Some problems of alpha-fetoprotein screening.

In two years of a routine screening programme for the detection of neural-tube defects over 6000 women had serum-alpha-fetoprotein (A.F.P.) measured. As a result, 80 women carrying live singleton fetuses were offered amniocentesis under ultrasound control. All neural-tube defects detected at birth were recorded and necropsies were done routinely on abortuses, thus permitting evaluation of the screening programme in terms of false negatives and false positives. 16 amniotic-fluid A.F.P.s were abnormal, 1 for an unaffected fetus; and 2 others were normal for fetuses which had closed lesions. At the serum-A.F.P. stage, 3 lesions were missed. The detection-rate compares favourably with those of other series. However, in a pilot study, routine ultrasonic assessment of gestation yielded a substantial proportion of cases where a serum-A.F.P. deemed normal or abnormal when gestation was assessed clinically would have been misleading. Loss of normal fetuses subsequent to amniocentesis (5 in this series) also has to be taken into account.