| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Technical Briefs |
a author for correspondence: fax 44 (0) 1708381486, e-mail Kevin_Spencer{at}Compuserve.com
Prenatal screening for trisomy 21 (Down syndrome) based on the
analysis of biochemical markers in maternal serum during the second
trimester of pregnancy is becoming an established part of obstetric
practice (1)(2). Several markers have been
investigated in the second trimester, and of these, a combination of
-fetoprotein and free ß-human chorionic gonadotropin (hCG) has
been shown in retrospective (3) and prospective studies
(4)(5) to achieve detection rates of 65–75%
(with a 5% false-positive rate) in routine practice. Similarly,
combinations including -fetoprotein, total hCG, and unconjugated
estriol have been shown in prospective studies to perform almost as
well, with detection rates on average of 65% (6).
Recent developments in this area of prenatal screening have focused on three specific areas. The first area is that of refinements to the risk algorithm, leading to improved detection efficiency by taking into account a variety of factors that influence the analyte concentrations, for example, multiple pregnancies (7), maternal weight (8)(9)(10), insulin-dependent diabetes mellitus (11)(12), gravidity (13), ethnicity (9), previous pregnancy results (14)(15), and smoking (16). Correction for these factors seeks to reduce the between-patient variance, leading to reductions in the false-positive rate. The second area involves the investigation of possible new markers of trisomy 21, such as dimeric inhibin A, which has been reported to add anything from 3% (17) to 22% (18) to the detection rate using existing double or triple marker approaches. In addition, beta core in urine (19) and free ß-hCG in urine (20) have both been proposed as possible markers of trisomy 21. Neither the urine approach nor the addition of dimeric inhibin A have yet been shown conclusively to be better than existing procedures, and indeed their use is still somewhat controversial. The third area of investigation is the development of screening programs in the first trimester, in which a combination of ultrasound (nuchal translucency) and the biochemical markers pregnancy-associated plasma protein A and free ß-hCG enabled detection rates of ~90% (21).
Recently, Lambert-Messerlian et al. (22) in this Journal reported increased concentrations of prostate-specific antigen (PSA) in the maternal serum of second trimester pregnancies affected by trisomy 21. Although PSA was originally thought to be a marker specific to the prostate, a variety of biological fluids and extracts of female tissues has now been shown to contain and produce a material that reacts positively in sensitive PSA immunoassays. The original observation leading to the study of PSA in cases of trisomy 21 was a study in amniotic fluid (AF) that observed low values of PSA in four of six cases (23).
In this study, we endeavored to reproduce these initial observations, using a similarly sensitive commercially available PSA assay (24), and to investigate the potential value of a sensitive PSA assay in screening for trisomy 21 in the second trimester of pregnancy.
The study population consisted of cases of trisomy 21 in women who
presented through the Neural Tube Defect and Trisomy 21 screening
program in this laboratory during the period 1994–1998. The serum
collected from each woman was stored as aliquots at -20 °C after
being analyzed in routine screening with assays for -fetoprotein and
free ß-hCG. Our local ethics committee approved the use of stored
serum banks for research. The samples plus analytical and clinical data
were collated on the basis of abnormal birth outcome, with karyotyping
or cytogenetic confirmation after midtrimester amniocentesis. A total
of 43 singleton pregnancies associated with trisomy 21 were identified
with maternal serum samples taken between 14 and 17 weeks of gestation.
Each case of trisomy 21 was matched for gestational age, length of
storage (±14 days), and number of freeze-thaw cycles with five samples
from unaffected singleton pregnancies (total, 215 controls).
Of the cases of trisomy 21, 32 of 43 were identified by the screening program to have a trisomy 21 term risk of <1 in 250, and 11 of 43 were identified as a result of a live- born trisomy 21 baby. The mean maternal ages were 28.4 years (range, 17–43 years) for the controls and 34.2 years (range, 18–43 years) for the trisomy 21 cases. The median gestational age was 115 days (range, 98–125 days) for the controls and 108 (range, 98–120 days) for the trisomy 21 cases. The mean storage time was 30.2 months (range, 4–49 months) for samples from controls and 30.0 months (range, 4–49 months) for samples from trisomy 21 cases.
AF samples were identified from those in a previous study (25)(26). A total of 38 AF samples from cases of trisomy 21 with gestational ages from 16 to 20 weeks were available for analysis. A total of 110 control samples with gestational ages from 16 to 20 weeks were identified as matched for the number of freeze-thaw cycles and length of storage (±1 month).
PSA concentrations were measured with an Immulite automated third generation ultrasensitive immunochemiluminescent assay (Diagnostic Products Corp., Los Angeles, CA). The assay has been shown by Ferguson et al. (24) to have comparable sensitivity and clinical performance to the assay used in the initial studies with cases of trisomy 21 (22, 23). The detection limit of the assay in our hands analyzing 30 replicates of the assay zero diluent was 2 ng/L (+2 SD of the measurement of the zero diluent). The between-run imprecision (CV) was 5.5% at 7786 ng/L, 2.6% at 2704 ng/L, and 3.5% at 32.4 ng/L. The precision and sensitivity are similar to those observed by Ferguson et al. (24).
To assess the suitability of using the Immulite third generation PSA assay to measure PSA concentrations in AF, which has ~1/10 of the protein concentration of serum, a series of dilution experiments were carried out in which five male serum samples were diluted in a both zero diluent and phosphate-buffered saline (to simulate the low protein concentrations in AF). The results of the assay validation showed parallel and almost identical 100% recovery in the buffer and protein matrix and indicated no problem with matrix effects when measuring samples with low protein concentrations.
Results for each analyte were expressed in multiples of the median (MoM) for unaffected pregnancies of the same gestational age, derived from the regressed log10 medians for the analyte when appropriate and from the overall control population median for analytes with concentrations unrelated to gestational age. Statistical analysis of data was performed using Astute, a statistical software add-in for Microsoft Excel 5 (DDU Software, University of Leeds, UK)
Table 1
shows the observed and regressed median data for AF PSA and the
observed medians for maternal serum PSA across the second trimester
period. For AF, the median values showed a statistically significant
increase (P <0.01, Mann–Whitney U-test) across
the second trimester period, whereas for maternal serum there was no
significant change (P >0.05) across this period. The
overall control population median in maternal serum was 9 ng/L, with
PSA 25- to 100-fold higher in AF. The median concentrations in AF are
comparable with those observed by Melegos et al. (23).
Lambert-Messerlian et al. (22) stated that the median
concentrations in maternal serum varied with gestational age (by
~20%); however, they showed no data. Our study cannot confirm a
variation of maternal serum PSA with gestational age.
|
In maternal serum, the median PSA MoM in the trisomy 21 group was 0.89
(95% confidence interval, 0.67–2.56); this was not significantly
different from the controls (P = 0.2543, Mann–Whitney
U-test). In the control group the 10th and 90th centiles
were 0.22 and 4.56, respectively, with a 5th to 95th centile of
0.11–13.0. In the trisomy 21 group the 10th and 90th centiles were
0.22 and 7.22, respectively. The distribution of PSA MoM in the trisomy
21 group with gestational age is shown in Fig. 1
B. The distribution of PSA MoM fitted a gaussian distribution
after log10 transformation in the control group and the
trisomy 21 group. The parameters of the distribution are mean
log10 (controls = -0.003; trisomy 21 group =
-0.104) and log10 SD (controls = 0.5635; trisomy 21
group = 0.5573).
|
In AF the median PSA MoM in the trisomy 21 group was 0.80 (95%
confidence interval, 0.49–1.68); this was not significantly different
from the control group (P = 0.4117, Mann–Whitney
U-test). In the control group the 10th and 90th centiles
were 0.23 and 4.15, respectively, with a 5th to 95th centile of
0.19–5.40. In the trisomy 21 group the 10th and 90th centiles were
0.16 and 7.18, respectively. The distribution of PSA MoM in the trisomy
21 group with gestational age is shown in Fig. 1A
. The distribution of
PSA MoM fitted a gaussian distribution after log10
transformation in both the trisomy 21 group and the controls. The
parameters of the distribution are mean log10
(controls = 0.000; trisomy 21 group = -0.112) and
log10 SD (controls = 0.4424; trisomy 21 group =
0.6722).
When the AF data in the trisomy 21 group were analyzed by fetal sex, the median MoMs were 0.81 and 0.73 for female and male fetuses, respectively (P = 0.3954, Mann–Whitney U-test). No difference could be established in the control group.
In conclusion, we have been unable to confirm the previous report of Lambert-Messerlian et al. (22) that maternal serum PSA is increased in pregnancies with trisomy 21 fetuses. PSA concentrations in the previous study (22) were less than the detection limit of the assay in 22% of the control cases, whereas in our study only 10% of control cases were below the detection limit. It is possible, therefore, that the study design and this higher percentage of controls with undetectable PSA concentrations deflated the median value in the earlier study (22), resulting in the trisomy 21 MoMs being overestimated.
Previously, Melegos et al. (23) from the same group showed low concentrations of PSA in the AF of four of six cases of trisomy 21, and in our study we can confirm that AF contains PSA and that PSA concentrations on average tend to be lower in cases of trisomy 21. Moreover, AF PSA changes significantly with gestational age. The lowering of AF PSA in our study is mirrored by a lowering of maternal serum PSA in the trisomy 21 group. However, the amount of the lowering of maternal serum PSA and the width of the distribution are likely to mitigate against PSA being of value in a screening context.
Acknowledgments
We thank the DPL Division of Euro/Diagnostics Products Corporation, Ltd., Llanberis, United Kingdom for providing the kits to carry out this work.
Footnotes
Endocrine Unit, Clinical Biochemistry Department, Harold Wood Hospital, Gubbins Lane, Romford, Essex, RM3 0BE United Kingdom
References
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
