The application of maternal plasma-extracted short cell-free DNA (cfDNA) has been widely accepted for screening common fetal aneuploidies [1, 2]. This approach is known as the noninvasive prenatal testing (NIPT), since the discovery of Y chromosome signals in maternal plasma by Lo et al. in 1997 [3]. By 2020, NIPT had been validated and recommended by professional committees, such as the American College of Obstetricians and Gynecologist (ACOG) and the American College of Medical Genetics and Genomics (ACMG) for general pregnancies in order to screen for prenatal trisomy21/trisomy18/trisomy13 (T21/T18/T13) [4, 5].

The accuracy of NIPT increased as sufficient fetal-derived cfDNA was captured during sequencing. The proportion of fetal-derived cfDNA in all circulating cfDNA from the maternal plasma is called the fetal fraction (FF). In maternal plasma, the fetal-derived cfDNA originates from the placenta, and the level of the FF generally ranges from 3 to 30% [6]. Most of the existing methodologies for NIPT do not analyze samples with a low FF, and the cut-off values are defined by the platform used, because analyzing samples below the threshold can result in increased false negative rates [7, 8]. Depending on the platforms used, the reported low FF from the first sample of maternal blood drawn ranges from 1 to 8% [9]. Known factors that can affect the level of the FF include the gestational age of the fetus, maternal body mass index (BMI), maternal age and the size and functional integrity of the placenta [10, 11]. However, these FF-related factors cannot be manipulated during gestational weeks 9 to 12, which is the optimal time window for NIPT. As the sequencing resolution is elevated, the screening targets for NIPT can extend beyond the common trisomies.

Sub-chromosomal imbalanced copy number variants (CNVs) can be associated with structural and developmental abnormalities in the newborns. Collectively, recurrent CNVs associated with genomic disorders can account for 1 to 2% of congenital abnormalities in newborn babies [12, 13]. CNVs-implicated genomic disorders, such as 1q21.1 microdeletion/microduplication as well as Williams-Beuren, Prader-Willi/Angelman and DiGeorge syndromes, are not lethal at birth but require continuous treatment throughout the postal period [14]. DiGeorge syndrome, also known as the 22q11.2 deletion syndrome, has a prevalence of one in 4000 live newborns. This multisystemic disorder manifests as one of the most common CNV-related syndromes [15]. The sensitivity and specificity of CNV detection by NIPT is correlated with the size of the CNV, the FF and the depth of sequencing. Increasing the sequencing depth of each sample can improve CNV detectability, but this approach is economically unwise, and a feasible option is to achieve a higher level of the FF. Several methods for FF enrichment had been reported, mainly through size selection and DNA repair. DNA repair was reported for the recovery of long cfDNA molecules which are more than 250 bp, and the total increase of the FF achieved was less than 5% [16]. Size selection by magnetic nanoparticles showed robust efficiency in FF enrichment, with an average of approximately 10% being achieved [17, 18]. At the same time, the magnetic beads option was able to cope with large scales as well as an automated DNA library construction operation system.

In this pilot clinical study, we set up a NIPT protocol with FF enrichment using magnetic nanoparticles. Compared to the preliminary method of enriching the FF using magnetic beads, the pipeline we adopted aimed to achieve a more effective enrichment in a fully automated, large-scale DNA extraction and library construction system. We compared the clinical performance of this approach for simultaneous detection of common trisomies and CNVs versus standard NIPT and identified a relationship between the effectiveness of the FF fraction enrichment and BMI.

The current clinical guidelines support the use of NIPT in detecting common fetal trisomies in all pregnant women [4, 5]. The use of fetal cfDNA screening for genome-wide fetal CNVs, pathogenic as well as likely pathogenic, has not as yet been recommended by the formal clinical guidelines. This was partly due to concerns about the lack of sufficient and specific experimental parameters (e.g., sensitivity, specificity and PPV) for the application of NIPT in different types of CNVs. However, in some updates and supplements to the formal guidelines, there have been thorough discussions regarding the use of NIPT beyond the screening of common trisomies. Local experiences in the clinical application of expanded NIPT with respect to CNVs have also been shared [22, 23]. Although expanding the target of NIPT will be likely to increase the rate of invasive diagnosis in pregnant women, for now, this is the only screening method of CNVs and SCAs offered that has the potential of detecting CNV-related syndromes that may not show distinct abnormalities on fetal ultrasound imaging [12, 24].

In this study, we tested the performance of NIPT with an additional step of FF enrichment using magnetic nanoparticles. Through parallel testing with clinical samples, we observed a robust increase in the FF by using this new approach, particularly in pregnant women with high BMIs. In addition, significantly fewer cases of low FF were observed in NIPT with enrichment. Samples that failed enrichment NIPT due to a low FF all had an extremely low fetal fraction when the standard NIPT was used. Such conditions may be related to CNVs in the Y chromosome, as we encountered in this study, or they may be associated with other complications during pregnancy [25]. As is recommended in standard NIPT, no-calls in NIPT with enrichment should be considered for prenatal diagnostic procedures. We observed a general decrease in DNA library concentration after adopting FF enrichment. As limited samples were included in this test for NIPT with enrichment, the failure rate of DNA library construction may be lower when pooling from larger experimental cohorts.

