| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Articles |
Departments of
1
Obstetrics and Gynecology and
2
Genetics, Yale University School of Medicine, New Haven, CT 06510.
a Address correspondence to this author at: Center for Women’s Health Research, Department of Obstetrics and Gynecology, University of New Mexico School of Medicine, 2211 Lomas Blvd., N.W., Albuquerque, NM 87131. Fax 505-272-6385; e-mail larry{at}hCGlab.com
 |
Abstract |
|---|
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-fetoprotein in the triple screen test, and with the addition of
inhibin-A in the quadruple marker test for detecting Down syndrome in
the second trimester of pregnancy. These tests have a limited detection
rate for Down syndrome: ~40% for hCG or free ß-subunit alone,
~60% for the triple screen test, and ~70% for the quadruple
marker test, all at 5%, or a relatively high, false-positive rate. New
tests are needed with higher detection and lower false rates.
Hyperglycosylated hCG (also known as invasive trophoblast antigen
or ITA) is a new test. It specifically detects a unique oligosaccharide
variant of hCG associated with Down syndrome pregnancies. We evaluated
this new Down syndrome-directed test in prenatal diagnosis. Methods: Hyperglycosylated hCG was measured in urine samples from women undergoing amniocentesis for advanced maternal age concerns at 14–22 weeks of gestation, 1448 with normal karyotype and 39 with Down syndrome fetuses.
Results: The median hyperglycosylated hCG value was 9.5-fold
higher in Down syndrome cases (9.5 multiples of the normal karyotype
median). The single test detected 80% of Down syndrome cases at a 5%
false-positive rate. Urine hyperglycosylated hCG was combined with
urine ß-core fragment (urine breakdown product of serum hCG free
ß-subunit), serum -fetoprotein, and maternal age-related risk.
This urine-serum combination detected 96% of Down syndrome cases at a
5% false-positive rate, 94% of cases at a 3% false-positive rate,
and 71% of cases at a 1% false-positive rate. These detection rates
exceed those of any previously reported combination of biochemical
markers.
Conclusions: Hyperglycosylated hCG is a new base marker for Down syndrome screening in the second trimester of pregnancy. The measurement of hyperglycosylated hCG can fundamentally improve the performance of Down syndrome screening protocols.
|
Introduction |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
and
ß, joined noncovalently. The -subunit is composed of 92 amino
acids with two N-linked oligosaccharide side chains. The ß-subunit
comprises 145 amino acids with two N-linked oligosaccharides in the
core of the subunit and four O-linked sugar structures on the
C-terminal extension. hCG usually is produced by villous
syncytiotrophoblast cells during pregnancy. It is also produced by
villous syncytiotrophoblast cells in hydatidiform mole or molar
pregnancy, and by non-villous intermediate cytotrophoblast and
intermediate syncytiotrophoblast cells in choriocarcinoma, cancer of
trophoblast cells. Recently, we examined the peptide and N-linked and
O-linked sugar structures of the separated - and ß-subunits of
purified hCG from normal pregnancies, molar pregnancies, and
choriocarcinoma (1). Although no significant difference was
observed in the peptide structure of the -subunit of hCG, variable
nicking or peptide bond cleavage at ß43–44, ß44–45, and ß47–48
was demonstrated on the ß-subunit of hCG. Significantly greater
nicking was observed in choriocarcinoma, compared with normal pregnancy
and molar pregnancy hCG preparations (1). Primarily mono-
and biantennary N-linked oligosaccharides, and tri- and
tetrasaccharide-type O-linked sugar units were found in the ß-subunit
of hCG in normal and molar pregnancy samples. A small proportion of
more complex triantennary N-linked oligosaccharides (0–30%) and
larger hexasaccharide-type O-linked sugar units (0–20%) were also
found in normal and molar pregnancy preparations. In contrast,
primarily triantennary N-linked oligosaccharides (up to 100%) and
hexasaccharide-type O-linked oligosaccharides (up to 100%) were found
in choriocarcinoma hCG (1). hCG with these larger N- and
O-linked oligosaccharides is called hyperglycosylated hCG (also known
as invasive trophoblast antigen, or ITA). One choriocarcinoma hCG preparation, hCG batch C5, had 100% hexasaccharide-type O-linked sugar units. In collaboration with Drs. S. Birken, A. Krichevsky, J. O’Connor, and R. Canfield of Columbia University, New York, NY, a monoclonal antibody was generated against C5 hCG (2). Using this antibody (monoclonal B152) an assay for hyperglycosylated hCG was developed (3). With this assay, O’Connor et al. (3) were able to show that primarily hyperglycosylated hCG was produced by trophoblast cells in the week after implantation in normal pregnancies. Using a similar assay, we showed that the proportion of hyperglycosylated hCG molecules declines as pregnancy advances. Using this assay, we found that the proportion of hyperglycosylated hCG detected was 26% of hCG molecules at 4–6 weeks, 11% at 6–8 weeks, and then 2.9% (third month) to 2.0% (third trimester) through the rest of pregnancy (Shahabi and Cole, unpublished data). Hyperglycosylated hCG accounted for virtually all hCG produced in five individuals with choriocarcinoma. The finding that hyperglycosylated hCG was produced by invasive or cancerous trophoblast cells in choriocarcinoma and by invasive normal pregnancy cells during aggressive implantation in the first week of gestation suggested that hyperglycosylated hCG is a product of separate trophoblast cells to normal hCG, and that it may be synthesized by invasive trophoblast cells (intermediate cytotrophoblast or fused cytotrophoblast cells). Furthermore, it was postulated that such cells are responsible for the small proportion of hyperglycosylated hCG molecules that are hyperglycosylated during most of the course of gestation.
