Objectives-To assess the impact of current serum fetoprotein (AFP) assays on the performance of screening for open neural tube defects and Down's syndrome. Methods-Maternal serum samples, collected between weeks 15 and 22 from 470 singleton pregnancies without neural tube defects or Down's syndrome, were assayed for AFP using an automated fluorometric immunoassay. The samples had been assayed for AFP using an in house radioimmunoassay with a lower precision ten years before. The variance of AFP using the radioimmunoassay was compared with that using the current fluorometric assay and then used to estimate the detection rates and false positive rates for neural tube defect and Down's syndrome screening. Results-Current serum AFP assays are more precise. Using a cut oV level of 2.5 multiples of the median, the false positive rate in screening for anencephaly and open spina bifida was 0.8% with the new assay compared with 2% using the previous assay. When screening for Down's syndrome, the false positive rate is reduced by about one percentage point without loss of detection. Recently, it has become apparent when using serum AFP measurement as a screening test for open neural tube defects, that the observed positive rate (that is the proportion of women screened with raised AFP levels) is lower than that expected on the basis of serum AFP distribution parameters estimated about 10 years ago.
Conclusion-ImprovementsA reduction in the serum AFP screen positive rate is not surprising. The eYcacy of screening using AFP measurement is determined by the distribution of serum AFP in aVected and unaVected pregnancies-the less overlap in the distributions, the better the test. A reduced overlap is reflected in a reduced standard deviation, and this will probably arise due to more precise serum AFP measurement. We examine the eVect of the improvement in precision on the standard deviation of AFP at 15 to 22 weeks of pregnancy, and we estimate the resulting improvement in the performance of antenatal screening for open neural tube defects and Down's syndrome.
MethodsEstimates of the standard deviation of maternal serum AFP (and the correlation between AFP and the other serum markers) in unaffected pregnancies at any given gestational age from 15 to 22 weeks of gestation have been published. These data were based on estimating gestational age from the first day of the last menstrual period, with and without maternal weight correction, and on a biparietal diameter scan measurement (again, with and without weight correction) using data from 970 white women screened at the Homerton Hospital, London, from 1989 to 1990. [3][4][5] Estimates for anencephaly and open spina bifida were based on data from the UK Collaborative AFP Study (177 cases), 6 7 and those for Down's syndrome from a series from Oxford (77 cases).3-5 They were also adjusted to take account of the method of estimating gestational age (dates or scan) and maternal weight.Four hundred and seventy serum samples from the dataset of 970 white women with unaVected...