In Order to Test the Hypothesis That the Dried Blood Spot Assay Being Compared in the Main

Supplement to Cowans et al:Evaluation of a dried blood spot assay measuring prenatal screening markers PAPP-A and free hCGβ

Supplement

In order to test the hypothesis that the Dried Blood Spot assay being compared in the main study measures a composite of free hCGβ that is found in serum and additional beta-subunits that have been released from intact hCG during the drying process, a statistical simulation was carried out.

Methods

The serum raw concentrations for free hCGβ and intact hCG from the main study were corrected for maternal ethnicity and smoking status(1-3). Models were selected to predict the log10 concentration of each analyte with explanatory terms up to second order gestational age and reciprocal maternal weight. Outliers with large standardized deletion residuals were investigated and removed, and the models were refitted without these data points. Expected molar concentrations for each analyte at the mid point of gestational weeks 10 to 13 were calculated. At each week, 100,000 affected and unaffected multivariate log10 MoMs for PAPP-A, free hCGβ, intact hCG and NT were randomly generated using SURUSS parameters(4) with revised estimates for NT (5). SURUSS free hCGβ and intact hCG MoMs were converted to molar concentrations by multiplying by the calculated expected molar concentrations, and then use to predict the serum-equivalent free hCGβ dried blood molar concentrations using the DBS free hCGβ model,from the main study. Median values were calculated and used to calculate the MoM of each observation. Median MoM and standard deviations were then determined at each gestational week.

To model the effect of using DBS free hCGβ or serum free hCGβ in simulated combined test screening programs, means, standard deviations and variance-covariance matrices were calculated at each week for log10 MoMs of the simulated serum PAPP-A, serum free hCGβ, serum-equivalent free hCGβ and NT, and these were used to obtain detection rates at fixed false positive rates using Monte Carlo methods. Random multivariate samples of 500, 000 observations were drawn for trisomy 21 and unaffected pregnancies and likelihood ratios were computed for each observation and then used to produce detection rates for each maternal age. Overall population detection rates were obtained by combining the maternal age-specific rates according to the maternal age distribution for trisomy 21(6) and the UK 2009-2010 (7, 8) maternal age distributions. MoM truncation limits for PAPP-A, free hCGβ and NT were taken from SURUSS, except for the lower NT truncation limit which used the MoM at the minimum likelihood ratio.

1

Supplement to Cowans et al:Evaluation of a dried blood spot assay measuring prenatal screening markers PAPP-A and free hCGβ

Results

Full step-wise forward and backward elimination selected the following models for raw serum concentrations corrected for maternal ethnicity and smoking status:

Log10 free hCGβ serum concentration =

Log10 intact hCG serum concentration =

where GA is gestational age in days and MW is maternal weight in kg.

Simulated log10 DBS freehCGβ MoMSD were 0.21 and 0.22 in affected and unaffected pregnancies at all weeks. The median freehCGβ MoM in simulated affected pregnancies at weeks 10 to 13 were 1.27, 1.60, 1.91 and 2.25 respectively.

Supplemental Table 1 shows estimated detection rates using serum free hCGβ and PAPP-A or DBS free hCGβ and PAPP-A with and without NT MoM. Overall, the detection rates between the methods with DBS and serum free hCGβ were similar, however, at weeks 10 and 11, using serum free hCGβ and PAPP-A produced slightly higher detection rates than DBS free hCGβ and PAPP-A, whereas the reverse was true in weeks 12 and 13. These differences were more notable with screening without NT MoM.

1

Supplement to Cowans et al:Evaluation of a dried blood spot assay measuring prenatal screening markers PAPP-A and free hCGβ

Supplementary Table 1

Test / FPR / Week 10 / Week 11 / Week 12 / Week 13
Serum free hCGβ, PAPP-A / 1% / 53.94% / 48.68% / 45.51% / 45.19%
3% / 68.28% / 63.60% / 60.92% / 61.02%
5% / 74.91% / 70.79% / 68.49% / 68.76%
DBS free hCGβ, PAPP-A / 1% / 50.77% / 47.26% / 46.33% / 49.06%
3% / 65.83% / 62.42% / 61.97% / 64.83%
5% / 72.85% / 69.78% / 69.58% / 72.39%
Serum free hCGβ, PAPP-A, NT / 1% / 69.66% / 72.55% / 73.56% / 74.42%
3% / 80.56% / 81.74% / 81.83% / 82.73%
5% / 85.15% / 85.66% / 85.62% / 86.45%
DBS free hCGβ, PAPP-A, NT / 1% / 67.95% / 72.62% / 74.54% / 77.09%
3% / 79.40% / 81.86% / 82.81% / 85.09%
5% / 84.27% / 85.84% / 86.52% / 88.60%

Estimated detection rates at 1, 3 and 5% false positive rates (FPR) determined by simulating screening using serum and DBS free hCGβ and PAPP-A, with and without NT. UK 2009-2010 maternal age distribution was used in the simulation

1

Supplement to Cowans et al:Evaluation of a dried blood spot assay measuring prenatal screening markers PAPP-A and free hCGβ

Reference List

(1) Cuckle HS, Arbuzova S. Multi-Marker Maternal Serum Screening for Chromosomal Abnormalities. In: Milunsky A, editor eds. Genetic Disorders and the Fetus; Diagnosis, Prevention and Treatment. 5 Ed. Johns Hopkins University Press; 2004. p. 820.

(2) Spencer K, Bindra R, Cacho AM, Nicolaides KH. The impact of correcting for smoking status when screening for chromosomal anomalies using maternal serum biochemistry and fetal nuchal translucency thickness in the first trimester of pregnancy. Prenat Diagn 2004;24:169-73.

(3) Spencer K, Heath V, El-Sheikhah A, Ong CY, Nicolaides KH. Ethnicity and the need for correction of biochemical and ultrasound markers of chromosomal anomalies in the first trimester: a study of Oriental, Asian and Afro-Caribbean populations. Prenat Diagn 2005;25:365-9.

(4) Wald NJ, Rodeck C, Hackshaw AK, Walters J, Chitty L, Mackinson AM. First and second trimester antenatal screening for Down's syndrome: the results of the Serum, Urine and Ultrasound Screening Study (SURUSS). Health Technol Assess 2003;7:1-77.

(5) Bestwick JP, Huttly WJ, Wald NJ. Distribution of nuchal translucency in antenatal screening for Down's syndrome. J Med Screen 2010;17:8-12.

(6) Cuckle HS, Wald NJ, Thompson SG. Estimating a woman's risk of having a pregnancy associated with Down's syndrome using her age and serum alpha-fetoprotein level. Br J Obstet Gynaecol 1987;94:387-402.

(7) Characteristics of mother 1, England and Wales, 2009. (Accessed Jun 2012).

(8) Characteristics of mother 1, England and Wales, 2010. (Accessed Jun 2012).

1