Discordant Results in Human Chorionic Gonadotropin Assays

Document Sample
Discordant Results in Human Chorionic Gonadotropin Assays Powered By Docstoc
					CLIN.CHEM.38/2.     263-270   (1992)

Discordant Results in Human Chorionic Gonadotropin Assays
Laurence A. Cole and Andrew Kardana

Discordance has been reported in human chorionic go-                   /3-subunit, however, is unique          and distinguishes        hCG
nadotropin (hCG) concentrations measured by different                  from these other hormones.
immunoassay kits. We examined the results for 40 serum                    Qualitative      measurements       of urine hCG are used to
samples assayed with 10 different hCG immunoassay                      diagnose   pregnancy.        Quantitative serum hCG measure-
kits. Results varied considerably. Individual sample re-               ments are used to monitor the progress of pregnancy;                 to
sults varied by as much as 58-fold. Average results for                assess the date of pregnancy; to screen for ectopic
different kits varied by as much as 1.4-fold for pregnancy             pregnancy,        hydatidiform    mole, and Down syndrome;
(20 samples) and 2.2-fold for trophoblast disease (20                  and to follow the course of trophoblast disease, testicular
samples) serum. We investigated the causes of this                     cancer, and certain nontrophoblastic malignancies. For
discordance. hCG or hCG/3 are general names for mix-                   >20 years, competitive radioimmunoassays (RIAs) have
tures of hCG, hCGa, or hCG/3 immunoreactive molecules                  been used to determine hCG concentrations in serum. In
in serum. These mixtures include regular hCG, nicked                   1972 Vaitukaitis et al. (1) developed the hCG/3 RIA,
hCG (missing peptide linkages at /344-45 or p347-48),                  using antisera to hCGf3-subunit. This was the first assay
carbohydrateariants of hCG, hCG missing the /3-subunit                 to distinguish hCG from luteinizing hormone. In the
C-terminal segment, free /3-subunit,13-corefragment, and               mid-1970s the hCG/3 RIA became the standard test, and
free a-subunit. We prepared standards for each of these                is still widely used today. The advent of monoclonal
major variants and measured their reactivities in the 10               antibodies facilitated the development of more sophisti-
hCG immunoassay         kits. Free 13-subunit reactivity varied        cated     and more specific immunoassays. Multiple-site
from nonrecognition (anti-f3:anti-a type kits; Hybritech               sandwich assays were developed in which one or more
Tandem-R and others) to overrecognition (one kit had                   immobilized        antibodies “capture” the hCG and either
five-fold greater affinity for free /3than for hCG). Kits with         radiolabeled       (immunoradiometric         assay, IRMA) or en-
antibodies to /3-subunit C-terminal segment (Organon                   zyme-labeled (immunoenzymometric                assay, IEMA) anti-
NML and others) failed to recognize hCG missing this                   bodies detect the bound molecules (2-4). In 1991 serum
segment, a component of serum hCG in trophoblast                       hCG could be measured by any one of five types of
disease. Kits with anti-hCG aptibodies (Serono MAIA-                   immunoassays,           with >50 different commercial           kits.
clone and others) had minimal recognition of nicked hCG                These types of assays were the hCG/3 RIAs and IRMAs or
(12%), a component of all serum hCG samples, and                       IEMA8 involving         two different /3-subunit (anti-I3:anti-/3)
consistently gave the lowest values with all serum sam-                antibodies, an a- and a /3-subunit (anti-/3:anti-a) anti-
ples. We conclude that differences in recognition of                   body, a /3 C-terminus and a /3-subunit (anti-/3 C-termi-
nicked hCG, free /3, nd these other hCG variants cause
                       a                                               nus:anti-/3) antibody, or an hCG and a /3-subunit (anti-
discordance in hCG immunoassay results.                                hCG:anti-/3) antibody.
                                                                          Discordant       results have been reported for the various
AddItIonal Keyphrases: variation, source        of    pregnancy        hCG immunoassays, with different kits giving different
immunoassay               d
                trophobiast isease
                                                                       values for the same serum sample. The problems and
                                                                       confusion      created by discordant      results are exemplified
  Human     chorionic  gonadotropin(hCG) is a glycopro-                in an article by Painter (5) and in several other publi-
thin hormone produced by trophoblast tissue in       preg-             cations (6-11). In the past, discordance was attributed to
nancy and trophoblast    disease and by testicular     and             differences in hCG standards [the 1st International
certain nontrophoblastic  cancers.’ hCG is composed    of a            Reference Preparation (IRP) for immunoassay                    (pure
/3-subunit (145 amino acids) and an a-subunit (92 amino                hCG), the 2nd International           Standard    (IS) for bioassay
acids) joined noncovalently.   The a-subunit    of hCG is              (crude urine extract), and othersJ and to interference by
similar   to that of pituitary     lut.einizing   hormone              pituitary     luteinizing hormone (6-11). Today, however,
(lutropin), follicle-stimulating      hormone      (follitropin),      most inununoassays            have little or no cross-reactivity
and thyroid-stimulating         hormone   (thyrotropin).      The      with luteinizing hormone and most assays use interre-
                                                                       latable standards        (1st IRP for immunoassay       and 3rd IS,
                                                                       both from CR119 hCG). Nonetheless,              discord still exists
   Department of Obstetrics and Gynecology,Yale University             between hCG values obtained with different immunoas-
School of Medicine, 333 Cedar St., New Haven, CT 06510.
   ‘Nonstandard      abbreviations: hCG,human choriomc gonadotro-      says.
pin (choriogonadotropin); IRP, International Reference Prepara-           We considered           reasons for discordant in hCG
tion; IS, International Standard; LEMA, immunoradiometric     assay;   results.   A     growingamount of data shows that the
IEMA, immunoenzymometric            assay; HLE, human leukocyte
                                                                       substance broadly described as hCG (or hCG(3) immuno-
elastase; and NICHHD, National Institute       of Child Health and
Human Development.                                                     reactivity in serum is a mixture    of hCG a- and (3-sub-
   Received June 7, 1991; accepted December 27, 1991.                  unit-related  molecules   (12-26). These include regular

