Discordant performances of non-invasive prenatal testing for foetal trisomy 21 screening in subgroups of pregnancies

Abstract Background Non-invasive prenatal testing (NIPT) has been widely adopted as an approach for foetal aneuploidy screening. This study was to evaluate the performance of NIPT for foetal T21 detection in subgroups of pregnancies and the correlation between Z-score and discordant positive predictive values (PPVs). Methods We retrospectively reviewed the NIPT results among 22361 pregnancies undergoing combined second-trimester screening (cSTS) previously. Sixty-four cases with positive NIPT results for foetal T21 were validated by invasive prenatal diagnosis. Results In pregnancies with cSTS-T21 low-, intermediate-, and high-risk, the PPVs at NIPT were 14.3%, 64.3%, and 86.4%, respectively. Mean Z-scores of positive NIPT cases with cSTS-T21 high- and intermediate-risk were comparable, while were higher than that of cases with pre-test low-risk. Furthermore, PPVs for positive NIPT cases at 3 < Z < 5, 5 ≤ Z < 9, and Z ≥ 9 were 16.7%, 63.2%, and 100.0%, respectively. Conclusions This study suggested that Z-score value of positive cases might be associated with discordant PPVs for T21 screening in subgroups of pregnancies. PLAIN LANGUAGE SUMMARY Non-invasive prenatal testing has been offered as a primary screening option to high-risk or general pregnancy. However, the accuracy of non-invasive prenatal testing in patients with various pre-test risks remained unveiled. The current study revealed that the true positive probability for foetal trisomy 21 screening in pregnancies with prior high-risk was higher than that in pregnancies of intermediate-risk, and both of them were much higher than that of those with pre-test low-risk. The average of Z-score for chromosome 21 of positive non-invasive prenatal testing case in high-risk group was comparable with that of intermediate-risk group, while was higher than that of low-risk group. There was also an upward trend for the true positive probability of foetal trisomy 21 screening with the increase of Z-score. Our study revealed that pre-test risk and Z-score for chromosome 21 were helpful for accurately interpreting the reliability of positive results for foetal trisomy 21.


Introduction
Non-invasive prenatal testing (NIPT) performed with massively parallel sequencing analysis of cell-free foetal DNA (cffDNA) fragment in maternal circulation has been adopted as an approach for foetal aneuploidy screening.Since the introduction into clinical practice, NIPT have demonstrated consistently high sensitivity and specificity in the prenatal screening for foetal trisomy 21 (T21), trisomy 18 (T18), and trisomy 13 (T13) (Chiu et al. 2011, Bianchi et al. 2012, Palomaki et al. 2012).A number of professional societies such as the International Society for Prenatal Diagnosis, Israeli Society of Medical Genetics, and the Royal College of Obstetricians and Gynaecologists have published committee opinions on the clinical use of NIPT for the screening of T21, T18, and T13 (Benn et al. 2013, Michaelson-Cohen et al. 2014, Royal College of Obstetricians & Gynaecologists 2014).In China, the medical guidelines recommend NIPT to be available to pregnant woman at intermediate-risk of serological screening results, while in high-risk pregnancy NIPT should be used with caution (NHFPC 2016).
In clinical practice, NIPT has been offered as a primary screening option to general pregnancy in some countries and regions.The American College of Obstetricians and Gynaecologists, Committee on Genetics, and Society for Maternal-Foetal Medicine have recently issued a joint committee opinion regarding the use of NIPT in general pregnancy (American College of Obstetricians and Gynecologists' Committee on Practice Bulletins-Obstetrics, Committee on Genetics, Society for Maternal-Fetal Medicine 2020).Furthermore, clinical studies further demonstrate that NIPT has similar sensitivity and specificity for aneuploidy detection between patients with pre-test high-or low-risk of foetal aneuploidy (Pergament et al. 2014, Zhang et al. 2015).Notably, positive predictive value (PPV) is more clinically relevant than sensitivity and specificity when assessing a report of NIPT clinically.Some studies show that NIPT has variable PPVs in subgroups pregnancies classified based on the risk derived from the first-trimester maternal serum screening (Gil et al. 2016, Miltoft et al. 2018).However, the defining on the risk of pregnancy and the screening algorithms do not represent the complex and variable screening methods currently used in clinical practice.In Mainland China, combined second-trimester maternal serum screening (cSTS) is the common approach to assess the foetus's risk for foetal T21 (Liu et al. 2011).
In addition, foetal aneuploidy determined by cffDNA testing relies on the Z-score per chromosome, of which the chromosomal ratio was calculated from the number of sequencing reads of the chromosome of interest relative to that of the reference chromosome.A Z-score of three is commonly used as the threshold of aneuploidy risk, although different values may be used in some other testings (Chiu et al. 2008, Fan et al. 2008).For an individual pregnancy, beyond the threshold value of Z-score in cffDNA testing is commonly considered positive for foetal aneuploidy.Several studies have shown that the value of Z-score might be associated with the true likelihood that a positive result indicate true foetal aneuploidy (Sikkema-Raddatz et al. 2016, Tian et al. 2018).Thus, the primary objective of this study was to evaluate the performance of NIPT for the screening of foetal T21 in pregnancies with high-, intermediate-, and low-risks calculated from cSTS, in which the assessment of pre-test risk was based on the maternal age, gestational age, maternal weight, and the maternal serum levels of free beta subunit of human chorionic gonadotropin (fbhCG) and alpha fetoprotein (AFP).The secondary aim was to explore the correlation between Zscore of positive NIPT results and discordant PPVs in subgroups of pregnancies classified based on the pre-test risk of T21 derived from cSTS.

