, just straight off the 6 table. I mean, once we set the sample size, it's going to 7 lock this in to what their allowable rate is to be sure of 8 a 90 percent confidence is not higher than our threshold. 9 Given the difference -- for example, let's look 10 at the 1.5 percent category. Given the difference between 11 the smallest number, the 243 sample size, and the unwieldy 12 542 independent patients over three years, that's a huge 13 number and it would cost probably a fortune to even try to 14 do it. 15 So I'm still, I think based on statistics and 16 -- I see I was looking at the 15 percent standard 17 deviation. But looking at the statistics and stuff, I 18 think that the sample size that we're actually asking for 19 is somewhere in the neighborhood of 250 or so. That's what 20 it looks like on this table. Whatever the number happens 21 to be, but I think asking for higher precision than that is 22 not reasonable. 23 DR. WEISS: So I think from what I understand 24 you're saying, yearly loss would somewhat be dependent on 25 the fact that most -- I would also agree. You don't want 221 1 to ask for than 250 to 300 patients. So that already locks 2 us into what we want our yearly loss to be. 3 DR. GRIMMETT: My hope is that with the 4 precision and the careful techniques that Dr. Edelhauser 5 described, if they can actually be implemented with care, 6 is that by lowering the true standard deviation, we'll have 7 much better precision than we would want, and that's got to 8 be hopeful. Controlling technicians is so important to 9 lower that standard deviation to give the power of the 10 study better precision. 11 DR. WEISS: Dr. Mathers? 12 DR. MATHERS: And our precision is going to 13 improve with time because as we monitor afterwards, 14 presuming that is the case, then monitoring for a longer 15 period of time improves our data on the loss rate. It's 16 not part of this table, but this doesn't get worse over 17 time. It gets better if you continue to monitor. 18 DR. WEISS: Dr. Bandeen-Roche? 19 DR. BANDEEN-ROCHE: Yes, I would just like to 20 bring up a little something about the safety and 21 effectiveness precision given a sample size of 300. This 22 was Attachment A, Section A.1, and by my calculation -- you 23 know, of course, zero events is the least that you can have 24 -- with a sample size of 300, that gave a 95 percent upper 25 confidence bound of .01. 222 1 Now, so that's a 1 percent, say, adverse event 2 rate, and I would just submit that for the panel's 3 consideration. I don't think that that can be argued as 4 meeting the .001 standard that was cited in the attachment 5 in the way that I feel is honest and candid. 6 DR. WEISS: What sample size would allow you 7 that rate? 8 DR. BANDEEN-ROCHE: Well, unfortunately, it's 9 very large. 10 DR. WEISS: Well, what is very large? 11 DR. BANDEEN-ROCHE: Three-thousand. 12 DR. WEISS: So in other words, we have to 13 change the rate. We might want that rate, but none of us 14 believe that a study with 3,000 patients can be done. 15 DR. BANDEEN-ROCHE: That's right, but maybe it 16 just supports the importance of postmarketing data. 17 DR. WEISS: Okay. So it supports our concern 18 for stringency. 19 Dr. Bullimore? 20 DR. BULLIMORE: And one of the continuing 21 limitations of the data we consider is we're presented, 22 bombarded, with event rates and, give complication rates or 23 adverse event rates, we choose to ignore the confidence 24 intervals or we're not presented with the confidence 25 intervals that you give you an indication of the precision 223 1 of those estimates, and if you really truly want to ensure 2 that the event rate is, say, less than 1 percent, you would 3 have to do as Dr. Bandeen-Roche suggested, enroll 4 considerably more patients, as was done, say, in recent 5 continuous wear contact lens studies. 6 We choose to ignore information, we sort of try 7 and meet targets, and we keep in the back of our minds 8 often what the precision of the estimate might be, but it's 9 not something we consider on a regular basis, and maybe we 10 should, but I'm not sure that we'd like the answer that 11 we'd get if we were presented with those on a regular 12 basis. 13 DR. WEISS: Dr. Bandeen-Roche? 14 DR. BANDEEN-ROCHE: Well, just stating it 15 another way, I mean, is the panel willing to live with 5 16 percent of studies claiming an event rate of .001 or less 17 when in fact it's higher than 1 percent? I mean, that's 18 the ramification. 19 DR. WEISS: I think the difficulty is in the 20 real world, if we required the number of patients we would 21 like to get the answer, it would take so many years by that 22 point the technology would be archaic. 23 Mr. McCarley? 24 MR. McCARLEY: Just one comment. There is 25 always an ongoing postmarket surveillance on products. 