                                                                                         CLINICAL CHEMISTRY, Vol. 38, No. 2, 1992 263
hCG, nicked hCG (missing peptide linkages at /344-45                               (90% nicked) was prepared           by treating hCG batch
or (347-48) (13-20), carbohydrate      variants of hCG (with                       CR127 with human           leukocyte     elastase    (HLE, EC
reduced sialic acid content or polyantennary       core oligo-            as previously described (27). Asialo hCG stan-
saccharide structure)   (15,21,22,25),       hCG missing the                       dard was prepared by treating hCG batch CR127 with
p-subunit C-terminal segment (20,23), hCG free p-sub-                              neursminidase       (EC as described        previously
unit (12,24), hCG /3-subunit core fragment (15,23,26),                             (22). The purification   and characterization        of the indi-
hCG free a-subunit (15), and possibly others (Table 1).                            vidual hCG preparations        C5 (100% nicked) and M1A
Here we consider the possibility that differences in                               (missing p-subunit C-terminal segment) were described
recognition of each of these hCG variants by the five                              previously (19). hCG p-subunit       standard     and a-subunit
types of hCG immunoassays are the cause of discord in
                                                                                   standard (both batches          CR129)      were     gifts from
reported values.                                                                   NICHHD, and hCG p-subunit core fragment standard
Materials and Methods                                                              batch SB455 was a gift from Dr. Steven Birken of
                                                                                   Columbia      University  College of Physicians         and Sur-
Serum Samples
                                                                                   geons. All standards were prepared in hCG-free human
   Serum was collected in the Obstetrics and Gynecology                            serum directly from amino acid analysis-calibrated
Infertility Clinic at Yale University from 10 women in                             stock solutions. Table 2 summarizes         the standards     and
early first-trimester     pregnancy (five to eight weeks                           individual     hCG preparations, their N-terminal              se-
since their last menstrual periods). Samples were also                             quence analysis, and other characteristics.
collected at the Obstetrics and Gynecology Perinatology
Clinic from 10 women with 14- to 20-week (second-
trimester) pregnancies and at the Gynecologic Oncology                             Physical Analyses
clinic from 10 women with hydatidiform      mole (preevac-                            The concentration of all standards       (individual hCG
nation) and from 10 women with persistent trophoblast                              preparations,    HLE-digested preparations,       and CR119
disease (prechemotherapy).                                                         and CR127 hCG) was established by amino acid analy-
   Studies    in our laboratory with electrophoresis    and                        sis. Aliquots were placed in tubes (containing internal
immunoblotting        and with immunoassays          showed                        standard) and analyzed by Dr. Steven Birken and Mary-
higher proportions of nicked hCG in nonseparated blood
                                                                                   Ann Gawinowicz at the Protein Core Laboratory                  at
samples shipped from patients’ homes at ambient tem-
                                                                                   Columbia University       College of Physicians and Sur-
perature and in serum samples kept refrigerated          for
                                                                                   geons with a Beckman        (Fullerton, CA) Model 6300A
longer than a week. For this study, all blood samples
                                                                                   automated     amino acid analyzer. Concentrations          (pico-
were placed in an ice bucket after collection and the
serum was separated and stored at -20 #{176}C. samples
                                               The                                 moles per liter) were calculated from the mean values
were thawed just before being assayed.                                             for each of six amino acids, divided by the number of
                                                                                   residues of that amino acid in hCG: Leu (-16), Ile (#{247}6),
Standards                                                                          Val (#{247}20), (#{247}29), (#{247}12), Ala (-13).
                                                                                                 Pro         Gly         and
   hCG standard batch CR127 and hCG standard batch                                    N-terminal peptide sequence      analysis was performed
CR119 (also known as the 1st IRP for iminunoassay and                              in the Protein Core Laboratory at Columbia by previ-
the 3rd IS, respectively) were kindly provided for this                            ously published methods (27). Sialic acid was released
study by Dr. Gabe Bialy and the Center for Population                              from hCG by the action of Clostridium             perfringens
Research of the National Institute  of Child Health and                            neuraminidase     and was quantified by the spectrometric
Human Development (NICHHD).        Nicked hCG standard                             periodate-.thiobarbituric    acid method (22).

