The application of expanded noninvasive prenatal screening for genome-wide chromosomal abnormalities and genetic counseling

Abstract

Objective

To evaluate the clinical application of expanded noninvasive prenatal screening (eNIPS) for genome-wide large copy number variation (CNV), i.e. chromosomal deletion/duplication >5 Mb, and aneuploidy; also to provide practical information for counseling eNIPS positive cases.

Method

We recruited 34,620 women with singleton pregnancy for genome-wide cell-free plasma DNA sequencing. Screening positive cases were verified by karyotyping and/or SNP array.

Result

A total of 461 (1.33%) positive cases were identified through our cfDNA screening including 209 cases of common trisomies (0.60%), 124 cases of sex chromosomal abnormalities (SCA) (0.36%), 71 cases of other autosomal anueploidies (OAA) (0.21%), and 57 CNVs larger than 5 Mb (0.16%). The predictive positive values (PPV) were 70.06% in general for common trisomies with as high as 91.67% for Trisomy21 (T21), 40.22% in general for SCAs with as high as 100% for Jacob Syndrome (XYY). The PPV for OAAs was 5.45%, and T7/T8/T16/T22 were the most frequent OAAs (n = 15, 9, 9, 8, respectively). The PPV for CNVs larger than 5 Mb was 51.22% (n = 57) with the CNV mostly detected on Chr5/Chr4/Chr2/Chr7 (n = 10, 8, 5, 5, respectively).

Conclusion

The expanded NIPS had shown promising PPVs for CNVs (large than 5 Mb), SCAs and common trisomies, yet this method required higher efficacy in screening for OAAs. The post-test genetic counseling for expanded NIPS should be tailored to the types of positive cases and also address the origin of abnormal signals (fetal vs. maternal).

Keywords:

• Aneuploidy
• chromosomal deletion/duplication
• noninvasive prenatal screening
• positive predicted value
• genetic counseling

Introduction

Human chromosomal abnormalities can result in a spectrum of diseases ranging from fetal demise (e.g. trisomy and monosomy) to adults with neurological disorders (e.g. copy number variant, CNV) [1,2]. In live births, the estimated prevalence of chromosomal abnormalities was 1 in 150, and this number is greater in early gestation when taking early birth loss into account [3]. The discovery of cell-free fetal DNA (cffDNA) in maternal blood started a new era of prenatal screening for fetal chromosomal abnormality, by the technique named noninvasive prenatal screening (NIPS) [4]. Since the early 20-first century, NIPS rapidly evolved to cover a broader range of human chromosomal abnormalities: from the basic three common trisomies, trisomy21/trisomy18/trisomy13 (T21/T18/T13), to sex chromosomal abnormalities (SCA), other autosomal aneuploidies (OAA), and copy number variants (CNV) [5–7]. The clinical validity and utility of basic NIPS in common trisomies had been well-established and their prenatal screening had been recommended by the American College of Obstetricians and Gynecologist (ACOG) and the American College of Medical Genetics and Genomics ACMG since 2015 [8,9]. Providing with a larger volume of sequence data and an advanced bioinformatic analysis pipeline, expanded NIPS (eNIPS) is able to screen for CNV, SCA, and OAA [6,10,11]. Varied performances had been reported for eNIPS so far [7,12], and its clinical validity has yet to be verified by larger cohort studies. We evaluated the eNIPS results based on a cohort of 34,620 singleton pregnancies and recorded eNIPS performance in prenatally screening for genome-wide chromosomal abnormalities. Our data is also clinical significant for genetic counseling the eNIPS reports.

Material and method

Supervision of the study

This clinical research was approved by the Ethics Committee of the Maternal and Child Health Hospital of Guangxi Zhuang Autonomous Region. Between October 2017 and March 2019, 34,620 women with singleton pregnancy from the general population who opted or referred for basic NIPS, in the Maternal and Child Health Hospital of Guangxi Zhuang Autonomous Region, were consecutively recruited for eNIPS. The written consent and pretest genetic counseling were offered to the pregnant women, covering the nature of the test, targets, limitations of the test, and options for follow-up. 5 ml peripheral blood was collected in a cfDNA storage tube (CoWin Bioscience, China) from each pregnant woman. Wet lab procedures were performed according to assay NIFTY (BGI, China) [13]. Pregnant women with positive results went through a posttest counseling that included a recommendation for invasive prenatal diagnostic testing. Invasive prenatal diagnostic outcomes were collected afterwards. Pregnant women with negative results were also counseled and monitored for routine pregnancy checkups. An after-birth follow-up applied for all participants with a negative screening result.