224 1 Every year we have an annual report in all products, and 2 especially implants, where we essentially divide the number 3 of adverse events we've had by the number of implants that 4 have taken place. So if we saw any increase in it, the FDA 5 would immediately take action or we'd have to justify why 6 that would be. 7 So I agree for the purpose of making an initial 8 decision for a PMA, you might not have all the information, 9 but you certainly have the mechanism in place to continue 10 to monitor any higher rates. 11 DR. WEISS: Dr. Grimmett? 12 DR. GRIMMETT: Dr. Grimmett. I would counter 13 that by saying that postapproval, there is probably 14 significant underreporting of adverse events. 15 DR. MATOBA: We're going to collect data on 16 both eyes, right? So for events, specific events, that 17 would become available to the FDA, wouldn't it? On twice 18 as many eyes potentially as 300? 19 MS. LOCHNER: Right, but the statistical 20 assumptions that, for example, would be inherent in this 21 table would then have to be adjusted. 22 DR. MATOBA: From a practical point of view, 23 but in terms of missing something terrible, it's not as bad 24 as she says. 25 MS. LOCHNER: I think from a practical 225 1 standpoint, I hear you. I mean, you will have more eyes 2 from a practical standpoint. 3 But I think the issue with phakic IOLs isn't 4 missing something catastrophic early on, but missing a slow 5 bleed that's occurring over time and approving it without 6 understanding that the rates could be higher. 7 DR. WEISS: Dr. Bullimore? 8 DR. BULLIMORE: Reasonable assurance of safety. 9 That's what we're asked for. 10 DR. WEISS: I guess that's the difference 11 between the 300 and the 3,000. 12 DR. BULLIMORE: Exactly. 13 I have one other issue on the endothelial cell 14 count which I've hinted at before and I'll come back to. 15 When these data are presented, I think it will be 16 appropriate not only to have the mean rate of loss, whether 17 you give that annually, but I think over a three-period, 18 knowing the proportion of eyes that have lost 10 percent, 19 20 percent, and 30 percent of endothelial cells -- I mean, 20 I'm sure a reviewer's going to ask for that information, 21 but prospectively it should be at the forefront of the 22 analyses. 23 DR. WEISS: I wanted to find out if the agency 24 had any other questions for the panel at this point. 25 MS. LOCHNER: No, just if there are any other 226 1 comments on any other sections of the guidance. 2 One of the things that I think I took away from 3 the earlier discussion on contrast sensitivity is that we 4 may need to provide to vision scientists some of the data 5 upon which we came to this conclusion about contrast 6 sensitivity, and so we may follow up with a homework 7 assignment to look at that because it's possible, first of 8 all, that we're misinterpreting what we're looking at, and 9 so we took those contrast acuity comments especially to 10 heart if we are in fact doing that. 11 DR. WEISS: Dr. Matoba had a comment. 12 DR. MATOBA: I had a question about the 13 guidance. Number 5, study population. This is phakic IOLs 14 for myopes, specified minimum uncorrected visual acuity 15 20/40 or worse, meaning you could have myopia uncorrected 16 visual acuity of 20/40 and then be eligible to get into the 17 myopic phakic IOL study? Twenty/forty doesn't seem 18 compatible with high myopia. 19 DR. EYDELMAN: Dr. Eydelman. This is for all 20 phakic IOLs. As Donna has mentioned previously, current 21 studies are not limited to high myopia. So we have phakic 22 IOLs for -2 and -3. 23 DR. WEISS: So would any members -- and I'm 24 going to regret asking this question. 25 (Laughter.) 227 1 DR. WEISS: Briefly, would any members of the 2 panel -- or actually, even more importantly, does the FDA 3 care whether the panel wants it to be 20/40 or not or it's 4 irrelevant? 5 MS. LOCHNER: We care. 6 DR. WEISS: You care. That's too bad. 7 (Laughter.) 8 DR. WEISS: So do any members of the panel have 9 any disagreement with doing a phakic IOL for someone who's 10 20/40? 11 DR. BULLIMORE: I'm having a senior moment. 12 You were talking about excluding patients with entering 13 visual acuity of worse than 20/40? 14 DR. WEISS: Twenty/forty uncorrected. It's 15 uncorrected visual acuity of 20/40. I want my 20/40 -- 16 DR. MATOBA: Would make you eligible to get in 17 the study. 18 DR. WEISS: Would make you eligible to have a 19 phakic IOL at this point. 20 DR. GRIMMETT: This is Dr. Grimmett. You're 21 using the 20/40 as a marker for your refractive error. 22 DR. WEISS: It's about a -1, isn't it? 23 DR. GRIMMETT: Yes. You're really asking the 24 question should patients with low myopic or low refractive 25 errors be entered into trials that have significant risks 228 1 that we've discussed today of cataracts, endothelial cell 2 loss, pigment dispersion, glaucoma, et cetera? 3 DR. WEISS: And at the present time, they are 4 being entered into this. 5 Dr. Mathers? 6 DR. MATHERS: I think they should not be 7 entered into this study. We should have a cutoff that is 8 much higher than that for patients to enter the study. 