                                      Table 1. hCG Immunoreactive Molecules                    in Serum and UrIne
                                                  Carbollydrats                      hCG missing
                                                   vedantsof                         p C-termInal       hCG free          p-core
           Source                    hCG              I.C0           NICkSdhCO         segment          p-subunit       fragment             Reterencee
  Pregnancy                      Pnncipalb       Trace              Secondary              ?          Trace           Trace            12, 15, 19, 23, 24
  Hydatidiform mole              Principal       Secondary          Secondary              ±          Secondary       Trace            12, 15, 19, 21-25
  Persistenttrophoblast          Secondary       Principal          Secondary              ?          Secondary       Trace            12, 15, 16, 19, 21-25

  Pregnancy                     Secondary        Trace               Secondary               ?         Trace           Pnnclpal         14-22,26
   Hydatldlform mole            Secondary        Secondary           Secondary               ±         Secondary Principal              14-16, 19-22
   Persistent trophoblast       Secondary Principal                  Secondary               ±         Secondary Secondary              13-16, 19-22 26
     Aslalo hCG and hCG w$thcore ollgoeaccharldeantennary and sialicacid variations.
   b Summary of observations reported In 10 publIcations. P’inc(oaIrefers to the pdnc,pal component of a serum or urine sample, secondary refers to one or more
secondary   components (>1% of total), trace refers to componentsaccounting for <1% of totallmmunoreactivfty, refers to components
                                                                                                                 t                       that occur sporadically,
and 7 means not determined.

264 CLINICALCHEMISTRY, ol. 38, No. 2, 1992
                                  Table 2. Analysis of IndMdual and Standard hCG Preparations
                Sample                                                                                                 Interpretation                     flefsrence
hCGstandard CR119 (1st IRPfor              pi sequence (Ser..              .),      1.0                      9% nicked In p44-49 region                       27
 immunoassay 3rd IS for
             and                           fl45 sequence (Leu         .    .     .), 0.05
  bioassay)                                fl48 sequence (VaI        . .       .),   0.04

hCGstandard CR127,from                     p1 sequence(Ser..               .),       1.0                     20% nIcked In p44-49 regIon                       19
 NICHHD                                    45 sequence (Leu           .    . .),      0.10
                                           48 sequence (Val . . .), 0.10
                                           Sialic acId, 14 nmol/nmol hCG

hCG standard CR127,                        SialicacId, <0.1 nmol/nmol hCG                                    Missing slalic acid                              22

hCG standard CR127, nicked by              flu sequence     (Ser..         .), 1.0                           90% nicked in p44-49 region                      27
  action of HLE                            fl45 sequence (Leu         .    . .),      0.54
                                           fl49 sequence (Val        . .       .),   0.36

hCGpreparationP1, from let                 flu sequence (Ser..             .),       1.0                     Notnicked                                        20
 trimester pregnancy urine                 fl48 sequence     (Val    . . .),          <0.01

hCG preparation    C5, from                $1 sequence      (Ser.. .), 1.0                                   100% nIcked in $44-49 region                     20
  trophoblast   disease urine              fl48 sequence     (Val . . .), 1.28
                                           Sialic acId,   12 nmoVnmol hCG

hCGpreparationM1A,from                     flu sequence     (Ser..         .), 1.0                           Notnicked, issing p-subunit
                                                                                                                      m                                       20
  trophoblast   disease urineb             fl48 sequence (Val . . .), 0.0                                          C-terminal peptld&’
                                           Slalic acId, 7.1 nmoL/nmol hCG

hCG standard CR127,from                    flu sequence (Ser.. .), 1.0                                       10% nicked in fl44-49 region                     27
 NICHHD                                    fl45 sequence (Leu .. .), 0.05
                                           fl49 sequence (Val . . .), 0.05

hCG p-subunit core-fragment              p-subunit residues     6-40 disulflde-linked   to                                                       26
  standard 55455                            residues 55-92
   The concentrationof each was determined by amino acid analysis, and the N-terminal sequence and carbohydrate content by the methods described here or
   Aminoacid content consistent wfth absence of p123-145. PreparatIon was recognized by antibody FBT11 (Immunogeri: l09-1                18) but not by antibody CIII

Immunoassays                                                                                  (National Institutes of Health, Bethesda, MD).
   The Tandem-R hCG kit (Hybritech, La Jolla, CA),                                               The nine amino acid analysis-calibrated       standards-
Organon NML kit (Organon Teknika, Rockville, MD),                                             CR119 hCG, CR127 hCG, nicked CR127 hCG, asialo
Serono MAlAclone kit, Biomerica hCG IRMA kit, Abbott                                          CR127 hCG, individual hCG M1A and C5, CR129 free a
hCG 15/15 kit (Abbott Labs., Abbott Park, IL), Amer-                                          and free p-subunit,       and SB455 p-subunit core frag-
sham (Arlington Heights, IL) Amerlex-M kit, and Diag-                                         ment-were       tested with each of the seven commercial
nostic Products Corp. (Los Angeles, CA) hCG kit were                                          hCG immunoassay         kits and each of the three in-house
all purchased from the manufacturers      or their distribu-                                  hCG assays. Each standard was tested in duplicate in at
tors and used within seven days of receipt. The methods                                       least six different concentrations in the range 0.25-200
described in the instruction     pamphlets were strictly                                      jtg/L. Standards     were diluted in bulk so that the same
followed except that the above-described amino acid anal-                                     dilution could be tested with each immunoassay,           elim-
ysis-calibrated   standards were used. We used our pub-                                       inating the possibility of kit-to-kit dilution errors. Val-
lished methods for the in-house 2119. 12/BP052 anti-aan-                                      ues within the standard       curve (four or five concentra-
ti-a IRMA, the in-house B109/BP052 anti-hCG:anti-f3 IEMA,                                     tions) were plotted for each assay: radioactivity (less
and the in-house CC11JHCO514 anti-a C-terminus:anti-                                          nonspecific radioactivity) vs hCG concentration for IRMA
IRMA (19,20). Antibody B109 was a gift from Drs. John                                         results, absorbance at 492 nm (less nonspecific absor-
O’Connor and Alex Krishevsky (Columbia University                                             bance) vs hCG concentration for iEM results, and per-
College of Physicians and Surgeons), antibody 2119.12                                         cent bound (B NSB)/(B0 - NSB) vs hCG concentra-

