FDG PET CT to Select Patients with Peritoneal ResearchGate

Document Sample
FDG PET CT to Select Patients with Peritoneal ResearchGate Powered By Docstoc
					Ann Surg Oncol
DOI 10.1245/s10434-009-0387-7


  F-FDG-PET/CT to Select Patients with Peritoneal
Carcinomatosis for Cytoreductive Surgery and Hyperthermic
Intraperitoneal Chemotherapy
                                                                                    ¨ ¨ ¨z,
Christina Pfannenberg, MD2, Ingmar Konigsrainer, MD1, Philip Aschoff, MD2, Mehmet O. Oksu MD2,
                 1                    1               3                  3
Derek Zieker, MD , Stefan Beckert, MD , Stephan Symons , Kay Nieselt, PhD , Jorg Glatzle, MD1,
Claus V. Weyhern, MD4, Bjorn L. Brucher, MD1, Claus D. Claussen, MD2, and Alfred Konigsrainer, MD1
                          ¨         ¨                                               ¨

 Department of General, Visceral and Transplant Surgery, University of Tuebingen, Tubingen, Germany; 2Department of
Radiology and Nuclear Medicine, University of Tuebingen, Tubingen, Germany; 3Center for Bioinformatics Tuebingen,
University of Tuebingen, Tubingen, Germany; 4Department of Pathology, University of Tuebingen, Tubingen, Germany
                          ¨                                                                       ¨

ABSTRACT                                                             Results. There was a strong correlation between the PCI
Background. Cytoreductive surgery followed by hyper-                 obtained with PET/CT and the surgical PCI with respect to
thermic intraperitoneal chemotherapy (HIPEC) is                      the total score (r = 0.951) as well as in the regional
associated with significantly longer survival in patients             analysis (small bowel, r = 0.838; other, r = 0.703). The
with peritoneal carcinomatosis (PC). So far, no morpho-              correlation was slightly lower for CT alone (total score,
logical imaging method has proven to accurately assess the           r = 0.919; small bowel, r = 0.754; other, r = 0.666) and
intra-abdominal tumor spread. This study was designed to             significantly lower (p = 0.002) for PET alone (total score,
predict tumor load in patients with PC using dual-modality           r = 0.793; small bowel, r = 0.553, other, 0.507).
   FDG-PET/CT and to compare the results with those of               Conclusions. Contrast-enhanced CT is superior compared
PET and CT alone by correlating imaging findings with                 with PET alone to predict the extent of PC. In our patient
intraoperative staging.                                              group, the combination of both modalities (contrast
Methods. Twenty-two patients with PC from gastrointes-               enhanced PET/CT) yielded the best results and proved to
tinal (n = 13), ovarian cancer (n = 8), and mesothelioma             be a useful tool for selecting candidates for peritonectomy
(n = 1) underwent contrast-enhanced 18FDG-PET/CT                     and HIPEC.
before surgery and HIPEC. In a retrospective analysis PET,
CT, and fused PET/CT were separately and blindly
reviewed for the extent of peritoneal involvement using the             Peritoneal carcinomatosis (PC) from gastrointestinal and
Peritoneal Cancer Index (PCI). Imaging results were cor-             gynecological cancer as well as primary tumors of the peri-
related with the intraoperative PCI using Pearson’s                  toneum have always been classified as extended tumor
correlation coefficient and linear regression analysis.               disease with no more than palliative treatment options. A
                                                                     change in paradigm has occurred since Sugarbaker demon-
                                                                     strated good long-term survival after peritonectomy with
                                                                     hyperthermic intraoperative chemotherapy (HIPEC).1
                                                                     Recent multicenter phase II studies and a single phase III
Christina Pfannenberg and Ingmar Konigsrainer equally contributing
authors.                                                             study that evaluated the usefulness of peritonectomy and
                                                                     perioperative intraperitoneal chemotherapy have been
Statement: This manuscript contains original material that has not   promising.2,3
been previously published or submitted to another journal.              The extent of peritoneal involvement and adequacy of
                                                                     cytoreduction are the most significant prognostic factors.4
Ó Society of Surgical Oncology 2009                                  Patients with a high PC Index or extensive small-bowel
First Received: 1 August 2008                                        involvement have a poorer prognosis, mainly because
                                                                     complete cytoreduction cannot be achieved. Therefore,
A. Konigsrainer, MD
e-mail: alfred.koenigsrainer@med.uni-tuebingen.de                    patient selection is greatly important to optimize results and
                                                                                                            C. Pfannenberg et al.

