Background
Cell-free DNA noninvasive prenatal screening for trisomies 21, 18, and 13 has been rapidly adopted into clinical practice. However, previous studies are limited by a lack of follow-up genetic testing to confirm the outcomes and accurately assess test performance, particularly in women at a low risk for aneuploidy.
Objective
To measure and compare the performance of cell-free DNA screening for trisomies 21, 18, and 13 between women at a low and high risk for aneuploidy in a large, prospective cohort with genetic confirmation of results
Study Design
This was a multicenter prospective observational study at 21 centers in 6 countries. Women who had single-nucleotide-polymorphism-based cell-free DNA screening for trisomies 21, 18, and 13 were enrolled. Genetic confirmation was obtained from prenatal or newborn DNA samples. The test performance and test failure (no-call) rates were assessed for the cohort, and women with low and high previous risks for aneuploidy were compared. An updated cell-free DNA algorithm blinded to the pregnancy outcome was also assessed.
Results
A total of 20,194 women were enrolled at a median gestational age of 12.6 weeks (interquartile range, 11.6–13.9). The genetic outcomes were confirmed in 17,851 cases (88.4%): 13,043 (73.1%) low-risk and 4808 (26.9%) high-risk cases for aneuploidy. Overall, 133 trisomies were diagnosed (100 trisomy 21; 18 trisomy 18; 15 trisomy 13). The cell-free DNA screen positive rate was lower in the low-risk vs the high-risk group (0.27% vs 2.2%; P <.0001). The sensitivity and specificity were similar between the groups. The positive predictive value for the low- and high-risk groups was 85.7% vs 97.5%; P =.058 for trisomy 21; 50.0% vs 81.3%; P =.283 for trisomy 18; and 62.5% vs 83.3; P =.58 for trisomy 13, respectively. Overall, 602 (3.4%) patients had no-call result after the first draw and 287 (1.61%) after including cases with a second draw. The trisomy rate was higher in the 287 cases with no-call results than patients with a result on a first draw (2.8% vs 0.7%; P =.001). The updated algorithm showed similar sensitivity and specificity to the study algorithm with a lower no-call rate.
Conclusion
In women at a low risk for aneuploidy, single-nucleotide-polymorphism-based cell-free DNA has high sensitivity and specificity, positive predictive value of 85.7% for trisomy 21 and 74.3% for the 3 common trisomies. Patients who receive a no-call result are at an increased risk of aneuploidy and require additional investigation.
Introduction
Noninvasive prenatal testing using cell-free DNA (cfDNA) to screen for fetal chromosomal aneuploidy has seen rapid uptake since 2011. , It was demonstrated to have high sensitivity and specificity , and be superior to standard maternal serum analyte-based screening. Currently, most professional societies recommend cfDNA as an option for primary aneuploidy screening.
Why was this study conducted?
There are limited data on the performance of cell-free DNA (cfDNA) screening for aneuploidy in low-risk populations.
Key findings
In women at low previous risk for aneuploidy, cfDNA has high sensitivity and specificity and a positive predictive value of 85.7% for trisomy 21 and of 74% for trisomies 21, 18, and 13 combined. Patients who receive a failed (no-call) result are at an increased risk of aneuploidy. An updated algorithm has a lower no-call rate while maintaining performance.
What does this add to what is known?
This is the first study to assess cfDNA screening performance using genetic confirmation in a prospective obstetrical population. It adds valuable information on test performance in women at a low risk for aneuploidy and in cases with failed cfDNA tests.
Despite this, the routine offer of cfDNA screening to all patients has not been uniformly adopted. Cost, loss of benefits associated with ultrasound-based screening, and limitations of existing studies in particular are a concern. In addition, some providers may feel that a benefit of primary cfDNA screening over contingency screening in low-risk patients has not been clearly demonstrated. Initial validation studies using genetic confirmation were conducted on small cohorts of pregnancies at a high previous risk for aneuploidy. , Conversely, studies on large cohorts that included all-risk populations have been limited by a lack of genetic confirmation. This left some doubt as to whether there was underreporting of trisomies and whether the measurement of sensitivity and positive predictive value, particularly in women at a low risk of aneuploidy, was accurate enough. , , In addition, previous studies have generally excluded cases with a noninterpretable (“no-call”) result, leaving questions about how this impacts the overall test performance. ,
The Single-nucleotide-polymorphism-based Microdeletion and Aneuploidy RegistTry (SMART) was a large prospective study designed to evaluate cfDNA performance for the 22q11.2 deletion syndrome and the common trisomies (trisomy 21 [T21], trisomy 18 [T18], and trisomy 13 [T13]) in a general referral population. A unique aspect of the SMART study was the confirmatory genetic testing requested in all cases through cytogenetic or cytogenomic analysis of fetal samples or chromosome microarray analysis (CMA) of newborn DNA samples, including analysis of cases with no-call cfDNA results. Here we report the results of the SMART study for the prenatal detection of T21, T18, and T13 in women at low vs high previous risk for aneuploidy.
Materials and Methods
Study design and participants
We enrolled pregnant women undergoing cfDNA screening for aneuploidy and 22q11.2DS at 21 centers in 6 countries ( Supplement #1 ). The study was approved by each site’s institutional review board or ethics committee, and all the participants provided written consent. Eligible women who requested and underwent screening for aneuploidy and 22q11.2 deletion syndrome were ≥18 years old, ≥9 weeks’ gestation, had a singleton pregnancy, and planned to deliver at a study site-affiliated hospital. Women were excluded if they received a cfDNA result before enrollment, had a history of organ transplantation, conceived using ovum donation, had a vanishing twin, or were unwilling or unable to provide a newborn sample. Women who had a serum screening result for aneuploidy or sonographic detection of fetal anomalies were eligible for inclusion. Women were considered to be at a high risk for aneuploidy if they had a previous positive serum-based (first trimester combined or second trimester triple or quadruple) screen for aneuploidy, fetal nuchal translucency (NT) ≥3.0 mm, an ultrasound-detected anomaly before enrollment, or if the maternal age was ≥35 years at delivery and no other screening results (eg, serum) were available. The participants did not receive remuneration for enrolling. The results of cfDNA screening were utilized by the providers and patients as part of clinical care.
Outcomes
The primary outcome was the test performance of single nucleotide polymorphism (SNP)-based cfDNA for detecting T21, T18, and T13 in participants with a low previous risk for aneuploidy than those at a high risk. The secondary outcomes included the rates of trisomies in cfDNA no-call cases and the test performance of an updated algorithm that was made available after enrollment completion.
Procedures
The sample preparation and analysis of cfDNA were performed as previously described (Natera Inc, Austin, TX). Noninvasive prenatal testing results indicating a risk of ≥1/100 for a trisomy were categorized as high-risk and those <1/100 were categorized as low-risk. In cases that did not yield a result, the patients were offered repeat testing and results after a second draw were included for analysis. During enrollment, the cfDNA laboratory protocol was modified once ; the results from both the periods were combined for analysis (original algorithm).
Independent of the study, the laboratory developed an updated algorithm optimized to improve the no-call rate at a low fetal fraction using a deep neural network, which utilizes an artificial intelligence approach. A deep learning (Tensorflow v1.15 [Google Inc., Mountain View, CA]) approach was used to optimally model noise using a deep mixture-of-experts neural network with multiple independent networks, combining the results into a probability score. This self-supervised algorithm leveraged 1.6 million sequenced mixtures of mother and fetus cfDNA samples, learning to harness linkage among the SNPs to make high-confidence calls for a larger proportion of samples. Deeper sequencing of high-risk calls was applied to lower false positive rates. This updated protocol was assessed after enrollment completion and was blinded to the outcomes.
The genetic outcomes were assessed by CMA through the analysis of DNA from fetal (chorionic villus sampling, amniocentesis, or products of conception) or infant (cord blood, buccal swab, or newborn blood spot obtained for state newborn screening) samples. Postnatal confirmatory samples were obtained at the end of pregnancy in all the cases regardless of the availability of previous prenatal diagnostic genetic testing.
CMA was performed by an independent laboratory (Center for Applied Genomics, Children’s Hospital of Philadelphia, Philadelphia, PA) and was blind to the clinical findings and cfDNA results. For CMA analysis, the DNA was prepared from cord blood, buccal smear, or a dried blood spot. Copy number variants were identified using the Illumina (San Diego, CA) SNP-based Infinium Global Screening Array (GSA) platform. The samples were genotyped on standard versions GSA-V1.0, GSA-V2.0, GSAMD-V1.0, or GSAMD-V2.0, which contain >700,000 SNPs from chromosome 1 to 22 or a custom-designed SMARTArray in which additional SNPs were added to the GSA backbone. In addition, positive samples underwent confirmation on the Omni 2.5-8V1-3 array and were reviewed by a clinical molecular cytogeneticist before results were generated.
If a postnatal sample for CMA confirmation was not available, results from pre- or postnatal clinical testing with karyotype, quantitative fluorescent polymerase chain reaction (QF-PCR), fluorescence in situ hybridization (FISH), or CMA were used for genetic confirmation, if available.
The cases with mosaicism were considered affected if >80% of cells were trisomic on confirmatory testing. Mosaicism identified only by chorionic villus sampling (CVS) was not considered as confirmation of genetic outcome. The study steering committee reviewed any discordance between the confirmatory tests blinded to the clinical outcome to adjudicate how the results should be interpreted and included in the analysis.
The neonatal DNA samples were mostly obtained in the form of dry blood spots from the States’ health departments; they were collected as a part of neonatal screening programs. For quality assurance, a concordance test was developed and designed to confirm that cfDNA results and newborn samples were correctly paired using alignment between SNPs in the 2 samples; any samples that could not be paired were excluded.
Data collection
Research coordinators at each site recorded clinical data using a secured computerized tracking system developed and managed by the Data Coordinating Center at The Biostatistics Center at the George Washington University, Washington, DC. We collected patient and obstetrical data, imaging reports, and aneuploidy serum screening and prenatal diagnosis results. In addition, information on pregnancy complications; genetic testing or ultrasound findings; and newborn features suggestive of genetic abnormality, major malformations, and other adverse outcomes was collected after delivery.
Study oversight
The study was a collaboration between the clinical investigators and the sponsor (Natera, Inc, Austin, TX). The first and last authors designed the protocol with the sponsor and had a majority vote in study design and data interpretation. All the laboratory analyses were blinded to the outcome data. The clinical and laboratory results were managed by the Data Coordinating Center, which independently matched the deidentified information and analyzed the results only after the pregnancy outcomes were available and testing was complete.
Statistical analysis
The trisomy analysis was a secondary analysis, and the sample size was calculated on the basis of confidence intervals for the 22q11.2 deletion syndrome, with a prevalence range of 1 per 1000 to 1 per 5000. This was more than adequate to assess the detection of T21, with an expected prevalence of 1 per 425, and it would provide a reasonable assessment of the detection rates of T18 (prevalence of 1/1000) and T13 (prevalence of 1/3000). The sensitivity, specificity, positive likelihood ratio, and positive and negative predictive values of cfDNA results were assessed in the entire cohort and within the risk groups. When appropriate, exact (Clopper–Pearson) 95% confidence intervals (CIs) were reported. Low- and high-risk groups were compared for test performance using the Fisher’s exact test. Participants without genetic confirmation were excluded from the analysis. The SAS Studio 9.04 software (SAS Institute, Cary, NC) was used for analysis. The MedCalc software was used to calculate the CIs for the positive likelihood ratios. Continuous variables were compared using the Wilcoxon test and categorical variables were compared using the chi-square or Fisher exact test. The McNemar test was used for paired analyses.
Results
Study participants
A total of 25,199 pregnant individuals were assessed for eligibility, and 20,194 (80.1%) were enrolled ( Figure ); 56.6% were enrolled in the US and 43.4% in Europe or Australia. Of the enrolled participants, 285 (1.4%) had pregnancy loss without genetic confirmation, 93 (0.5%) withdrew consent, 1085 (5.4%) were lost to follow-up; in 603 (3.0%), a sample for genetic confirmation of aneuploidy was not obtained, and in 277 (1.4%) the confirmation test failed laboratory quality control. The latter group included 48 cases in which the neonatal sample could not be genetically paired with a cfDNA sample. After all exclusions, the study cohort included 17,851 (88.4%) women for whom both cfDNA results and DNA analysis of the fetus or newborn were available.
The baseline characteristics of the entire study cohort stratified by risk groups are outlined in Table 1 . The median maternal age was 34.3 years (interquartile range [IQR], 30.2–37.4), and the median gestational age was 12.6 weeks (IQR, 11.4–13.9). A total of 13,043 cases (73.1%) were considered low-risk for aneuploidy, including 3,873 that were ≥35 years old but had a low-risk result on a screening test before enrollment. The remaining 4808 (26.9%) were categorized as high-risk ( Table 1 ). Most high-risk women (4010, 83.4%) were ≥35 years old with no previous serum screening; 616 (12.8%) were high-risk on the basis of the results of traditional serum analyte-based screening, 112 (2.3%) had cfDNA screening following the detection of a fetal abnormality on ultrasound, and 101 (2.1%) had a cystic hygroma or a NT ≥3 mm. Participants at a high risk for aneuploidy were enrolled at an earlier gestational age, were more likely to be enrolled in Europe, and were more likely to have conceived using in vitro fertilization. Compared with non-US participants, the US participants were younger (median 32.6 vs 35.9; P <.0001), had a higher median body mass index (BMI) (26.1 vs 24.1; P <.0001), and enrolled at a later mean gestational age (13.7 week vs 12.8 week; P <.0001).
| Variable | Full cohort (n=17,851) | Low risk (n=13,043) | High risk (n=4808) | P value Low vs high risk |
|---|---|---|---|---|
| Maternal and gestational characteristics | ||||
| Median maternal age (IQR) – y | 34.3 (30.2–37.4) | 32.5 (28.8–35.7) | 37.6 (35.8–39.7) | <.001 |
| Nulliparity | 7876 (44.2) | 6283 (48.2) | 1593 (33.4) | <.001 |
| Median BMI (kg/m 2 ) (IQR) | 25.0 (22.3–29.1) | 25.0 (22.3–29.3) | 25.0 (22.4–28.8) | .699 |
| Race/ethnicity | <.001 | |||
| Asian | 1532 (8.6) | 1260 (9.7) | 272 (5.7) | |
| Black | 1569 (8.8) | 1300 (10.0) | 269 (5.6) | |
| White | 10,811 (60.6) | 7283 (55.8) | 3528 (73.4) | |
| Hispanic | 3331 (18.7) | 2704 (20.7) | 627 (13.0) | |
| Other/unknown | 608 (3.4) | 496 (3.8) | 112 (2.3) | |
| Median gestational age at enrollment (IQR)—wk | 12.6 (11.4–13.9) | 12.7 (11.9–14.0) | 11.7 (10.4–13.6) | <.001 |
| Pregnancy through assisted reproductive technology | 904 (5.1) | 582 (4.5) | 323 (6.7) | <.001 |
| Current smoker | 314 (1.8) | 257 (2.0) | 57 (1.2) | <.001 |
| Enrollment site | <.001 | |||
| United States | 10,105 (56.6) | 8345 (64.0) | 1760 (36.6) | |
| Europe | 7331 (41.1) | 4401 (33.7) | 2930 (60.9) | |
| Australia | 415 (2.3) | 297 (2.3) | 118 (2.5) | |
| Prenatal screening and testing | ||||
| Positive first trimester screen before cfDNA testing | 509 (2.9) | 509 (10.6) | ||
| NT>3 mm before cfDNA testing | 101 (0.9) | 101 (2.1) | ||
| Positive second trimester before cfDNA testing | 107 (0.6) | 107 (2.2) | ||
| Major anomaly before cfDNA testing | 112 (0.6%) | 112 (2.3%) | ||
| No call – % | 287 (1.6) | 207 (1.6) | 80 (1.7) | .717 |
| Mean cfDNA fetal fraction (SD) | 9.9 (4.1) | 9.9 (4.1) | 9.7 (4.2) | <.001 |
| Diagnostic testing (CVS and amniocentesis)—% | 544 (3.1%) | 283 (2.2) | 261 (5.4) | <.001 |
| Any trisomy (T13, 18, 21) | 133 (0.8%) | 29 (0.2) | 104 (2.2) | <.001 |
| Pregnancy and delivery outcome | ||||
| Delivery outcome | <.001 | |||
| Miscarriage | 49 (0.3%) | 15 (0.1) | 34 (0.7) | |
| Elective abortion | 159 (0.9%) | 64 (0.5) | 95 (2.0) | |
| Live birth | 17,600 (98.7%) | 12,935 (99.3) | 4665 (97.1) | |
| Stillbirth | 30 (0.2%) | 19 (0.2) | 11 (0.2) | |
| Neonatal death | 29 (0.2%) | 16 (0.1) | 13 (0.3) | .036 |
| Median gestational age at delivery (IQR) – wk | 39.4 (38.4–40.3) | 39.4 (38.6–40.3) | 39.3 (38.3–40.1) | <.001 |
| PTB<34 wk | 459 (2.6%) | 262 (2.0) | 197 (4.1) | <.001 |
| Preeclampsia | 711 (4.1%) | 519 (4.1) | 192 (4.1) | .846 |
| Small for gestational age | 1546 (8.9%) | 1158 (9.1) | 388 (8.3) | .135 |
| Mean birthweight (SD) g | 3353 (555) | 3347 (544) | 3371 (586) | <.001 |
| Apgar 1 < 7 | 797 (5.1%) | 587 (4.9) | 210 (6.2) | .002 |
| Apgar 5 < 7 | 154 (1.0%) | 106 (0.9) | 48 (1.4) | .006 |
| Median days to newborn discharge (IQR)—d | 2.0 (2.0–3.0) | 2.0 (2.0–3.0) | 3.0 (2.0–4.0) | <.001 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
