Non-invasive prenatal testing for fetal aneuploidy: charting the course from clinical validity to clinical utility
In this issue of the Journal, we see further contributions to the rapidly growing body of literature on non-invasive prenatal testing (NIPT) for fetal aneuploidy using sequencing of maternal plasma DNA. Ashoor et al. provide data on a chromosome-selective method for the non-invasive prenatal detection of trisomy 13 and the effect of fetal fraction on test performance. These papers join a wave of clinical validation studies published in the past 18 months; since 2011, there have been at least 11 such publications analyzing the test performance of NIPT for aneuploidy in high-risk women[1, 3-12]. It is now established that trisomy 21 can be detected reliably from 10 weeks' gestation in high-risk singleton pregnancies with superior sensitivity (> 99%) and specificity (false-positive rate < 1%) compared with conventional screening methods (Table 1). Trisomy 18 and 13 are also detectable, but test accuracy is less consistent than that for trisomy 21.
Table 1. Summary of major non-invasive prenatal testing studies since 2011 using sequencing of maternal plasma cell-free DNA
| || || || || || || || || || |
|High-risk singleton pregnancies|
|Palomaki (2011)||Multicenter nested case–control||1971||283||212||—||—||Sn 98.6% Sp 99.8%||—||—|
|Palomaki (2012)||Multicenter nested case–control||1971||283||—||59||12||—||Sn 100% Sp 99.7%||Sn 91.7% Sp 99.03%|
|Sehnert (2011)||Multicenter cross-sectional validation||119||26||13||8||1||Sn 100% Sp 100%||Sn 100% Sp 100%||—|
|Ehrich (2011)||Blinded prospective||449||39||39||—||—||Sn 100% Sp 99.7%||—||—|
|Chiu (2011)||Diagnostic accuracy with prospectively collected and archived serum||753||156||86||—||—||Sn 100% Sp 97.9%||—||—|
|Bianchi (2012)||Blinded prospective multicenter with nested case–control||532||221||89||36||14||Sn 100% Sp 100% ||Sn 97.2% Sp 100%||Sn 78.6% Sp 100%|
|Ashoor (2012)||Nested case–control||397||100||50||50||—||Sn 100% Sp 100%||Sn 98% Sp 100%||—|
|Ashoor (2013)||Case–control with training and validation datasets||2116||21||—||—||10||—||—||Sn 80% Sp 99.95%|
|Sparks (2012)||Nested case–control with training and validation sets||338||88||36||8||—||Sn 100% Sp 99.2% ||Sn 100% Sp 99.2% ||—|
|Chen (2011)||Diagnostic accuracy with prospectively collected and archived serum||392||149||—||37||25||—||Sn 91.9% Sp 98%||Sn 100% Sp 98.9%|
|Norton (2012)||Prospective cohort||3228||119||81||38||—||Sn 100% Sp 99.97%||Sn 97.4% Sp 99.9%||—|
|Unselected first-trimester screening population|
|Nicolaides (2012)||Cohort||1949||10||8||2||—||Sn 100% Sp 99.9%||Sn 66% Sp 99.9% ||—|
|Mixed-risk screening population|
|Dan (2012)||Multicenter in routine clinical setting||11 105 (3000 karyotyped)||180||Cases with karyotype 139||Cases with karyotype 41||—||Sn 100% Sp 99.96%||Sn 100% Sp 99.96%||—|
|Lau (2012)||Case series||12||1||1 discordant||—||—||Sn 100% Sp 100%||—||—|
|Canick (2012)||Twin cases from primary study reported by Palomaki (2011)||27 (25 twins, 2 triplets)||8||5 discordant2 concordant||—||1 discordant||Sn 100% Sp 100%||—||Sn 100% Sp 100%|
|Sehnert (2011)||Twin cases within primary study||5||2||1 discordant1 concordant||—||—||Sn 100% Sp 100%||—||—|
What does this paper add to current knowledge?
