PRENATAL DIAGNOSIS Prenat Diagn 2005; 25: 484–488. Published online in Wiley InterScience (www.interscience.wiley.com). DOI: 10.1002/pd.1176 SHORT COMMUNICATION Weight adjustment of serum markers in early first-trimester prenatal screening for Down syndrome Tina Sørensen*, Severin Olesen Larsen and Michael Christiansen Department of Clinical Biochemistry, Statens Serum Institute, Denmark Objective To assess whether existing weight correction formulas for PAPP-A and free-beta-hCG developed for weeks 11 to 14 can be applied to pregnancies in weeks 8 to 10. Methods Development of formulas based on limited data sets of 8- to 10-week pregnancies and comparison with existing formulas. Calculation of median MoMs adjusted with different formulas for weight correction. Results Weight correction formulas for the gestational age of 11 to 14 weeks were not appropriate in the 8- to 10-week gestational age interval for PAPP-A, whereas existing weight correction formulas could be applied to free-beta-hCG, independent of gestational age interval. Conclusion If PAPP-A is used in different gestational age intervals, weight corrections should be developed for the interval. Copyright 2005 John Wiley & Sons, Ltd. KEY WORDS: prenatal; Down; screening; maternal; weight INTRODUCTION The maternal serum concentration of screening markers is weight dependent. A serum marker is more diluted in high-weight pregnancies compared to that in low- weight pregnancies. As shown in Neveux et al. (1996), weight correction is performed by dividing the unad- justed MoM with the expected MoM value according to weight. The expected MoM formulas are developed from data using the serum marker MoM values and the corresponding maternal weight. Weight adjustment of MoM values in a prenatal screening program using the combined test (nuchal translucency (NT), pregnancy- associated plasma protein-A (PAPP-A) and free-beta- human chorionic gonadotrophin (free-beta-hCG)) to cal- culate the risk of a Down syndrome pregnancy has been examined by Spencer et al. (2003) in the OSCAR pro- gram including 32 010 pregnancies from 11 to 14 weeks. The median weight and median MoM were calculated for maternal weight groups of 10 kg ranging from 35 to >124.9 kg. Two methods were investigated: a linear regression of the logarithmic median MoM against the median weight (log-lin regression) and a linear regres- sion of median MoM against the reciprocal median weight (Rec-lin regression). Both regressions fit the data well for both markers, the Rec-lin regression, however, being slightly better for PAPP-A and the log-lin being slightly better for free-beta-hCG. A suggested weight correction formula is appropriate for a particular population when the median MoM of *Correspondence to: Tina Sørensen, Department of Clinical Bio- chemistry, Statens Serum Institute, Artillerivej 5, DK-2300S. E-mail:
[email protected] the serum marker, before and after weight adjustment, is very close to 1.00. If new regression constants are required, 1000 data points are enough to provide reliable constants (Neveux et al., 1996). When expanding the gestational age interval of first- trimester prenatal screening for Down syndrome from 11 to 14 weeks to 8 to 14 weeks, it is prudent to investigate the maternal weight adjustment relationship of the applied biochemical serum markers. In this study, we investigate the impact of using the OSCAR program’s weight correction formulas (Spencer et al., 2003) on unadjusted MoM values from pregnan- cies with gestational age intervals from 8 to 10 weeks, 11 to 14 weeks and 8 to 14 weeks, respectively. METHODS Samples were collected by general practitioners or hos- pitals and sent to Statens Serum Institute (SSI) by mail along with gestational age information. Results from ultrasound scanning, performed after sample collection, were collected by phone or fax. The samples were centrifuged within 2 days of collection and PAPP-A and free-beta-hCG concentrations measured using the KRYPTOR platform (BRAHMS, Germany). MoM values were calculated by dividing the serum marker concentration with the median serum marker concentration. The median serum marker concentra- tion was calculated from a regression based on 5902 pregnancies. We grouped the marker