Study design and setting
This was a retrospective analysis of NIPT results in a cohort of 22361 pregnancies who underwent prenatal screening between January 2015 and February 2020 at the Centre for Genetic Medicine of Xuzhou Maternity and Child Health Care Hospital Affiliated to Xuzhou Medical University.The study was approved by the Ethics Committee of the Xuzhou Maternity and Child Health Care Hospital Affiliated to Xuzhou Medical University (no.2020-06).Enrolled patients signed the informed consent.Only those pregnant women with a cSTS risk value of T21 undergoing a contingent screening by NIPT were included in the study.Additionally, patients with a twin pregnancy or vanishing pregnancy were excluded.

Maternal serum screening
Pre-test risk for foetal T21 was estimated on the determination of second-trimester maternal serum levels of fbhCG, and AFP in combination with maternal age, gestational age, and maternal weight.Pregnant women with cSTS-T21 highrisk for foetal T21 were offered the option of invasive diagnostic testing, cffDNA testing, or no further testing.Besides, pregnant women with cSTS-T21 intermediate-or low-risk were offered the option of cffDNA testing or no further testing.

Non-invasive prenatal testing
All pregnant women opting for cffDNA testing signed informed consent after pre-test counselling.A maternal blood sample (5 mL) of each pregnant woman was drawn into an ethylenediaminetetraacetic acid (EDTA) tube or Cell-Free DNA BCT TM tube (Streck) for cffDNA testing.NIPT was carried out in three high-throughput sequencing platforms, including BGISEQ-500, Illumina NextSeq CN500, and Ion proton semiconductor sequencer, according to the methods described previously (Suo et al. 2018, 2020, Gou et al. 2020, Qu et al. 2021).Combined GC-bias correction and Z-score calculation were used to estimate the risk of foetal aneuploidy.The risk estimation of foetal aneuploidy was categorised according to the criteria of a Z-score � 3 or Z-score � −3.

Invasive diagnostic testing and clinical follow-up
Each pregnant woman received post-test counselling after NIPT, and those with a positive result were advised to undergo confirmatory diagnostic testing by amniocentesis and karyotype analysis.Information regarding pregnancy outcome, newborn physical examination, and cytogenetic testing results were collected through hospital records database and telephone interviews.Positive NIPT cases demonstrating the relevant foetal aneuploidy or mosaic aneuploidy in confirmatory karyotype analysis were categorised into true positive cases.Pregnancies with unknown karyotype, termination, miscarriage, stillbirth, or a missing outcome, were excluded from assessing the performance of NIPT.Karyotype analysis or clinical follow-up results were used as the gold standard to calculate the PPV of cffDNA testing in the study cohort.