Confirmation of the eNIPS positive cases

Pregnancies with eNIPS positive result were offered invasive prenatal diagnostic testing. Whole chromosomal aneuploidy positive cases were confirmed by SNP array (HumanCytoSNAP-12, Illumina) analysis or karyotyping. CNV positive cases were verified by SNP array or CNV-Seq. Following the guideline of National Health Commission of China, NIPS was offered for women after 12 weeks of gestation, and the confirmatory prenatal diagnosis was mostly accomplished through amniocentesis.

Bioinformatics analysis

0.1X genome-wide sequence data generated by a polymerase chain reaction (PCR) amplification were used for bioinformatics analysis. K-mer (K denote the sequencing reads length) coverage on each chromosome was introduced to remove the GC bias; K-mer coverage also applied on Y chromosome for fetal gender and SCA determination [13].

Statistical analysis

Chi-square test was performed to estimate the statistical significance between two categorical variables. A p-value less than .05 was considered significant.

Results

Participants’ demographics

The maternal age of the participants ranged from 13 to 54 years of age (mean 31.5 years), and the fetal gestational age ranged from 12 to 33 weeks (mean 17.1 weeks) (Supplementary Table 1). There were 11,358 (32.81%) pregnant women with advanced maternal age (≥35 years old). The majority of our participants (72.31%) had a Body Mass Index (BMI) value which indicated normal. The fetus we screened in this study were mostly conceived without Assisted Reproductive Technology (ART) (64.6%). 5,852 (16.90%) of our consecutive participants had abnormal results in maternal serum biomarker screening, and 550 (1.59%) had abnormal fetal ultrasound results, including soft markers. The participants’ fetal DNA fraction ranged from 3.48% to 50.19% with a mean of 9.94%. 270 (0.78%) participants required blood sampling for second time in order to achieve informative results.

Screening for common trisomies and SCAs

In a cohort of 34,620 singleton pregnancy, 461 (1.33%) had positive screening results by eNIPS (Figure 1). There were 209 positive cases for common trisomies: T21 (n = 121), T18 (n = 44), and T13 (n = 44) (Table1). 108 of T21, 31 of T18 and 35 of T13 were subjected for diagnostic confirmation, as results, 99, 16, and 9 cases were confirmed for T21, T18, and T13, respectively (mosaicism included). Therefore the calculated PPV was 91.67% for T21, 51.61% for T18, and 23.68% for T13. Meanwhile there were 124 positive cases for SCA, including 45, X (n = 54), 47, XXX (n = 24), 47, XXY (n = 24), 47, XYY (n = 5), and unclassified other SCA (n = 17) (Table1). There were 9 confirmed true 45,X and 31 unconfirmed; 8 confirmed true 47,XXX and 7 unconfirmed; 14 confirmed true 47,XXY and 2 unconfirmed; and all five 47,XYY positive cases were confirmed. 15 other SCA positive cases had normal karyotype, and only one with 47, XXY. The calculated PPV was 22.50% for 45, X; 53.33% for 47, XXX; 87.50% for 47, XXY; 100% for 47, XYY; and 6.25% for other SCA.

Figure 1. The flowchart of genome-wide expanded noninvasive prenatal screening (eNIPS) results and clinical outcomes. T21: Trisomy 12; T18: Trisomy 18; T13: Trisomy 13; SCA: sex chromosome abnormality; OAA: other autosomal aneuploidy; CNV: copy number variant.

