To NIPT or Not to NIPT

After the discovery by Lejeune et al. in 1959 that patients with ‘mongolism’, i.e. Down syndrome, had 3 instead of 2 copies of chromosome 21, a surge of karyotyping resulted in the identification of an aberrant chromosome number as the cause of numerous genetic disorders [Lejeune et al., 1959; Lejeune and Turpin, 1961]. By culturing cells from amniotic fluid samples and subsequent karyotyping, trisomies 21, 18, 13, and other numerical aberrations were readily ascertained during the second trimester of pregnancy [Schwarz, 1975]. In 21 European countries, surveys over the past 20 years of karyotyping pregnancies in women aged 35+ have shown a prevalence per 10,000 births of 22.0 (95% CI = 21.7-22.4) for trisomy 21, 5.0 (95% CI = 4.8-5.1) for trisomy 18, and 2.0 (95% CI = 1.9-2.2) for trisomy 13 [Loane et al., 2013]. Since amniotic fluid sampling may increase the risk of a miscarriage, this procedure is only performed on pregnancies which are deemed to be ‘at risk’ [Morris et al., 2012]. The discovery of fetal reticulocytes and of cell-free fetal DNA (cffDNA) in peripheral blood samples of pregnant women suggested novel ways to circumvent the risk of a miscarriage associated with an amniocentesis or chorionic villi sampling (CVS) [Herzenberg et al., 1979; Thomas et al., 1995; Lo et al., 1999].

Upon recovery of intact fetal cells or cffDNA from maternal blood samples, noninvasive prenatal genetic diagnosis can be performed with either FISH or real-time PCR [Bischoff et al., 2002]. Using FISH, fetal aneuploidies were detected with 50-75% accuracy, while real-time PCR assays proved to be 4× more sensitive. This study also suggested that the cffDNA and the intact fetal cells may be of different biological origin [Bischoff et al., 2002]. The fetal cells in maternal blood were transferrin receptor expressing fetal reticulocytes, while the cffDNA most likely originated from placental cyto- and syncytiotrophoblastic cells [Bianchi et al., 1994; Flori et al., 2004]. While FISH on intact cells limits the diagnostic scope to detect fetal aneuploidy and sex, real-time PCR-based assays on cffDNA can cover all types of mutations [Wright and Burton, 2009]. With the advent of massively parallel sequencing noninvasive prenatal testing (NIPT), the testing of all pregnancies became technically feasible and economically affordable [Bianchi and Wilkins-Haug, 2014]. However, a number of recent observations should caution us against a widespread and indiscriminate use of NIPT [Mao et al., 2014; Hochstenbach et al., 2015a, b; Snyder et al., 2015].

The NIPT result of a pregnancy indicated a trisomy 21, whereas a trisomy 18 and diploidy for chromosome 21 was detected by karyotyping amniotic fluid cells [Mao et al., 2014]. The authors found no evidence for maternal mosaicism, microchimerism or a vanishing twin with a trisomy 21 that could have explained the discordant NIPT result. Examination of multiple placental biopsies showed both a trisomy 21 and a trisomy 18 mosaicism. The authors suggested that confined placental region(s) with higher proportions of trisomy 21 cells may have preferentially released fetal DNA fragments into the maternal circulation. These findings indicate that placental mosaicism may potentially give rise to inaccurate NIPT results [Mao et al., 2014].

In a pregnancy in which ultrasonographic imaging at 16^1/7 weeks of gestation demonstrated severe growth retardation, an omphalocele, a univentricular heart, and clenched fists, altogether indicative of trisomy 18, CVS and NIPT were performed on a peripheral blood sample of the pregnant woman [Hochstenbach et al., 2015a]. A short-term culture of cytotrophoblast cells of the CVS showed a 46,XY karyotype, consistent with the NIPT result, while a long-term culture of mesenchymal core cells revealed a non-mosaic 47,XY,+18 karyotype. After termination of the pregnancy, cultured umbilical cord fibroblasts showed a 47,XY,+18 karyotype. Analyzing 10 placental samples by FISH with centromere 18 (D18Z1), centromere 11 (locus D11Z2, control for diploidy) and DYZ3 (control against maternal cells) probes, in 7 out of 10 cytotrophoblast biopsies, the authors found a low percentage of trisomy 18 cells (average 3.6%, range 0-13%) and in 3 biopsies a high level of trisomy 18 (varying from 40 to 80%). In all 10 mesenchymal biopsies, trisomy 18 was predominant (average 74%, range 59-90%). These findings agree with a study of a large series of consecutive patients analyzed by CVS, indicating a false-negative NIPT result for trisomy 13, 18, or 21 of 1 in 107 cases [Grati et al., 2012]. The concordance between NIPT and cytogenetic results of cytotrophoblast cells unambiguously demonstrates that the cffDNA in the maternal circulation is mainly, maybe even exclusively, derived from the cytotrophoblast [Mao et al., 2014; Hochstenbach et al., 2015a].

Investigating 4 pregnancies in which NIPT indicated a trisomy 18 in healthy babies, Snyder et al. [2015] found in 2 cases that the maternal chromosomes 18 contained numerous duplications. Since a converse constellation, maternal deletions combined with a fetal trisomy resulting in a ‘diploid NIPT’, is also conceivable, the authors modeled the likelihood that maternal CNVs may interfere with NIPT outcomes. Their model suggests that nonpathogenic CNVs ranging from 0.3 to 0.6 Mb on chromosomes 13, 18, and 21 may be sufficient to cancel out fetal aneuploidies. Thus, a duplication of 0.5 Mb of chromosome 18 explained the false-positive NIPT result in one of their patients [Snyder et al., 2015]. Likewise, a loss of 0.3 Mb would give rise to a normal NIPT result in a case of a trisomy 21. Therefore, they recommend to discard the portions of NIPT data that overlap with a maternal CNV or to group NIPT data into bins of fixed sizes and to discard bins that prove to be outliers based on z-test scores. This means that a NIPT for trisomies should always be accompanied by a high-resolution array-CGH analysis of the maternal genome.

A review of 15 clinical validation studies, comprising over 21,000 cases, showed that 3 out of 835 cases were false negative (0.36%) for trisomy 21, 6 out of 315 for trisomy 18 (1.90%) and 4 out of 60 for trisomy 13 (6.67%) [Liao et al., 2014]. Nevertheless, in 2 cases with a trisomy 13 and one with a trisomy 18, respectively, Hochstenbach et al. [2015b] could not find an explanation for false-negative NIPT results. Reviewing the literature, they found another 12 case reports published during the past 2 years with false-negative NIPT results. Although the numbers of cases in these reports are vanishingly small in comparison with the large validation cohorts, they still confer a disquieting message. Since the cffDNA used for NIPT is not directly of fetal origin, but most likely derived from the cytotrophoblast, factors such as placental mosaicism or a vanishing twin, etc. may confound the outcome of the NIPT.

These findings emphasize discrepancies between NIPT and the true fetal karyotype. This problem was first encountered when CVS was introduced some 30 years ago. Again, placental mosaicism is the most probable confounding factor, causing discrepant outcomes between NIPT and classical karyotyping. In clinical practice, cases with normal NIPT results and an ultrasonographic anomaly should be followed up by karyotyping or array-based aneuploidy screening. Conversely, in cases with an unsuspicious ultrasonographic result, but an abnormal NIPT outcome, amniocentesis followed by karyotyping becomes mandatory. During counseling, pregnant women, being offered NIPT, should always be made aware of the inherent limitations of this test and the small, but finite, likelihood of a false-negative or false-positive result.