Statistical analysis
The statistical software package SPSS V21.0 and GraphPad Prism 8.0 were used for data analysis.Descriptive data were presented as mean ± SD, median or minimum and maximum value.Categorical variables were expressed as n (%).PPV was calculated as the number of true positive cases divided by the number of positive NIPT cases, multiplied by 100.Chisquare test was used for the categorical variables' analysis, and t-test was performed for the analysis of continuous variables.95% confidence intervals (95% CI) were calculated using the Clopper-Pearson method.Logistic regression analysis was performed to analyse the relations between the Z-scores and the PPV of positive NIPT results.A P value less than 0.05 was considered as statistical significance.

Study patients
During the study period, a total of 22361 pregnant women with prior risk for foetal T21 derived from cSTS result underwent further cffDNA testing.Of these patients, 89 cases underwent re-sampling owing to failed quality control, assay failure, or insufficient foetal fraction.Among them, 13 pregnancies failed to produce informative results and were classified as a test failure.The non-invasive prenatal testing and invasive prenatal diagnosis for foetal T21 were summarised in a flowchart (Figure 1).Moreover, the demographic characteristics including maternal age and gestational age of the pregnancies were summarised in Table 1.Maternal ages ranged from 17 to 34 years, with a mean age of 27.6 ± 3.6 years.Gestational age at the time of NIPT ranged from 15 þ3 to 24 þ5 weeks, with a mean of 18.9 ± 1.6 weeks (Table 1).

Performance of NIPT in foetal T21 screening
In the cohort studied, positive NIPT results for foetal T21 were found in 76 pregnant women.After post-test counselling, 65 positive NIPT cases of foetal T21 opted for invasive karyotype analysis.Of these, normal karyotype was found in 17 cases, in which all pregnant women chose to continue the pregnancy.The other 48 foetuses were diagnosed with T21, including 1 foetus of mosaic T21.Among these positive NIPT samples, 11 cases declined the invasive diagnostic testing, in whom clinical outcomes were obtained through the review of pregnant outcomes.There were 3 cases choosing elective abortion due to the abnormality in foetal ultrasound structure, while the pregnant outcomes were unknown due to the loss of contact in other 8 positive cases (Table S1).Pregnant women making decision to terminate pregnancy under the conditions with no chromosomal analysis and those cases lost to follow-up were excluded from the performance analysis of NIPT.The overall PPV for the screening of foetal T21 by NIPT was 72.3% (95% CI, 61.1-83.5)(Table 2).
Among these cases, 6933 samples were found with a high-risk for foetal T21 at cSTS, including 305 cases with a risk of > 1:30, 2135 cases with a risk between 1:30 and 1:100, and 4493 cases with a risk between 1:101 and 1:270.Moreover, 10785 pregnancies were found with an intermediate-risk, and 4643 pregnancies were found with low-risk of T21.The positive rate for foetal T21 detection at NIPT in pregnancies with cSTS-T21 risk > 1:30, between 1:30 and 1:100, and between 1:101 and 1:270 were 3.0%, 0.7%, and 0.6%, respectively.Of these positive NIPT samples, 44 pregnant women opted for further invasive diagnostic testing.Consequently, 37 cases were diagnosed with foetal T21, while 6 cases had normal karyotype (Table 2).Moreover, 1 case with no further investigations chose to end the pregnancy due to the abnormal ultrasound result.Other five cases declined further invasive diagnostic testing, and the pregnant outcomes were unknown due to the loss of follow-up (Table S1).
For the positive NIPT pregnancies with pre-test intermediate-risk, 9 cases had confirmatory foetal karyotype of T21, while 5 cases had normal karyotype.Two cases with  NIPT: non-invasive prenatal testing; SD: standard deviation.

