Comparison of urinary hyperglycosylated human chorionic gonadotropin concentration with the serum triple screen for Down syndrome detection in high-risk pregnancies

OBJECTIVE: Both modest screening performance and declining patient and physician acceptance have stimulated interest in alternative markers to the triple screen for the detection of Down syndrome. Our purpose was to compare the concentration of a single urinary analyte, hyperglycosylated human chorionic gonadotropin, with the serum triple screen (α-fetoprotein, human chorionic gonadotropin, and unconjugated estriol concentrations combined with age) for second-trimester Down syndrome detection. STUDY DESIGN: Urine and blood were obtained from pregnant women in the second trimester undergoing genetic amniocentesis. Urinary hyperglycosylated human chorionic gonadotropin concentration and serum triple-screen values were measured. Individuals undergoing amniocentesis because of abnormal triple-screen results were excluded. Individual Down syndrome risks on the basis of urinary hyperglycosylated human chorionic gonadotropin concentration plus maternal age and on the basis of the triple-screen results were calculated. For each algorithm the sensitivity and false-positive rate for Down syndrome detection at different risk thresholds were determined. From these values receiver operating characteristic curves were constructed, and the area under the curve was determined for each algorithm. Finally, the performance of a new combination in which urinary hyperglycosylated human chorionic gonadotropin concentration replaced serum human chorionic gonadotropin concentration in the triple screen was ascertained. RESULTS: We studied 24 pregnancies complicated by Down syndrome and 500 unaffected pregnancies between 14 and 22 weeks' gestation in a mostly white (93.5%) population undergoing amniocentesis primarily because of advanced maternal age. The sensitivity and false-positive rate for urinary hyperglycosylated human chorionic gonadotropin concentration were 75.0% and 5.6%, respectively, whereas those for the triple screen were 75.0% and 33.2%, respectively. Urinary hyperglycosylated human chorionic gonadotropin concentration was superior to the triple screen (area under the curve, 0.9337 vs 0.7887; P = .02). The substitution of urinary hyperglycosylated human chorionic gonadotropin concentration for serum human chorionic gonadotropin concentration in the triple screen resulted in a 91.7% sensitivity at a 10.0% false-positive rate, versus a 54.2% sensitivity for the traditional triple screen at the same false-positive rate. CONCLUSION: The performance of urinary hyperglycosylated human chorionic gonadotropin concentration was statistically superior to that of the serum triple screen in a high-risk population. The use of urinary hyperglycosylated human chorionic gonadotropin concentration as an alternative test or substitution of this measurement for serum human chorionic gonadotropin concentration in the triple screen would improve diagnostic accuracy and address many current concerns related to the triple screen.