Immunoassay biomarkers of first and second trimesters: a comparison between pregnant Syrian refugees and Turkish women

Introduction

The first trimester double screening test, performed on weeks 11–14, uses hormone parameters of pregnancy associated-plasma protein-A (PAPP-A) and free beta subunit of human chorionic gonadotropin (fβ-hCG) while second trimester triple screening test, performed on weeks 15–20, uses hormone parameters of total human chorionic gonadotropin (tβ-hCG), unconjugated estriol (uE3) and alpha-fetoprotein (AFP) together with demographic characteristics such as gestational week, maternal age and weight, and sonographic fetal measurements for estimation of fetal aneuploidy risk via a software.

The results of prenatal double and triple screening tests are interpreted as positive or negative according to cut-off value via complex mathematic algorithms in the software. The screening tests determine risk for birth of mental retarded infant and serve as a guide to refer invasive diagnostic modalities that can establish definitive diagnosis [1].

Maternal serum AFP, free and total ß-hCG, and uE3 values also provide information about pregnancy-related complications specific to a certain pregnancy. The elevated AFP levels in the second trimester are associated to fetal neural tube defects, congenital anomalies, low birth weight, fetal or perinatal losses [2]. Low PAPP-A level was shown to be associated with premature birth and preeclampsia [3]. It was also shown that the levels of these hormones are affected by several factors such as maternal weight [4], race and ethnicity [5] and insulin-dependent diabetes mellitus [6]. It was reported that data retrieved from ethnic/race groups are needed to improve performance of prenatal screening tests [7].

In this study, it was aimed to evaluate maternal serum hormone values used for aneuploidy screening, demographic characteristics and fetal measurements in pregnant Syrian refugees (PSRs) and to compare with data from pregnant Turkish women (PTW).

Material and methods

We retrospectively reviewed data from pregnant Syrian and Turkish women who underwent prenatal double and triple screening test between January, 2018 and June, 2018. The study protocol was approved by the Institutional Review Board and Ethics Committee of Erciyes University (Approval number: 2019/410). Maternal hormone data, demographic characteristics and measurements of nuchal translucency and crown-rump length (CRL) were recorded in singleton, non-diabetic pregnancies. In the estimation of median values, data from pregnant women with multiple pregnancy, those congenital fetal malformation, fetal neural tube defect, hydrops fetalis and perinatal fetal loss and those with increased risk for neural tube defect defined by AFP>2.5 MoM and those with 1:250 mid-trimester risk ratio for Down syndrome were excluded. Gestational age was calculated according to Hadlock II method. Gestational week used for sonographic measurements and median value calculations were rounded (e.g. gestational week 11 included days from 11⁺⁰ to 11⁺⁶). All median values for hormone parameters and sonographic measurements were calculated for each gestational week.

The maternal serum free and total ß-hCG, PAPP-A, uE3 and AFP values were analyzed by IMMULITE^® 2000 XPi system (Siemens Healthcare Diagnostics, Los Angeles, USA) using chemiluminescence immunoassay method.

Throughout study period, the assay stability was achieved by checking mean and median MoM values in study population and between groups for each analyte and using test reagents, control sera and calibrators from same lot under supervision of same biochemistry specialist. In each study, control sera for different levels (low, moderate or high values) and test-based criteria for intra- and inter-assay variations as recommended by Westgard were used [8]. The analytic sensitivities of the assays are approximately 1 ng/mL for free β-hCG, 0.025 mIU/mL for PAPP-A, 0.4 mIU/mL for total β-hCG, 0.01 ng/mL for uE3 and 0.24 ng/mL for AFP. The within-run coefficients of variations were around 6%, 3.6%, 7%, and 8%, and 2.5%, respectively.

The multiplies of the median (MoM) values were recorded, which is obtained via dividing test result by median value of population, and used to assess hormone result of a pregnant woman relative to normal population [9]. The MoM values were calculated for each gestational week. Median values, MoM values, demographic characteristics and sonographic parameters used in the prenatal screening test via Priska software were compared between two groups.

All statistical analyses were performed using IBM^® SPSS^® v. 24.0. The significance level was accepted as α=5%. Normal distribution was assessed by Shapiro-Wilk test. Descriptive statistics are given as median and % 25–75 interquartile range (IQR). The independent-sample Mann-Whitney U test was used to compare variables between groups. Percent change in demographic and sonographic measurement data and median and MoM values of hormones in PTW relative to those from PSR (inter-group percent bias) was calculated using following formula:

Bias %=TPQ2 or MoM value−SRPQ2 or MoM valueSRPQ2 Or MoM value∗100

(Q₂=median value calculated for each gestation week and analyte)

The formula non-standardized and standardized ß coefficients for age and weight variables were separately calculated for PSR and PTW groups.