Maternal obesity was associated with failures in NIPT due to a low FF, and some guidelines have previously suggested a specific weight limit [26, 27]. Due to the rapid turnover of adipocytes, high concentrations of cfDNA have been linked to obesity [28]. Thus, there is usually an excess of maternal-origin cfDNA present in the peripheral blood plasma of obese pregnant women. The length of fetal-origin cfDNA typically peaks at 143 bp, while maternal cfDNA peaks at 166 bp [29]. The platform we adopted for FF enrichment further selected DNA fragments below 160 bp using magnetic beads. This size threshold was designed to strike a balance between cost, labor and effectiveness. With the implementation of FF enrichment, a process that effectively removes excess maternal and preserves fetal cfDNA, overweight and obese women had an increased likelihood of obtaining more stable results from NIPT [28, 29]. However, the specific weight and BMI limits for NIPT with enrichment remains to be discovered by future studies with larger and different populations.

NIPT with enrichment was reported with enhanced z-scores and sensitivity [18, 30]. Our study reflected similar outcomes, with more distinguished z-scores for the trisomy positive samples (Supplementary Fig. 1) and a significantly increased number of CNVs calls (Table 3). Even though the enrichment method successfully reduced the low FF rate in NIPT, false positive calls for trisomy and MCA still remained. MCA was found to be associated with maternal tumors [31], and false positive calls for trisomy in NIPT were mainly due to feto-placental mosaicism [32]. In addition, maternal background and placental origin interferences are still technical issues that NIPT cannot readily address.

Based on this pilot study, NIPT with enrichment showed satisfactory performance in detecting common trisomies and CNVs. NIPT with enrichment did not outperform the standard NIPT in terms of the abilities to detect the common trisomies (T21/T18/T13). However, due to the higher FF obtained, the z-scores of the positive cases were higher in the enriched NIPT making them more easily distinguishable. For CNVs, we reported a PPV of 75% for NIPT with enrichment and the detection of more concordant and smaller CNVs: fragments with size of 3 Mb and above in general, and 1 Mb and above for the 22q11.2 deletion. With the FF enrichment, we were able to screen out three cases of 22q11.2 deletion syndrome, one case of 16p11.2-p12.2 duplication syndrome and several other clinically significant pathogenic CNVs. It is noteworthy that the NIPT estimated size for a 2.5 Mb 22q11.2 deletion was 1.72 Mb in the region of chromosome 22q11.21, which suggested there is still a discrepancy between the estimated size of the CNVs using the screening method and the size measured by diagnostic method. This size difference may be due to the bioinformatics analysis of the NIPT platform that we chose. It may also be due to the fact that the NIPT technique was analyzing placental-originated cfDNA [32, 33].

Low percentage of mosaicisms cannot be detected by NIPT even with the inclusion of FF enrichment. A 20–40% mosaicism of a 3.7 Mb duplication at 22q11.1q11.21 (as was seen in sample no. 1009 in Table 4) showed negative results in NIPT with and without enrichment. The PPV for CNVs in our selected samples was 75%, which compared favorably to other studies, which ranged from 19 to 47% [12, 34]. The introduction of CNV screening into NIPT would likely result in additional invasive procedures and genetic counseling. According to this study, CNV screening resulted in a minimum false positive rate of 25%, assuming there were no false negatives in our screened negative cohort. We suggest that this enhanced version of NIPT should only be recommended when comprehensive counseling is available. Clinical validation of this enrichment approach should be further carried out with larger and unselected samples. Our recruitment was limited in number, and therefore, does not represent the general population.

Another study by Liang et al. [35] also performed enrichment of the FF during prenatal screening. However, their study differed from this one as it focused mainly on aneuploidies whereas we were also interested in CNVs. In addition, the technique used for FF enrichment was gel electrophoresis, while we employed magnetic nano-beads, which are more compatible with a large-scale DNA extraction and library construction system. In addition, our interest was focused on the clinical performance of FF enrichment, especially in terms of BMI and its effect on the enrichment obtained.

The data from this small-scale pilot study demonstrated the clinical usefulness of NIPT with FF enrichment. The implementation of FF enrichment requires little additional cost and labor, although it is likely to render the results from NIPT more reliable in obese pregnant women and it is capable of detecting fetal CNVs as small as 3 Mb. In a similar way to standard NIPT, restriction for mosaicism and maternal cancer are retained in NIPT with enrichment. Patients with negative results should always follow routine ultrasound tests during their pregnancies. Comprehensive pre- and post-test genetic counseling should be provided for individual cases. This study will increase the awareness of NIPT with FF enrichment and encourage researchers to conduct future studies with larger and different populations. The combination of an advanced version of NIPT and ultrasound may eventually serve as an effective clinical tool to identify a wider range of fetal sub-chromosomal CNVs.

The supporting data of findings in this study are available from the author, YC, upon reasonable request.

YC, JY, XH, SY, YT and JS carried out the experiments, data collection and data analysis. YL and HW made substantial contributions to the conception and design of the project. YC designed and performed the experiments. YC prepared the manuscript. YS provided substantial contributions to the conception of the manuscript and revised it critically for important intellectual content. All authors contributed to editorial changes in the manuscript. All authors have participated sufficiently in the work and agreed to be accountable for all aspects of the work. All the authors discussed the results and approved the manuscript for submission.

This retrospective clinical study was approved by the Ethics Committee of the Maternal and Child Health Hospital of Guangxi Zhuang Autonomous Region, with the approval number of [2019]-2-17. All of our standard NIPT patients gave their written consent for participation in this de-identified research.

We thank Dr. Jingsi Luo from the Maternal and Child Health Hospital of Guangxi Zhuang Autonomous Region for administrative support for the experiments and manuscript production. The authors also thank Dr. Dev Sooranna of Imperial College London for editing the language of this manuscript.

This study was supported by The Guangxi Zhuang Autonomous Region Health Commission [project number Z20190827].

The authors declare no conflict of interest.