to avoid explorative laparotomy in patients who will not        TABLE 1 Clinicopathological characteristics of 22 patients with
benefit from this invasive strategy. To date, the ‘‘gold stan-   peritoneal carcinomatosis scheduled for cytoreductive surgery and
                                                                intraperitoneal chemotherapy
dard’’ for classification of PC is surgical exploration.
Laparoscopic staging of PC is rarely performed because          Characteristics                                   No. of patients
adhesions from previous operations might hamper further
                                                                Male/female ratio                                 8/14
surgical procedures in the majority of patients. So far, no
                                                                Mean age (yr)                                     58
reliable noninvasive parameter has been proven to ade-
                                                                Range                                             37–75
quately select candidates for potentially complete
cytoreductive surgery. Various imaging modalities have
                                                                   Colorectal carcinoma                           7
been analyzed for the detection and preoperative evaluation
                                                                   Ovarian carcinoma                              7
of patients suffering from PC, including CT, MRI, and
                                                                   Pseudomyxoma                                   4
recently PET and PET/CT in small series.5–9 At present,
                                                                   Appendix carcinoma                             3
contrast-enhanced CT is the modality of choice for staging
                                                                    Mesothelioma                                  1
and restaging patients with suspected PC, although CT has
limited accuracy in detecting small peritoneal lesions and      Mucinous
involvement of bowel/mesentery. Positron emission                  Yes                                            9
tomography (PET) with the tracer fluorodeoxyglucose (18F-           No                                             13
FDG) is a functional modality that has an increasingly          Postoperative CC score
important role in the diagnosis, staging, and treatment            0                                              13
monitoring of malignant diseases.10 The combination of             1                                              1
functional PET data and detailed anatomical information            2                                              2
provided by CT in dual-modality PET/CT further improves            3                                              3
staging accuracy and has proven to be superior to standalone    HIPEC                                             19
CT and PET imaging.11                                           Palliative operation                              3
   This study was designed to assess the preoperative           HIPEC hyperthermic intraoperative chemotherapy
prediction of tumor load in patients with PC using dual-
modality 18FDG-PET/CT (= PET/CT) and to compare the
results of PET/CT with those for PET and CT evaluated              The extent of peritoneal carcinomatosis was assessed
individually by correlating imaging findings with intraop-       according to the Peritoneal Cancer Index (PCI) reported by
erative results (Peritoneal Cancer Index).                      Sugarbaker at the time of surgical exploration.4,13 All
                                                                positive findings were documented by histology of each
PATIENTS AND METHODS                                            resected specimen. After cytoreductive surgery, residual
                                                                tumors were recorded using the CC Index by Sugarbaker
   From July 2005 to February 2008, 22 patients with            (CC = complete cytoreduction). Patients with no residual
peritoneal carcinomatosis and absence of extra-abdominal        tumor (CC-0) and with residual tumor nodules \0.25 cm
metastases scheduled for cytoreductive surgery and HIPEC        (CC-1) were scored as having achieved adequate cytore-
were enrolled in this analysis (8 men; mean age, 58 (range,     duction, whereas patients with residual tumors [0.25 cm
37–75) years). All patients had been operated on before for     (CC-2) or [2.5 cm (CC-3) were regarded as having
their primary tumor; 13 patients suffered from gastroin-        achieved suboptimal cytoreduction.
testinal cancer, 8 patients had ovarian cancer, and 1 had
malignant mesothelioma. Histologies revealed 9 mucinous         Peritoneal Cancer Index
and 13 nonmucinous tumors. No patient was excluded from
surgical exploration because of imaging results. All               The PCI combines tumor distribution with a lesion size
patients underwent dual-modality contrast-enhanced PET/         score (LSS) and divides the abdomen/pelvis into 13
CT for clinical reasons before surgery, both performed in       regions: central, right upper, epigastrium, left upper, left
our hospital. The median interval between PET/CT and            flank, left lower, pelvis, right lower, right flank, upper
surgery was 13.7 (range, 2–40) days; patients were exclu-       jejunum, lower jejunum, upper ileum, and lower ileum
ded from evaluation if the interval exceeded 40 days. All       (Fig. 1). For each region the maximum size of tumor
patients were operated on by the same surgeon following         deposits was categorized into four groups: LSS-0, no
the Sugarbaker strategy. The details of the surgical tech-      detectable disease; LSS-1, tumor size \0.5 cm; LSS-2,
niques are described elsewhere.12 Informed consent for          tumor size C 0.5 cm but B 5 cm; LSS-3, tumor
PET/CT and surgical exploration was obtained from each          size [5 cm. The lesion sizes were summed for all regions,
patient. Patient characteristics are summarized in Table 1.     giving a numeric score from 0 to 39 for each patient.
PET/CT in Peritoneal Carcinomatosis

                                                                 reconstructed slice thickness/increment of 5/5 mm (axial)
                                                                 and 3/2 mm (coronal). Patients were positioned on the
                                                                 scanning table with their arms raised to reduce beam-
                                                                 hardening artefacts. To obtain diagnostic CT data, in all
                                                                 patients a multi-phase CT protocol with intravenous
                                                                 administration of 120 ml iodinated contrast agent (Ultravist
                                                                 370, Schering, Berlin, Germany) at a flow of 2–3 ml per
                                                                 second was performed. The PET/CT examination protocol
                                                                 started with a nonenhanced low-dose CT scan for attenua-
                                                                 tion correction, which was followed by a bolus-triggered
                                                                 arterial phase thorax/liver scan, portal-venous abdomen/
                                                                 pelvis scan, and whole-body PET. To prevent contrast-
                                                                 induced artifacts, we optimized the injection protocol with a
                                                                 40-ml saline chaser. Patients were asked to stop breathing in
                                                                 normal expiration during the contrast-enhanced CT scans
                                                                 for optimal coregistration of PET and CT.
FIG. 1 Abdominopelvic regions 0–12 according to the Peritoneal
Cancer Index (PCI)                                               Image Reconstruction