Screening for OAAs and CNVs

In this study we reported 71 positive OAA cases including 15 different types of trisomy (n = 67) and one type of monosomy (n = 4) (Table 1). 55 OAA positive cases went through confirmatory testing. Three out of the 55 OAA cases confirmed to be mosaic trisomy: 47,XN,+2[4]/46.XN[48]; 47,XN,+7[4]/46,XN[46]; and 47,XN,+22[4]/46,XN[36]. The rest OAA positive cases had normal karyotypes. Thus, the combined PPV for OAA was 5.66%. From these positive calls of other aneuploidies, the top 4 frequently observed other aneuploidies were T7 (n = 15), T8 (n = 9), T16 (n = 9), and T22 (n = 8); their PPVs were 9.09%, 0.00%, 0.00% and 25.00%, respectively. 41 out of 57 positive cases for CNV ≥5 Mb were subjected for diagnostic confirmation. 21 had consistent results with eNIPS, whereas 20 were discordant. Therefore, the PPV for CNVs ≥ 5 Mb was 51%. Positive calls for CNVs were frequently observed in chromosome 5 (n = 10), chromosome 4 (n = 8), chromosome 7 (n = 5) and chromosome 2 (n = 5); their PPVs were 25.00%, 66.67%, 100%, and 100%, respectively.

Table 1. Performance characteristic of expanded noninvasive prenatal screening (eNIPS) for fetal chromosomal abnormalities screening in 34,620 singleton pregnancies.

Fetal chromosomal abnormalities	Positive	PR (%)	Concordant	Discordant	Unknown	PPV (%)
Traditional trisomies	209	0.60	124	53	32	70.06
Trisomy 21	121	0.35	99	9	13	91.67
Trisomy 18	44	0.13	16	15	13	51.61
Trisomy 13	44	0.13	9	29	6	23.68
SCAs	124	0.36	37	55	32	40.22
45,X	54	0.16	9	31	14	22.50
47,XXX	24	0.07	8	7	9	53.33
47,XXY	24	0.07	14	2	8	87.50
47,XYY	5	0.01	5	0	0	100.00
Others	17	0.05	1	15	1	6.25
OAAs	71	0.21	3	52	16	5.45
Trisomy 2	1	0.00	1	0	0	100.00
Trisomy 3	5	0.01	0	3	2	0.00
Trisomy 4	2	0.01	0	1	1	0.00
Trisomy 5	3	0.01	0	2	1	0.00
Trisomy 7	15	0.04	1	10	4	9.09
Trisomy 8	9	0.03	0	7	2	0.00
Trisomy 9	1	0.00	0	1	0	0.00
Trisomy 10	3	0.01	0	3	0	0.00
Trisomy 11	3	0.01	0	2	1	0.00
Monosomy 14	4	0.01	0	3	1	0.00
Trisomy 14	2	0.01	0	2	0	0.00
Trisomy 15	3	0.01	0	3	0	0.00
Trisomy 16	9	0.03	0	9	0	0.00
Trisomy 17	1	0.00	0	1	0	0.00
Trisomy 20	2	0.01	0	2	0	0.00
Trisomy 22	8	0.02	1	3	4	25.00
CNVs(≥5 Mb)	57	0.16	21	20	16	51.22
Chromosome 1	4	0.01	0	4	0	0.00
Chromosome 2	5	0.01	4	0	1	100.00
Chromosome 3	3	0.01	1	0	2	100.00
Chromosome 4	8	0.02	4	2	2	66.67
Chromosome 5	10	0.03	2	6	2	25.00
Chromosome 6	1	0.00	1	0	0	100.00
Chromosome 7	5	0.01	3	0	2	100.00
Chromosome 8	2	0.01	0	2	0	0.00
Chromosome 9	2	0.01	0	0	2	N/A
Chromosome 10	3	0.01	0	2	1	0.00
Chromosome 11	2	0.01	1	0	1	100.00
Chromosome 12	1	0.00	0	1	0	0.00
Chromosome 15	3	0.01	2	1	0	66.67
Chromosome 16	1	0.00	1	0	0	100.00
Chromosome 18	4	0.01	2	1	1	66.67
Chromosome 20	2	0.01	0	1	1	0.00
Chromosome 22	1	0.00	0	0	1	N/A

All verified other autosomal aneuploidies (OAA) cases were mosaicism, their karyotypes were: 47,XN,+2[4]/46.XN[48]; 47,XN,+7[4]/46,XN[46]; and 47,XN,+22[4]/46,XN[36]. PR: positive rate; PPV: positive predictive value; SCA: sex chromosome abnormality; CNV: copy number variant.