concentration of the 5902 pregnancies according to weeks and then made a logarithmic-linear regression of median marker concentration against the median day in each week. Copyright 2005 John Wiley & Sons, Ltd. Received: 26 October 2004 Revised: 11 February 2005 Accepted: 1 April 2005 WEIGHT ADJUSTMENT IN FIRST-TRIMESTER SCREENING 485 Table 1—Median weight and unadjusted MoMs for data based on pregnancies from 8 to 14, 8 to 10 and 11 to 14 weeks 8–14 weeks 8–10 weeks 11–14 weeks Weight Weight BHCGa PAPP-A Weight BHCG PAPP-A Weight BHCG PAPP-A Group n Median MoM median MoM median N Median MoM median MoM median n Median MoM median MoM median 35.0–44.9 11 44.0 1.18 1.25 5 44.0 1.19 1.18 6 43.5 1.21 1.28 45.0–54.9 283 52.0 1.18 1.33 165 52.0 1.18 1.29 118 52.0 1.16 1.32 55.0–64.9 1406 60.0 1.08 1.15 787 60.0 1.08 1.15 619 60.0 1.09 1.12 65.0–74.9 1159 69.0 1.01 1.00 687 69.0 1.01 1.00 472 69.5 1.00 0.98 75.0–84.9 515 78.0 0.90 0.85 315 79.0 0.91 0.88 200 78.0 0.85 0.78 85.0–94.9 204 89.0 0.86 0.72 120 89.0 0.82 0.73 84 88.0 0.92 0.67 95.0–104.9 123 99.0 0.83 0.61 76 98.2 0.89 0.71 47 99.2 0.79 0.54 105.0–114.9 45 108.0 0.70 0.59 25 108.0 0.68 0.65 20 108.5 0.82 0.53 115.0–124.9 15 116.0 0.56 0.49 10 117.0 0.84 0.73 5 119.0 0.47 0.42 >124.9 11 133.0 0.65 0.38 5 131.0 0.63 0.57 6 138.0 0.76 0.27 35.0–>124.9 3772 66.0 1.01 1.02 2195 66.0 1.02 1.02 1577 65.0 1.01 0.99 a free-beta-HCG Logarithmic-linear regressions were carried out on the data set of pregnancies from 8 to 14 weeks, sub-data sets with gestational age from 8 to 10 weeks (3645 pregnan- cies) and gestational age from 11 to 14 weeks (2255 pregnancies). For 3772 of the 5902 pregnancies, maternal weight on the date of collection was available. The data were sorted in three data sets according to gestational age interval (8–10, 11–14 and 8–14 weeks). After calculating the unadjusted MoM values of these pregnancies, they were grouped according to weight, as described by Spencer et al. (2003). Using the median MoM and weight listed in Table 1, we made weight adjustment formulas for 8 to 10 weeks, 11 to 14 weeks and 8 to 14 weeks of gestation. We made weighted regressions by minimizing the sum of the weighted residual (SWR) using the solver tool for equation optimization in Excel: SWR = m∑ i=1 (wi (yi − f (x)))2, wi = nwi m is the number of weight groups, n is the number of data points in each weight group, w is the weighting factor, f (x) is the result of a proposed weight equation and yi is the data point to which the best fit should be achieved (Hayes, 1970). We recalculated data from previous publications (Neveux et al., 1996 and Spencer et al., 2003) to ensure the validity of the formula. We were also able to closely reproduce the regressions found by Neveux et al. (1996) for AFP, hCG and uE3 from our second-trimester data sets, which were similar to the sets presented here in amount and distribution (results not shown). We used the weight regressions to calculate the weight-adjusted MoMs by dividing the unadjusted MoM by the expected MoM. After weight adjustment, the median MoM was calculated for each data set (Table 2). RESULTS The median weight of the presented populations were within 2 kg of the weight of the population investigated by Spencer et al. (2003), whereby the published weight correction formula should be applicable to the pre- sented data (Neveux et al., 1996). However, we found that the median PAPP-A MoM was further from 1.00 after weight adjustment than before weight adjustment (Table 2). This caused us to fully investigate the data for PAPP-A. From the regressions of the logarithmic marker concentration versus gestational age in days, we calculated the unadjusted MoM values. We calculated logarithmic-linear median regressions for the 8- to 10- and 11- to 14-week intervals and a polynomial regres- sion for 8- to 14-week intervals. The results presented in Table 1 and 2 are based on these regressions shown in Figure 1 along with the regression presented by Spencer et al. (2000). We also calculated the average and stan- dard deviation of log MoMs in each weight group for the 8- to 10- and 11- to 14-week populations and found significant difference between the 8- to 10- and 11- to 14-week populations’ log MoMs for weight groups above 95 kg (u > 2). When calculating weight regressions, for PAPP-A (from data in Table 1), for the 8- to 10-week interval, we found significantly different regression constants from the 11- to 14-week regression, and from the formulas published by Spencer et al. (2003). Figure 2 shows weight regressions for each gestational age interval, along with the median MoMs for each weight group. The log-linear weight regression from Spencer et al. (2003) and the log-linear regression for 8 to 14 weeks for free-beta-hCG are shown. These regressions are very similar and represent the data equally well. The median MoMs for PAPP-A for pregnancies from 8 to 14 weeks show a small discrepancy from the Rec-linear regression developed by Spencer et al. (2003), based on pregnancies from 11 to 14 weeks. The 8- to 10- week weight adjustment formula is shown along with median MoMs for 8- to 10-week pregnancies and a Copyright 2005 John Wiley & Sons, Ltd. Prenat Diagn 2005; 25: 484–488. 486 T. SØRENSEN ET AL. Table 2—The medians of MoMs adjusted with different weight regressions. Unadjusted MoMs were calculated using median regressions for each gestational age interval. Adjusted MoMs are shown for weight regressions based on 8- to 14-, 8- to 10- and 11- to 14-week pregnancies. In comparison, the adjusted MoM medians are shown when the weight regressions developed by Spencer et al. (2003) are applied Median regression Median weight adj. MoM Weeks Analyte Median MoM Weight regression. Rec-lin Log-lin 8–10 PAPP-A 1.02 Spencer 1.09 1.07 (8–14) 1.04 1.03 (8–10) 1.01 1.00 BHCG 1.02 Spencer 1.01 1.00 (8–14) 1.01 1.00 (8–10) 1.00 1.00 11–14 PAPP-A 0.99 Spencer 1.00 0.99 (8–14) 0.97 0.96 (11–14) 1.01 0.98 BHCG 1.01 Spencer 1.04 1.00 (8–14) 1.01 1.00 (11–14) 1.01 1.00 8–14 PAPP-A 1.02 Spencer 1.06 1.04 (8–14) 1.01 1.00 BHCG 1.01 Spencer 1.04 1.00 (8–14) 1.01 1.00 2 2,2 2,4 2,6 2,8 3,2 3,4 3,6 3,8 3 4 50 70 90 110 Gestational age (days) Lo g(P AP P- A) Figure 1—Filled circles show the PAPP-A concentration median for each week. The logarithmic-linear regressions for weeks 8 to 10 (solid line marked with *) and week 11 to 14 (solid line marked with �) are shown along with the polynomial regression for weeks 8 to 14 (solid line) and the regression presented by Spencer et al. (2000) significant discrepancy from the Rec-linear regression of Spencer et al. (2003) is seen. The 11- to 14-week weight adjustment regression for PAPP-A is very similar to the regression provided by Spencer et al. (2003). From Figure 2, it can be concluded that the log- linear regression for free-beta-hCG provided by Spencer et al. (2003) is very similar to the regression developed from the data in Table 1, even though the regressions are based on pregnancies from different gestational age intervals. For PAPP-A, however, the reciprocal- linear regressions (and also the logarithmic-linear, which are not shown in Figure 1) are very different when comparing regressions for 11- to 14-week pregnancies and 8- to 10-week pregnancies. Unadjusted MoMs were calculated for the 8- to 10-, 11- to 14- and 8- to 14-week intervals and the medians are shown in Table 2. The weight-adjusted MoMs were calculated using the weight-correction formulas devel- oped from data in Table 1 and formulas from Spencer et al. (2003). We found that weight-adjusted PAPP-A MoMs from the 8- to 10-week pregnancies popula- tion, adjusted using the Spencer Rec-lin equation, had a median MoM of 1.09 (average and standard devi- ation of log MoM was 0.036 and 0.260 respectively. The average log MoM was significantly different (u > 2) from the average log MoM of the weight-adjusted 11- to 14-week population (average and standard deviation of log MoM was 0.0016 and 0.239 respectively). When weight regressions developed from the data set shown in Table 1 were used, the median weight-adjusted MoM values were 1.01 and there was no significant differ- ence between the average log MoMs of the different populations (just as there was no significant difference between the unadjusted populations). When using the log-lin regressions, the 8- to 14-week formula was appli- cable to both 8- to 10- and 11- to 14-week populations, whereas the Spencer log-lin was free-beta-hCG was also adjusted using the regressions provided by Spencer et al. The median weight-adjusted MoM for all three data sets were very close to 1.00 for the logarithmic-linear regres- sion (which was the most optimal regression formula for free-beta-hCG, Spencer et al. (2003)). Using the reciprocal-linear regression from Spencer et al. (2003) gave an elevated median weight adjusted MoM for the 8- to 10-week data set of 1.04. Copyright 2005 John Wiley & Sons, Ltd. Prenat Diagn 2005; 25: 484–488. WEIGHT ADJUSTMENT IN FIRST-TRIMESTER SCREENING 487 0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 0 0,2 0,4 0,6 0,8 1 1,2 1,4 1,6 1,8 0 50 100 150 Weight (kg) 0 50 100 150 Weight (kg) 0 50 100 150 Weight (kg) 0 50 100 150 Weight (kg) BH CG (M oM ) ( 8-1 4 w ee ks ) PA PP -A (M oM ) ( 8-1 4 w ee ks ) PA PP -A (M oM ) ( 8-1 0 w ee ks ) PA PP -A (M oM ) ( 11 -14 w ee ks ) Figure 2—Weight regressions shown with the individual median MoMs (♦) (Table 1) for which they are the best fit. Logarithmic-linear regressions are shown for free-beta-hCG and reciprocal-linear regressions are shown for PAPP-A because these were the optimal regressions found for each serum marker in Spencer et al. (2003). The solid line is the regressions from Spencer et al. (2003). The broken line shows the regressions developed in this work. For 8 to 10 weeks, the expected PAPP-A MoM = (59.331/weight) + 0.128 DISCUSSION Early first-trimester screening is an attractive option, but the performance is poorer than the performance of integrated screening (Wald et al., 1999a). With the introduction of new markers, which, like PAPP-A, perform well in very early pregnancy (Laigaard et al., 2003; Qin et al., 1997; Christiansen et al., 1999; Wald et al., 1999b), new first-trimester screening set-ups are possible. As an example, it may be contemplated to screen in 8 to 10 weeks and use contingent testing (Christiansen and Larsen 2002). Furthermore, future serum markers may very well be similar to PAPP-A, in that a low MoM may favor a high risk, which means that weight adjustment becomes essential because the dilution effect for the markers will not counteract each other. We have shown the necessity of developing dif- ferent weight correction formulas for PAPP-A, when working in different gestational age intervals and we reported a weight-correction formula suitable for early first trimester. REFERENCES Christiansen M, Larsen SO. 2002. An increase in cost-effectiveness if first trimester maternal screening programmes for fetal chromosome anomalies is obtained by contingent testing. Prenat Diagn 22: 482–486. Christiansen M, Oxvig C, Wagner JM, et al. 1999. The proform of eosinophil major basic protein: a new maternal serum marker for Down’s syndrome. Prenat Diagn 19: 905–910. Hayes JG (ed). 1970. Numerical Approximation to Functions and Data. Athlone press: London. Laigaard J, Sørensen T, Fro¨hlich C, et al. 2003. ADAM12: a novel first-trimester maternal serum marker for Down syndrome. Prenat Diagn 23: 1086–1091. Neveux LM, Palomaki GE, Larrivee DA, Knight GJ, Haddow JE. 1996. Refinements in managing maternal weight adjustment for interpreting prenatal screening results. Prenat Diagn 16: 1115–1119. Copyright 2005 John Wiley & Sons, Ltd. Prenat Diagn 2005; 25: 484–488. 488 T. SØRENSEN ET AL. Qin QP, Christiansen M, Nguyen TH, Sørensen S, Larsen SO, Nørgaard-Pedersen B. 1997. Schwangerschaftsprotein 1 (SP1) as maternal serum marker for Down’s syndrome in first and second trimester. Prenat Diagn 17: 101–108. Spencer K, Bindra R, Nicolaides KH. 2003. Maternal weight correction of maternal serum PAPP-A and free beta-hCG MoM when screening for trisomy 21 in the first trimester of pregnancy. Prenat Diagn 23: 851–855. Spencer K, Ong CYT, Liao AWJ, Nicolaides KH. 2000. The influence of ethnic origin on first trimester biochemical markers of chromosomal abnormalities. Prenat Diagn 20: 491–494. Wald NJ, Watt HC, Hackshaw AK. 1999a. Integrated screening for Down’s syndrome based on tests performed during the first and second trimester. N Engl J Med 341: 461–476. Wald NJ, Watt HC, Nørgaard-Pedersen B, Christiansen M. 1999b. SP1 in pregnancies with Down syndrome in the first trimester of pregnancy. Prenat Diagn 19: 517–520. Copyright 2005 John Wiley & Sons, Ltd. Prenat Diagn 2005; 25: 484–488.