Funding sources: The work on this project was supported by the California Technology Assessment Program.
Fetal aneuploidy detection by maternal plasma DNA sequencing: a technology assessment
Article first published online: 17 MAY 2013
© 2013 John Wiley & Sons, Ltd.
Special Issue: Noninvasive Prenatal Testing Using Maternal Plasma DNA: Part I
Volume 33, Issue 6, pages 514–520, June 2013
How to Cite
Walsh, J. M. E. and Goldberg, J. D. (2013), Fetal aneuploidy detection by maternal plasma DNA sequencing: a technology assessment. Prenat. Diagn., 33: 514–520. doi: 10.1002/pd.4109
Conflicts of interest: None declared
- Issue published online: 17 MAY 2013
- Article first published online: 17 MAY 2013
The American College of Obstetricians and Gynecologists currently recommends that all pregnant women be offered screening for chromosomal abnormalities, regardless of maternal age. Traditional screening tests have detection rates ranging from 85% to 90% and false-positive rates of 3% to 5%. A woman with an abnormal noninvasive test is offered a diagnostic test, but diagnostic tests are associated with a risk of pregnancy loss. Recently, analysis of cell-free fetal DNA (cffDNA) in maternal blood has been shown to have potential for the accurate detection of some of the common fetal autosomal aneuploidies. As part of a technology assessment for the California Technology Assessment Forum, we critically reviewed the evidence for the use of cffDNA as a prenatal screening test. We evaluated the evidence for its use as either a ‘primary’ or an ‘advanced’ screening test and for its use in screening for three different trisomies: 21, 18, and 13. We evaluated whether the use of cffDNA met established technology assessment criteria and established conclusions about evidence-based use of this new technology. © 2013 John Wiley & Sons, Ltd.
The American College of Obstetricians and Gynecologists (ACOG) recommends that pregnant women be offered screening for chromosomal abnormalities, regardless of maternal age. The noninvasive tests typically include measurement of maternal serum markers that are interpreted in the context of maternal age and many also include nuchal translucency. ACOG recommends that all patients should be provided with an individualized risk acknowledging that different patients have different concepts of what would be considered a personal positive result and that all pregnant women have the option of diagnostic testing, regardless of test results.
Traditional noninvasive screening methods have detection rates of about 85% to 90% and false-positive rates of 3 to 5%, when used in real-world clinical practice,[2-5] but all provide a combination of noninvasive testing followed by an offer of diagnostic testing if the screening test is positive.
The currently available diagnostic tests are amniocentesis and chorionic villus sampling (CVS). Both of these tests provide actual chromosomal analyses and can either confirm or disconfirm an abnormal screening test but are associated with a small risk of pregnancy loss.
Cell-free fetal DNA (cffDNA) in maternal blood can be used to detect fetal chromosomal abnormalities. There are at least two techniques for isolating fetal DNA from maternal DNA. Massively parallel signature sequencing (MPSS) precisely quantifies cffDNA fragments for fetal trisomy detection. The other technique is directed DNA analysis, which selectively sequences relevant chromosomes using digital analysis of selected regions (DANSR).
The potential role of cffDNA in prenatal diagnosis would be either as a primary screening test (replacing the currently available noninvasive tests) or as an ‘advanced screening test’. As an advanced test, it would be used in women who have a ‘screen positive’ result on a traditional screening test before proceeding to invasive diagnostic testing; confirmation of positive results by a diagnostic test would still be required. The potential advantage of using cffDNA as an advanced screening test would be to reduce the number of invasive procedures and the resulting loss of normal fetuses.
The California Technology Assessment Forum (CTAF) is a forum that serves the public good by assessing new and emerging medical technology (www.ctaf.org). The mission of CTAF is to identify medical technologies that improve health. Using an established set of technology assessment (TA) criteria, CTAF evaluates approximately 15 medical technologies every year. The Blue Shield of California Foundation (BSCF) spearheads CTAF. Neither BSCF nor CTAF is a revenue-generating organization, consultant organization, endorser of specific technologies, advocacy organization, or an organization that determines health plan benefit coverage. CTAF evaluated the use of cffDNA for prenatal diagnosis in June of 2012 and then again in October 2012. In the initial assessment, CTAF assessed the use of cffDNA but did not specifically look at high-risk versus low-risk women. In the most recent assessment, CTAF evaluated the role of cffDNA as a primary or advanced screening test for fetal aneuploidy in high-risk versus low-risk women and as a primary versus an advanced screening test for selected trisomies. Descriptions of the CTAF process and analysis are described as follows.
