In 1997, Lo et al. found the presence of fetal DNA in maternal plasma (Lo et al., 1997). Then cell-free DNA-based non-invasive prenatal screening (NIPS) has been established to detect fetal aneuploidies (Chiu et al., 2011; Ehrich et al., 2011; Palomaki et al., 2011, 2012; Bianchi et al., 2012; Norton et al., 2012). This technology has been widely used and benefited a lot of women (Chandrasekharan et al., 2014). Previous studies have shown that NIPS has high sensitivity and specificity in detecting common fetal trisomies (trisomies 21, 18, and 13) (Ehrich et al., 2011; Bianchi et al., 2012; Porreco et al., 2014; Gil et al., 2017). A recent meta-analysis estimated that the sensitivity of NIPS was more than 99% for trisomy 21 (T21), and more than 97% for trisomy 18 (T18) and trisomy 13 (T13), with a high specificity of more than 99% for these trisomies, even for low-risk pregnancies (Rezaei et al., 2019). The positive predictive values (PPV) of NIPS was higher for T21 and T18 than for T13, but PPVs vary in different studies (Chiu et al., 2011; Bianchi et al., 2014; Norton et al., 2015; Petersen et al., 2017; van der Meij et al., 2019). Both the ACMG guideline in 2016 and the updated ACOG guideline in 2020 recommend NIPS as the most sensitive and specific screening for T21, T18 and T13 (Gregg et al., 2016; Rose et al., 2020).

However, most studies focused on the NIPS in singleton and high-risk pregnancies, such as pregnancies of advanced maternal age and high-risk standard screening (Chiu et al., 2011; Palomaki et al., 2011, 2012; Norton et al., 2012). The updated ACOG guideline recommends NIPS be offered to all pregnant women regardless of maternal age or risk. Besides, it suggests NIPS be performed in twin pregnancies, and its performance for T21 was encouraging (Rose et al., 2020). However, the performances of NIPS in low-risk and twin pregnancies were not well understood up to now because of the limited positive cases of aneuploidies, especially the limited positive cases of T18 and T13 in low-risk and twin pregnancies. Meanwhile, few studies have compared the low-risk with the high-risk pregnancies on the detective efficiency of NIPS, and a similar problem exists in singleton and twin pregnancies. Thus, the clinical efficacy research of NIPS in low-risk and twin pregnancies is of great significance for the effective application of NIPS at the present stage.

In the current investigation, 66,172 pregnant women were enrolled in a single-centered tertiary university hospital. The first objective was to evaluate the clinical efficiency of NIPS for fetal T21, T18, and T13 in the overall patients. The second objective was to compare the performance between singleton and twin pregnancies, as well as between high- and low-risk pregnancies. This study will contribute to a more comprehensive understanding of the clinical efficiency of NIPS and reduce unnecessary worries for expanding the scope of application.

In this retrospectively single-centered study, we found that NIPS had high sensitivity, specificity and NPV for detecting common fetal trisomies in pregnancies with all risk levels, which was similar to a multicenter prospective study in China (Zhang et al., 2015). The PPV of NIPS was highest for T21 (90.6%), moderate for T18 (55.8%), and relatively low for T13 (11.8%). Our study included singleton and twin pregnancies, as well as pregnancies of all risk levels. We found that the performance of NIPS in singleton pregnancies was similar to that in twin pregnancies. With the higher incidence of T21 and T18 in high-risk pregnancies, the PPV was higher in high-risk than in low-risk pregnancies, but the high sensitivity, specificity and NPV were shown in both high- and low-risk pregnancies. Two false-negative cases of T21 were identified in pregnancies with abnormal ultrasonography, indicating that the pregnancies with abnormal ultrasonography should receive the prenatal diagnosis. The confirmed mosaic T21 indicates that fetal mosaicism is an important cause of false-negative NIPS.

In the current study, the high-risk patients were the main population for NIPS, indicating that NIPS has mainly been applied for or chosen by high-risk pregnant women, and this is inconsistent with the 2020 ACOG guidelines. One reason for this might be that patients who were at high risk would like to choose the NIPS first. The second reason might be the marked increase in pregnant women with advanced maternal age due to the two-child policy in China (Tian et al., 2020). NIPS had a similar performance on sensitivity, specificity, PPV and NPV in different maternal ages in the current observation, and it was consistent with a previous report, but they only analyzed the sensitivity (Dar et al., 2014).