Discussion
The present study mainly explored the accuracy of NIPT for the screening of foetal T21 in pregnancies with distinct risks of T21 derived from cSTS.The findings showed that the accuracy of NIPT in foetal T21 detection depending on the condition's risk in the test population.At NIPT, the PPV for foetal T21 detection in pregnancies with pre-test high-risk was higher than that of in pregnancies of intermediate-risk, and both of them were much higher than that of cases with pre-test low-risk values for foetal T21.Furthermore, binary logistic regression analysis demonstrated a significant association for positive NIPT results between Z-score and true positive result for foetal T21.The results demonstrated that the average of Z-score of positive NIPT cases with prior low-risk were much lower than that of pregnancies with high-risk or intermediate-risk.The lower Z-score of positive NIPT results might be related to the lower PPV for foetal T21 screening by NIPT in pregnancies with cSTS-T21 low-risk values.
Evidence shows that cffDNA testing exhibits similar sensitivity and specificity for foetal aneuploidy screening in pregnancies with pre-test low-or high-risk of foetal aneuploidy (Bianchi et al. 2014, Pergament et al. 2014, Zhang et al. 2015, Gil et al. 2016).Moreover, there are many studies supporting the clinical utility of NIPT for foetal aneuploidy screening in general pregnancies (Gil et al. 2013, Gadsbøll et al. 2020).To be noted, sensitivity and specificity are important metrics for NIPT, whereas PPV is more clinically relevant for medical choices making by clinicians and patients in clinical settings.Furthermore, pregnant women were commonly classified into high-, intermediate-, and low-risk populations based on the results of traditional prenatal screening in clinical practice (Maiz et al. 2009, Torres Aguilar et al. 2021).According to China's guidelines for NIPT, NIPT is recommended to pregnant women with intermediate-risk of maternal serum screening results, while in high-risk pregnancy NIPT should be used with caution (NHFPC 2016).Some studies show that the PPV is lower in pregnancies at low risk as compared with those with increased risk of foetal aneuploidy (Gil et al. 2016, Miltoft et al. 2018).However, the number of cases who underwent NIPT and invasive testing was small in some previous study.Moreover, the defining on the risk of pregnancy and the screening algorithms varied in clinical practice, thus, more definite experimental evidence on the PPVs in populations with distinct pre-test characteristics are needed.Currently, we investigated the utility of NIPT in the screening for pregnancies with pre-test high-, intermediate-and lowrisk of foetal T21 derived from cSTS.In the cohort studied, 100% of foetal T21 cases were detected by cffDNA testing, with an overall PPV of 72.3%.The PPVs for T21 detection in pregnant women of subgroups were also compared and analysed.Our results showed that PPVs in pregnancies with pretest high-risk and intermediate-risk of foetal T21 were much higher than that of cases with prior low-risk.These results further demonstrated that PPV for foetal T21 screening by NIPT was dependent on the pregnancy's pre-test risk.
In China, the medical guidelines recommend that the high-risk cut-off value for foetal T21 detection in maternal serum screening is 1:270.In the present cohort undergoing NIPT, 1.4% (305/22361) pregnant women had a cSTS-T21 risk > 1:30, 9.5% (2135/22361) cases had a cSTS-T21 risk between 1:30 and 1:100, and 20.1% (4493/22361) cases had a cSTS-T21 risk value between 1:271 and 1:1000.At NIPT, the positive rate for foetal T21 detection in pregnancies with cSTS-T21 risk > 1:30 was 2.95%, which was much higher than those with cSTS-T21 risk between 1:30 and 1:100 or between 1:101 and 1:270.Nonetheless, there were no significant differences for PPVs among these three groups.Some studies report that NIPT should not be recommended to high risk pregnancy (Beulen et al. 2017, Zhu et al. 2021).In clinical practice, some cSTS high-risk pregnant women still considered it to be an acceptable alternative to invasive diagnostic testing.In cSTS high-risk pregnancies, NIPT might decrease the false positives and the invasive prenatal diagnosis rates.