In 1997, Cole et al. (4) used lectin affinity chromatography to demonstrate increased proportions of hyperglycosylated hCG molecules in Down syndrome (trisomy 21) pregnancies. A large prospective study was carried out in which urine samples from women undergoing amniocentesis or chorionic villous sampling for maternal age concerns were collected. Over a 31-month period, 1157 normal karyotype and 23 Down syndrome urine samples were collected and tested in the hyperglycosylated hCG assay (5). Cases were from 11 to 22 weeks of gestation. The test detected 79% of Down syndrome cases, with a 5% false-positive rate (5). This is an extremely high detection rate for a single analyte test for Down syndrome pregnancies. The reported detection rate is close to double that of other individual analytes such as normal hCG, hCG free ß-subunit (free ß-subunit), or inhibin-A (6)(7)(8).
The prospective urine study included only 2 Down syndrome and 98 normal karyotype cases from the first trimester of pregnancy (11–14 weeks). There thus are too few first-trimester samples to validate the performance of this test during early pregnancy. Here, we briefly reexamine the second-trimester cases used in this study (14–22 weeks of gestation) and show that the regression curve was adversely skewed by the first-trimester cases. A completely new study is presented with 389 normal karyotype and 18 Down syndrome cases to validate the prospective study. The combined set of second-trimester cases is examined (1448 normal and 39 Down syndrome cases). For the first time, trisomy 18 pregnancies are also investigated, as are combinations of urine hyperglycosylated hCG and serum markers.
Commercial kits are now being produced for detecting hyperglycosylated hCG, to replace or compliment the current triple and quadruple tests used for Down syndrome screening. We examine the needed conditions for shipping urine samples to laboratories for hyperglycosylated hCG measurements and for storage of urine samples in freezers.
|
Materials and Methods |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Three groups of normal and Down syndrome cases are described here. All were tested in the hyperglycosylated hCG assay. The first group consisted of the second-trimester cases (1059 normal and 21 Down syndrome cases) from our original (1134 normal and 23 Down syndrome cases) Down syndrome study population. These samples were collected prospectively over a 31-month period between 1996 and 1998 (5) and included six Down cases collected by the Prenatal Diagnosis Service postamniocentesis. The screening performance for this group of samples was corrected, reanalyzed, and accumulated here. The second group consisted of new second-trimester cases collected between July 1998 and April 1999. There were 389 normal karyotype and 18 Down syndrome second-trimester samples, which constituted the confirmation study. This included three Down cases collected by the Prenatal Diagnosis Service postamniocentesis. The third group combined the original and confirmation studies, with 1448 normal and 39 Down syndrome cases. In the third group, a total of six trisomy 18 cases were also identified (cases compiled 1996–1999).
In the majority of our cases undergoing genetic amniocentesis for
advanced maternal age concerns, blood for -fetoprotein was drawn at
the time of the procedure. This is useful as a marker of pregnancy
outcome. We identified 692 subjects (21 with Down syndrome and 671 with
normal karyotype) in the combined group that had blood taken at the
time of amniocentesis for -fetoprotein for determination.
-Fetoprotein concentrations were retrieved from the -fetoprotein
laboratory and added to our spreadsheet.
Samples (12 mL) were thawed overnight in the refrigerator and tested for hyperglycosylated hCG. For the original study, samples were tested sequentially in the spring of 1998. For the confirmation study, they were tested sequentially in the spring of 1999. In both studies, testing, calculation, and recording of results were carried out in a blind fashion, without knowledge of the karyotype.
The hyperglycosylated hCG test is a two-step sandwich-type ELISA. In brief, 96-well microtiter plates (Nunc Immulon-1; Fisher Scientific) are coated by incubation 16–24 h at 4 °C with capture antibody (0.2 mL per well of a solution containing 2.5 mg/L antibody B152 in 0.25 mol/L NaHCO3 and 0.1 mol/L NaCl). Plates are then washed three times with water and blotted dry, and wells are blocked with phosphate-buffered saline, pH 7.4 (Life Technologies), containing 10 g/L bovine serum albumin and 0.4 g/L sodium azide (both from Sigma). After incubation for 1 h at ambient temperature, plates are again washed three times with water, blotted dry, and used for the assay. The total assay volume is 0.2 mL: 0.1 mL of sample or calibrator and 0.1 mL of phosphate-buffered saline containing 1 g/L bovine serum albumin and 0.4 g/L sodium azide. C5 hCG (100% hexasaccharide-type O-linked oligosaccharides), the immunogen for antibody B152, that has been calibrated by amino acid analysis is used as the calibrator. C5 hCG at concentrations of 0, 60, 12, and 2.4 µg/L is added to quadruplicate wells of the plate. Urine samples are added at two- and fivefold dilutions. Buffer is added, and the plates are incubated 4 h at ambient temperature on an orbital plate shaker. Plates are again washed three times with water and blotted dry. Tracer antibody [0.2 mL of peroxidase-labeled anti-hCGß batch 4001 (Medix Biotech), 1:5000 titer in Tris, pH 7.3 (Sigma) containing 1 g/L bovine serum albumin and 1.9 g/L CaCl2 · 2 H2O] is added to each well. After an additional 2-h incubation at ambient temperature on the plate shaker, plates are again washed three times with water and blotted dry. Finally, 0.2 mL of substrate [TMB reagent (cat. no. T8665; Sigma) diluted 1:1 with water] is added to each well. After a 15-min incubation at ambient temperature, the reaction is stopped by the addition of 0.050 mL of 2 mol/L HCl. The plates are read on a microtiter plate reader at 450 nm, and the calibrators are plotted. The points best fit a cubic function, which was used to calculate sample values.