PET/CT IMAGING                                                      Diagnostic CT scans were reconstructed with 3-mm
                                                                 slice thickness in the coronal and 5-mm in the axial plane
Patient Preparation                                              (reconstruction increment 2-mm and 5-mm, respectively).
                                                                 In addition, postprocessing reconstructions (MIP, maxi-
   All patients fasted overnight (at least 6 hours) before       mum intensity projection; high-resolution reconstruction)
intravenous administration of the radiotracer 18F-FDG. The       were performed. The nonenhanced CT data were used for
injected dose of 18F-FDG varied between 350 and                  attenuation correction of PET emission images. PET ima-
450 MBq depending on the patient weight. Blood glucose           ges were reconstructed using an iterative algorithm
was measured before injection of the tracer to ensure blood      (ordered-subset expectation maximization: 2 iterations, 8
glucose levels \150 mg/dl. During the uptake phase of 55–        subsets). The reconstructed PET, CT, and fused images
65 minutes, the patients were instructed to rest comfort-        were displayed with commercially available software in
ably. All patients were asked to drink 1,000–1,500 ml of         axial, coronal, and sagittal planes using a matrix of 128 x
mannitol 2% as a negative oral contrast agent before             128 pixels for the PET and 512 x 512 pixels for the CT.
scanning to distend the bowel. Bowel preparation was
supplemented by rectal administration of 300–500 ml of           Image Analysis
water and intravenous administration of a spasmolytic drug
(BuscopanÒ, Boehringer-Ingelheim, Germany) immedi-                  PET-only and CT-only image volumes were evaluated
ately before the CT scan.                                        retrospectively and independently by a radiologist (CP) and
                                                                 a nuclear medicine physician (PA), both having more than
PET/CT Imaging Protocol                                          10 years of experience in oncology imaging. The two
                                                                 readers were aware that all patients suffered from perito-
    PET/CT imaging started 55 to 65 minutes after admin-         neal carcinomatosis but were blinded to surgical results and
istration of 18F-FDG. The examinations were performed            the findings of the other imaging studies. In a second
using the Hi-Rez Biograph 16 (Siemens Medical Solutions,         reading session after 6 to 8 weeks, fused PET/CT images
Knoxville, TN), consisting of a high-resolution three-           were evaluated by the same reader team in consensus on a
dimensional LSO PET and a state-of-the-art 16-row multi-         lesion-by-lesion basis. In CT, PET, and PET/CT the size
detector CT (MDCT). The PET scanner has an axial field of         and distribution of intraperitoneal tumor implants were
view (FOV) of 15.5 cm and a slice thickness of 4.25 mm.          recorded in 13 abdominopelvic regions and categorized in
Emission data were acquired for 6–8 bed positions, typi-         four size-related groups according to the above-described
cally from the base of the skull to the upper thigh with 3-      standardized scoring system (PCI) used for intraoperative
minute acquisition time per bed position. CT was operated        documentation of peritoneal tumor deposits. All images
at a peak voltage of 120 kVp, tube current of 120–250 mAs,       were evaluated digitally on a multimodality workstation
rotation time of 0.5 seconds, collimation of 0.75 mm (tho-       connected to a PACS workstation and reviewed in three
rax) and 1.5 mm (abdomen), table feed of 12/24 mm, and           orthogonal planes (axial, coronal, sagittal).
                                                                                                          C. Pfannenberg et al.

   Before starting the retrospective data analysis, the         analysis was performed for the small bowel (region 9–12)
radiologist and nuclear medicine specialist practiced doc-      in addition to overall assessment.
umentation of imaging findings according to the PCI                 The PCI difference (e.g., individual PCIPET minus
together with the peritoneal surgeon for consistency of         individual PCIsurgery) and the absolute PCI difference (i.e.,
scoring based on data sets independent of this study.           absolute value of the PCI-difference, also denoted PCI
                                                                error) between PET, CT, PET/CT results, and surgical
FDG-PET                                                         findings were calculated and averaged to obtain the mean
                                                                PCI-difference and the mean absolute PCI-difference.
     F-FDG distribution was evaluated visually and semi-        Differences were tested for significance using the t test, and
quantitatively using the standard uptake value (SUV).           p values were corrected using Holm’s method.14 A cor-
Peritoneal tumor was suspected on FDG-PET if the fol-           rected p value of \0.05 was considered statistically
lowing tracer abnormalities were identified: (1) irregularly     significant. Furthermore, for comparison of predictions,
distributed focal uptake exceeding normal regional tracer       Pearson’s correlation coefficient was calculated. Finally,
accumulation and located randomly within the abdomen,           linear regression was used to predict the intraoperative
unrelated to solid viscera or nodal stations, (2) curvilinear   tumor load. For accuracy calculation, results were sum-
uptake along the liver or in the left subphrenic space, and     marized: true-positive (TP) means a lesion rated as
(3) diffuse low-grade uptake spreading uniformly                suggestive of peritoneal tumor was confirmed as positive
throughout the abdomen and pelvis obscuring visceral            by the intraoperative result and histology; false-negative
outlines. For assessment of lesion size in PET, we used an      (FN) occurred when one of the modalities failed to detect
automatically generated three-dimensional isocontour at         the peritoneal carcinomatosis or when a lesion was falsely
50% of the maximum FDG-uptake as lesion border                  classified as benign but the lesion was found to be perito-
(FWHM = full width at half maximum).                            neal carcinomatosis at histology; false-positive (FP) means
                                                                a modality classified a lesion as suggestive of peritoneal
                                                                carcinomatosis, but the lesion was negative on histology;
Computed Tomography
                                                                true-negative (TN) means a lesion/region was rated nega-
                                                                tive for peritoneal carcinomatosis (LSS 0) by a modality
   CT scans were suggestive of peritoneal carcinomatosis
                                                                and also was classified as negative by the intraoperative
if they demonstrated (1) a nodular, plaque-like, or infil-
                                                                and histological results. From these values, sensitivity,
trative soft-tissue lesion in the peritoneal fat or on the
                                                                specificity, positive predictive value (PPV), and negative
peritoneal surface with variable density and enhancement
                                                                predictive value (NPV) were calculated. All statistical
pattern, (2) a smooth or nodular parietal peritoneal thick-
                                                                analyses were performed using the statistical software
ening or enhancement, or (3) a thickening, distortion, or
                                                                package R version
obstruction of small bowel and mesenteric root.