Results

The study assessed data from 2697 pregnant women who underwent triple and double screening test, including 1738 Turkish pregnant women and 959 pregnant Syrian women. Of these, the data from double screening test in 1108 pregnant women (n[PTW]/n[PSR]=749/359) and from triple screening in 1589 pregnant women (n[PTW]/n[PSR]=989/600) were available.

Mean total age and weight were found to be significantly different and lower in the PSR group in all gestational ages evaluated (p<0.001; p<0.001). In both groups, descriptive statistics for maternal age and weight, fetal sonographic measurements and free and total ß-hCG, PAPP-A, uE3 and AFP were performed for each gestational week at first and second trimester. The percent bias was calculated for each analyte and groups were compared. Accordingly, fetal median values obtained for weeks 11–13 were higher in PSR group but the difference did not reach statistical significance (p>0.05). The mean age was significantly lower in PSR group in all gestational weeks evaluated (p<0.05). As similar, mean weight was found to be significantly lower in PSR group in all gestational weeks other than weeks 12, 16 and 20 (p<0.05).

We identified 74 women in PTW and 30 women in PSR groups in addition to patients with adverse pregnancy outcomes whose data were not used in calculations of median values. The proportion of these patients to original study populations were 4.1% and 3.9%, respectively.

When compared to PTW group, median fß-hCG values in PSR group had negative biases of 38% and 24.5% for weeks 11 and 13, respectively (p<0.019 and p<0.001). Again, the median MoM values in PSR group had negative biases of 28.6% for both weeks (p<0.016 and p<0.015). Although maternal age in PSR group had negative bias of 13% on week 12, there was no significant difference in median and MoM values of fß-hCG and PAPP-A between groups (p>0.05; Table 1).

In the first trimester, median fß-hCG values of PTW group were found to be higher by 38% and 24.5% on weeks 11 and 13, respectively. When hormone parameters were assessed in the second trimester, it was seen that tß-hCG values had negative biases of 11.8% and 21.2% in PSR group on weeks 16 and 19 (p<0.001). When compared to PTW group, tß-hCG MoM values in PSR group were found to be lower by 7.6%, 19.1% and 25% on weeks 16, 17 and 19, respectively (p<0.001). Mean MoM values obtained at all gestational weeks in which tß-hCG values showed median differences were found to be significantly difference with a direction similar to median values (Table 2).

The coefficients of hormone analyte formula were calculated as fβ-hCG, for TP: 73.611 and for SRP: 61.271; tβ-hCG, for TP: 29965.780 and SRP: 30001.398. In the regression model, the maternal weight had significant negative effect on all hormone parameters in PSR group while it had significant negative effect in all parameters other than uE3 in PTW group. The maternal age had no significant effect in binary regression models created for both groups. Although mean maternal age was significantly lower in PSR group for all gestational weeks evaluated, it had no significant effect on hormone parameters in both groups.

Discussion

In this study comparing PSR and PTW groups, there were significant differences in demographic characteristics, median and MoM values of maternal serum free and total ß-hCG, PAPP-A, uE3, AFP, and fetal sonographic measurements in both first and second trimesters. In the literature, this is the first study demonstrated significant difference in maternal hormone parameters, particularly in ß-hCG, during first and second trimesters of pregnancy between two race/ethnicity groups.

Previously, several studies were conducted, which compared birth management, fetal weight, preeclampsia [10], infectious seroprevalence [11] and results of neonatal auditory screening [12] between PSR and PTW.

It has been advocated that maternal weight-specific median hormone values should have to be used in ethnic groups [13]. The negative correlations between maternal weight and hormone parameters of screening tests could be interpreted as a result of higher plasma volume in overweight pregnant women when compared to those with normal body weight. Relatively higher hormone biases and regression coefficients estimated for first trimester can be considered as greater impact of ethnic origin on biochemical markers in first trimester when compared to second trimester. In this context, fß-hCG appeared as the maternal hormone which was associated with greatest impact of ethnic diversity in the first trimester.

In a previous study from Turkey, it was found that birth weight and obstetric complications were comparable in PSRs [14]. However, in another study, birth weight was found to be significantly lower among PSRs [10]. In our study, the finding that there was no significant difference in fetal sonographic measurements which are markers of fetal development is in agreement with those reported by Güngör et al.