Plates included a quality control. The concentration was 21 µg/L or approximately in the middle of the calibration curve. The interplate/interassay variance was calculated. The mean result was 21 ± 1.8 µg/L, indicating an interassay variance (CV) of 8.9%.
The specificity of the hyperglycosylated hCG assay was investigated.
Eight antigens were tested at multiple dilutions (Table 1
and Fig. 1
). These included five preparations of pure intact hCG
(non-nicked, nicked, or hyperglycosylated) and samples of pure
hyperglycosylated hCG free ß-subunit, pure free ß-subunit missing
the C-terminal extension, and pure human luteinizing hormone (hLH). All
preparations were calibrated by amino acid analysis. The peptide and
carbohydrate structures of the five intact hCG preparations have been
determined (Table 1
) (1). The assay clearly discriminates
the two hyperglycosylated hCG preparations (C7 hCG and C5 hCG from
choriocarcinoma) from all other hCG-related antigens (Fig. 1
). These
hCG preparations have 57% and 48% triantennary N-linked
oligosaccharides, and 68% and 100% hexasaccharide-type O-linked
oligosaccharides (Table 1
). The relative immunoreactivities of these
two hCG preparation (93% and 100%, respectively) are much greater
than those of the nonhyperglycosylated hCG or nicked hCG preparations
(P8 and P3 hCG from normal pregnancies, 10% and 12%, respectively).
We correlated the structural features of the 5 hCG preparations with
the relative immunoreactivities. The most significant relationship was
between immunoreactivity and the percentage of hexasaccharide type
O-linked oligosaccharides (r2 = 0.94).
The low activities of P8 and P3 hCG correlated with the low
hexasaccharide content in these preparations (13% and 12%,
respectively). Minimal activity was detected with C5 hCG free
ß-subunit (10%), and no measurable activity was detected with hLH.
It is inferred that the assay is specific for hyperglycosylated hCG,
possibly for molecules with hexasaccharide-type O-linked
oligosaccharides. It is by this means that it has poor recognition of
pregnancy hCG (P3 and P8 hCG) and near total recognition of
choriocarcinoma hCG molecules (C7 and C5 hCG). A partial response was
detected with molar pregnancy hCG (M4 hCG).
|
|
hCG ß-core fragment concentrations were determined by a method similar to that for the hyperglycosylated hCG assay. The only difference was the use of a different coating antibody, B210 (gift from Drs. S. Birken and R. Canfield, Columbia University, NY), and a different calibrator (P13 ß-core fragment). The ß-core fragment assay detected hCG ß-core fragment. Although this assay had 100% activity with the hLH ß-core fragment calibrator, no measurable activity was found with hCG free ß-subunit or any of the intact-hCG calibrators.
Results were normalized to spot urine creatinine concentrations. Creatinine was determined using a commercial kit, cat. no. 555A (Sigma), and a microtiter plate adaptation of the protocol. Calibrators (0, 2.5, 1.5, 0.5, and 0.2 g/L creatinine) and urine samples (0.053 mL per well, in triplicate) were added to a 96-well microtiter plate. Alkaline picrate reagent was prepared fresh (5 parts of solution plus 1 part sodium hydroxide) and added (250 µL) to the wells. The plates were incubated 15 min at ambient temperature. The absorbance was measured at 492 nm by a plate reader, and the calibrators were plotted. The points best fit a cubic equation, which was used to calculate sample concentrations (g/L).
In the original study with 1134 normal karyotype samples, hyperglycosylated hCG results were first normalized for urine concentration. The hyperglycosylated hCG concentration (µg/L) was divided by the spot creatinine concentration (µg/g creatinine). A relationship was observed between the normalized hyperglycosylated values and the creatinine concentration. Whereas samples with low creatinine concentrations were giving unduly high creatinine-normalized values, those with high creatinine concentration were giving unduly low creatinine-normalized values (5). An equation was derived to correct this error: c' = (0.877c) + 0.107, where c is the actual creatinine concentration and c' is the corrected value. A correction algorithm was not needed with creatinine-normalized intact hCG and ß-core fragment concentrations.
