Clinical Perspectives

Peter Benn, DSc

NIPT: Post-test risk estimate

Two online calculators are available to estimate individual patient risk following a positive NIPT result (1,2). These calculators compute a risk value by combining and the patient’s prior risk (based on maternal age or other screening) with published estimates of detection rates and false positive rates derived from clinical trials. This result is described as a “positive predictive value (PPV)” because the calculation is essentially the same as PPV except that the patient’s prior risk is substituted for the population prevalence.  It might be better to describe these calculated results as a “patient-specific risk estimates” to avoid confusion with PPV that is based on the overall assessment of the test’s performance in whole populations. The Perinatal Quality Foundation (PQF) (1) calculator uses meta-analysis data from a large number of trials using different study groups and laboratory methods (2). The UNC (3) calculator separately considers the performance of the four major US laboratory providers of NIPT.  
There are a number of important factors that users should consider before using these calculators:

  1. The trials that established NIPT performance were based on selected cases. There was heterogeneity in trial entry criteria and exclusions; performance is different for the various NIPT laboratories, and test protocols have evolved since the trials were conducted.  Most laboratory providers have not extensively verified performance in actual clinical practice. 
  2. Small differences in false-positive rates can have a large effect on the estimates of the post-test risk estimate.  For some of the calculations, the confidence interval for the false-positive rate is sufficiently large that there will be a wide range for the post-test risk.
  3. The calculations assume that prior risk is independent of the determination that the NIPT result is positive.  This is not necessarily true.  For example, maternal age specific risk has already been incorporated into the determination that a case is test positive or test negative for some NIPT algorithms.  Moreover, some causes of false-positive results (e.g., confined placental mosaicism with the abnormality arising from meiotic error, maternal cancer, and maternal somatic cell sex chromosome aneuploidy) may all be expected to be more frequent in older women.
  4. Some laboratories provide a "risk score” on their report.  This is can be considered an estimate of the probability that aneuploidy was present in the placenta, not the fetus, and there is data to show that a very high risk score is more likely to be a true-positive compared to a lower positive score.  As currently structured, the risk score cannot be used in the calculators.
  5. For the UNC calculator (2), at the time of writing this article, there is an inconsistency in the use of truncation of detection rates and false-positive rates, and this artificially creates the impression of higher post-test risk estimates for one laboratory provider relative to others.

1. Perinatal Quality Foundation.
2. Gil et al. Ultrasound Obstet Gynecol. 2015;45:249-266. 3. Grace et al., Am J Obstet Gynecol. 2015;213:30-32


NIPT: Use of microarray for quantification

Stokowski et al., report on the performance of a counting based NIPT using a microarray to measure the number of DNA fragments present.  The detection rates were: trisomy 21 107/108 (99.1%), trisomy 18 29/30 (96.1%), trisomy 13 12/12 (100%) and the false-positive rate for all three trisomies was 0/641 (0%) for all three trisomies.  The no-result rate due to insufficient DNA was 8/799 (1%).  The results were consistent with data using sequencing for the quantification. 

Stokowski et al.  Prenat Diagn. 2015. Eprint ahead of publication.

NIPT: Maternal cancer

Osborne et al., (1) reported in 2013 that maternal cancer could be a reason for false-positive results using counting-based NIPT. Several recent reports have now further documented this finding (2-4).  In the largest study, Bianchi et al., (4) identified 10 cases of maternal cancer in 3,757 women with positive NIPT findings.  Under the crude assumption that perhaps 50% of the cases with positive findings are false-positives, this translates to over 1 in 200 false-positives potentially explainable by maternal cancer. In cases where NIPT is positive for more than one aneuploidy, 7/39 (16%) were attributable to maternal cancer.  
Although many maternal cancer cases could be identified through imbalances involving multiple region-specific gains and losses, some malignancies have a relatively simple abnormal karyotype (such as trisomy 21 as a sole abnormality in acute myeloid leukemia and myelodysplasia). Apparent inconsistencies in the “fetal” fraction across regions or analysis of SNPs that differ between maternal and fetal genotypes could also be used to distinguish between fetal aneuploidy and maternal cancer. However, these approaches would each involve analyses beyond that usually required for fetal aneuploidy assessment. Diana Bianchi suggests that patients are counseled and consent prior NIPT if they wish receive information that may reflect maternal as well as fetal health (5). These considerations would not be applicable to a NIPT format that only returns information related to the fetus (6).   

1. Osborne et al., Prenat Diagn. 2013;33:609-611
2. Vandenberge et al., Lancet Hematol. 2015;2:e55-e65.
3. Amant et al. JAMA Oncol. 2015;1:814-819.
4. Bianchi et al.  JAMA 2015;314:162-169.
5. Bianchi.  Nature 2015;522:29-30.6. 6.     Gross et al. Nature 2015;523:290.

NIPT: Economics

Several new studies have revisited the economics of NIPT for the full US pregnancy population. Fairbrother et al., (1) modeled the expected economic benefits of NIPT for t21, t18 and t13. After allowing for the additional cases of trisomy detected as result of the use of NIPT, they estimated that, if the NIPT were offered to the general pregnancy population, the cost would need to be $665, or less, for it to be advantageous over the first trimester Combined test.  Walker et al., (2) considered NIPT cost for the detection of trisomy 21, 18 and 13 from three different perspectives, societal (includes cost of testing, all direct and indirect lifetime costs of affected individuals), governmental (testing, medical and educational costs), and payer (cost of testing only).  They concluded that offering NIPT to all women would be advantageous from the societal perspective, but contingent NIPT would be preferred by government and payers.  Benn et al., (3) also included monosomy-X in the analysis and considered the costs associated with second trimester follow-up screening tests (sequential quadruple testing, ultrasound) for women receiving the Combined test.  For NIPT to be applied to the general pregnancy population, the cost would need to be $744 or less.  
An entirely different conclusion is drawn by Kaimal et al. (4) who reported that cf-DNA as a primary screening approach only becomes the optimal screening strategy at maternal age 40 and older, and that at younger ages multiple marker screening (MMS, first and second trimester serum markers and nuchal translucency measurement) is preferable.  This study attempted to consider all chromosome abnormalities present including variants of uncertain significance (VOUS) identified through chromosome microarrays. Lifetime costs associated with affected liveborns were not included but maternal quality-adjusted life-years (QALYs) were considered.  A cost of $1,796 was used for NIPT and some other costs, rates and components differed markedly from the other studies. The analysis did not appear to fully consider the additional costs (follow-up ultrasound, counseling, array studies on parents) or consequences of VOUS detection.  Based on the rates and assumptions used by the authors, for every 100,000 pregnancies screened by MMS, there would be 88 pregnancy terminations where there would be fetal VOUS, but no intellectual disabilities would occur if the pregnancies continued. There would, therefore, be far more common than procedure-related losses.        

1. Fairbrother et al.  Mat Fetal and Neonat Med. 2015; [Epub ahead of print]
2. Walker et al.  PLOS One 2015; 10:e0131402.
3. Benn et al.  PLOS One 2015; 2015;10:e0132313.
4. Kaimal et al. Obstet Gynecol. 2015;126:737-746.

NIPT: Microdeletions

NIPT is being extended to screening for selected microdeletion syndromes.  Helgeson et al.,(1) reported results for microdeletions and “subchromosomal events” in 175,393 women using a DNA fragment counting-based approach. For 22q11.2 deletions, they found 32 (0.02%) positive results including 31 true or suspected true positive, 0 false-positive, and 1 unknown outcome. Of the 32 cases, 20 (1/8,770) were maternal deletions, and an additional 12 (1/14,616) were fetal deletions. Based on expected prevalence (2,3), only a proportion of the deletions were detected. For 1p36, 15q, and 5p deletions, they found 20 (0.01%) positive tests including 16 true- or suspected true-positive, 3 false-positive, and 1 unknown outcome. The 20 cases included 4 (1/43,845) maternal deletions and 13 (1/13,492) fetal deletions.
Using SNP-based NIPT, Gross et al., (3) report initial results of screening for the 22q11.2 microdeletion. In 21,948 samples referred for testing, there were 95 (0.5%) positive results that included 11 true-positive and 50 false-positive (PPV 18%). There were also two maternal deletions identified. Ultrasound anomalies were present for 81.8% of true-positive and 18.0% of false-positive cases. Overall incidence of all 22q11.2 fetal deletions in the referral population was estimated to be at least 1/946.  Reflex retesting of positive cases at a high depth of sequencing could raise the PPV from 18% to 42%.
Expanding NIPT to include microdeletion syndromes was the subject of a debate at the Washington ISPD meeting.  The arguments presented will be summarized in an article to be published in Prenatal Diagnosis (4).

1. Helgeson et al. Prenat Diagn. 2015 Epub ahead of print
2. Grati et al., Prenat Diagn. 2015;35:801-9
3. Gross et al. Ultrasound Obstet Gynecol. 2015 Epub ahead of print
4. Rose et al.  Prenat Diagn. In Press.

Q fever and pregnancy outcomes

Jeroen Pennings, PhD

From 2007 to 2010, a community outbreak of Q fever occurred in the Netherlands. In total, there were over 4,000 notified human cases, mainly in the south of the country in areas with intensive dairy goat farming. Current literature data provide no clear answer whether a Q fever  infection during pregnancy leads to an increased risk for adverse outcomes. Marit de Lange and colleagues at the National Institute for Public Health and the Environment (Bilthoven, the Netherlands) analyzed national registry data for postal-code areas with two or more Q fever notifications and areas not affected by any Q fever notification. Data on adverse pregnancy outcomes (preterm delivery, small for gestational age, perinatal mortality) were compared between affected and unaffected areas for data collected before (2003 - 2004) and during the epidemic. This allowed adjusting for maternal characteristics (age, ethnic background, smoking behavior, SES) as well as area characteristics (urbanization degree and farm animal densities). Of the adverse outcomes examined, the authors only found a weak significant association between residing in a Q fever affected area and small for gestational age (adjusted OR 1.06). However, there was no clear dose-response relation with increasing Q fever notifications, and the population-attributable fraction was low due to stronger factors such as heavy smoking. 

Considering these results, and the uncertainties about the efficacy and the adverse effects of antibiotic treatment, the authors conclude that mass screening of pregnant women for early detection of infection is not justified. Instead, a case-by-case approach of pregnant women with acute Q fever is recommended. (De Lange MM, et al, BMJ Open. 2015 Apr 10;5(4):e006821).

Pre-eclampsia and aspirin treatment

Jeroen Pennings, PhD

Felicity Park from the University of Sydney examined the effect of screening and treatment with low-dose aspirin on the rate of pre-eclampsia (PE) leading to delivery before 34 or 37 weeks gestation. In a first – observational – cohort, 3,066 women were screened to determine whether algorithms developed to screen for pre-eclampsia at 11-13+6 weeks gestation could be applied. In a second – interventional – cohort, 2,717 women were screened and women with a high (> 2%) risk were offered aspirin (150 mg/day) prophylaxis. There were 12 cases (0.4%) of early-onset (< 34 weeks) PE in the observational cohort and 1 (0.04%) in the interventional cohort. For all preterm PE (<37 weeks) the number of cases were 25 (0.83%) and 10 (0.37%), respectively. The authors conclude that a strategy of first trimester screening coupled with prescription of aspirin to the high-risk group appears to be effective in reducing the prevalence of early-onset PE. (Park F, et al, Ultrasound Obstet Gynecol. 2015 in press. doi: 10.1002/uog.14819).

A study by Gaea Moore from the University of Colorado determined whether low-dose aspirin (60 mg/ day) was beneficial when initiated <17 weeks gestation. The cohort studied comprised 523 women at high-risk for PE based on clinical characteristics. Aspirin was associated with a lower rate of late (⩾34w) PE (17% vs 24% for placebo), with a 41% reduction in women with chronic hypertension (18% vs 31%). There were no significant differences in early-onset (< 34w) PE or other pregnancy outcomes. The reduced risk for late PE supports early initiation of aspirin in high-risk women. (Moore GS, et al, J Perinatol. 2015 May;35(5):328-331).

To determine the minimum absolute risk for PE at which low dose aspirin prophylaxis is justified, Emily Bartsch from the Western University (London, Ontario, Canada) applied two different approaches. Starting either from the number needed to prevent and the prevalence of PE in aspirin treatment groups, or the number needed to treat and the relative risk reduction by aspirin, the authors conclude that eligible women need not be at high risk for PE to warrant aspirin prophylaxis, but rather at some modestly elevated absolute risk of 6 - 10%. (Bartsch E, et al, PLoS One. 2015 Mar 19;10(3):e0116296).