Ashoor et al. build on previous work by applying their chromosome-selective technique to trisomy 13, having already published their results on trisomy 21 and 18. One of the early concerns with NIPT was the exclusive focus on trisomy 21. With this publication, data are now available on trisomy 13 using both chromosome-selective and whole-genome massively parallel sequencing (MPS) techniques, completing coverage of the three autosomal trisomies commonly targeted in conventional screening programs.
The chromosome-selective technique developed by this group uses a highly multiplexed assay, called digital analysis of selected regions (DANSR), to improve the mapping efficiency, sequencing cost and throughput of MPS. In this method, locus-specific oligonucleotides bind to cell-free DNA sequences unique to chromosome 13 present in maternal plasma. The hybridized oligonucleotides are then amplified using universal polymerase chain reaction (PCR) primers and the PCR reaction products are sequenced. The results are analyzed in an algorithm with patient data on maternal and gestational age to provide an individualized trisomy 13 risk assessment for each sample.
The major advantage of this technique is the substantial reduction in the depth of sequencing required compared to whole genome sequencing, thus making widespread screening a much more feasible prospect. The main disadvantage of targeted sequencing is that off-target abnormalities will not be detected.
The detection rate of trisomy 13 in this study was 8/10 (80%), reflecting the technical challenges of detecting dosage aberrations of this chromosome. Other clinical validation studies using whole genome MPS detected trisomy 13 at rates of 11/14 (79%) and 11/12 (92%). The relatively poorer performance of NIPT for trisomies 13 and 18 compared with trisomy 21 has been attributed to the relatively low guanine and cytosine content of chromosomes 13 and 18, leading to sequencing bias. Statistical normalization to remove the bias can help improve detection rates. An additional challenge is that these trisomies are much less common than is trisomy 21, making any estimate of test performance less precise. Other speculated mechanisms contributing to the lower detection rate for trisomy 13 include a greater incidence of confined placental mosaicism, which would reduce the amount of excess pregnancy-derived chromosome 13 fragments in maternal blood. Furthermore, trisomy 21 appears to be associated with a 25% increase in fetal fraction (proportion of fetal-derived DNA in total maternal plasma cell-free DNA) compared with euploid pregnancies, while the other trisomies are not, which may provide a technical advantage for testing performance of this chromosome. Overall, the clinical impact of the lower detection rate for trisomy 13 is likely to be small. Fetuses with trisomy 13 have a high rate of major structural anomalies, many of which are detectable at 11–13 weeks, which would prompt invasive testing independently. Trisomy 13 infants are also much less likely to be liveborn due to the high pregnancy loss rate.
There must be some degree of caution when interpreting the results of this two-phase case–control study, however. First, the number of trisomy 13 cases in the validation set (n = 10) is smaller than that in previously published studies reporting test performance for this aneuploidy. Second, the validation samples on which the blinded testing was performed contained trisomy 13 cases at a statistically significantly more advanced gestational age than were the euploid controls (median gestational ages 21 vs 13 weeks). This was associated with a significantly higher median fetal fraction in cases vs controls (14% vs 10%), which may have provided a technical advantage for successful analysis of the trisomic samples that would not be present in a clinical setting.
The non-invasive era of fetal aneuploidy detection is here
NIPT has now moved well and truly beyond exclusive focus on Down syndrome to cover the three most common autosomal trisomies currently targeted in conventional screening programs. Individual studies have also provided data on sex chromosomal abnormalities and test performance in multiple pregnancies[5, 13, 14]. Large-scale data on NIPT in unselected first-trimester and mixed-risk screening populations are now available, providing even more food for thought on the role of NIPT in pregnancy care.
Where does this deluge of publications leave the general obstetric clinician? While these studies establish the high accuracy of NIPT for Down syndrome, it is clear that it is not ready to replace invasive testing. A small but clinically significant false-positive rate necessitates confirmation of abnormal results with an invasive diagnostic test, such as chorionic villus sampling or amniocentesis. The California Technology Assessment Forum (CTAF), the International Society of Prenatal Diagnosis (ISPD) and the United States National Society of Genetic Counselors (NSGC) all state that NIPT is not diagnostic and does not replace the gold standard of fetal karyotyping from conventional samples.