Results were analyzed using the multiple of the median methods of Royston and Thompson (9). Creatinine-normalized hyperglycosylated hCG and ß-core fragment concentrations were each plotted against gestational age. Weekly median values were determined for normal karyotype samples, and a regression equation was calculated that best fit the median values. Using the equation, multiples of the calculated median (MoM) were determined for all samples. In all three groups of cases, median values best fit a simple logarithmic equation. In all three groups, probability plots with lines defined by log-gaussian distribution were used to show that MoM values fit a log-gaussian distribution for both normal pregnancy and Down syndrome data.
To assess screening performance, MoM values, log MoM values, median values, and log mean and log SD (estimated by the 10th–90th centile difference of the log MoM values, divided by 2.56) were determined for both Down syndrome and normal pregnancies. The detection rates were determined from the proportion of Down syndrome pregnancies exceeding a specific centile of the normal karyotype population. ROC curves were used to compare detection rates (percentage of Down syndrome hyperglycosylated hCG results exceeding a specified centile of normals) and false-positive (100 - specified centile of normals) rates, and to determine the extent of discrimination between affected and unaffected pregnancies. Univariate and multivariate gaussian models were used to predict detection rates for hyperglycosylated hCG and combinations of hyperglycosylated hCG and other biochemical markers, and hyperglycosylated hCG and maternal age-related risk, considering the general age distribution of the population of the United States (10).
|
Results |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
). Second-trimester (14–22 weeks of gestation) results were
plotted against gestational age (Fig. 2
A). Second-trimester median values best fit a simple logarithmic
equation: median = 6383 x 0.706ga. We
used this equation to calculate MoM values and centiles. MoM values for
both normal and Down syndrome cases fit a log-gaussian distribution
between the 5th and 95th centiles. With the new equation, the median of
the Down syndrome cases was 8.4-fold higher than that of the normals
(median MoM = 8.4). The Down syndrome case results (mean log
MoM ± SD, 1.05 ± 0.51) were very significantly different
from normals (-0.001 ± 0.43; t-test,
P = 3.5 x 10-26).
Seventeen of 21 (81%) Down syndrome cases exceeded the 95th
centile of the normal karyotype cases. From the ROC curve, 81%
detection was indicated at a 5% false-positive rate (Table 2
). The
area under the ROC curve was 0.96.
|
|
The second-trimester only results were superior to the combined first- and second-trimester data, The attempt to combine the two trimesters with one median equation may have diminished the observed screening performance of hyperglycosylated hCG (median MoM, detection rate, and area under the ROC curve). Therefore, the two trimesters should be evaluated separately.
The finding of a single analyte test detecting 
80% of Down syndrome
cases is exceptional. A blind repeat study was needed to confirm these
finding. We collected urine samples from 389 normal karyotype and 18
Down syndrome cases undergoing amniocentesis at 14–22 weeks of
gestation for advanced maternal age concerns (Table 2
).
Hyperglycosylated hCG results were plotted against gestational age
(Fig. 2B
). Weekly medians were determined. Weekly median values best
fit a simple logarithmic equation: median = 6050 x
0.716ga. We used this equation to calculate MoM
values and centiles. MoM values fit a log-gaussian distribution between
the 5th and 95th centiles. With this equation, the median of the Down
syndrome cases was 9.9-fold higher than normals (median MoM =
9.9). Again, a very significant difference was observed between Down
syndrome (mean log MoM ± SD, 0.96 ± 0.42) and normal cases
(-0.046 ± 0.43; t-test, P = 5.2
x 10-16). Fourteen of 18 (78%) Down syndrome
cases exceeded the 95th centile of normal karyotype cases. From the ROC
curve, 82% detection was indicated at a 5% false-positive rate (Table 2
). The area under the ROC curve was 0.96.
The results of the two studies (original study - second trimester
only, and the confirmation second-trimester study) virtually overlaid
each other. The regression equations, the variation of normal samples
expressed as log SD values, the detection rates, and the areas under
the ROC curves were all either the same or close to being the same
(Table 2
). We interchanged the samples, combining the confirmation
study normal samples with the original study Down syndrome samples, and
vice versa. After substitution, the same median MoM values were noted
for Down syndrome cases, and the same proportion of sample exceeded the
95th centile. We inferred that the confirmation study verified the
original study and that it was appropriate to combine the two groups of
samples.
The two groups of samples were combined, for a total of 1448 normal
karyotype and 39 Down syndrome cases from 14 to 22 weeks of gestation.
Hyperglycosylated hCG results were again plotted against gestational
age (Fig. 2C
). Weekly medians were determined. Weekly median values
again best fit a simple logarithmic equation, median = 6180
x 0.710ga. We used this equation to calculate
MoM values and centiles. MoM values fit a log-gaussian distribution
between the 5th and 95th centiles. With this equation, the median of
the Down syndrome cases was 9.5-fold higher than that for the normals
(9.5 MoM). A very significant difference was observed between Down
syndrome (mean log MoM ± SD, 1.02 ± 0.47) and normal
karyotype cases (-0.019 ± 0.43; t-test, P
<0.001). Thirty-one of 39 (79%) Down syndrome cases exceeded the 95th
centile of the normals. From the ROC curve, 80% detection was
indicated at a 5% false-positive rate (Table 2
and Fig. 3
). The area under the ROC curve (0.96) again indicated 96%
discrimination between normal and Down syndrome cases.
