Diagnostic value of maternal alpha-fetoprotein variants in biochemical screening for trisomies 21 and 18 in the second trimester

Trisomy 21 and trisomy 18 are the most frequently observed chromosomal abnormalities in neonatal congenital malformations. After birth, children with these abnormalities cannot support themselves, which creates serious economic and social demands on families and society. Therefore, early prenatal diagnosis and intervention with certain measures are particularly important. The use of highly sensitive and specific markers for prenatal screening in early and mid-pregnancy has become a recent goal in research on prenatal diagnosis and screening. Previous studies have found low sensitivity of maternal serum AFP screening for fetuses with Down syndrome 21 and 18 in the second trimester.30. In the present study, we evaluated the effect of combining free β-hCG after replacing AFP with AFP-L2 or AFP-L3 on the prediction of trisomies 21 and 18.

Research has shown that AFP-L2 levels of trisomy 21 and 18 screening of fetuses in the second trimester were higher than in control subjects (1.48 and 1.46 versus 0.83 MoM, respectively) (P17 also showed that the average serum AFP-L2 level in fetuses with Down syndrome was higher than in normal subjects. The results of our research also revealed that the levels of AFP-L3 of gravidas carrying trisomy 21, 18 fetuses were increased compared to those of control subjects (1.73 MoM, 1.66 MoM vs. 0.85 MoM , respectively) (all P31 demonstrated that the ratio of serum AFP-L3 to AFP in pregnant women with a DS fetus was higher than in pregnant women carrying a healthy fetus. However, Huai et al.32 showed that the level of serum AFP-L3 MoM in normal control subjects was significantly higher than that of gravidas with DS fetuses (PP19. Similarly, Wu et al.33 demonstrated that AFP-L2 and AFP-L3 concentrations in women with healthy fetuses were lower than in DS fetuses, which resembled the data from the current study.

In the current research, we determined the levels of AFP-L2 and AFP-L3 by ELISA, and we showed that the screening of ASC for trisomy 21, 18 fetuses using AFP-L2 was superior to that with AFP for trisomy 21, and lower than that for trisomy 18 (0.785 and 0.775 vs 0.613 and 0.869, respectively). The AUCs of trisomy 21, 18 fetus screening with AFP-L3 were 0.758 and 0.754, respectively, which were higher than the AUCs using AFP (0.613 and 0.869). These results showed that the diagnostic values ​​of AFP-L2 and AFP-L3 for trisomy 21, 18 were better than that of AFP. After denaturation of AFP, it can be easily refolded into the two recombinantly produced AFP forms under the conditions of dilution and redox reaction, and it is detectable by ELISA34.

Compared to Long’s study, our AUC for AFP-L3 screening for trisomy 21 in our study was higher than the AFP-L3 to AFP ratio (0.758 vs 0.710)32, but lower than the concentration of AFP in gravidas as measured by a combined liquid-phase assay. Yamamoto found that the AUCs of AFP MoM (0.750), AFP-L3% (0.868), L3 MoM (0.949) and L3 MoM/AFP MoM (0.946), respectively, as determined by liquid phase binding test35. This revealed that different detection methods can produce different diagnostic values. However, the AUC for trisomy 21 as predicted by AFP-L2 and AFP-L3 in the present study showed a tendency to be lower than that predicted in our preliminary study (0.891, 0.824)19. The apparent discrepancy between the former and the latter may have been due to our subsequent increase in the number of control cases, or it may have been related to our modeling after calibrating for gestational age and weight in the current study.

AFP levels in the control group were higher than in gravid trisomy 21, 18 fetuses (all PP

Yamamoto et al. postulated that the placental transfer of the AFP-L3 component in women carrying a fetus with trisomy 21 may be relatively high, which could be one of the causes of the elevated serum AFP-L3 levels in these pregnant women36. Additionally, Yamamoto37 did not detect a correlation between maternal serum AFP-L3 and AFP MoMs (r = 0.006) and did not observe a significant correlation between serum AFP level and AFP-L3 percentage . No significant correlation was observed between serum AFP level and AFP-13 percentage (r = 0.160) by Khien et al.38.

In our latest study, the MoM values ​​of AFP-L2 and AFP-L3 were calculated by indirect simulation based on the AFP MoM values ​​of 21,656 mid-pregnancy maternal serum samples in our laboratory.24. We therefore used the 569 maternal serum samples in this study to construct the risk model construct of AFP-L2 and AFP-L3. These results all refuted any correlation between AFP-L2/AFP-L3 and AFP values. In support of this, our data also suggests that the combination of AFP-L2 and AFP-L3 was better than that of AFP, AFP-L3 and AFP-L2 in as unique indicators to predict fetal trisomies 21, 18.

For trisomy 21, the results revealed that substitution of AFP-L2 for AFP improved IDI and NRI by 9.56% and 26.50%, respectively, than substitution of AFP- L3 to AFP increased IDI and NRI by 12.34% and 26.70%, respectively. For trisomy 18, replacing AFP with AFP-L2 decreased IDI and NRI by 8.12% and 13.84%, and replacing AFP with AFP-L3 lowered the IDI and NRI by 1.52% and 8.54% in Table 4, respectively.

The NRI is an approach of classifying patients into risk categories and determining how the new model reclassifies patients into risk categories compared to previous models39. The calculation of the IDI is another way of assessing reclassification which is not based on a predefined risk category but represents an ongoing measurement40.41. Adding AFP-L2 and AFP-L3 to the traditional model improved the ability to predict T21 and T18 fetuses assessed by ROC and to reclassify subjects into different risk categories by NRI and IDI. It shows that AFP-L2 and AFP-L3 can effectively increase the accuracy of previous studies.

The data in Table 4 also suggest that the utility of AFP-L3 and AFP-L2 alone in predicting trisomy 21 was better than that with AFP, while for trisomy 18 the ranked sequence was AFP-L3 > AFP > AFP-L2. When combined with the DR, FPR, and +LR rating indicators, Model I was optimal, followed by Model F and Model E.

When false positive and false negative results are unavoidable, it behooves us to find a way to maximize net benefit, which is a clinical utility issue. Therefore, we introduced DCA to assess the predictive effects of different models for trisomies 21 and 18. It is generally assumed that the threshold probability range of the abscissa of the risk threshold is 0 to 1. However, if a specific situation is consistent in clinical practice and the probability threshold reaches a certain value (e.g. 40%), intervention measures should be taken42 and, therefore, a risk threshold >0.4 is of little significance. Figure 3 illustrates the order classified as Model A > Model G > Model I. The main models are Model B, Model E, Model I, and Model D using DCA prediction for trisomy 18.

We note that these results were different from previous AUC, IDI, and NRI assessment indices, which may be due to the small number of trisomy 18 cases in the current study.18.19.

Comments are closed.