Abstract
OBJECTIVE: To evaluate the accuracy of Z-scores of noninvasive prenatal screening (NIPS) in predicting 21, 18 trisomy, and X chromosome aneuploidy. METHODS: A total of 39,310 prenatal women were recruited for NIPS from September 2015 to September 2020. Interventional prenatal diagnosis was applied to verify the diagnosis of NIPS-positive results. Logistic regression analysis was employed to relate the Z-scores to the positive predictive value (PPV) of NIPS-positive results. Using receiver operating characteristic (ROC) curves, we calculated the optimal cutoff value of Z-scores to predict fetal chromosome aneuploidy. According to the cutoff value, NIPS-positive results were divided into the medium Z-value (MZ) and high Z-value (HZ) groups, and PPV was calculated to access the accuracy of Z-scores. RESULTS: A total of 288 effective values of Z-scores were used as the final data set. The logistics regression analysis revealed that Z-scores were significantly associated with true-positive results for 21 trisomy (T21) and 18 trisomy (T18) (P < 0.05), whereas the same was not observed for X chromosome aneuploids (P > 0.05). The optimal cutoff value of the Z-score for T21, T18, XO, XXX, and XXY indicated by ROC curve analysis were 5.79, 6.05, -9.56, 5.89, and 4.47, and the area under the curve (AUC) were 0.89, 0.80, 0.48, 0.42, and 0.45, respectively. PPV in the HZ group was higher than that in the MZ group, and the application of the cutoff value reduced the false discovery rate (FDR), which was only 2.9% in the HZ group compared with 61.1% in the MZ group for T21 and T18. The difference in total PPV between the MZ and HZ groups for X chromosome aneuploids was statistically significant. Moreover, the PPV for XXX and XXY seemed to increase with Z-scores but not for XO. CONCLUSION: The Z-score is helpful for the accurate judgment of NIPS results and for clinical prenatal counseling. Especially for T21 and T18, Z-scores have an excellent clinical association, which is superior to that seen with X chromosome aneuploids. In addition, using Z-scores to judge NIPS results offers a certain reference value for XXX and XXY but not for XO.