Maternal age and positive rate

Five groups of different maternal ages were assessed for the association of eNIPS positive rate and maternal age (Supplementary Table 2). The majority of our recruitment were adult between18 and 40 years old (96.20%), fair equally distributed between 18–29 yrs (34.86%), 30–34 yrs (31.71%), and 35–40 yrs (29.63%) groups. Using chi-square test, we found statistical significance in positive rate for T21 in the 35–40 yrs and 41–54 yrs group versus other age groups. Statistical significance was also found in positive rate for T18/13 in the 41–54 yrs group versus other age groups. Positive rate for CNVs in the adolescent 13–17 yrs group was significantly higher than other ages (Supplementary Table 2).

Sizes and origins of CNV positive cases

We grouped CNV positive cases based on CNV sizes: group 1 (5–10 Mb), group 2 (10–20 Mb) and group 3 (>20 Mb) (Supplementary Table 3). For group 1, 92.31% (24/26) of the CNVs indicated maternal signal, i.e. signals extremely far deviated from the baseline and most likely due to maternal chromosome abnormality. Group 1 yielded a PPV of 60.00% (12/20). For group 2, 12.50% (2/16) CNVs involved maternal signal and its PPV was 41.67% (5/12). No maternal related CNV was detected in group 3, and the PPV for group 3 was 44.44% (4/9). A total of 26 maternal-signal related CNVs were detected, the PPV was 57.89 (11/19). 31 CNVs were called with fetal-signal only, the PPV was 45.45% (10/22).

Discussion

Noninvasive prenatal screening for fetal T21/T18/T13 using cfDNA had been well-accepted worldwide. It remains controversial if this approach can be expanded to detect other chromosomal abnormalities [14]. Several pilot studies had attempted for microdeletions, other aneuploidy, and genome-wide CNVs [7,12,15,16], but few studies dealt with large sample sizes. Recent studies reported 19–50% of PPVs for CNVs [7,15]. In our study, from a cohort of 34,620 singleton pregnancies, we reported a PPV of 51.22% for CNV larger than 5 Mb, which indicated a consistently promising performance of eNIPS in screening for genome-wide CNVs. The most recent guidelines had recognized cell-free DNA screening as ‘the most sensitive and specific screening test for the common fetal aneuploidies [17], mostly because of consistent high PPVs, which our platform had generated comparable and even better numbers such as 91.67% for T21, 51.61% for T18, and 23.68% for T13 [15,17]. The reason that the current guidelines have not yet recommended cell-free DNA screening for SCA, microdeletion, or genome-wide CNVs is mainly due to lack of clinical validations and uncertain performance characteristics (e.g. detection rate, false-positive rate) [8,17]. Our study provided some important data in this regard. Certainly eNIPS would increase demands for confirmatory tests, both for numerical chromosomal abnormalities and CNVs. The compliancy related with the increased cost and perceived risk of invasive sampling would be affected. Yet a reasonably high PPV as we reported here would be the most critical factor to justify the expanded screening. From our follow-up data so far, we have not received any report on false negative (FN) cases for the common trisomies. There was one case classified as low risk but turned out carrying a pathogenic 5.22 Mb 13q33.3-34 deletion. This FN case was identified due to abnormal amniotic fluid by routine ultrasound checkup after eNIPS, and this pregnancy was subsequently terminated. Unfortunately, we did not get any placenta biopsy to further investigate this case. It is possible that the chromosomal abnormalities maybe confined to fetus, not in placenta in this case, as such scenarios had been previously indicated [7,18,19]. Our follow-up data may not be able to reveal all the FN CNV cases; therefore, the upper limit of our eNIPS detection rate for CNV (larger than 5 Mb) was 95.45%. On the other hand, we reported a PPV of 5.45% for other aneuploidies, and the most frequent positive cases coming from T7/T8/T16/T22 (Table 1). Many false positive cases of other aneuploidies had been frequently encountered in our study as well. The origin of cffDNA is mainly derived from the cytotrophoblast but not the fetus. The scenario of confined placental mosaicism (CPM), or abortuses with growth failure [20–22] could both contribute to the false positive cases. We also noted that CNVs most frequently observed by eNIPS were on Chr5/Chr4/Chr2/Chr7 (Table 1). This may be due to high incidence of recurrent structural abnormalities associated with Cri-du-chat on Chr5 and Wolf-Hirschhorn on Chr4 or CNV polymorphisms on these chromosomes [23,24]. More data is needed to better understand the phenomena.

