Positive Predictive Values and Outcomes for Uninformative Cell-Free DNA Tests: An Italian Multicentric Cytogenetic and Cytogenomic Audit of DiagnOstic Testing (ICARO Study)

Methods for diagnosis of genetic abnormalities have recently undergone substantial improvement. Noninvasive prenatal testing (NIPT) is performed early in pregnancy, typically at the end of the first trimester using cell-free DNA (cfDNA) to identify 3 commonly occurring trisomies (13, 18, and 21). In addition, the test now includes identification of sex chromosome abnormalities, recurrent microdeletions, or other large genomic imbalances. Previous studies have shown NIPT to be an effective initial tool when followed up with other diagnostic measures and that it has high sensitivity and specificity. The purpose of this study was to extend the previous literature on effectiveness by establishing the positive predictive values (PPVs) of NIPT on a large scale in a national population, to understand if the PPV would differ based on follow-up method or NIPT methodology, and to study the implications of uninformative cfDNA tests. This study was a retrospective analysis of deidentified test results from 39 cytogenetics and cytogenomics laboratories in Italy and Italian territories between 2013 and March 2020. Samples included NIPT data and results from amniotic fluid (AF) cells, chorionic villus sampling (CVS), and newborn blood. Confirmatory diagnostic results after a high-risk NIPT result were included for 1327 pregnancies. More than half of NIPT high-risk results were in patients who underwent testing because of maternal age, with only 5.3% due to abnormal first-trimester combined screening. A small percentage (1.1%) of high-risk results were due to an abnormality identified by ultrasound. Each of the trisomies had a significantly different PPV (P < 0.0001), with T21 at 624/671 (93%; 95% confidence interval [CI], 91%-95%), T18 at 91/124 (73.4%, 95% CI, 65%-80%), and T13 at 26/84 (31%; CI, 19%-38%). For each of the 3 trisomies, there was a significantly higher PPV associated with follow-up testing done by CVS rather than AF. Methodology for CVS analysis did not make a significant difference in PPV value, and mosaicism was uncommon for each of the trisomies. A trisomy was detected in 19/139 (13.9%) noninformative cases. For sex chromosome abnormalities, PPVs for XXX, XXY, and XYY were significantly higher than for monosomy X (P < 0.0001). In these cases, confirmatory diagnostic method did not have a significant impact; PPVs for CVS and AF were similar. Variations between each of the abnormalities and between methodology in any aspect were insignificant. Positive predictive values also remained low for rare autosomal trisomies, with each less than 30% with 2 exceptions: T22 (57.1%) and T9 (33.3%). It is notable, however, that the number of samples for confirming rare autosomal trisomies was small, with n < 15 in every case. To date, this is the largest study of its kind confirming PPVs for cfDNA screening. These findings are consistent with pre-vious literature and with reports made or released by NIPT laboratories. This study allowed for more analysis of rarer abnor-malities, although statistical power remains small because of the sample size associated with rarer abnormalities. Particularly for T21, T18, and T13, PPVs were based off of confirmatory tests from mesenchyme or cytotrophoblasts. Chorionic villus sampling PPVs were also significantly higher than PPVs from AF. These findings show that NIPT PPVs vary according to condition and are also dependent on methods used for confirmatory diagnostic tests. In general, this demonstrates the need for case-specific testing and counseling to provide the best care and counsel to patients.