|
The true-positive Down syndrome cases included eight of the nine cases collected by the Prenatal Diagnosis Service postamniocentesis at 17–22 weeks of gestation. No significant difference was observed between the postamniocentesis (log MoM, 0.98 ± 0.43) and regularly collected (log MoM, 1.05 ± 0.52) cases.
clinical utility of hyperglycosylated hCGMEASUREMENTS
The original and confirmation studies described above examined
mostly high-risk cases for Down syndrome, i.e., older individuals
[mean age 36 ± 3.0 years (original group) and 36.9 ± 3.1
year (confirmation group)] undergoing amniocentesis for advanced age
concerns. Gestational age was determined by ultrasound measurements.
Down syndrome screening tests are most commonly used for low-risk or
younger individuals. These tests use less precise last menstrual
period-based measurements of gestational age. We investigated how
maternal age and the accuracy of the measurement of gestational age
influence hyperglycosylated hCG Down syndrome screening utility.
We used the combined group of samples to investigate the relationship between hyperglycosylated hCG values and maternal age. Linear regression indicated the significant absence of a relationship between maternal age and MoM values (r2 = 0.0002). We divided the cases into two groups: low-risk/younger women (19–34 years; 169 normal and 7 Down syndrome cases) and high-risk/older women (35–49 years; 1279 normals and 32 Down syndrome cases). We examined the mean log MoM ± SD of the two normal karyotype groups (-0.016 ± 0.43 and -0.011 ± 0.43, respectively). By t-test, the two groups were statistically indistinguishable (P = 0.994). We examined the mean log MoM ± SD of the two Down syndrome groups (0.922 ± 0.49 and 0.983 ± 0.47, respectively). By t-test, no clear difference was observed (P = 0.77). We examined the screening statistics of the two groups. Eighty-six percent of the low-risk/younger group and 78% of the high-risk/older group of Down syndrome cases exceeded the 95th centile of normal karyotype cases. It was concluded that hyperglycosylated hCG measurements are independent of maternal age.
We investigated normalization of hyperglycosylated hCG results to last menstrual period-based gestational age calculations. Hyperglycosylated hCG results were plotted against gestational age, weekly medians determined, and a new regression equation determined. MoM statistics were calculated and centiles determined. As with ultrasound-based data, 31 of 39 (79%) Down syndrome cases exceeded the 95th centile. No loss of sensitivity was indicated. It is inferred that hyperglycosylated hCG is not maternal age specific and can be used with last menstrual period-base gestational age calculations, and so may be suitable for general low-risk screening for Down syndrome.
Regular hCG is the principal or base analyte test for current Down
syndrome screening protocols. Regular serum hCG detects ~40% of Down
syndrome case at a 5% false-positive rate
(6)(7). Regular serum hCG is combined with
maternal age-related risk, serum -fetoprotein, unconjugated estriol,
and more recently, with a fifth variable, inhibin A, to screen for Down
syndrome pregnancy. Taking all five screening markers together (the
quadruple test), screening performance did not reach the sensitivity
observed for hyperglycosylated hCG. We considered using
hyperglycosylated hCG as the principal or base analyte test and
combining it with other markers. Using ROC analysis, hyperglycosylated
hCG alone detected 80% of Down syndrome cases at a 5% false-positive
rate, 67% at a 3% false-positive rate, and 49% and a 1%
false-positive rate (area under ROC curve, 0.97; Fig. 3
). Previously,
we have shown that urine ß-core fragment complements
hyperglycosylated hCG measurements (5). We combined
hyperglycosylated hCG results with age-related risk and ß-core
fragment measurements. This combination detected 92% of Down syndrome
cases at a 5% false-positive rate, 79% at a 3% false-positive rate,
and 53% and a 1% false-positive rate (area under ROC curve, 0.97).
Serum -fetoprotein is routinely determined to evaluate risk for
neural tube defect. Serum -fetoprotein data were found for 671 of
the normal and 21 of the Down syndrome cases. We combined the two urine
tests with age-related risk and serum -fetoprotein. This combination
detected 96% of Down syndrome cases at a 5% false-positive rate,
94% at a 3% false-positive rate, and 71% at a 1%
false-positive rate (Fig. 3
). The area under the ROC curve was
0.98.
Trisomy 18 is a less common genetic abnormality in second-trimester pregnancies. Over 3.5 years, we accumulated six cases of trisomy 18. The median for the hyperglycosylated hCG values was 10-fold lower than that for normals (0.10 MoM of normal karyotype samples). All six cases had results below 0.6 MoM. Four of the six cases had values below both the fifth (0.14 MoM) and the third (0.11 MoM) centiles. Hyperglycosylated hCG may also be useful for identifying trisomy 18.
shipping and storage conditions
We investigated the stability of hyperglycosylated hCG
immunoreactivity in urine samples from normal and Down syndrome cases
to asses the impact on shipping and storage conditions. Ten normal and
10 Down syndrome case urine samples refrigerated after collection and,
once frozen, were rapidly thawed and tested. Aliquots (1 mL) were
stored for 3 days in a refrigerator (4 °C), in a bench drawer
(22 °C), or in a heating block (37 °C). As shown in Table 3
, no significant difference was observed between the results of
normal karyotype or Down syndrome samples after 3 days at room
temperature (95% ± 8.7% and 96% ± 4.9% recovery of control
immunoreactivity, respectively) or after 3 days in a refrigerator (95%
± 6.9% and 98% ± 2.3%, respectively). In addition, no significant
difference was observed between the losses observed in normal karyotype
and Down syndrome samples. These losses were within in the range of the
interplate/interassay variance of the hyperglycosylated hCG assay
(8.9%). More significant losses in immunoreactivity occurred after 3
days at 37 °C: 81% ± 19% of normal and 83% ± 10% of Down
syndrome sample immunoreactivity was recovered (t-test,
22 °C vs 37 °C, P = 0.05 for normal and
P = 0.0017 for Down syndrome samples). It is inferred
that urine samples can be shipped or stored for 3 days at 4 or 22 °C
before assay.
|
In the original and confirmatory Down syndrome screening studies,
samples were frozen in -20 °C freezers, 12 mL of urine in 15-mL
vials. Sample were later thawed overnight (slowly) in a refrigerator
and tested for hyperglycosylated hCG. We examined the effect of these
freezing and thawing procedures on hyperglycosylated hCG
immunoreactivity (Table 4
). When urine samples from 84 normal karyotype pregnancies were
refrozen, thawed in the refrigerator, and assayed again, there was an
overall loss in hyperglycosylated hCG immunoreactivity (86% ± 39%
remained of prior hyperglycosylated hCG activity). When urine from six
Down syndrome cases was similarly refrozen and assayed again, there was
a much greater loss of immunoreactivity (48% ± 12% of the original
activity remained; t-test, Down syndrome vs normal
karyotype, P <0.0001). We examined the normal karyotype
samples with the lowest and with the highest concentration of
hyperglycosylated hCG (Table 4
). Those with lowest concentration had
the better recovery (95% ± 7.5% of the original activity remained;
P = 0.02), and those highest concentration had the
poorer recover (62% ± 37% of the original activity remained;
P = 0.08). The poor recovery found in large aliquots of
urine from Down syndrome pregnancies that had been frozen and then
thawed may be related to the very high concentration of
hyperglycosylated hCG.
|
In a further experiment, urine samples were tested fresh (1–2 days in refrigerator) before any freezing and then after an initial freeze-thaw cycle. A similar loss in hyperglycosylated hCG immunoreactivity was detected (47% ± 22% of the original activity was recovered; t-test, Down syndrome vs normal karyotype, P <0.01). It was inferred that freezing and thawing 12-mL samples of urine led to a loss of hyperglycosylated hCG immunoreactivity, particularly in samples with a high concentration of hyperglycosylated hCG.
Further studies were carried out to confirm these findings and to develop a solution for the freeze-thaw problem. Experiments were repeated with 12-mL volumes of urine and similar procedures but with 200 mL/L glycerol, neutral pH buffer (50x concentrated Tris, pH 7.5), and antibiotics (50x concentrated penicillin-streptomycin-fungizone) added to the urine samples. None of the additives significantly changed or improved the freeze-thaw problem. Believing that the slow freezing of packed racks of 12-mL urine samples in a regular -20 °C freezer and slow thawing of the tubes overnight in a refrigerator may be responsible for the losses, we repeated the experiment with smaller urine aliquots. Ten urine samples from normal and 10 from Down syndrome cases were rapidly thawed, tested for hyperglycosylated hCG, and then refrozen again in 1-mL aliquots. Three days later, the small aliquots were rapidly thawed in a water bath and retested. No significant loss was noted in hyperglycosylated hCG immunoreactivity (97% ± 5.1% and 96% ± 4.5%). No difference was apparent between normal and Down syndrome cases. It was inferred that the freezing of racks of 12-mL samples, as was carried out in the Down syndrome screening studies described above, might have been detrimental to the results. The clear implication is that better screening data might have been obtained if the original measurements had been performed before freezing the samples. Conversely, if urine samples must be frozen, they should be stored in small aliquots (<1 mL) in noncrowded vials and thawed rapidly in a water bath before assay.
|
Discussion |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
-fetoprotein
(12) and unconjugated estriol (13). To these
three tests were added maternal age-related risk, a fourth marker, to
optimize Down syndrome screening performance
(14)(15). This triple biochemical marker test
became widely used for screening for Down syndrome between 15 and 22
weeks of gestation. This triple test is far from perfect. It detects
~60%, or misses ~40%, of Down syndrome cases
(6)(15)(16). It has a 5%
false-positive rate, so that a relatively large number of amniocentesis
procedures must be performed to identify one case with a Down syndrome
fetus. The triple-marker is criticized by physicians and patients
because of its poor screening performance. Some laboratories are now
adding a fifth marker (or a fourth biochemical test) to the hCG-based
mixture, inhibin A. This test may, depending on the report, boost
performance by an additional 10%
(7)(17)(18). It is still, however,
an unsatisfactory screening test. New tests are needed, not just to
replace the triple-marker test or inhibin-containing quadruple-marker
test, but also to supplant the hCG assay, the base test at the root of
these combinations. Recently, we described a large prospective study examining the screening performance of an independent marker, urine hyperglycosylated hCG, in the first and second trimesters of pregnancy (5). In this report, we started by realizing the restraints of this data set and showing better performance when limiting it to the second trimester of pregnancy, 14–22 weeks of gestation. An 81% detection rate was indicated at a 5% false-positive rate. This is the highest detection rate ever reported for a single screening marker (without adding maternal age-related risk). We now describe a blind confirmation study. An 82% detection rate was indicated at a 5% false-positive rate. The original data set for 14-22 weeks of gestation and the new confirmatory data sets overlapped. Down syndrome and normal cases were interchangeable with no decline in screening performance. When the two data sets were combined, we had 1448 normals and 39 Down syndrome cases with 80% detection at a 5% false-positive rate. This one urine test outperforms all current individual analyte tests and all currently available combinations of four or five screening markers.