There are currently three commercial providers of NIPT in the USA who have received Clinical Laboratory Improvement Amendments (CLIA) certification and are providing analysis in the form of a laboratory developed test (LDT). Sequenom released its MaterniT21 test in the USA in October 2011, soon followed by MaterniT21 Plus in February 2012 (adding coverage of trisomies 18 and 13), Verinata released the verifi Prenatal Test in March 2012 (trisomies 21, 18 and 13 and, optionally, monosomy X) and Aria Diagnostics the Harmony Prenatal Test in May 2012 (trisomies 21, 18 and 13). Market uptake has been extremely rapid, with Sequenom currently testing at an estimated run rate of more than 70 000 samples per year in the USA. A fourth company, Natera, is also poised to enter the huge USA market, where an estimated 750000 women are at increased risk of aneuploidy annually.
From the relatively unregulated environment of the US, NIPT has now expanded to the UK, Europe and Asia via licensing agreements from the US providers. Recent data on MPS for routine screening in China reveal its widespread use in relatively resource-poor health systems. However, even in countries where licensing has not yet been sought, international shipping of samples means that lack of local availability is not an absolute barrier to testing. It is clear that even with the current high costs (approximately $1000–2000 per test), inconsistent private insurance reimbursements and the absence of formal recommendations from professional bodies, NIPT is here to stay.
Clinical utility vs clinical validity
Clinical validity refers to the ability of a test to predict the risk of an outcome and to categorize patients with different outcomes into separate risk classes. While the clinical validity of NIPT has now been assessed in multiple trials, this is only the beginning of what should be a considered process of integrating this test into existing healthcare systems. The clinical utility of a test refers to its ability to improve health outcomes for patients. Specifically, the test results must change clinical decision-making, and these changes should be beneficial to patients. There are currently no studies on the impact of DNA-based NIPT on patient management or outcome.
Part of the difficulty with integrating NIPT into existing models is the fact that NIPT is currently neither a first tier-screening test nor a diagnostic test. It is currently being used as an ‘advanced’ screening test for women identified as being at high risk for fetal aneuploidy by conventional methods (advanced maternal age, an abnormal serum screen, personal or family history or abnormal ultrasound). Its major perceived benefit is to reduce invasive testing rates via its high negative predictive value (false-positive rates of < 1% in high-risk women). In October 2012, the CTAF concluded that the use of cell-free fetal DNA as a prenatal advanced screening test for fetal trisomy 21 and trisomy 18 in high-risk women met all five CTAF criteria for safety and efficacy and improvement in health outcomes. These five criteria are: 1) the technology must have final approval from the appropriate government regulatory bodies (CLIA); 2) the scientific evidence must permit conclusions concerning the effectiveness of the technology regarding health outcomes; 3) the technology must improve net health outcomes; 4) the technology must be as beneficial as any established alternative(s); and 5) the improvement must be attainable outside the investigational setting. This recommendation for the targeted application of NIPT as an advanced screening test is consistent with earlier statements issued by the NSGC and the ISPD, the only other organizations to have issued official position statements to date[18, 19].
However, the CTAF concluded that NIPT for trisomy 13 did not meet criteria 4 and 5, but this assessment was performed prior to availability of data from Ashoor et al. and was hindered by the small numbers of trisomy 13 cases. The CTAF also recommended against the use of NIPT for primary screening of average- and high-risk women because it did not meet criteria 3–5 for safety, efficacy and improvement in health outcomes. Again, this conclusion was reached without the benefit of more recent data[15, 16].