CTAF technology assessment criteria
The CTAF uses five TA criteria to evaluate all technologies (Table 1). These criteria have been slightly modified from those developed and used by the Blue Cross Blue Shield Technology Evaluation Center.
|TA criterion 1||The technology must have approval from the appropriate governmental bodies.|
|TA criterion 2||The scientific evidence must permit conclusions concerning the effectiveness of the technology regarding health outcomes.|
|TA criterion 3||The technology must improve the net health outcomes.|
|TA criterion 4||The technology must be as beneficial as any of the established alternatives.|
|TA criterion 5||The improvement must be attainable outside the investigational setting.|
For the first assessment of cffDNA, the Medline database, Cochrane clinical trials database, Cochrane reviews database, and the Database of Abstracts of Reviews of Effects were searched using the key words ‘aneupoloid’ or ‘down syndrome’ or ‘chromosome disorders’ or ‘trisomy and prenatal diagnosis’ or ‘fetal diseases’ or ‘fetus’ and ‘cell free system’ or ‘cell free’ or ‘DNA’ or ‘RNA’ or ‘maternal plasma DNA’ or ‘maternal blood DNA’. The search was performed from database inception to May 2012. The bibliographies of systematic reviews and key articles were manually searched for additional references. The abstracts of citations were reviewed for relevance, and all potentially relevant articles were reviewed in full. Inclusion criteria were that the study had to evaluate cffDNA as a prenatal screening test in pregnant women, had to compare cffDNA with a gold standard, included only humans, and published in English as a peer-reviewed article.
Our initial search included articles published through May 2012 and revealed 265 potentially relevant articles. We reviewed all the titles and identified 16 potentially relevant abstracts. Abstracts were reviewed in full, and we identified seven studies as potentially relevant for inclusion. We later updated our search to include articles published through 31 August 2012 and identified one additional relevant study.
All laboratory tests (except those for research) performed on humans in the United States are regulated by the Centers for Medicaid and Medicare Services through the Clinical Laboratory Improvement Amendments (CLIA). There are three companies with products for fetal aneuploidy detection via maternal plasma that are performed in CLIA-certified laboratories (Table 2).
|Company||LDT name||Trisomy identified||Testing approach|
|T21 (Down syndrome)||T18 (Edwards syndrome)||T13 (Patau syndrome)||Directed analysis||Random sequencing (MPSS)|
|Ariosa Diagnostics||Harmony™ Prenatal Test||X||X||X|
|Verinata Health||verifi™ prenatal test||X||X||X||X|
Strength of scientific evidence
A total of eight studies evaluated the use of cffDNA for screening for fetal aneuploidy. These studies are described in Table 3. Although some of the studies were performed prospectively, in all but two studies, not all samples in a cohort were analyzed for trisomy status.[8, 9] The majority of studies selected trisomy cases and additional controls for analysis from the overall cohort.
|Author, year||Study type||N total||N with abnormal karyotype||Trisomy targeted||Test||Inclusion criteria|
|Palomaki, 2011 Palomaki, 2012||Multicenter nested case control||1971||283 212 (T21) 59 (T18) 12 (T13)||21 18 13||MPSS||High risk for Down syndrome: maternal age, family history, or positive screening test|
|Sehnert, 2011||Multicenter cross-sectional validation study||119||53||21 18||MPSS||18 years and older and pregnant|
|Ehrich, 2011||Blinded prospective||480||39||21||MPSS||Advanced maternal age, positive screening test, personal or family history of Down Syndrome|
|MELISSA Bianchi, 2012||Blinded prospective; multicenter with nested case control||532||221||21 18 13||MPSS||Women undergoing an invasive prenatal procedure; aged 38 years and over; positive screening test for aneuploidy or prior aneuploidy fetus|
|Chiu, 2011||Diagnostic accuracy with prospectively collected or archived serum||753||86||21||MPSS||Women with ‘clinical indications for CVS or amniocentesis’|
|Ashoor, 2012||Nested case control||400||100||21 18||DANSR||Stored plasma High-risk women where risk for aneuploidy was >1/300|
|Sparks, 2012||Nested case control with training and validation sets||338||88||21 18||DANSR||At least 18 years of age; at least 10 weeks pregnant; singleton pregnancy (subset of larger prospective group chosen based on ploidy status)|
|Prospective cohort||3228||119 81 (T21) 38 (T18)||21 18||DANSR||At least 18 years of age; at least 10 weeks pregnant; singleton pregnancy; planning to undergo invasive procedure for any reason|
Impact on health outcomes
Among the eight studies that evaluated the use of cffDNA for screening for fetal aneuploidy, five used the MPSS technology,[9-14] and three used the DANSR technology.[6, 15] All eight studies evaluated cffDNA as a screening test for trisomy 21 (T21),[6, 9-15] six studies evaluated it as a screening test for trisomy (T18),[6, 10-12, 14, 15] and two studies evaluated it as a screening test for trisomy 13 (T13). In all the studies, the true chromosomal state of the fetus was known from either amniocentesis or CVS. Two studies were validation studies where part of the sample was used as the training set, and the other part of the sample was used as the validation set.[6, 12] Although some of the studies were performed prospectively, in many cases, all samples in a cohort were not analyzed for trisomy status. These studies selected trisomy cases and additional controls for analysis from the overall cohort. In one study that compared two protocols for MPSS, all samples (regardless of ploidy status) were analyzed with MPSS with one of the MPSS protocols, but only about 146 of the 571 euploid fetuses were evaluated with one of the MPSS protocols. One large prospective multicenter cohort study tested all subjects for aneuploidy status.