This study was limited to urine specimens. A preliminary study has now
been completed indicating that hyperglycosylated hCG can also be
measured in serum samples (8). Unfortunately, gel separator
tubes, like those used in tiger top or similar tubes, can interfere
with hyperglycosylated hCG detection (Cole, unpublished observations).
Most libraries of normal and Down syndrome serum samples have been
accumulated using gel separator tubes. This has slowed down the
evaluation of hyperglycosylated hCG as a serum marker. Thus, to date,
our studies have been carried out with more easily obtainable urine
samples. Urine hyperglycosylated hCG may be an effective replacement
for serum analytes in Down syndrome screening. We considered the
combination of urine hyperglycosylated hCG, urine ß-core fragment
(renal degradation product of serum hCG free ß-subunit), and maternal
age-related risk as a screening test for Down syndrome pregnancies.
This combination detected 92% of Down syndrome cases at a 5%
false-positive rate. The area under the ROC curve was 0.97. Although
collection of urine is less invasive and may be preferred by patients,
venipuncture must be carried out for serum -fetoprotein
determination, which is needed for detection of neural tube defects
(19). We considered the addition of serum -fetoprotein to
the package of age and two urine screening markers. This urine-serum
triple marker combination detected 96% of Down syndrome cases at a 5%
false-positive rate. The area under the ROC curve was 0.98.
Ninety-six percent is an extremely high detection rate for Down syndrome pregnancies. We examined the ROC curve for the urine-serum triple marker combination and the use of different false-positive rates. Ninety-four percent detection was indicated at a 3% false-positive rate and 71% detection at a 1% false-positive rate. The use of a 3% or even a 1% false-positive rate would lead to a major reduction in the number of amniocenteses that must be performed because of biochemical screening to identify a single Down syndrome case. This would produce major reductions in the miscarriages of normal fetuses as a result of the amniocentesis procedure and in the overall cost of prenatal screening. It could also renew the confidence of both physicians and patients in the screening process.
All studies were completed with women undergoing amniocentesis for advanced maternal age concerns. This is a mostly older group of women than that usually tested by biochemical screening methods. Furthermore, gestational age was determined by ultrasound, a more accurate method than extrapolation from the date of last menstrual period used in biochemical screening methods. We investigated the effects of maternal age and methods of determination of gestational age on hyperglycosylated hCG screening statistics. No loss of sensitivity was found when the combined group of samples was normalized to the last menstrual period-based gestational age. In the combined group of samples, no statistically significant relationship was present between hyperglycosylated hCG and maternal age. The mean MoM value was compared in normal cases under and over 35 years of age. The two groups were statistically indistinguishable. It is inferred that hyperglycosylated hCG is suitable for general low-risk biochemical screening for Down syndrome pregnancies.
The serum triple-marker test can also be used for detecting trisomy 18 (20). Unduly low hCG concentrations (less than the fifth centile of normal cases) are detected in approximately two-thirds of second-trimester pregnancies with trisomy 18. We investigated the use of hyperglycosylated hCG as a marker of trisomy 18. In four of six cases, hyperglycosylated hCG MoM values were below the fifth centile. Thus, hyperglycosylated hCG may also be useful for identifying trisomy 18.
The effect of storage or shipping at different temperatures was investigated. Although no significant losses of immunoreactive hyperglycosylated hCG were noted after 3 days of storage at -4 or 22 °C, losses were noted after 3 days at 37 °C. Although no significant losses were noted when urine was frozen and thawed in 1-mL aliquots, losses were found after urine was frozen in large 12-mL aliquots. We conclude that urine samples should either be shipped to laboratories with refrigerant or shipped frozen in 1-mL or smaller vials. Samples should be stored at the testing center either in a refrigerator or at ambient temperature (up to 22 °C) for no more than 3 days if they have never been frozen, or kept in 1-mL or smaller aliquots if they are shipped frozen.
In conclusion, the combination of urine hyperglycosylated hCG,
urine ß-core fragment, serum -fetoprotein, and maternal
age-related risk can detect 96% of Down syndrome cases at a 5%
false-positive rate or 94% of cases at a 3% false-positive rate.