In a study comparing second trimester serum hormone levels of Caucasian women with Afro-Caribbean, South Asian, Oriental and Hispanic women, it was shown that AFP and HCG levels were higher in African women by 15% when compared to Caucasian women [15]. In this current study, for second trimester hormones, the highest difference (21.2%) was detected for median tß-hCG values on gestational week 19 while no significant difference was detected in median AFP and uE3 values between groups. In another study, it was shown that median values of second trimester hormones in pregnant women from North-West Indian were comparable to those from Belgium, Canada, Germany and UK [16]. In a study comparing median values in pregnant women from Egypt, it was found that median uE3 levels were lower by almost one-half when compared to median values from Asia and Western countries while median AFP levels were lower than those from Korea [17]. Given these median AFP, ßhCG and uE3 values reported to be higher in Asian women, median values for fß-hCG and tß-hCG hormones in PTW group were higher than PSR group in our study.

There was no significant difference in median tß-hCG values at gestational week 17 while a significant difference was found in tß-hCG Mom values between groups. In fact, MoM values are the result of normalization process used to determine the distance of the result of a certain patient to population median. As it is the case in our study, the fact that comparable median values correspond to different MoM values indicates the need for determining median values specific to various ethnic/racial populations served. Thus, negative influence on performance of prenatal screening test is inevitable when median values and MoM values used to determine risk for fetal aneuploidy are not specific to ethnicity or race. Based on studies comparing median values, it is recommended to use different cut-off thresholds for remarkable ethnic variations in false-positive rates [18]. Since marked differences in median hCG values could lead false-positive results for Down syndrome in PSRs, an adjusted cut-off value should have to be considered. Thus, in the center where our study was conducted, the threshold ratio for referral to advanced invasive procedures is recommended as 1: 270.

ß-hCG is a maternal hormone and its expression represents somatic genotypes. It has been reported that ß-hCG production is affected by genetic diversity [19]. The CGB5 is one of the six genes that regulate hCG expression. It was shown that genetic variation of CGB5 gene plays role in recurrent, spontaneous abortion and that the gene has population-based variations in its synthesis [20].

As the maternal weight but not maternal age showed significant negative correlation with hormonal variations in both groups and the maternal weight alone could not explain the difference in coefficients estimated for study groups, we emphasize that there is a need for further descriptive studies for genetic expression of ß-hCG which has been reported to be regulated by some ethnicity- and race-related genetic polymorphism and was found to have significant differences between PSR and TPW groups.

The covariates proposed for pregnancy such as in vitro fertilization, intrauterine insemination, hormone concentrations in previous pregnancy, gravity, parity or fetal gender) are beyond scope of this study. The improvement that will achieve by addition of clinical covariates to screening tests should have to be arranged based on real-world practice about data collection.

The pregnant women carrying fetuses with Down syndrome often have low AFP and ß-hCG concentrations together with low uE3 concentrations [21]. In our study, the fact that PSRs had significantly lower median values for ß-hCG can lead decrease in pool of median of the region served for Turkish population and misleadingly results of risk-free infant during reporting for Down syndrome in prenatal screening tests. This may result in questioning of major performance criteria such as test sensitivity and specificity for laboratories providing prenatal screening test services.

Despite lack of number of infants with aneuploidy confirmed by chromosomal assays due to retrospective nature of the study, our results suggests to plan multicenter, prospective studies investigating regional/ethnic variations in the prevalence of Down syndrome known to be linked to ethnicity [22].

As shown in our study, although median values for maternal hormones used in double and triple screening tests are significantly different, it should be suggested that assessment for same ethnicity in commercially available software used for risk reporting for infant with aneuploidy will have negative impact on the test performance for both groups. In particular, the increase in presentation rates of PSRs to laboratories can result in decrease in mean median values for the region. Lower median values and low free and total ß-hCG MoM values resulting from this heterogeneity can lead false-negative screening test results in Turkish population. Similarly, median/MoM values with positive bias for PSRs will be calculated due to this heterogeneity; in other words, it will lead reporting false-positive results for increased Down syndrome risk. Thus, race/ethnicity-specific coefficients should have to be used in the software used to estimate fetal aneuploidy risk for PRSs in screening laboratories. The risk reports based on validated risk estimations using median values found in our study can provide more effective results in both PSRs and TPW who may face skewed median values and risk assessment.

In addition, given the ethnic similarities among Arabian women (Arab panethnicity), the variations in median values when compared to TPW can also be valid for other countries in the region.

The Prisca software does not allow input of descriptive data specific to Syrian refugees. As shown in our study, the presence of significant differences in median values warrants such modification. The biochemistry specialist reporting results of screening tests and the obstetricians providing prenatal care to PSRs take the responsibility of being attentive to this population while take potential genetic and ethnic variations that affects results of prenatal screening reports into consideration.

As it is the case in our country, given the changes in migration dynamics, validated reference hormone ranges specific to regions or even centers should be identified and updated median values specific to ethnic populations should be defined as PSRs in the pregnant populations served may have potential effects on median distribution of maternal hormones.