The number of fetuses with abnormal karyotypes varied in the studies but ranged from 39 to 283 in the largest study. The number of T21 cases in each study ranged from 39 to 212. The sensitivity rate for the detection of T21 ranged from 98.6% to 100%. Specificity ranged from 97.9% to 99.97%.
Six of the studies evaluated cffDNA for the detection of T18. Two of those studies were validation studies, and so, the total number of T18 fetuses in the validation set was only eight and seven, respectively.[6, 12] Both of these validation studies reported a 100% sensitivity for the detection of T18. In one study that had 59 T18 cases, the reported sensitivity of cffDNA was 100%, and in another study that included 36 T18 cases, the sensitivity was 97.2%. Finally, in a multicenter prospective study that included 38 evaluable cases of T18, the sensitivity was 97%.
Only two studies evaluated the use of cffDNA for the detection of T13.[11, 14] These studies included a total of 26 T13 cases. The detection rate was 78.65% in one study and 91.7% in the other study, but the ability to draw conclusions is limited by the small number of cases. A recent case–control study, published after the cutoff date for our literature search, showed that the majority of T13 cases could be detected but was based on only 11 cases of T13.
Description of individual studies
The results of the eight studies using cffDNA to screen for fetal aneuploidy are described in Table 4. In the MatErnal Blood IS source to Accurately diagnose fetal aneuploidy (MELISSA) study, investigators collected blood samples in a blinded nested case–control study from 2882 women that were undergoing prenatal diagnostic procedures at 60 different US sites. MPSS was performed, and the results were compared with the karyotypes as determined by amniocentesis or CVS. There were 532 samples, 221 of which had abnormal karyotypes. Sensitivity for T21 was 100%, for T18 was 97.8%, and for T13 was 78.6%, although this estimate was based on only 14 cases. There were no false positive, resulting in specificities for T21, T18, and T13 of 100% each. Because this was a nested case–control study, and did not reflect true population prevalence of the fetal aneuploidies, positive and negative predictive values (NPVs) cannot be calculated.
|Author, year||N total||N with abnormal karyotype||Main results: T21||Main results: T18||Main results: T13||Comments|
|Palomaki, 2011 Palomaki, 2012||1971|| |
98.6% detection rate
100% detection rate
91.7% detection rate
Overall detection rate 98.9% for common aneuploidies
17 samples could not be interpreted: three were T18
Very few T13 cases
|Sehnert, 2011||119||53||100% sensitivity||100% sensitivity||N/A|| |
Training and validation set: Training set – 71 samples (26 abnormal karyotypes) and validation set – 48 samples (27 abnormal karyotypes)
13 T21 and 8 T18 samples
|Ehrich, 2011||480||39|| |
|N/A||N/A||Only 39 T21 samples|
|MELISSA Bianchi, 2012||532||221 Total 89 (T21) 36 (T18) 14 (T13)||100 sensitivity||97.2% sensitivity||78.6% sensitivity||No false positives for autosomal euploides|
|Chiu, 2011||753||86|| |
|N/A||N/A||Two different methods tested (two-plex and eight-plex). two-plex method had better sensitivity|
|Ashoor, 2012||400||100 50 (T21) 50 (T18)||100% sensitivity 100% specificity||98% sensitivity 100% specificity||N/A||Specificity calculated on a total of 300 euploid fetuses, although three samples were not available for analysis|
|Sparks, 2012||338||88||100% sensitivity||100% sensitivity||N/A|| |
36 T21 cases in validation set
Seven T18 cases in validation set
|3228||119 81 (T21) 38 (T18)|| |
False-positive rate 0.03% (95% CI = 0.002–0.20%)
100% sensitivity (95% CI = 95.5–100%)
|97% sensitivity (95% CI = 86.5–99.9%) False-positive rate 0.07% (95% CI = 0.02–0.25%)||NA||4.6% of samples were not able to be analyzed either due to low fetal DNA fraction or assay failure|
Chiu and colleagues used prospectively collected or archived serum from 753 women at high risk for T21. All women were undergoing or had undergone a diagnostic procedure. A total of 86 women had a fetus with T21. MPSS, according to two different protocols, was performed on all samples. The two-plex protocol was performed on samples from 314 pregnancies, and the eight-plex protocol was performed on samples from 753 pregnancies. The two-plex protocol detected T21 fetuses with 100% sensitivity and 97.9% specificity, resulting in a positive predictive value (PPV) of 96.6% and an NPV of 100%. The eight-plex protocol detected 79.1% of the T21 cases and had a specificity of 98.9%, resulting in a PPV of 91.9% and an NPV of 96.9%.