was a gift from Dr. Keith May (Unipath Inc., Bedford,                                         tion for RIA results (where NSB = nonspecific binding).
U.K.), and antibody CC11 was a gift from Dr. H.-C. Chen                                          In a separate experiment,       the hCG content of 40

                                                                                                            CLINICAL CHEMISTRY, Vol. 38, No. 2, 1992                265
serum samples was determined with the seven commer-                                                                        sham    Amerlex-M), and two anti-fl C-terminus:anti-fl
cial hCG kits and the three in-house immunoassays.                                                                         sandwich (Organon NML and in-house CC11IHCO514)
hCG standards CR119 and CR127 contain both nicked                                                                          types of immunoassay.     As Figure 1 shows, wide varia-
and intact hCG (Table 2) and may not be recognized                                                                         tion was apparent in results for single serum samples
equally by nick-sensitive and nick-insensitive assays. A                                                                   determined    by different hCG kits. In the pregnancy
new nick-free standard was needed. We chose P1 hCG, a                                                                      samples (Fi-FlO and Si-Sb),       values for an individual
preparation purified from pregnancy urine that is 100%                                                                     sample varied by as much as 2.2-fold (serum Fl: 707
intact (not nicked) and has no a-subunit or fl-subunit                                                                     pgfL by Serono MAIAclone anti-hCG:anti-fl assay, 1575
N-terminal heterogeneity (20). P1 hCG standards were                                                                       g/L by in-house 2119-12/BP052 anti-fl:anti-a assay). In
prepared  and calibrated by amino acid analysis.                                                                           trophoblast   disease samples (Mi-Mb        and P1-PlO),
                                                                                                                           variation was greater and values for an individual
 Results                                                                                                                   sample varied by as much as 58-fold (serum Mi: 3470
   Forty serum samples-lO       from women in the first                                                                     tgfL by the Organon NML anti-p C-terminus:anti-fl
trimester of pregnancy (Fi-FlO), 10 from women in the                                                                      assay and 201 600 og/L by the Amersham         Amerlex-M
second trimester of pregnancy    (Si-Sb),  10 from women                                                                   anti-fl RIA). Overall, the two anti-hCG:anti-f3 type as-
with hydatidiform mole (Mi-Mb),        and 10 from women                                                                   says (Serono MAlAclone and in-house B109/BP052)
with persistent trophoblast disease       (P1-P10)-were                                                                    gave the lowest values (Table 3). Other types of immu-
tested in seven commercial and three in-house hCG                                                                          noassay gave, on average, as much as 1.4-fold higher
iinmunoasaays     (Figure 1). These assays included two                                                                    values for serum hCG in first- and second-trimester
anti-franti-a   sandwich  (Hybritech Tandem-R and in-                                                                      pregnancy and as much as 2.2-fold higher values in
house 2119-121BP052), two anti-hCG:anti-/3 sandwich                                                                        trophoblast  disease.
(Serono MAlAclone and in-house B109/BP052),            two                                                                    We considered reasons      for this discordance in hCG
anti-$:anti-$ sandwich     (Abbott 15/15 and Biomerica                                                                     assay  results. As indicated in Table 1, hCG is not a single
hCG), two anti-fl RIA (Diagnostic Products and Amer-                                                                       molecule but rather a mixture of hCG a- and fl-subunit-




                                                                                                                            80jI             52204840


                                                         545-106                                                      000fI400011



                                                                                                                                      ihtIthiiEthfl                                     2325-3184
                                                                                                                                                                                                2160-2866 1316-2616
     SampleFl            F2            f3       F4          F5      ES           F7     F8          F9     FlO             Sample Si              S2      S3          S4         S5        S6       S7       SO           S9       Sb

               3470-201600                                                                                                 40000-

                                                                                                                           25000!            6060-23140

               1 I I              I   35000-76900                                                                    20000     I 111904)5-4495)   I    13470-21890


        hiI Ii I                                          30846-70500

                        I ih iIi Iio                                                                                                                       L,
-‘                                                                 24700-63600                                             15000                                           6170-13400
     #{176}#{176}#{176}III                                                  25840-62160
                                                                                                                                                           IIIJg]                     30004600