Will NIPT eventually replace ultrasound and maternal serum testing as a primary screening test for all pregnancies? The recently published UK trial of NIPT in a routinely screened first-trimester population suggests that it performs equally well in these women, with similar detection rates of > 99% and a false-positive rate of 0.1%. This is a significant improvement on the 85–90% detection rate and 5% false-positive rate associated with existing screening strategies[21, 22]. The real limiting factor now is cost ($1000–$2000 per test) and the turnaround time of 8–10 days. If the costs come down, as they are expected to do with improvements in targeted sequencing techniques and improved throughput, there is a very real prospect of incorporating NIPT into routine prenatal screening protocols.
Can we afford NIPT?
Early evidence suggests that NIPT has significant clinical advantages when incorporated into existing practice as an advanced screening test, but the cost benefits remain uncertain. Using data generated by one clinical trial, an economic model found that NIPT as a secondary screening test could result in a 72% reduction in invasive tests and a subsequent 66% reduction in procedure-related miscarriages. Cost savings were estimated at 1%, assuming the cost of NIPT was no more than $1200 per test.
Authors of another clinical trial performed their own economic model of NIPT as a secondary screening test. This analysis concluded that NIPT testing of 100 000 high-risk women, followed by invasive testing in only those with positive results, would detect 2958 cases of trisomy 21, miss 42 cases and avert 480 pregnancy losses for an additional cost of $3.9 million.
There is a consensus that NIPT is currently too expensive for primary screening. Future technical advances may bring costs down to a level at which primary screening could be considered. Furthermore, a significant driver of currents costs is the initial screening test prior to eligibility for NIPT. If this is subtracted from the costs, NIPT may very well become cost effective. This would have profound implications for the current models of prenatal diagnosis, affecting serum screening laboratories, cytogenetic and molecular genetic laboratories and obstetric ultrasound practice.
What then will be the future of the first-trimester nuchal translucency ultrasound? One of the current disadvantages of NIPT is the 8–10 day turnaround time, which could potentially cause a woman to lose the opportunity for nuchal translucency assessment if the NIPT fails for technical reasons (which occurs in 1–5% of samples). It has been suggested that all women could have NIPT at 10 weeks followed by a 12-week ultrasound examination and consultation, allowing them to access combined first-trimester screening if non-invasive testing fails. Although this would maximize screening choices for women, it would almost certainly prove too costly for most healthcare systems. It remains to be seen if other emerging obstetric indications for the first-trimester ultrasound scan may eventually provide adequate economic and health justification for its routine use independent of aneuploidy screening[24-26].
The way forward
The combined commercial and consumer pressure to embrace NIPT has taken many clinicians by surprise. Fortunately, this test has not followed non-invasive prenatal diagnosis for fetal sex and paternity down the road of direct-to-consumer marketing. There appears to be general consensus from current industry providers, clinicians and the public that NIPT testing should remain a physician-ordered test due to the counseling requirements and need for integration into existing pregnancy care.
Health-professional education is an urgent priority. In July 2012 the National Coalition for Health Professional Education in Genetics and the NSGC released an NIPT fact sheet for practitioners. This will go some way to preparing practitioners for encounters with patients requesting NIPT. However, clinicians also need to be provided with recommendations regarding appropriate use of NIPT in their specific healthcare system, particularly given the cost and the intensive marketing of these products by industry.
The role of professional organizations is very important in countries in which there is no central government regulation of new diagnostic methods. In the USA, as an LDT, NIPT is not currently regulated by the Food and Drug Administration. There have been many calls to integrate various stakeholder perspectives into the introduction of NIPT for aneuploidy in the USA. In such a rapidly changing field, it can be difficult for clinicians to reach consensus and make timely statements on the role of new technologies. It is therefore not surprising that most professional colleges, including the American College of Obstetricians and Gynecologists, have not formally addressed this urgent issue.
Key clinical services in industrialized countries should develop national best practice guidelines to ensure that NIPT is used only within agreed clinical pathways and that audit and monitoring processes are established. Ideally, modeling of different integration scenarios with current screening programs, full economic evaluation and studies of clinical impact should be performed before its widespread implementation into routine practice. Only then can we hope to grasp accurately the scope of the clinical and societal impact of NIPT for fetal aneuploidy.