In pregnancies with cSTS-T21 risk between 1:271 and 1:1000, the positive rate for foetal T21 screening by NIPT was 0.19%, which was similar to that of cases with cSTS-T21 risk between 1:1001 and 1:2000.However, the PPV in pregnancies with cSTS-T21 risk between 1:271 and 1:1000 was 64.3%, which was much higher than that of cases with cSTS-T21 risk between 1:1001 and 1:2000.Thus, in consideration of the high cost of NIPT, the increased economic burdens and high screening efficiency of this technique might be balanced when evaluating the implementation of NIPT in pregnancies with cSTS-T21 risk between 1:1001 and 1:2000.In the current study, the cut-off value of intermediate-risk for foetal T21 detection in cSTS is set as 1:1000.According to the medical guidelines of China, NIPT is recommended to pregnant woman at intermediate-risk of serological screening results.The current study confirmed that NIPT exhibited high efficiency for foetal T21 screening in pregnancies with cSTS-T21 intermediate-risk.However, our study demonstrated that the PPV for foetal T21 screening at NIPT in pregnancies with cSTS-T21 low-risk was much lower than that of cases with cSTS-T21 high-or intermediate-risk.Thus, genetic counselling and estimation on the individual's risk for T21 of NIPT positive results needed to refer the pregnancy's pre-test risk value.However, the defining on high-risk pregnancy and screening algorithms do not represent the complex and variable screening methods currently used in clinical practice.Therefore, large-scale clinical studies investigating the performance of NIPT for foetal T21 detection in population with pre-test risk values are warranted.Z-score analysis of the mapped sequencing reads is used to estimate an individual likelihood of foetal aneuploidy at cffDNA testing.For an individual woman, NIPT result was considered to be positive for aneuploidy if a Z-score was beyond the risk threshold.In clinical practice, Z-score is currently used as an outcome of NIPT to predict the risk of foetal aneuploidy for genetic counselling.Several studies further found that the true likelihood depends not only upon the Zscore of NIPT results but also on specific test characteristics belonging to the test patients (Tian et al. 2018, Sikkema-Raddatz et al. 2016, Morain et al. 2013).In the present study, we explored the association between Z-score of positive NIPT results and accurate judgement of NIPT results for foetal T21 detection.First, we evaluated the Z-score in pregnancies with distinct cSTS-T21 risk values.For positive NIPT results, our study found that the average of Z-score in cases with cSTS-T21 low-risk was much lower than that of pregnancies with prior high-or intermediate-risk values.Secondly, logistic regression analysis revealed a significant association between Z-score and true positive results for foetal T21.Lastly, an upward trend for the PPV of T21 with the increase of Z-score was observed.These findings demonstrated that Z-score was associated with the accuracy of cffDNA testing.The higher the Z-score of chromosome 21, the more likely the result of foetal T21 was correct.Moreover, our findings also showed that the lower Z-score distribution of positive NIPT cases might be related to the lower accuracy of cffDNA testing in populations with T21 low-risk identified by cSTS.
In conclusion, Z-score and pre-test risk were helpful for accurately interpreting the reliability of T21 positive NIPT results and genetic counselling.At NIPT, the PPV for foetal T21 screening in pregnancies with pre-test high-risk was higher than that of pregnancies of intermediate-risk, and both of them were much higher than that of cases with pretest low-risk values for T21.Lower distribution of Z-score for T21 positive NIPT results might be associated with the lower PPV in pregnancies with pre-test low-risk of T21.These findings also highlight that for a pregnancy with T21 positive NIPT result, it would be preferable for clinicians to manage the individual pregnant woman using the risk assessment from cffDNA testing, and modify this by the addition of pretest risk and Z-score.

Table 1 .
Characteristics of pregnancies undergoing NIPT for foetal aneuploidy screening.

Table 2 .
Performance of NIPT in detecting foetal T21 in subgroups of pregnancies classified according to the risk derived from combined second-trimester screening.

Table 3 .
Logistic regression analysis of the association between Z-scores and true positive results at NIPT.