   PET/CT results were interpreted as positive for perito-
                                                                   In 19 of 22 patients, cytoreductive surgery with HIPEC
neal tumor under the following conditions: (1) areas with
                                                                was completed. In three patients only an explorative lap-
increased FDG uptake corresponding to a CT abnormality
                                                                arotomy was performed because of diffuse intestinal tumor
and not related to physiologic uptake, (2) CT abnormalities
                                                                involvement. One patient underwent surgery and HIPEC
with strong suspicion of tumor even in absence of a cor-
                                                                primarily for malignant ascites and an intraoperative PCI of
responding FDG uptake, (3) abnormal FDG uptake with
                                                                2 was confirmed. Partial peritonectomy was subsequently
strong suspicion of tumor even in absence of a corre-
                                                                performed. The postoperative CC score was CC-0 in 13
sponding CT abnormality. For the assessment of lesion size
                                                                patients, CC-1 in one patient, CC-2 in two patients, and
on PET/CT scans, the CT score was used where applicable.
                                                                CC-3 in three patients. The follow-up period was 5 to 36
                                                                (mean, 12.4) months.
Statistical Analysis                                               Mean PCI was 18.6 ± 11.6 (range 2–39) for surgery,
                                                                11.8 ± 7.9 (range 1–33) for FDG-PET, 18.2 ± 11.3 (range
   Intraoperative and histological findings were regarded as     0–39) for CT, and 18.5 ± 11.4 (range 0–39) for PET/CT.
the ‘‘gold standard’’ for presence, size, and localization of   PCI details relating to region 0–8 (overall abdomen/pelvis)
peritoneal tumor deposits. Because involvement of small         and region 9–12 (small bowel and mesentery) are sum-
bowel and its mesentery is greatly important for surgical       marized in Table 2. Tumor involvement of the small bowel
and treatment outcome of PC patients, the statistical           was found in 17 patients (77%) during surgery, in 9
PET/CT in Peritoneal Carcinomatosis

TABLE 2 Mean PCI values for surgery, PET, CT, and PET/CT in               (p = 0.002). The error was smaller in PET/CT than in CT,
total and for the subgroups region 0–8 and 9–12                           but this difference was not statistically significant
           Total                 Region 0–8a           Region 9–12b       (p = 0.17). This was true for total abdominopelvic
                                                                          assessment as well as for small-bowel involvement
Surgery 18.6 ± 11.6 (2–39) 14.7 ± 8.2 (1–27) 4.1 ± 4.2 (0–12)
                                                                          (Fig. 2). There was a significantly greater difference
PET          11.8 ± 7.9 (1–33) 10.2 ± 6.2 (0–21) 1.6 ± 2.9 (0–12)         between surgical PCI and PET alone (p \ 0.01). In 18 of
CT         18.2 ± 11.3 (0–39) 14.3 ± 8.1 (0–27) 3.9 ± 4.1 (0–12)          22 patients (82%) the extent of peritoneal carcinomatosis
PET/CT 18.5 ± 11.4 (0–39) 14.9 ± 8.3 (0–27) 3.6 ± 3.9 (0–12)              was underestimated by PET. When comparing the differ-
Data are presented as mean ± standard deviation                           ences between preoperative and intraoperative PCI in
    Region 0–8, overall abdominopelvic region                             relation to tumor histology, all imaging methods show
    Region 9–12, small bowel and mesentery                                greater discrepancies in mucinous tumors with the highest
                                                                          error rate in PET alone. The mean absolute PCI differences
patients (40%) on FDG-PET, in 15 patients (68%) on CT,                    for PET, CT, and PET/CT were 12.3, 3.9, and 3.3,
and in 14 patients (63%) on PET/CT. Figure 2 illustrates                  respectively, for mucinous tumors and 3.9, 3.1, and 2.1,
the errors in PCI determination by different imaging                      respectively, for nonmucinous lesions.
methods. The error (mean absolute difference) was larger                     In the Pearson’s correlation analysis, we found a strong
in PET compared with CT (p = 0.008) and PET/CT                            and statistically significant correlation between the results
                                                                          for PCI-PET/CT and for PCI surgery (reference standard)
    Error (Mean
    Absolute Difference)                                                  in overall assessment (r = 0.951) as well as in regional
    8                                                        PET          analysis (region 0–8, r = 0.703; region 9–12, r = 0.838).
                                                             CT           Correlation was lower for CT (total, r = 0.919; region 0–8,
                                                                          r = 0.666; region 9–12, r = 0.754) and PET (total,
                                                                          r = 0.793; region 0–8, r = 0.507; region 9–12,
                                                                          r = 0.553). With respect to histological type of tumor,
    4                                                                     correlation was weakest in pseudomyxoma for PET and
                                                                          best for all three methods in colorectal tumors.
    2                                                                        In the prediction of intraoperative tumor load, evaluated
                                                                          by linear regression analysis, PET/CT yielded the best
                                                                          results with a coefficient of determination of r2 = 0.904,
                Total           Region 0–8         Region 9–12
                                                                          and a much greater accuracy at higher tumor loads. CT
FIG. 2 Errors in Peritoneal Cancer Index (PCI) values (mean               (r2 = 0.845) and PET (r2 = 0.628) followed, both under-
absolute difference, i.e., the mean of absolute values of differences     estimating high tumor loads (Fig. 3). Sensitivity,
between surgery PCI and imaging PCI) for PET, CT, PET/CT versus           specificity, and positive and negative predictive values of
surgery. Errors for total PCI in PET/CT are consistently lower than
for CT (p = 0.1012) and significantly lower than for PET
                                                                          the different methods for diagnosis of peritoneal carcino-
(p = 2 * 10-5). For mean errors in region 0–8 (p = 0.239/                 matosis are summarized in Table 3. With regard to the
2 * 10-4) and 9–12 (p = 0.2048/0.0133), the same is observed              various regions, sensitivity was lowest in the region 9–12

           PCI -



                                 r2 = 0.63                                r2 = 0.85                              r2 = 0.9
                                 p = 10-5                                 p = 10-9                               p = 10-11

                0       10      20           30   40     0       10      20           30   40   0     10        20           30   40
                             PCI - PET                                PCI - CT                             PCI - PET/CT

FIG. 3 Correlation between surgical PCI and imaging PCI. For PET/CT, a low scattering, especially at higher PCI values, is apparent where CT
and PET show a higher deviation, leading to gross underestimation of the PCI. All correlations were highly significant (p B 10-5)
                                                                                                                         C. Pfannenberg et al.

TABLE 3 Sensitivity, specificity, positive predictive value, and negative predictive value of PET, CT, and PET/CT in detecting peritoneal
carcinomatosis in different regions
                Total                                           Region 0–8a                              Region 9–12b
                Sens      Spec       PPV        NPV             Sens     Spec        PPV       NPV       Sens     Spec        PPV        NPV

PET             63        89         93         52              68       83          93        45        46       95          91         62
CT              84        82         91         70              85       77          92        61        80       88          88         80
PET-CT          88        85         93         76              89       81          94        69        85       90          91         84
Data are presented as percentages (%)
Sens sensitivity, Spec specificity, PPV positive predictive value, NPV negative predictive value
    Region 0–8, overall abdominopelvic region
    Region 9–12, small bowel and mesentery

    Percent                                                                   chemotherapy offers better survival in patients with peri-
    Tumors Detected                                                           toneal carcinomatosis from gastrointestinal or ovarian
    100     PET                                                               cancer.3,16 However, treatment-related mortality rates of up
                                                                              to 13% and morbidity rates of more than 50% as well as the
     80     PET/CT
                                                                              high costs of this therapy warrant reliable preoperative
                                                                              prognostic indicators to identify the patients most likely to
                                                                              benefit from this strategy. Because completeness of cyto-
     40                                                                       reductive surgery as the most prognostically significant
                                                                              predictor is dependent on the extent and distribution of
     20                                                                       peritoneal involvement, its preoperative radiological
                                                                              assessment plays an important role in the patient selection
              < 0.5cm          0.5cm – 5cm            >5cm                    process.6,7,17 For all noninvasive methods, the diagnosis of
              (n = 69)           (n = 41)            (n = 87)                 peritoneal carcinomatosis is challenging due to the small
                               Tumor Size                                     size of peritoneal implants.
FIG. 4 Influence of lesion size on tumor detection rate (%) for PET,              Various imaging modalities have been used to detect
CT, and PET/CT in 22 patients with 197 peritoneal lesions                     peritoneal carcinomatosis, with CT being the state-of-the-
                                                                              art imaging modality.18–20 MRI is useful for detecting thin
(small bowel/mesentery) ranging between 85% (PET/CT)                          intraperitoneal tumor layers and has some advantages in
and 45% (PET). Tumor size strongly influenced the tumor                        differentiating tumor from intestine, but in general MRI
detection rate in all three methods, with lowest values                       offers no advantage compared with CT.18,21,22 Reported
when tumor implants were \0.5 cm (49%–72%). The                               sensitivities of CT in detecting peritoneal carcinomatosis
detection rate increased to 79%–99% when tumor size                           vary from 25–90% depending on the size, site, and mor-
was [5 cm (Fig. 4).                                                           phology of tumor deposits and adequacy of bowel
   The three patients who underwent exploration only pre-                     opacification. Several investigators agree that CT is not
sented intraoperatively with a PCI of 39, 28, and 39,                         reliable for detecting low-volume tumor.5,6,19 De Bree et al.
respectively. The preoperative PCI evaluation by imaging                      reported a CT sensitivity of 59–67% for tumor
demonstrated a PCI-PET of 12, 16, and 24, a PCI-CT of 36,                     implants [5 cm, which decreased to 9–24% for tumor
30, and 39, and a PCI-PET/CT of 39, 33, and 39, respec-                       deposits \1 cm.6 Jacquet et al. described a sensitivity of
tively. PET underestimated the tumor load in all three                        90% for tumor diameter [5 cm, which decreased to 28%
patients, most likely due to the mucinous histology of the                    when tumor nodules were \0.5 cm.5 In addition to that,
primary tumors in these patients. PET/CT accurately asses-                    sensitivity varied depending on the particular anatomic
sed all PCI subscores in two patients. In the third patient, CT               region and was lowest in the pelvis. In the study by Coakley
and PET/CT slightly overestimated the small-bowel                             et al. in 64 patients suffering from peritoneal dissemination
involvement.                                                                  of ovarian cancer the overall sensitivity of spiral CT was
                                                                              85–93% with 25–50% sensitivity in metastases \1 cm.19
DISCUSSION                                                                    The good sensitivity of CT in detecting peritoneal carci-
                                                                              nomatosis in our study (overall, 84%; lesion \0.5 cm, 65%;
  The new comprehensive treatment approach of cytore-                         lesion [5 cm, 98%) can be explained by different factors.
ductive surgery and intraoperative intraperitoneal                            First, all images were analyzed by the same qualified reader
PET/CT in Peritoneal Carcinomatosis