Hyperglycosylated hCG-based tests and multiparameter algorithms may be
able to entirely replace hCG-based tests (triple-marker test and
quadruple-marker test incorporating inhibin A) and their algorithms in
screening for Down syndrome.
|
Acknowledgments |
|---|
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
- and ß-subunits of human chorionic gonadotropin from normal and aberrant pregnancy and choriocarcinoma. Endocrine 1997;7:15-32.
[ISI][Medline]
[Order article via Infotrieve]
-fetoprotein and fetal chromosome abnormalities. Am J Obstet Gynecol 1984;148:886-894.
[ISI][Medline]
[Order article via Infotrieve]
-fetoprotein, unconjugated estriol and hCG. J Clin Immunoassay 1990;13:30-33.
The following articles in journals at HighWire Press have cited this article:
|
|
 |
|
  L. Valmu, H. Alfthan, K. Hotakainen, S. Birken, and U.-H. Stenman Site-specific glycan analysis of human chorionic gonadotropin {beta}-subunit from malignancies and pregnancy by liquid chromatography--electrospray mass spectrometry Glycobiology, December 1, 2006; 16(12): 1207 - 1218. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 U.-H. Stenman, A. Tiitinen, H. Alfthan, and L. Valmu The classification, functions and clinical use of different isoforms of HCG Hum. Reprod. Update, November 1, 2006; 12(6): 769 - 784. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 G. E. Palomaki, G. J. Knight, L. M. Neveux, R. Pandian, and J. E. Haddow Maternal Serum Invasive Trophoblast Antigen and First-Trimester Down Syndrome Screening Clin. Chem., August 1, 2005; 51(8): 1499 - 1504. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
M. J.N. Weinans, U. Sancken, R. Pandian, J. M.W. van de Ouweland, H. W.A. de Bruijn, J. P. Holm, and A. Mantingh Invasive Trophoblast Antigen (Hyperglycosylated Human Chorionic Gonadotropin) as a First-Trimester Serum Marker for Down Syndrome Clin. Chem., July 1, 2005; 51(7): 1276 - 1279. [Full Text] [PDF]
|
|
|
|
|
|
R. Rej Clinical Chemistry through Clinical Chemistry: A Journal Timeline Clin. Chem., December 1, 2004; 50(12): 2415 - 2458. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. E. Palomaki, L. M. Neveux, G. J. Knight, J. E. Haddow, and R. Pandian Maternal Serum Invasive Trophoblast Antigen (Hyperglycosylated hCG) as a Screening Marker for Down Syndrome during the Second Trimester Clin. Chem., October 1, 2004; 50(10): 1804 - 1808. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Pandian, L. A. Cole, and G. E. Palomaki Second-Trimester Maternal Serum Invasive Trophoblast Antigen: A Marker for Down Syndrome Screening Clin. Chem., August 1, 2004; 50(8): 1433 - 1435. [Full Text] [PDF]
|
|
|
|
 |
|
 J.-L. Frendo, J. Guibourdenche, G. Pidoux, M. Vidaud, D. Luton, Y. Giovangrandi, D. Porquet, F. Muller, and D. Evain-Brion Trophoblast Production of a Weakly Bioactive Human Chorionic Gonadotropin in Trisomy 21-Affected Pregnancy J. Clin. Endocrinol. Metab., February 1, 2004; 89(2): 727 - 732. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
G. E. Palomaki, G. J. Knight, M. M. Roberson, G. C. Cunningham, J. E. Lee, C. M. Strom, and R. Pandian Invasive Trophoblast Antigen (Hyperglycosylated Human Chorionic Gonadotropin) in Second-Trimester Maternal Urine as a Marker for Down Syndrome: Preliminary Results of an Observational Study on Fresh Samples Clin. Chem., January 1, 2004; 50(1): 182 - 189. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. Pandian, J. Lu, and J. Ossolinska-Plewnia Fully Automated Chemiluminometric Assay for Hyperglycosylated Human Chorionic Gonadotropin (Invasive Trophoblast Antigen) Clin. Chem., May 1, 2003; 49(5): 808 - 810. [Full Text] [PDF]
|
|
|
|
|
|
S. Birken, P. Berger, J.-M. Bidart, M. Weber, A. Bristow, R. Norman, C. Sturgeon, and U.-H. Stenman Preparation and Characterization of New WHO Reference Reagents for Human Chorionic Gonadotropin and Metabolites Clin. Chem., January 1, 2003; 49(1): 144 - 154. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. C.W. Poon, T. S.K. Mok, A. T.C. Chan, C. M.L. Chan, V. Leong, S. H.T. Tsui, T. W.T. Leung, H. T.M. Wong, S. K.W. Ho, and P. J. Johnson Quantification and Utility of Monosialylated {alpha}-Fetoprotein in the Diagnosis of Hepatocellular Carcinoma with Nondiagnostic Serum Total {alpha}-Fetoprotein Clin. Chem., July 1, 2002; 48(7): 1021 - 1027. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 |
) and P8 hCG
(•) were from normal first-trimester pregnancies, M4 hCG (
) was
from a case of hydatidiform mole, and C5 hCG (
) and C7 hCG (
)
were from patients with choriocarcinoma. The characteristics of the hCG
preparations are described in Table 1
), a free ß-subunit
missing the C-terminal peptide (
), and hLH, NIH batch I-1 (
).