      500001111                                                                                                            10000                                    .181     -                   3530-5950 3226-9710
      2500                            Ifl                                              22I6-26100
                                                                                          IIl i.
                                                                                          liii i1kiQii
     Sample     Ml           M2        M3           M4      M5       MS          Mi      M8          M9     M10            Sample     P1          P2      P3          P4         P5        P6        P7       P8          P9       PlO

Fig. 1. hCG concentrations In 40 serum samples determined by 10 dIfferenthCG Immunoassays
Samples  FI-FlO are from women in first-trimester pregnancy, samples SI-Sb are from women In second-trimester pregnancy, samples M1-MIO are from
women with hydatidifom, mole, and samples Pb-PlO are fromwomen with persistent trophoblast disease. Above each set of bars are the ranges of results for
that sample. From left So right for each sample, the results are shown for In-house assay 21 19-12BPO52, Hybritech Tandem-A. in-house assay B109#BP052,
Serono MAlAclone, Abbott hCGP 15/15, Biomeiica hCG, Diagnostic Products Corp., Amersham Amertex-M, Organon NML, and In-house assay CCI 1/HC0514.
Note differences In otstnate ranges

266 CLINICAL CHEMISTRY, Vol. 38, No. 2, 1992
Table 3. Average hCG ConcentratIon (pg/L) In Four Groups of Serum Samples as DetermIned by 10 DIfferent hCG
                                                                        MedIan (and mien) fern      -   lOin   iiti   group

                                           let tolmsstsr            2nd trImester                                                tset          -
         hCG Immunousay                                                                           Hyda5WSo mole
In-house 2119-12/8P052                      477(612)                3317(4308)                     50950(52106)                          7650 (8273)
  (anti-:antI-a      IEMA)
HybntechTandem-A                            388 (487)               2702 (3318)                    49600(50888)                          7820 (8678)
In-houseB1 09/BP052                         404(483)                2573 (3284)                    33250(34346)                          4900(6737)
  (anti-hCG:anti-p     IEMA)
SeronoMAlAclone                             418(469)                2638(3114)                     39890(40540)                          7185(7384)
  (anti-hCG:anti-p     IEMA)
Abbott CGfl 15/15                           415(477)                3400 (4039)                    65550(65641)                         10920(13736)
  (anti-p:anti-p     IRMA)
BlomencahCG                                 430(571)                2652 (3273)                    50350(52049)                          8360 (8770)
  (antI-p:anti-p IRMA)
DiagnosticProducts(anti-p RIA)              414 (470)               2672(3418)                     55950(53179)                          8395 (9226)
Amersham Amerlex-M                          429(560)                3142(3747)                     55800 (67953)                        10080(11443)
Organon NML (anti-p                         539(581)                3150(3910)                     45200 (47895)                         9405(11245)
  C-terminus:antl-p IRMA)
In-house CC1 1/HC0514             452 (455)                         3646(4638)                     47700(51066)                          7100(8291)
Range                         388-539 (469-612)                2573-3646(3114-4638)        33250-65550(34346-67953)           4900-10920       (6737-13736)
          n-fold                   1.4(1.3)                           1.4(1.5)                          2.0(2.0)                           2.2 (2.0)
 a   Values are normally distributed (medianless than mean).

related molecules, which may be recognized differently by                         others (6-11) is of great concern, mpking it difficult to
the five different types of hCG immunoassays. To investi-                         interpret results from experiments                repeated      with a
gate this possibility, we prepared standards for the major                        different commercial        immunoassay.        Comparing results
a- and fl-subunit-related molecules (Fable 2) and tested                          from different clinical centers is problematic,                and the
them with the five different types of hCG immunoassay                             establishment     and interpretation of clinical cutoff val-
(Figure 2). None of the immunoassays detected hCG free                            ues and normal pregnancy hCG values has been hin-
a-subunit (data not shown). Assays varied greatly in their                        dered.
recognition    of hCG free fl-subunit,    from <2% (anfi-                            hCG, or hCGfl immunoreactivity,              material is a mix-
anti-a   type assays) to 490% (Abbott /3hCG 15/15 anti-fl:                        tore of hCG a- and fl-subunit-related           molecules (12-26).
anti-fl assay) of CR119 hCG (1st IRP for immunoassay)                             These include nicked hCG, carbohydrate-variants                       of
reactivity. The two immunoassays       involving an antibody
                                                                                  hCG, and hCG missing the p-subunit C-terminal seg-
to the p-subunit C-terminus (Organon NML and in-house
                                                                                  ment. As illustrated in Figure 3, nicked hCG has pep-
CC11/BP052)      had poor recognition of hCG missing the
                                                                                  tide linkages missing at fl44-45 or fl47-48. The nicks
C-terminssl segment (-33% of CR119 hCG activity); one of
the assays with an antibody to the fl-subunit C-terminus                          open an intercysteine         loop on the fl-subunit,          thereby
(Organon NML) had reduced recognition of asialo hCG                               altering   the tertiary      structure      of hCG and ablating
(-55% of CR119 hCG activity); and the two anti-hCG:                               gonadotropic    activity (13-20). Elsewhere we have sug-
anti-fl assays (Serono MAJAclone and in-house B109/                               gested that nicked hCG accounts for about a quarter of
BP052) had only minimal recognition of mcked hCG (-6%                             the hCG-related      molecules       in serum from women with
and -12% of CR119 hCG activity, respectively).                                    pregnancy     or trophoblast        disease (19). Carbohydrate
                                                                                  variants of hCG have either a decreased                    sialic acid
DIscussIon                                                                        content or polyantennary             oligoeaccharide      core struc-
   We collected 40 individual      serum samples, 20 from                         tures (15,21,22,25).        Carbohydrate       variation can occur
women with pregnancy       and 20 from those with tropho-                         at any of the eight glycosylation sites on hCG a- and
blast disease, and tested them with 10 separate kite, two                         p-subunit (Figure 3). Carbohydrate variants are major
of each of the five major types of hCG immunoassay.        We                     components of hCG in women with trophoblast disease
found that individual     values can vary by as much as                           (21, 25). hCG molecules             missing    all or part of the
58-fold, depending on the assay used. The results for the                         p-subunit C-termin50l segment have also been detected
10 first- and 10 second-trimester         pregnancy   samples                     in women with trophoblast           disease (20). Other hCG and
varied by as much as 1.4-fold; those for 10 hydatidiform                          a- and fl-subunit-related        molecules are hCG free p-sub-
mole, and 10 persistent       trophoblast     disease samples                     unit (uncombined        fl-subunit)     and p-subunit core frag-
varied even more (2.0 and 2.2-fold, respectively).       This                     ment (a posttranslational          processing     product of fl-sub-
wide variation in interassay       results found by us and                        unit). hCG free p-subunit,          although    a trace component