with longstanding experience in gastrointestinal tumor            Sensitivity of CT alone was 82% and decreased to 57% for
imaging, including peritoneal carcinomatosis. Several             PET/CT. In their study the extent of PC was underesti-
studies reported only poor interobserver agreement in the         mated by CT alone in 70% of patients and in 80% of
detection of peritoneal carcinomatosis, which represents a        patients by PET/CT, and there was only a poor correlation
main weakness of CT as a standard diagnostic tool for             (0.12) between PET/CT and surgery. The results of our
detection of PC.19,23 Second, all examinations were per-          study differ from those of Dromain et al.29 We found that
formed using the same state-of-the-art multidetector CT           dual-modality FDG-PET/CT achieved the highest overall
technology to obtain thin sections throughout the whole           sensitivity of 88% for detecting PC and offered the best
abdomen/pelvis for detection of subcentimeter implants and        results for predicting intraoperative tumor load
generation of high-resolution multiplanar images, which           (r2 = 0.904). Comparison with intraoperative PCI showed
potentially improves detection accuracy. This is in accor-        a significant correlation (p \ 0.01) between PET/CT and
dance with recent results reported by Baski et al., namely a      open staging during surgery. This also was true for the
CT sensitivity and specificity in detecting PC of 83% and          subgroups representing different anatomic regions,
97%, respectively, using a 64-row MDCT.24 Accordingly,            including the small bowel. One explanation for the dis-
the low CT accuracy reported by previous studies were             crepant results could be the different examination protocol
possibly caused partially by the use of older CT technology       used in the study by Dromain et al. including low-dose,
using only axial scans on film for scoring.6,7 Third, in all our   nonenhanced CT only, without utilizing the full potential
patients special attention was given to optimal bowel             of combined PET/CT.
preparation with oral and rectal filling as well as spasmo-           The 63% overall sensitivity of FDG-PET alone in our
lytic drug application to improve the detection of bowel          study was significantly inferior to that of CT and combined
involvement by PC.                                                PET/CT. This is in accordance with several other published
   Although the effectiveness of FDG-PET has been                 studies.8,30,31 One explanation is the different primary
demonstrated in a variety of neoplasms, the role of FDG-          histology in our patient cohort, including 9 of 22 patients
PET in evaluating PC has not yet been established. Tanaka         with mucinous histologies, which often present only low
et al. reported superior sensitivity of FDG-PET (88%)             metabolic activity.30,32 The other limitation of PET is the
compared with CT (38%) but gave no detailed information           low spatial resolution of this method. The strong influence
about tumor volume in their patients.25 In their study, the       of tumor size on the detection accuracy of PET was con-
smallest peritoneal lesion detected by PET-only was               firmed by our data, which showed limited sensitivity in
15 mm in diameter. Turlakow et al. demonstrated a limited         peritoneal deposits \0.5 cm. To a lesser degree, this also
sensitivity for detection of PC with FDG-PET and CT               was true for CT and the combined modality.
(57% and 43%, respectively), both used as stand-alone                Taken together, we can conclude that the intraperitoneal
modalities.26 When both modalities were evaluated toge-           tumor load is underestimated systematically by PET alone.
ther, sensitivity increased to 78% and a high PPV of 95%          In all measures (total PCI, PCI subscore for region 0–8,
was achieved. Similarly, Suzuki et al. found the sensitivity      extent of small-bowel involvement), the accuracy of PET/
and PPV of combined evaluation of FDG-PET and CT                  CT (including regression analysis and analysis of the
(reviewed side-by-side) to be superior to those of CT alone       individual difference of the scores) was better than that of
(sensitivity: 67% vs. 22%; PPV: 92% vs. 50%).27 The               CT alone; however, given the already high accuracy of CT
limitations from low spatial resolution of PET-only can be        alone, the statistical proof of PET/CT being superior to CT
compensated somewhat by coregistration of functional              would require a larger sample of patients. Both methods,
PET images with morphologic CT data using an integrated           CT and FDG-PET, complement one another in detecting
PET/CT system.10 It has been shown that the use of dual-          PC by increasing confidence on the fused scans. The
modality PET/CT significantly improves the accuracy of             malignant nature of a small nodule in CT, especially near
staging compared with PET or CT alone.11 There are only a         bowel loops, can be confirmed by PET, and mucinous
few reports on the results of combined dual-modality PET/         tumors with low metabolic activity are well depicted by CT
CT in detecting PC using various examination protocols            (Figs. 5, 6).
and different standards of reference. Mangili et al. com-            The good correlation of PCI-PET/CT with intraopera-
pared contrast-enhanced CT and PET/CT (including                  tive PCI may help the surgeon preoperatively estimate
nonenhanced CT protocol) in 32 patients with suspected            tumor load and facilitate patient selection for peritonecto-
ovarian cancer recurrence and reported a higher staging           my and HIPEC. However, to explore the full benefit of this
accuracy for PET/CT (91%) compared with CT alone                  dual-modality examination the distinct knowledge of var-
(62%).28 In contrast, Dromain et al. did not find any benefit       ious peritoneal tumor morphologies in CT and of different
of combined PET/CT vs. CT alone in detecting PC in 30             metabolic patterns in PET is crucial for accurate image
patients with colorectal and appendiceal carcinoma.29             interpretation.
                                                                                                                        C. Pfannenberg et al.

FIG. 5 A 69-year-old man with colorectal carcinoma. a Axial FDG-         enhanced CT scan, the corresponding lesions could not be differen-
PET demonstrates multiple, irregularly distributed, hypermetabolic       tiated from adjacent bowel and mesentery, which is only possible on
foci in the right and central abdomen suggestive of peritoneal           the fused PET/CT image (c)
carcinomatosis (confirmed by surgery). (b) On axial contrast-

FIG. 6 A 37-year-old man with pseudomyxoma peritonei from                a scalloping effect on the liver and spleen. (a) The axial FDG-PET
mucinous adenocarcinoma of the appendix. b The axial contrast-           and (c) the fused PET/CT image demonstrate a diffuse faint FDG
enhanced CT scan shows widespread heterogeneous peritoneal               uptake of the mucinous masses, which is too low to be differentiated
locules and intraperitoneal mucin displacing the bowel and producing     from normal liver and bowel uptake

   In centers where PET/CT is not available, CT with a                    4. Harmon RL, Sugarbaker PH. Prognostic indicators in peritoneal
dedicated protocol and experienced readers may provide a                     carcinomatosis from gastrointestinal cancer. Int Semin Surg
                                                                             Oncol. 2005;2:3.
practicable alternative for evaluating the extent of PC.                  5. Jacquet P, Jelinek JS, Steves MA, Sugarbaker PH. Evaluation of
                                                                             computed tomography in patients with peritoneal carcinomatosis.
ACKNOWLEDGMENTS The authors thank their technicians,                         Cancer. 1993;72:1631–6.
Henriette Heners, Agnetha Burklin, Sylvia Stotz, Tina Brutschy, and                                    ¨
                                                                          6. De Bree E, Koops W, Kroger R, van Ruth S, Witkamp AJ, Zo-
Diana Feil, for helpful assistance in the acquisition of PET/CT data,        etmulder FA. Peritoneal carcinomatosis from colorectal or
and gratefully acknowledge the continuing support of their colleagues        appendiceal origin: correlation of preoperative CT with intraop-
from the Radiopharmacy Department.                                           erative findings and evaluation of interobserver agreement. J Surg
                                                                             Oncol. 2004;86:64–73.
REFERENCES                                                                7. Yan TD, Haveric N, Carmignani CP, Chang D, Sugarbaker PH.
                                                                             Abdominal computed tomography scans in the selection of
                                                                             patients with malignant peritoneal mesothelioma for compre-
 1. Jacquet P, Stephens AD, Averbach AM, et al. Analysis of mor-             hensive treatment with cytoreductive surgery and perioperative
    bidity and mortality in 60 patients with peritoneal carcinomatosis       intraperitoneal chemotherapy. Cancer. 2005;103:839–49.
    treated by cytoreductive surgery and heated intraoperative intra-     8. Turlakow A, Yeung HW, Salmon AS, Macapinlac HA, Larson
    peritoneal chemotherapy. Cancer. 1996;77:2622–9.                         SM. Peritoneal carcinomatosis: role of (18)F-FDG PET. J Nucl
 2. Glehen O, Kwiatkowski F, Sugarbaker PH, et al. Cytoreductive             Med. 2003;44:1407–12.
    surgery combined with perioperative intraperitoneal chemother-        9. Dromain C, Leboulleux S, Auperin A, et al. Staging of peritoneal
    apy for the management of peritoneal carcinomatosis from                 carcinomatosis: enhanced CT vs. PET/CT. Abdom Imaging.
    colorectal cancer: a multi-institutional study. J Clin Oncol.            2007;33:87–93.
    2004;22:3284–92.                                                     10. Blodgett TM, Meltzer CC, Townsend DW. PET/CT: form and
 3. Verwaal VJ, van Ruth S, de Bree E, van Slooten GW, van Tintere           function. Radiology. 2007;242:360–85.
    H, Boot H, Zoetmulder FA. Randomized trial of cytoreduction          11. Antoch G, Saoudi N, Kuehl H, Dahmen G, Mueller SP, Beyer T,
    and hyperthermic intraperitoneal chemotherapy versus systemic            et al. Accuracy of whole-body dual-modality fluorine-18-2-flu-
    chemotherapy and palliative surgery in patients with peritoneal          oro-2-deoxy-D-glucose positron emission tomography and
    carcinomatosis of colorectal cancer. J Clin Oncol. 2003;21:3737–         computed tomography (FDG-PET/CT) for tumor staging in solid
PET/CT in Peritoneal Carcinomatosis

      tumors: comparison with CT and PET. J Clin Oncol.                         of patients with colorectal peritoneal carcinomatosis for cytore-
      2004;22:4357–68.                                                          ductive surgery and hyperthermic intraperitoneal chemotherapy.
12.   Sugarbaker PH. Peritonectomy procedures. Cancer Treat Res.                EJSO. 2006;32:65–71.
      1996;82:235–53.                                                     24.   Baski M, Marin D, Catalano C, et al. Diagnostic accuracy of 64-
13.   Jacquet P, Sugarbaker PH. Clinical research methodologies in              slice CT with isotropic multiplanar reconstructions (MPRs) in the
      diagnosis and staging of patients with peritoneal carcinomatosis.         pre-operative diagnosis of peritoneal carcinomatosis: correlation
      Cancer Treat Res. 1996;82:359–74.                                         with intraoperative and histopathologic findings. Eur Radiol.
14.   Holm S. A simple sequentially rejective multiple test procedure.          2008;18:145.
      Scand J Stat. 1979;6:65–70.                                         25.   Tanaka T, Kawai Y, Kanai M, Taki Y, Nakamoto Y, Takabayashi
15.   R Development Core Team. R: A language and environment for                A. Usefulness of FDG-positron emission tomography in diag-
      statistical computing. Vienna, Austria: R Foundation for Statis-          nosing peritoneal recurrence of colorectal cancer. Am J Surg.
      tical Computing; 2007.                                                    2002;184:433–6.
16.   Bijelic L, Jonson A, Sugarbaker PH. Systematic review of cy-        26.   Turlakow A, Yeung HW, Salmon AS, Macapinlac HA, Larson
      toreductive surgery and heated intraoperative intraperitoneal             SM. Peritoneal carcinomatosis: role of 18F-FDG PET. J Nucl
      chemotherapy for treatment of peritoneal carcinomatosis in pri-           Med. 2003;44:1407–12.
      mary and recurrent ovarian cancer. Ann Oncol. 2007;18:1943–50.      27.   Suzuki A, Kawano T, Takahashi N, et al. Value of 18F-FDG PET
17.   Jacquet P, Jelinek JS, Chang D, Koslowe P, Sugarbaker PH.                 in the detection of peritoneal carcinomatosis. Eur J Nucl Med
      Abdominal computed tomographic scan in the selection of                   Mol Imaging. 2004;31:1413–20.
      patients with mucinous peritoneal carcinomatosis for cytoreduc-     28.   Mangili G, Picchio M, Sirini S, et al. Integrated PET/CT as a
      tive surgery. J Am Coll Surg. 1995;181:530–8.                             first-line re-staging modality in patients with suspected recur-
18.   Raptopoulos V, Gourtsoyiannis N. Peritoneal carcinomatosis. Eur           rence of ovarian cancer. Eur J Nucl Med Mol Imaging.
      Radiol. 2001;11:2195–206.                                                 2007;34:658–66.
19.   Coakley FV, Choi PH, Gougoutas CA, et al. Peritoneal metas-         29.   Dromain C, Leboulleux S, Auperin A, et al. Staging of peritoneal
      tases: detection with spiral CT in patients with ovarian cancer.          carcinomatosis: enhanced CT vs. PET/CT. Abdom Imaging.
      Radiology. 2002;223:495–9.                                                2007;33:87–93.
20.   Pannu HK, Bristow RE, Montz FJ, Fishman EK. Multidetector           30.   Stahl A, Ott K, Weber WA, Becker K, et al. FDG PET imaging of
      CT of peritoneal carcinomatosis from ovarian cancer. Radio-               locally advanced gastric carcinomas: correlation with endoscopic
      Graphics. 2003;23:687–701.                                                and histopathological findings. Eur J Nucl Med Mol Imaging.
21.   Low RN, Duggan B, Barone RM Saleh F, Song T. Treated                      2003;30:288–95.
      ovarian cancer: MR imaging, laparotomy reassessment, and            31.   Whiteford MH, Whiteford HM, Yee LF, et al. Usefulness of
      serum CA-125 values compared with clinical outcome at 1 year.             FDG-PET scan in the assessment of suspected metastatic or
      Radiology. 2005;235:918–26.                                               recurrent adenocarcinoma of the colon and rectum. Dis Colon
22.                                 ¨
      Ricke J, Sehouli J, Hach C, Hanninen EL, Lichtenegger W, Felix            Rectum. 2000;43:759–67.
      R. Prospective evaluation of contrast-enhanced MRI in the           32.   Berger KL, Nicholson SA, Dehdashti F, Siegel BA. FDG PET
      depiction of peritoneal spread in primary or recurrent ovarian            evaluation of mucinous neoplasms: correlation of FDG uptake
      cancer. Eur Radiol. 2003;13:943–9.                                        with histopathologic features. Am J Roentgenol. 2000;174:1005–
23.                              ¨
      De Bree E, Koops W, Kroger R, van Ruth S, Verwaal VJ, Zo-                 8.
      etmulder FA. Preoperative computed tomography and selection

Shared By: