Allowing for ethnic group in antenatal screening for Down's syndrome

Maternal weight and ethnic status influence the concentration of serum markers used in antenatal screening for Down's syndrome. These are usually allowed for in screening by adjusting the concentration of the screening marker. For example, the values in light women are, on average, higher, and those in heavy women, on average, lower than values in women of average weight. Adjustment aims to modify the concentration so the average in lighter women and heavier women is similar to that in women of average weight. Women from an ethnic group that typically has lower screening marker concentrations would have these concentrations adjusted upwards. The adjustment can be more complicated because some ethnic groups (for example, South Asians) tend to be lighter, so there is an interaction between a woman's weight and her ethnic status.

Previous publications provide adjustment factors for various ethnic groups in either the first or the second trimesters of pregnancy (for example, Watt et al. ¹ ); however, corresponding adjustment factors for second trimester markers are not available for oriental women (women of Chinese, Korean, Japanese or South East Asian origin). Using data collected from the screening programme of the Wolfson Institute of Preventive Medicine, we here provide these, and update results for black and South Asian women previously reported from the programme, ¹ using data from a larger number of pregnancies. In this way we provide a comprehensive summary of ethnic status adjustments for all currently used serum screening markers.

Data were collected relating to women screened using either first trimester serum markers (free β human chorionic gonadotrophin [hCG] and pregnancy-associated plasma protein-A [PAPP-A]), second trimester serum markers (alphafetoprotein [AFP], unconjugated estriol [uE₃], hCG [total or free β] and inhibin-A), or Integrated first and second trimester serum markers, between August 1994 and December 2011. Serum marker values were expressed as multiples of the median (MoM; marker concentration divided by expected concentration for the same gestational age), derived from regressions of median marker concentration against median gestational age in one-week categories of gestational age. Log linear regression was used for first trimester free β-hCG and PAPP-A and second trimester AFP, and negative exponential regression for second trimester total and free β-hCG. ² Log quadratic regression was used for uE₃ ³ and log cubic for inhibin-A. ⁴ MoM values were calculated with and without correction for maternal weight (MoM divided by expected MoM for the same maternal weight, derived from a log-linear regression). Data for white women were used as the reference set for the median values at each gestational age and for the expected median MoM values for each maternal weight.

The table shows that, without weight adjustment, black women had first trimester free β-hCG levels 6% higher (95% confidence interval 4% to 7% higher) and PAPP-A levels 55% higher (54% to 57%) than white women (results for second trimester markers are similar to those previously reported ¹ ); South Asian women had first trimester free β-hCG levels 5% lower (6% to 4% lower) and PAPP-A levels 14% higher (13% to 16% higher) (results for second trimester markers are similar to those previously reported ¹ ). The Table also shows that without weight adjustment, oriental women had first and second trimester markers between about 10 and 30% higher than white women.

With weight adjustment, the black–white differences increased (see table). For example, for PAPP-A the difference increased from 55% higher in black women compared with white women without weight adjustment, to 72% higher with weight adjustment. This arises because the black women tended to be heavier than the white women (means 70 and 65 kg, respectively). In South Asian women, with weight adjustment, the South Asian-white differences decreased for both free β-hCG and PAPP-A, because the South Asian women were, on average, lighter than the white women (means 60 v 65 kg respectively). Oriental women were also lighter (mean 55 kg) than white women, so weight adjustment reduced all marker values, bringing the median closer to that of white women, for example, from 14% higher to 7% higher second trimester inhibin-A levels.

The first trimester marker results here are similar to those in previous publications; for example, Spencer et al. ⁵ showed that not-weight-corrected free β-hCG and PAPP-A levels were 14 and 23% higher, respectively, in oriental women than in white women (compared with 17% and 31% in this study) and weight corrected levels were 6% and 9% higher (compared with 6% and 7% higher in this study). Spencer et al. ⁵ also showed results similar to ours for Afro-Caribbean and Asian women. Kagan et al. ⁶ also showed similar weight corrected free β-hCG and PAPP-A levels for Afro-Caribbean, South Asian and East Asian women.

Marker levels can be adjusted for maternal weight using separate regression equations of the marker concentration plotted against gestational age and the MoM value against maternal weight for each ethnic group. This, however, requires sufficient data within each ethnic group (for example, at least 200 women) which may not always be practical. With fewer data relating to a particular ethnic group, this can be done as previously described, ¹ but using updated estimates in Table 1. This is done as follows:

Table 1

The ratio of serum markers in black, South Asian and oriental women to white women with and without maternal weight adjustment

				Ratio of median MoM (95% CI)
		Number of		Specific ethnic group/white women
Ethnic group	Serum marker	White women	Women in other ethnic group	Not-weight adjusted (a)	Weight adjusted (b)	Ratio of weight-adjusted MoM to not-weight-adjusted MoM (c=b/a)
Black women	1st trimester
	Free β-hCG	142,506	18,679	1.06 (1.04–1.07)	1.11 (1.00–1.13)	1.05 (1.04–1.05)
	PAPP-A	151,634	19,670	1.55 (1.54–1.57)	1.72 (1.70–1.74)	1.11 (1.09–1.11)
	2nd trimester
	AFP	163,574	28,006	1.10 (1.10–1.11)	1.15 (1.14–1.15)	1.05 (1.04–1.05)
	uE3	162,881	27,961	0.96 (0.96–0.97)	0.98 (0.98–0.99)	1.02 (1.02–1.02)
	Total hCG	10,156	3729	1.15 (1.12–1.17)	1.19 (1.16–1.21)	1.03 (1.03–1.03)
	Free β-hCG	152,681	24,232	1.07 (1.06–1.08)	1.12 (1.11–1.13)	1.05 (1.04–1.05)
	Inhibin-A	150,325	25,898	0.95 (0.95–0.96)	0.98 (0.97–0.99)	1.03 (1.03–1.03)
South Asian women	1st trimester
	Free β-hCG	142,506	23,293	0.95 (0.94–0.96)	0.89 (0.88–0.90)	0.94 (0.94–0.94)
	PAPP-A	151,634	24,262	1.14 (1.13–1.16)	1.01 (1.00–1.02)	0.89 (0.89–0.89)
	2nd trimester
	AFP	163,574	19,852	1.01 (1.00–1.02)	0.95 (0.95–0.96)	0.94 (0.93–0.94)
	uE3	162,881	19,782	1.14 (1.13–1.14)	1.11 (1.11–1.12)	0.97 (0.97–0.97)
	Total hCG	10,156	2205	1.12 (1.09–1.16)	1.06 (1.02–1.09)	0.95 (0.94–0.95)
	Free β-hCG	152,681	17,570	0.99 (0.98–1.00)	0.93 (0.92–0.94)	0.94 (0.94–0.95)
	Inhibin-A	150,325	19,782	1.11 (1.10–1.12)	1.06 (1.05–1.07)	0.95 (0.95–0.95)
Oriental women	1st trimester
	Free β-hCG	142,506	6578	1.17 (1.15–1.19)	1.06 (1.04–1.08)	0.91 (0.90–0.91)
	PAPP-A	151,634	7123	1.31 (1.29–1.33)	1.07 (1.06–1.09)	0.82 (0.80–0.82)
	2nd trimester
	AFP	163,574	5392	1.10 (1.09–1.11)	1.01 (0.99–1.02)	0.92 (0.92–0.93)
	uE3	162,881	5383	1.16 (1.15–1.17)	1.11 (1.10–1.13)	0.96 (0.96–0.96)
	Total hCG	10,156	564	1.30 (1.19–1.39)	1.18 (1.09–1.25)	0.91 (0.90–0.91)
	Free β-hCG	152,681	4817	1.23 (1.20–1.26)	1.10 (1.08–1.13)	0.89 (0.89–0.90)
	Inhibin-A	150,325	5382	1.14 (1.12–1.15)	1.07 (1.06–1.08)	0.94 (0.94–0.94)

MoM = multiple of the median for white women of the same gestational age; CI =confidence interval; hCG = human chorionic gonadotrophin; PAPP-A = pregnancy-associated plasma protein-A; uE₃ = unconjugated estriol; AFP, alphafetoprotein

Adjustment factors are in bold; marker concentration divided by factor in (a) then MoM calculated, and if maternal weight is available, marker MoM divided by factor in (c) then weight adjusted MoM calculated (see text for more detail)

Regardless of whether maternal weight is available:

The marker concentration (in mass units) for a woman of a particular ethnic group is divided by the appropriate adjustment factor in column a of the table (the ratio of the median not-weight-adjusted MoM in the ethnic group compared with white women). This makes the marker concentration equivalent to that of a white woman.

The adjusted concentration from (1) is divided by the median of the concentrations in all women (concentrations adjusted for non-white women as in (1) of the same gestational age to yield a MoM value.

Then if maternal weight is not available the calculated MoM value from (2) is used in estimating the risk of a Down's syndrome pregnancy. If maternal weight is available the following is done:

The MoM value calculated in (2) is divided by the appropriate adjustment factor in column c of the table (the ratio of weight-adjusted MoM to not-weight-adjusted MoM). This removes the part of the adjustment in (1) that is due to maternal weight differences between ethnic groups, so that the correction is purely due to ethnic group differences.

The adjusted MoM value calculated in (3) is then divided by the expected MoM for the woman's weight for all women, to yield the final weight-adjusted MoM.

The main reason for following these four steps is that results from all women (rather than using an ethnic group as the reference category) can be used to derive the expected median concentration for a particular gestational age and the expected median MoM for a particular maternal weight. This will also ensure that the median MoM is 1.0 across all women. As an example, if an oriental woman has an AFP concentration of 30 IU/mL, taken at 16 weeks and three days gestation, and weighs 65 kg: (1) using the adjustment factor from column a of the table, her adjusted concentration is 30/1.10 = 27.3 IU/mL. (2) If the expected median AFP at 115 days is 30 IU/mL then her AFP MoM is 27.3/30 = 0.91. (3) Using the adjustment factor from column c of the table her adjusted MoM is 0.91/0.92 = 0.99. (4) If the expected median AFP MoM for a woman weighing 65 kg is 1.02, then her weight-adjusted AFP MoM is 0.99/1.02 = 0.97.

The effect of adjusting for ethnicity in antenatal screening for Down's syndrome will have little impact on overall screening performance (detection rate for a specified false-positive rate and vice versa), particularly if the number of non-white women is small relative to the number of white women, ^1,5 but it can result in similar screening performances between ethnic groups, and greater accuracy in estimating a woman's risk of having an affected pregnancy.