                                                                                                   CLINICAL CHEMISTRY, Vol. 38, No. 2, 1992 267
               ‘8w’         Amersham                                                                           Organon NML
                      \                                    a

 ‘              U3Q\\\

       ,                                         #{149}0

                                      P                                                                        in house CCII/HCOS14


     I                            to
                           .ttIgo,t    @g,t)

                                                                                       nUCGnti.RMA)           j

Fig.2. Dose-response      curves for hCG immunoreactive molecules inthe various assays
Curves arestandardhCG batches CR119 (let IRP for Immunoassay) () and CR127 (A), deslalylated hCG batch CR127 (#{149}),    and nicked hCG batch CR127 (#{149}).
Other curvesare forindividual hCG samples 100% nIcked (0) or missing the ft-subunit C-terminal segment (*) and for hCG ft-subunit standard batch CR129 (#{149})
and p-subunit core-fragment standard 88455 (G). Cross-reactivity values are given for deviant dose-response lines. Vekies are percent of CR119 dose
determined at 50% radIoactivity bound (ED,). Approximate (-) and less than (<) values are given for dose-response lines that do not achieve 50% maxImum
response. NSB. nonspecific binding

of pregnancy    serum (<0.5% of hCG concentrations), is a                        oligosaccharide    antennae (because changes in the num-
significant  component (as much as 26% of hCG concen-                            ber of antennae      are normally associated  with changes in
tration) of trophoblast    disease  samples   (12, 15, 24).                      sialic acid content). We observed the following
p-Subunit   core fragment is a trace component    (<0.5% of                         1. Anti-hCG:anti-$     type assays (Serono MAlAclone
hCG concentration) of untreated     serum from pregnancy                         kit and in-house B109/BP052       assay) poorly detect or fail
or trophoblast  disease.                                                         to detect nicked hCG (-12% and -6%, respectively, of
      We consideredthe possibility that differences in rec-                      the reactivity of CR119 hCG). This could explain why
ognition of the various hCG and a- and p-subunit-                                the two anti-hCG:anti-fl    assays gave the lowest average
related molecules by the five types of hCG immunoassay                           values for the 20 pregnancy samples (12% below me-
were the cause of discordant reported values. We tested                          dian) and the 20 trophoblast          disease  samples (28%
10 different hCG kits with standards       for each of the                       below median).
major hCG a- and fl-subunit-related     molecules.  Asialo                          2. Although anti-fl:anti-a    type assays did not detect
hCG was included as a standard for sialic acid variation                         free subunits, other assays     (Abbott flhCG 15/15) over-
and possibly also for differences in the number of core                          detected   (4.9-fold greater reactivity than hCG) the free

268            CLINICALCHEMISTRY, ol. 38, No. 2, 1992
                                                                          results  for pregnancy and trophoblast disease serum
                                                                          samples,   other causes, even though they may not affect
                                                                          two-site sandwich-type assays, should be kept in mind.
                                                                             For the first time, a feasible explanation is presented
                                                                          for the discordance in hCG immunoassays       results. First,

