LAUSR

With interest, we read the article by Fiorentino et al. suggesting that there is clinical utility in expanding maternal plasma cell-free DNA (cf-DNA or NIPS) analysis to identify all rare aneuploidies and partial chromosomal imbalances in the general obstetric population. Their conclusion is based on 12 114 patients who received cf-DNA analysis using massively parallel shotgun sequencing. Confirmation of presence or absence of abnormality was based on karyotyping for all high-risk cases and on telephone interview or neonatal examination for all low risk cases. The authors claimed 100% sensitivity for trisomy 21 (88/88), trisomy 18 (15/15), trisomy 13 (12/12), sex chromosome abnormalities (36/36), rare mosaic and non-mosaic aneuploidies (10/10), and for segmental imbalances (8/8). The study design precludes accurate determination of analytic sensitivity. This is because measuring sensitivity requires the assessment of the total number of abnormal cases, and that can only be reliably obtained through comprehensive laboratory testing of the entire cohort. This includes apparently normal liveborns and those pregnancies that ended in spontaneous abortion. Studies on placental tissues would also be needed to determine the number of cases with a confined placental mosaicism (CPM). Based on the results of chorionic villus samples (CVS), a false-negative rate of ~1/40 for cytogenetically visible partial imbalances can be expected because of true fetal mosaicism with a normal cytotrophoblast (type 5 mosaicism). This latter condition is a biological unavoidable source of false negatives that can be expected by cf-DNA testing. In addition, because sex chromosome abnormalities, some partial imbalances, and low level autosomal mosaic aneuploidies will not necessarily be apparent from observing the neonatal phenotype, long-term follow-up would be needed to assess clinical sensitivity. Moreover, the low numbers of abnormal cases detected in the study population strongly suggest that only a proportion of segmental imbalances and rare aneuploidies was identified. The authors reported 13 positive cases for a segmental imbalance of which five were false-positives and one had a mosaic sex chromosome aneuploidy (that should have been identified through the existing NIPT testing). The seven autosomal true-positive cases included two cases where the reason for referral was fetal ultrasound abnormalities. The yield of large (cytogenetically visible) autosomal segmental imbalances in the low risk population without obvious fetal sonographic abnormalities therefore appeared to be less than 0.05% (5/9607), which is below that expected (0.096%; 1/1042). Additionally, submicroscopic CNVs would be expected to be present in as many as 1.65% of low risk cases. Although the authors do not claim that the test is designed to detect submicroscopic copy number variation (CNVs), only three partial imbalances less than 7 Mb (equivalent to one chromosome band) were found despite a relatively high depth of sequencing. In addition, only one case of 22q11.2 deletion was detected (1/12 078), which is below the expected number based on the prevalence in low risk prenatal population (1/1000). These findings suggest low test sensitivity for small segmental imbalances. The authors also reported 17 (1/710) cases NIPS positive for a rare autosomal trisomy (RAT) of which seven were non-mosaic and resulted in fetal loss, seven were not confirmed in the fetus and likely to be CPM (with one of them having a fetal UPD15), and three were mosaic aneuploidies with termination of pregnancy or no information on pregnancy outcome. The number of RATs was therefore substantially less than that expected from studies on chorionic villus specimens considering mosaic cases only (0.65%; 1/154). It is also much lower than that reported in another study that has evaluated expanded NIPT that included RATs. When considering viable pregnancies only, the rate of detection of RATs (1/1208) is far lower than the actual RAT prevalence in the low risk population without obvious fetal sonographic abnormalities (0.172%; 1/582). Finally, a cohort of this size should also be expected to contain approximately four cases of triploidy, but none were reported. We therefore conclude that there was an underascertainment of pregnancies with chromosome abnormalities and sensitivity is substantially less than 100%. In the study of Fiorentino et al., ultrasound abnormalities were present in at least 3.9% of the fetuses. Previous studies have demonstrated that cf-DNA testing has low clinical utility in evaluating high-risk pregnancies with fetal anomalies