9 DR. WEISS: Okay. So what would your cutoff 10 be? 11 DR. MATHERS: Minus 8. 12 DR. WEISS: That's pretty high. 13 DR. MATHERS: Maybe -6. I mean, -6 is very 14 treatable with most LASIK procedures. 15 DR. WEISS: So you would come down to a -6. 16 DR. MATHERS: Yes. 17 DR. WEISS: Dr. Swanson, do you have an opinion 18 on this? 19 DR. SWANSON: Well, I have an opinion on most 20 things, but I agree that we're talking about something that 21 has -- we want to determine what the risks are, so it makes 22 sense to look at the population that's supposedly to be 23 served by this risky procedure. 24 DR. BULLIMORE: I have a question related to 25 the question. I think if we start prefacing entry criteria 229 1 and say, well, this population can be adequately served by 2 other technology, we're actually entering a very dangerous 3 bias zone. 4 A question for the folks who do this kind of 5 thing. In terms of the safety of the device, are there any 6 a priori reasons why endothelial cell count, contrast 7 sensitivity loss, and lens opacifications would expect to 8 be greater in a low myope compared to a high myope or vice 9 versa? 10 DR. WEISS: I don't think they would be, but I 11 think the concern is why make the cutoff at 20/40? Why not 12 do it at 20/25? 13 DR. BULLIMORE: Yes. Well, I agree that -1 is 14 perhaps a little too conservative, but I don't think we 15 should say, well, we approved LASIK up to -6. That should 16 be our cutoff. 17 DR. WEISS: You know what? I think what you're 18 hearing, and obviously this discussion could go on for a 19 while, but I think some members of the panel have a concern 20 that the low myopes, the risk/benefit ratio might not be 21 the same as in the high myopes, and where you would draw 22 that line would be up to discussion. Perhaps it would it 23 be appropriate for these IDEs to first do a higher group of 24 myopes, and when there is proven to be some sort of 25 clinical safety and efficacy, then expand the trial to the 230 1 lower myopes. 2 Dr. Eydelman? 3 DR. EYDELMAN: Malvina Eydelman. That is 4 exactly what I was trying to make a point of, that we 5 usually allow brand new phakic IOLs only in the higher 6 degrees of myopia, and once the sponsor obtains enough 7 safety information on the high myopes and submits it to 8 FDA, then internally we review it and decide that is 9 sufficient, and we allow lower ranges. Again, depending on 10 safe we assess it to be, that's the degree of myopia that 11 we allow it to go down to. 12 DR. WEISS: Mr. McCarley? 13 MR. McCARLEY: Just very quickly, I agree with 14 it. I think that it's prudent to study higher myopes, 15 develop a level of confidence and safety, and then move 16 down, but I would ask I guess a question about LASIK, 17 another refractive technology that apparently is now safe 18 and effective, though from what we heard yesterday morning 19 or at the beginning of this session, it may not be 20 completely true when you have large numbers of patients. 21 Aren't there lasers approved right now for -15, 22 for instance? I think so. 23 DR. WEISS: There are, but I don't think 24 they're being used for it. 25 MR. McCARLEY: They're approved for it. That's 231 1 what I'm saying. So it's sort of a double standard and I 2 agree we all think of phakic intraocular lenses as treating 3 high myopia, and in fact, if you look at the means of the 4 data that's presented at the American Academy of 5 Ophthalmology and ASCRS, you'll see that that's up around 6 the 12, 13. 7 But in fact, this may be a replacement 8 technology. There may be benefits we don't know over 9 LASIK. 10 DR. WEISS: Dr. Swanson? 11 DR. SWANSON: Good. Thanks. I've been 12 promoted. 13 Well, I think the one question to consider 14 there is, in terms of effectiveness, one of the 15 effectiveness criteria is percentage of eyes that achieve 16 uncorrected visual acuity of 20/40 or better. So if there 17 are a lot of people enrolled that are just worse than -- 18 that are 20/50, that effectiveness is not going to mean as 19 much. So that's something in terms of study design. The 20 safety may not be different across eyes, but the 21 effectiveness should be considered. 22 DR. WEISS: Does the agency have any other 23 questions? 24 (No response.) 25 DR. WEISS: I want to thank the panel and the 232 1 presenters and the agency for all their work and excellent 2 preparation, and Sally will have some closing comments 3 before we end the meeting. 4 MS. THORNTON: I, too, would like to add my 5 thanks to the panel, and to Drs. Werner, Edelhauser, and 6 McCarey for being with us today. It's been quite a 7 contribution you've given to our proceedings, and I thank 8 the panel for all their hard work for yesterday as well. 9 I will be letting you know about mid-September 10 what the story is for the November 14-15 tentative panel 11 meeting schedule. So stay in touch with your website. 12 DR. WEISS: The meeting is closed. 13 (Whereupon, at 2:03 p.m., the meeting was 14 adjourned.) 15 16 17 18 19 20 21 22 23 24 25 233 1 2