        4                                                                 there is variable recognition of nicked hCG by different
                                                                          types of hCG assays. Anti-hCG:anti-fl     type kits (IEMA or
                                                                          IRMA) are common; in our study, both assays of this type
                                                                          either poorly detected or failed to detect nicked hCG, a
                                                                          major     component   of all hCG mixtures,      leading       to unduly
                                                                          low hCG values in serum samples from pregnancy and
                                                                          trophoblast     disease.   We suspect that other anti-hC-
                                                                          G:anti-fl assays may act likewise. Second, different hCG
Fig. 3. The primary structure of hCG a- (left) and $- (right) subunits,   kits vary in their recognition of hCG free /3-subunit.
indicating disulfide assignment (-S-S-) (28, 29), nicks at p44-45 and     Although kits of the anti-a:anti-fl    type did not detect
p47-48 (2,     oligosaccharide sites (-oligo) (21, 22), and position of   free /3-subunit, the Abbott hCG 15/15 anti-fl:anti-fl type
the fl-subunitC-terminalsegment (light-shaded area)
                                                                          assay detected p-subunit 4.9-fold more than it did hCG;
                                                                          this led to unduly high values with trophoblast     disease
p-subunit. The overdetection of free p-subunit by this                    samples.     Most anti-ftanti-fl assays detect free p-sub-
immunoassay      may explain the greater results obtained                 unit, usually to the same extent as hCG. The Abbott
for hydatidiform mole (median 65550 zgfL, with values                     assay may be an exception. Third, anti-fl C-terminu-
for the nine other kits ranging from 33 250 to 55950                      s:anti-f3-type assays poorly detect hCG missing          the
tg/L) and persistent trophoblast disease samples (medi-                   p-subunit C-terminal segment, a variable component of
an 10920 1zg/L, with values for the nine other kits                       trophoblast disease samples. One of 10 hydatidiform
ranging from 4900 to 10080 ,ug/L).                                        mole samples tested was very poorly recognized by this
   3. Assays also varied in their recognition of hCG                      type of assay (values were 2.6% of the median value for
missing the p-subunit C-terminal segment, from -35%                       the other eight assays). We infer that assays with
(of CR1 19 hCG reactivity) for the two anti-fl C-terminus:                exaggerated      recognition of hCG free fl-subunit     and
anti-fl assays to 160% in the Amersham              Amerlex-M             those with an anti-fl C-terminus       antibody should be
anti-fl RIA. hCG missing        the /3-subunit C-terminal                 avoided for testing trophoblast disease samples.
segment has so far been detected only in trophoblast                         How can interkit discordance be avoided? Manufac-
disease serum     and urine. The poor recognition        of this          turers should test kits with nicked hCG, hCG missing
hCG-related     molecule explains the very poor recogni-                  the C-terminal    segment, and free /3-subunit standards.
tion of hCG in serum sample Ml (Figure 1) by these two                    Kits could then be labeled as specific for total hCG (all
assays (38-fold lower than the median            for the other            forms of hCG), intact hCG (nonnicked        molecules), and
eight assays). Overrecognition        of free p-subunit      and          (or) free /3-subunit. This would help establish hCG cutoff
hCG missing the C-terminal       segment (170% and 160%                   values and ease the interpretation      of hCG concentra-
of CR119 hCG reactivity,       respectively)    by the Amer-              tions reported by different laboratories or determined    by
sham Amerlex-M assay may also explain the increased                       different immunoassays.
results obtained with this assay for hydatidiform          mole
(median 55800 pg/L) and persistent        trophoblast   disease             Research was supported in part by National     Institutes     of Health
samples (median 10080 tgfL).                                              grants CA44131 and CA46828 to L.A.C.
   4. Only one of the 10 kits (Organon NML, anti-                         References
p-subunit C-terminal segment:anti-fl assay) had signif-                   1. Vaitukaitis JL, Braunstein GD, Ross GT. A radioimmunoassay
icantly decreased activity     with asialo hCG (-55% of                   which specifically measures human chorionic gonadotropin     in the
CR119 hCG reactivity). Sensitivity to asialo hCG indi-                    presence of human luteinizing    hormone. Am J Obstet Gynecol
cates sensitivity to differences     in carbohydrate      struc-          1972;113:751-8.
                                                                          2. Armstrong EG, Ehrlich PH, Birken 5, et al. Use of a highly
ture, e.g., structures missing sialic acid and structures                 sensitive and specific immunoradiometric    assay for detection of
with   polyantennary       core   oligosaccharide      components.        human choriomc gonadotropin     in urine of normal, nonpregnant,
Such structures      have been demonstrated       on hCG mole-            and pregnant individuals. J Clin Endocrinol Metab 1986;59:867-
cules in trophoblast     disease (15,21,22,25).                           3. Canfleld RE, O’Connor JF, Birken 5, Krichevsky A, Wilcox AJ.
   A recent review by Hussa (6) listed hCG-like sub-                      Development of an assay for a biomarker of pregnancy and early
stances, luteinizing    hormone cross-reactivity, albumin-                fetal loss. Environ Health Perspect 1987;74:57-66.
a-subunit   complexes, rheumatoid        factor, immunoglob-              4. Bellet DH, Ozturk M, Bidart J-M, Bohoun CJ, Wands JR.
                                                                          Sensitive and specific assay for human chorionic gonadotropin
ulins, proteases, lipids, and nonspecific serum factors as                (hCG) based on antipeptide and anti-hCG monoclonal antibodies:
other possible causes of discordant        hCG immunoassay                construction  and clinical implications. J Clin Endocrinol Metab
results. Although the causes identified here for discor-                  1986;63:1319-27.
                                                                          5. Painter PC. Discordant hCG results   in   pregnancy:       a method in
dance (nicked hCG, free p-subunit, hCG missing the                        crisis. Clin Casebook 198927:20-4.
C-terminal segment, and carbohydrate          variants of hCG)            6. Hussa RO.The clinical marker hCG. New York: Praeger Press,
do explain the differences we find in immunoassay                         1987.

                                                                                           CLINICALCHEMISTRY, ol. 38, No. 2, 1992 269
7. Chen HC, Matsuura S. Ohashi M. Limitations and problems of                 deglycosylated and enzymatically desialylated states. Biochemis-
hCG-speciflc antisera. In: Sega! SJ, ed. Chorionic gonadotropin.              try 1989;28:9239-43.
New York: Plenum Press, 1980:231-52.                                          19. Cole LA, Kardana A, Andrade-Gordon P, et al. The heteroge-
8. Hussa    RO,Rinke ML, Schweitzer PG. Discordant            human chori-    neity of hCG: ifi. The occurrence, biological and immunological
onic   gonadotropin   results:   causes   and   solutions.   Obstet Gynecol   activities of nicked hCG. Endocrinology 1991;129:1559-67.
1985;65:211-9.                                                                20. Kardana A, Elliott ME, Gawinowicz MA, Birken S, Cole LA.
9. Hussa RO. Procedure  for discordant hCG results [Letter]. Clin             The heterogeneity of hCG: I. Characterization of peptide varia-
Chem 1984;30:809-1O.                                                          tions in 13 individual preparations of hCG. Endocrinology 1991;
10. Thomas CMG, Segers MFG. Discordant results for choriogo-                  129:1541-60.
nadotropin: a problem caused by lutropin beta-subunit interfer-               21. Mizouchi T, Nishimura R, Taniguchi      T, Kobata A. Compari-
ence [Letter]. Clin Chem 1985;31:159.                                         son of carbohydrate   structures between human chorionic gonado-
11. Rzasa PJ, Caride VJ, Prokop EK. Discordant inter-kit results              tropin present in urine of patients with trophoblastic disease and
in the radioimmunoaasay   for choriogonadotropin in serum. Clin               healthy individuals. Jpn J Cancer Res 1985;76:752-9.
Chem 1984;1240-3.                                                             22. Cole LA. O-Glycosylation of proteins   in the normal and neo-
12. Ozturk M, Bellet D, Manil L, Hennen G, Frydman R,                         plastic trophoblast. Trophoblast Res 1987;2:139-48.
Physiological studies of human chorionic gonadotropin (hCG),                  23. Kardana A, Cole LA. Serum hCG p-core fragment is masked
ahCG, and $hCG as measured by specific monoclonal irnmunora-                  by associated macromolecules. J Clin Endocrinol Metab 1990;71:
diometric assays. Endocrinology 1987;120:549-58.                              1393-5.
13. Nishimura R, Ide K, Utsunomiya T, Kitajima         T, Yuki Y,             24. Kardana   A, Cole LA. Polypeptide nicks cause erroneous re-
Mochizuki M. Fragmentation    of the p-subunit of human chorionic             sults in assays of human chorionic gonadotropinfree p-subunit.
gonadotropin   produced    by choriocarcinoma.      Endocrinology             Clin Chem 1992;38:34-8.
1988;123:420-5.                                                               25. Yazaki K, Yazaki C, Wakabayashi K, Igarashi M. Isoelectric
14. Bidart J-M, Puisieux A, Troalen F, Foglietti MJ, Bohuon C,                heterogeneity of human chorionic gonadotropin:     presence of cho-
BeRet D. Characterization of the cleavage product in the human                riocarcinoma specific components. Am J Obstet Gynecol 1980;138:
choriogonadotropin  p-subunit. Biochem Biophys Res Commun                     189-94.
1988;154:626.-32.                                                             26. Birken 5, Armstrong EG, Kolks MAG, et a!. Structure           of
15. Cole LA, Kardana    A, Birken S. The isomers, subunits and                human chorionic gonadotropin p-subunitcore fragment from preg-
fragments of hCG. Serono Symposia Publ 1989;65:59-78.                         nancy urine. Endocrinology    1988;123:572-83.
16. Puisleux A, Bellet D, Troalen F, et al. Occurrence of fragmen-            27. Birken S, Gawinowicz MA, Kardana A, Cole LA. The hetero-
tation of free and combined forms of the        subunit of human              geneity of hCG: II. Characteristics   and origins of nicks in hCG
chorionic gonadotropin. Endocrinology 1990;126:687-94.                        reference standards. Endocrinology 1991;129:1551-8.
17. Sakakibara   R, Miyazaki S, Ishiguro M. A nicked p-subunit of             28. Mise T, Bahl OP. Assignment of disulfide bonds in the p subunit
human chorionic gonadotropin purified from pregnancy urine. J                 of human chorionic gonadotropin. J Biol Chem 1981256: 6587-92.
Biochem 1990;107:858-82.                                                      29. Mise T, Bahl OP. Assignment of disulfide bonds in the a-sub-
18. Lustbader JW, Birken S, Pileggi NF, eta!. Crystallization and             unit of human choriomc gonadotropin. J Biol Chem 1980255:
characterization  of human choriomc gonadotropin in chemically                8516-22.

270    CLINICAL CHEMISTRY, Vol. 38, No. 2, 1992

Shared By: