Provision of screening for Down's syndrome

Is maternal serum total hCG a marker of trisomy 21 in the first trimester of pregnancy?

In a study of 130 first trimester cases of trisomy 21 and 959 controls we have shown that the median MoM for alpha-fetoprotein (AFP) is lower (0.82) and that for total human chorionic gonadotrophin (hCG) is higher (1.31) than in the control group. For AFP 15.3% of cases were below the 5th centile and for total hCG 19. 8% were above the 95th centile. The median shift observed for AFP and total hCG is poorer than that for pregnancy associated plasma protein-A (PAPP-A) or free beta-hCG and together with maternal age, AFP and total hCG could only be expected to detect 40% of cases. In combination with PAPP-A, total hCG would identify 52% of cases, somewhat less than the 67% observed with free beta-hCG and PAPP-A. However, we have demonstrated for total hCG a significant temporal change in median MoM with gestational age. Before 70 days the median MoM was less than 0.5, between 70 and 83 days this increased to 1.13, and between 84 and 97 days this increased to 1.52. This median shift has significant implications for interpreting previous studies and even more significant implications for detection rates. When population parameters specific to the gestational age in question are used, detection rates with total hCG and PAPP-A increase from 47% at 70-83 days to 60% at 84-97 days. This observation explains much of the confusion around total hCG in the first trimester and shows the importance of selecting analyte pairs and population parameters appropriate to the time in gestation when screening is performed.

Accuracy of Down's syndrome risks produced in a prenatal screening program

Down's syndrome screening in the second trimester using a variety of combinations of maternal serum markers has become an established part of antenatal care over the past decade. The methodology of calculating Down's syndrome risk using statistical models based on the Gaussian distribution functions of the various markers used is often considered daunting, and many screening centres rely on specific screening computer software to undertake the relevant calculations. Such software is often a closed system which does not allow the user to check or change any of the parameters within the model. It is clear from a recent UK NEQAS survey of Down's syndrome screening laboratories (Spencer, Ellis, Seth, 1997, unpublished data) that many centres are unsure what method of calculating risk and what population parameters are used in their software, and how appropriate these are for the population being screened. This confusion has led to uncertainty over the reliability of the risk estimates produced in Down's syndrome screening programs.

Variation in biochemical screening for Down's syndrome in the United Kingdom

A questionnaire survey was undertaken of all 73 laboratories performing Down's syndrome screening in 1995. An estimated 352,000 tests were performed representing 47% of maternities. Three-quarters of these tests have ultrasound dating information at the time of testing. The majority of laboratories (70%) commenced screening at 15 weeks of gestation or later, and there was considerable variation in the upper limit of screening (17 to 24 weeks). Eighty-six percent of laboratories screened all women regardless of age. The reported Down's syndrome risk was based on term in 85% of laboratories. There was an inconsistent approach to determining and reporting high risk for trisomy 18 (Edwards' syndrome): 5% reported risks on report forms and 42% notified the clinicians if the risk was considered to be raised.

The influence of smoking on maternal serum AFP and free beta hCG levels and the impact on screening for Down syndrome

I have studied the impact of maternal smoking on the levels of maternal serum alpha-fetoprotein (AFP) and free beta human chorionic gonadotrophin (hCG) during the second trimester in a large series (30,727) of self-reported smokers and non-smokers whose pregnancy resulted in the birth of a normal fetus. I have confirmed in a smaller subset of this population that self-reporting is an accurate method of assessing smoking status as confirmed by biochemical (serum cotinine) assessment. In addition, I have investigated marker levels among 195 pregnancies affected by Down syndrome with smoking status confirmed by measurement of serum cotinine. In both unaffected and Down syndrome groups, the incidence of smoking was 19 per cent but a considerable variation was observed with maternal age when the incidence in younger women (under 25) was 32 per cent. AFP median levels in unaffected smokers were 3 per cent higher, whilst in the Down syndrome group smoker medians were increased by 10 per cent compared with the non-smoker group. Free beta hCG levels in unaffected smokers were reduced by 14 per cent, whilst in the Down syndrome group smoker median levels were decreased by 16 per cent. In the smoker group, the Down syndrome detection rate was 10 per cent lower than in the non-smoker group, whilst the false-positive rate was also 2 per cent lower. Correcting for smoking status would redress this inequality and produce an overall 2 per cent increase in the detection rate for a 0.4 per cent increase in the false-positive rate. This increase in screening performance may be worth building into screening programmes, particularly in populations with a high smoking incidence.

Between-pregnancy biological variability of maternal serum alpha-fetoprotein and free beta hCG: implications for Down syndrome screening in subsequent pregnancies

In women who have an increased Down syndrome risk in a first pregnancy there is a five-fold greater chance of also having an increased Down syndrome risk in a second subsequent pregnancy. Similarly, in women who have a high alpha-fetoprotein (AFP) level in the first pregnancy, suggesting an increased risk of a neural tube defect (NTD), there is also a five-fold greater chance of them also having a high result in a second subsequent pregnancy. Such a biological association of serum marker levels between pregnancies suggests that there are additional maternal or genetic factors influencing the levels of these serum markers, other than the physiological factors which in themselves are poorly understood. In theory, it is possible to correct for high/low marker results in a previous pregnancy and to do so, I estimate, would reduce the overall Down syndrome screening false-positive rate by about 0.2 per cent. There are insufficient data to make any prediction regarding the impact on detection rates. The small reduction in false-positive rate is unlikely to be a feature worth implementing in Down syndrome screening programmes.

Nomograms to help inform women considering Down's syndrome screening

OBJECTIVE

To derive graphical information for use in counselling women considering whether or not to have maternal serum screening for Down's syndrome.

DESIGN

Statistical modelling of the frequency distribution of estimated Down's syndrome risk for four marker combinations.

RESULTS

Nomograms are provided showing for each maternal age: (a) the detection and false-positive rates, and (b) the proportion of pregnancies with different estimated risks.

CONCLUSION

When screening is offered, clinicians need to have information readily available on test accuracy and the likely result, which is specific to the individual.