Observed and Modeled Positive Predictive Values Using Cell-free DNA Testing for Fetal Trisomy in a Clinical Laboratory Population

Abstract

<b><i>Introduction:</i></b> The objective of this study was to explore different approaches to communicating the positive predictive value (PPV) of cell-free DNA screening for fetal trisomy. <b><i>Methods:</i></b> PPV was established for 4 maternal age-groups (&#x3c;30, 30–34, 35–39, and &#x3e;39 years) from clinical laboratory data and compared to the modeled PPV from an online calculator. In women under 35, PPV was compared between 2 subsets, high risk and low risk, classified based on the diagnosis codes that were provided to the laboratory. <b><i>Results:</i></b> In 503 high-probability trisomy 21 results, the observed PPVs in the 4 age-groups were 97.0% (&#x3c;30), 98.9% (30–34), 99.5% (35–39), and 96.3% (&#x3e;39), all higher than those from the calculator, which ranged from 53 to 95%. Likewise, PPVs were 77.4–97.0% observed versus 16–78% modeled in 131 trisomy 18 cases and 30.4–80.0% observed versus 6–61% modeled in 80 trisomy 13 cases. In women under 35, PPV for the trisomies combined was 90.4% in the higher-risk group compared to 79.7% in the lower-risk group. <b><i>Conclusion:</i></b> Modeling PPV based on maternal age will provide an underestimate in a clinical population. Although the PPV is higher for the samples with higher-risk diagnosis codes, the information that accompanies clinical samples is too general to model PPV for a specific patient.