Generally, the incidence of fetal trisomies was higher in the high-risk than in the low-risk pregnancies (Nicolaides et al., 2012). In our study, the incidence of fetal T21 was higher in the high-risk than in the low-risk standard screening, and the same was found for T18 in standard screening and ultrasonography cohorts. The higher incidence might explain the higher PPV in high-risk pregnancies because the PPV is associated with the number of positives cases in the population (Neufeld-Kaiser et al., 2015; Wang et al., 2015). In our study, the PPVs for T21, T18, and T13 were lower than a multicenter prospective study in China (Zhang et al., 2015), and lower than a national study in the Netherlands in which the PPV was 96% for T21, 98% for T18 and 53% for T13, respectively (van der Meij et al., 2019). The possible reason is that the first study was performed begin 2012 when the enrolled cases were mainly high-risk pregnancies. Therefore, they had a higher incidence of aneuploidies (1/155, 1/663, and 1/5,121 for T21, T18, and T13, respectively) and higher PPVs. The second study had a higher follow-up rate than ours, especially for cases with a high-risk NIPS. While in our cohort, some cases that had a high-risk NIPS rejected karyotyping or were lost to follow-up, and these cases might carry fetal trisomies. Considering those, the incidence of trisomies in our cohort may be lower than it was, so as PPV. Even though the higher PPV in high-risk pregnancies, we could not directly conclude that NIPS performs better in high-risk pregnancies because the ability of the test is determined by the precision of the technology and fetal fractions in the maternal blood rather than the incidence of trisomies in the population (Ashoor et al., 2012; Nicolaides et al., 2012). Additionally, the high specificity and NPV in each group of the three cohorts as well as in low-risk pregnancies showed the excellent ability of NIPS in excluding the unaffected pregnancies at any risk levels.

Herein, we proved the high sensitivity of NIPS in singleton and twin pregnancies, which is similar to previous reports (Gil et al., 2015, 2017). Due to the similar performance of NIPS in twin and singleton pregnancies (Fosler et al., 2017; Niemchak et al., 2020), some researchers considered that it could be used as a first-line screening in twin pregnancies (Le Conte et al., 2018; Yu et al., 2019). However, on account of the limited cases of trisomies in twin pregnancies, especially T18 and T13, the performance of NIPS may be unrepresentative and it needs to be further evaluated (Gil et al., 2019).

Two cases of false-negative T21 were identified in pregnancies with abnormal ultrasonography, and it lowed the NPV in this population. These discordant results highlight the irreplaceable status of early ultrasound in prenatal screening strategies (Hui et al., 2017), and NIPS is not a diagnostic test with inevitable false-positive and false-negative results (Allyse and Wick, 2018). More importantly, for those with abnormal ultrasonography, prenatal diagnosis instead of NIPS is essential, which has been issued in a previous study (Beulen et al., 2017), and it has been reported that the rate of pathogenic chromosome abnormalities missed by NIPS in the population of abnormal ultrasonography was 8% (Benachi et al., 2015), which was higher than ours (4%), but we only estimated the common trisomies. Causes of false-negative NIPS include low fetal fraction, which is caused by maternal obesity, multiple gestations causing low fetal fraction per fetus, maternal medical condition or treatment affecting the quality of circulating DNA, certain fetal chromosomal aneuploidies and confined placental mosaicism (Bianchi and Chiu, 2018). In this study, fetal mosaicism was confirmed by the karyotyping, causing the false-negative NIPS of case 1, and a similar finding has been reported in another study (Grati et al., 2014). For case 2, the fetal fraction was 7%, and it was a singleton pregnancy with a BMI of 24.8 and a normal karyotype. Meanwhile, there was no maternal medical condition or treatment during that time. A similar finding has been reported in an obvious study in which placental samples of two false-negative cases of T18 were obtained and both showed low-level confined placental mosaicism of T18 (Zhang et al., 2015). So, confined placental mosaicism may be the cause of false-negative NIPS of case 2.

There are several strengths in this study. Firstly, we assessed the performance of NIPS for common trisomies in a large population with different risk levels. Secondly, we make comparisons of singleton and twin pregnancies and confirmed a similar performance. Thirdly, we make comparisons of low- and high-risk populations in singleton pregnancies, and in which, the analysis was based on a single cohort of maternal age, conventional Down screening, or ultrasonography instead of the combined risk factors (Zhang et al., 2015). This study is the first large-scale retrospective study to assess the performance of NIPS in twin and low-risk pregnancies since the publication of the updated ACOG guideline.

There are also limitations in our study. Firstly, more than 4,400 patients lack information on maternal age, and some did not receive standard and ultrasound screening. So, we could not assess the risk levels of aneuploidy for these patients accurately. Secondly, we did not enroll cases that had no-call NIPS results and cases that had intrauterine fetal demise or selective termination of pregnancy, but they may be high risk for aneuploidy (Norton et al., 2015). From this point of view, our results may be little different from reality.

NIPS has excellent clinical efficiency for common fetal trisomies in the overall patients at different risks, and its performance in singleton pregnancies was similar to that in twin pregnancies. However, NIPS cannot replace the invasive prenatal diagnosis, and fetuses with abnormal ultrasonography should undergo prenatal diagnosis.

The datasets presented in this study can be found in online repositories. The names of the repository/repositories and accession number(s) can be found below: doi: 10.6084/m9.figshare.14407616.

This study was approved by the Ethics Committee of Women’s Hospital, Zhejiang University School of Medicine, and all the patients provided their informed consent.

MD and YaX: conception and design. YaX: statistical analysis and manuscript writing. MD: supervising and manuscript revising. All authors collected the data and approved the final version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.