NIPS was once recommended only for advance maternal age pregnant women due to a high prevalence of T21 in this cohort [25]. As the data generated from larger cohorts of non-advanced age groups also demonstrated high detection rates, NIPS had now been recommended for common trisomy screening for all pregnant women [6,17]. Importantly to note, the risk of fetal subchromosomal abnormalities is independent of maternal age, and are more frequent than T21 among women under the age of 36 [26]. Our data revealed that the positive rates for CNV remained steady in our four adult maternal age groups (18–54-year-old) and the high positive rates of T21 and T18/T13 in advanced maternal age groups were confirmed (Supplementary Table 2). But unlike the prevalence of fetal T21 which sharply increases after a maternal age of 35, the risks for fetal T18 and T13 rapidly increased at approximately 40 years of maternal age [27,28]. In addition, the CNV detection rate was significantly higher in the adolescent pregnancies, whether or not an increased risk of fetal central nervous system anomalies and preterm birth in teenage pregnancies [29,30] is associated with this observation requires further investigation.

As pointed out by the 2017 ACMG guidelines, if sufficient and proper pretest counseling was provided, target expanding NIPS can be offered to patients in need [31]. Genetic counselors should explain the advantages, limitations and performance characteristics in simple language before offering eNIPS. For posttest counseling, approaches for common aneuploidies NIPS reports had been well established [17]. As for OAA positive reports, before mentioning the confirmatory test, we suggest counselors provide patients with the correspondent PPV and also take patient’s latest sonographic test result into considerations. A non-mosaic true fetal OAA usually cannot make it into the second trimester without noticeable clinical or sonographic abnormalities. The PPV for OAA in this study with the NIFTY platform was 5.45%. This value is too low to justify using the NIFTY platform as a routine screening method for OAA. Other platforms generated PPVs ranged from 9.09% to 28.6% for rare trisomies [7,15]. The methodology using cell-free DNA and high throughput sequencing was in general less efficient in screening fetal OAAs. Therefore we do not suggest routinely reporting OAA in eNIPS at this point. In this study, screening of OAA generated a positive rate of 0.21%. We did not receive any miscarriage reports related to the invasive diagnostic procedures. So far, OAA prenatal screening is mainly relied on ultrasonography which is highly dependent on the sonographer personal experience and judgment. Except for eNIPS, there is no other robotic screening test emerged for rare chromosomal disorders. Future technological advancement may improve the eNIPS efficacy of OAA screening to be comparable to those of SCA or CNV, we may consider it as a routine screening object then. Although screening of OAA generates additional eNIPS positive cases need to be diagnosed, invasive procedures are safer than we used to think [32,33]. For SCA positive report, the PPV increased as more Y chromosomes involved (Table 1). For CNV positive report, CNVs of 5–10 Mb in size were more reliable but most likely due to maternal chromosome effect (Supplementary Table 3). To confirm such maternal signal involved CNVs reports, microarray analysis for both fetal and maternal samples was recommended. The maternal and fetal CNVs were not always identical (Supplementary Table 4). For negative eNIPS reports, always inform the expectant parents about the limitations of this screening test, and encourage routine sonographic examination for the duration of their pregnancy.

In conclusion, this study involved a large prospective group of pregnant women from consecutively clinical visits of mixed characteristics. Our data demonstrate promising clinical utility of genome-wide cell-free DNA profiling not only for common trisomies but also for SCAs and CNVs larger than 5 Mb. This study mainly report PPVs due to a lack of follow-up for fetus’ karyotypes from participants with negative eNIPS results. Further studies in larger cohorts with different clinical characters are required to fully evaluate other essential parameters of eNIPS and delineate corresponding PPVs for different type of genomic imbalances. Those information will help counselor to provide a nuanced and informative counseling session. With the advancement of methodology and accumulation of clinical experience, the combination of genome-wide eNIPS and periodically ultrasound tests are believed to serve as a practical option for routine screening of fetal chromosomal abnormalities.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This study was supported by the Guangxi Zhuang Autonomous Region Health Commission [project number Z20190827] and by Guangxi Zhuang Autonomous Region Clinical Research Center for Fetal Diseases.