The Predictive Value of Prenatal Cell-Free DNA Testing for Rare Autosomal Trisomies: A Systematic Review and Meta-analysis

Abstract

ABSTRACT Noninvasive prenatal testing has been a reliable method for prenatal screening for the common autosomal trisomies for the last decade. It has been analyzed in many prospective studies for accuracy in screening for the most common trisomies—13, 18, and 21—and the technology has continued to progress. With high accuracy for common trisomies, advances are attempting to expand noninvasive prenatal testing for identifying rare autosomal trisomies, such as trisomies 7, 15, 16, and 22; sex chromosome aneuploidies; and segmental copy number variants. Although cell-free DNA (cfDNA) technology is now available to screen for these abnormalities, accuracy and detection rates have not been well studied. This systematic review analyzes and reports the diagnostic accuracy of cfDNA screening for the rare autosomal trisomies. This review included case series with 10 or more cases that reported on the accuracy of cfDNA in reporting rare autosomal trisomies that were confirmed by diagnostic results or postnatal genetic testing; the final review included 31 studies. Methodological analysis found the risk of bias to be high in most studies, with only 5 rated as high quality and low risk in all domains. All studies included in the final analyses were published between 2017 and 2022, and study populations ranged from 14 to 153,575 patients. Indications for screening and characteristics of the patients varied from study to study, with 12 studies in high-risk women, 10 studies in the general obstetric population, and 9 studies without inclusion of underlying risk factors and indications. Whole-genome sequencing was used in all studies as testing methodology. Testing timing varied and was performed anywhere from gestational week 8 to week 39. In the 3 studies that analyzed sensitivity and specificity of cfDNA testing, sensitivity ranged from 87.2% to 100%, and specificity ranged from 90.7% to 99.9%. In all 3 of these studies, there were significant limitations that led authors to acknowledge the difficulty of drawing clear conclusions about sensitivity and specificity from their data. Meta-analysis performed for positive predictive value (PPV) provided a pooled estimate of 11.46% (95% confidence interval [CI], 7.80%–15.65%). Sensitivity analysis including only studies at low risk of bias gave a PPV of 9.13% (95% CI, 2.49%–18.76%). Analysis was performed again with groups divided into high or low background risk, with a slightly higher PPV in the higher background risk group, but no significant difference between these groups ( P = 0.595). Statistical heterogeneity was not explained by year of publication or baseline risk of participants but could be largely explained by rate of diagnostic follow-up. Studies with a follow-up rate of less than 50% showed a PPV of 23.69% (95% CI, 9.21%–41.94%), whereas studies with a follow-up rate greater than 50% reported a PPV of 8.80% (95% CI, 6.08%–11.90%). Based on the results of this systematic review and meta-analysis, it is likely that previously reported PPVs of cfDNA's diagnostic accuracy for detecting rare autosomal trisomies are overestimated. A reason for this could be lower follow-up rates as well as testing bias in women with high-risk cfDNA results. One reason for the low prevalence of rare autosomal trisomies is that many confirmed cases result in miscarriage before 12 weeks of gestation. Pregnancies that continue and that are not mosaic often experience major complications. The occurrence of mosaicism may impact the PPV, as mosaicism carries some risks but is sometimes not detected from cfDNA samples. This study was limited by the details included in the studies analyzed; this caused significant statistical heterogeneity, which limits the conclusions that can be drawn. In addition, detection rates may differ by type of cfDNA test, and this analysis did not stratify by type of test. The results of this analysis are consistent with previous literature on detection of rare autosomal trisomies in the general population and show that even with a high specificity and sensitivity for cfDNA testing in general, care should be taken in making clinical decisions on the part of both providers and patients in instances where a rare autosomal trisomy is suspected.