Prospective evaluation of prenatal maternal serum screening for trisomy 18

Our goal was to evaluate the performance of prenatal serum screening for trisomy 18.

All 40,762 samples for maternal serum testing (August 1991 to June 1994) with a trisomy 18-positive screen (n = 175, alpha-fetoprotein < or =0.75 multiples of the median, unconjugated estriol < or =0.60 multiples of the median, human chorionic gonadotropin < or =0.55 multiples of the median) were analyzed. Results of all amniocenteses, ultrasonographic studies, and birth or death certificate information were obtained from the Iowa Expanded Serum Screening Program, the Iowa Department of Public Health, and the Iowa Birth Defects Registry.

We obtained the expected screen-positive rate for trisomy 18 (0.43%, 175/40,762). Fourteen samples from outside the state were excluded, which left 161 cases with outcome data obtained through amniocentesis (n = 121), birth certificates (n = 34), telephone contact (n = 2), or a sonogram indicating a nonviable gestation (n = 4). Of 121 screen-positive women undergoing amniocentesis, 119 had a normal karyotype and 2 had an abnormal karyotype: 69,XXY and 47,XY,+18. Of 36 who declined amniocentesis, none had findings consistent with aneuploidy on clinical neonatal examination. Of the 103 patients who had a detailed ultrasonographic study at the University of Iowa, 27 had a subtle fetal abnormality or growth alteration. Both cases with aneuploidy were in this group. An additional 7 cases of trisomy 18 without the typical trisomy 18 maternal serum screening pattern were diagnosed during this period either at amniocentesis performed because of increased Down syndrome risk indicated by serum screening (n = 1), by elevated alpha-fetoprotein level (n = 1), or by advanced maternal age (n = 2) with serum for screening drawn coincidentally, or they were diagnosed postnatally (n = 3). Three of the 7 cases had early second-trimester ultrasonographic examinations, and all showed abnormalities.

The detection rate of trisomy 18 among patients offered amniocentesis was significantly lower (p < 0.05) than the expected rate (10/161 on the basis of published data). Combining serum screening with detailed ultrasonographic evaluations may improve predictive value by more precisely targeting amniocentesis toward those at highest risk.