The Noninvasive Chromosomal Evaluation Study, a multicenter prospective cohort study for detection of fetal T21 and T18, was conducted in three countries. Among women undergoing invasive prenatal diagnosis for any reason, blood was collected prior to the invasive procedure. A total of 4002 women were eligible for inclusion: of the 3228 participants eligible for analysis, 2888 of them had normal karyotypes, 84 had T21 karyotype (results obtained in 81), 42 had T18 karyotype (results obtained in 38), and 73 had other chromosomal abnormalities. The sensitivity for T21 was 100% [95% confidence interval (CI) = 95.5–100%]. The specificity for T21 was 99.97% (95% CI = 99.8–99.99%), resulting in a false-positive rate of 0.03% (95% CI = 0.002–0.20%). For T18, the sensitivity was 97.4% (95% CI = 86.5–99.9%). The specificity for T18 was 99.93% (95% CI = 99.75–99.98%). Overall, the PPVs for T21 and T18 were 98.8% and 94.9%, respectively. The NPVs were 100% and 99.96% for T21 and T18, respectively. An important limitation is that 148 out of 3228 cases (4.6%) could not be analyzed because of either assay failure or low fetal DNA fraction. The sensitivity, specificity, and predictive values are reported only for those samples that could be analyzed. Nevertheless, this large prospective study provides important information about test performance in a population of high-risk women undergoing invasive procedures.
The majority of studies have evaluated the role of cffDNA in high-risk women, but one recent study that was published after our date cutoff attempted to address the utility of cffDNA screening in a routinely screened first trimester population. In a cohort of 2049 women undergoing prenatal screening, there were eight cases of T21 and three cases of T18. Fetal karyotyping was performed in 86 cases: 75 were normal, there were eight cases of T21 and three cases of T18. The remaining 1963 cases resulted in phenotypically normal live births and were thus not karyotyped further. For all eight cases of T21, the trisomy risk score was >99%. For two of the three cases of T18, the risk score was >99%. The overall incidence of T21 in this study was higher than the general population, suggesting that this is not really an average-risk population. This and the small number of T21 and T18 cases in this study, the fact that not all pregnancies underwent karyotyping, and that not all samples could be given trisomy risk scores limit the conclusions that can be drawn from this study, but it does suggest that cffDNA might have potential utility in average-risk women and that it should be evaluated in future studies.
Despite its reported high sensitivity and specificity, its accuracy is not that of the gold standard of fetal karyotyping; thus, cell-free DNA cannot currently replace either CVS or amniocentesis. In addition, karyotyping can provide information about other conditions besides fetal aneupoloidy status, whereas cffDNA currently cannot.
Existing screening strategies have an accuracy ranging from 90% to 95% for the detection of fetal aneuploidy, with a false-positive rate of 3% to 5%.[2-5] cffDNA screening has a higher reported accuracy than the currently available screening tests, but to date, cffDNA has not been evaluated as a primary screening test.
Current studies support the role for cffDNA testing is as an ‘advanced’ screening test, which could be performed on women deemed at high risk after undergoing conventional noninvasive screening tests for fetal aneuploidy. These women could then undergo cffDNA testing for further risk refinement; some could avoid many invasive procedures and the potential loss of normal fetuses.
False-positive tests could lead to unnecessary diagnostic procedures, with their related risks. False-negative tests could lead to a fetus with Down syndrome being carried to term when the mother would have either preferred to prepare herself for this or may have chosen to the terminate the pregnancy had she known the true ploidy status of the fetus. In the one large prospective cohort study that allows an estimate of population prevalence, in a population of high-risk women, the false-positive rate for T21 was 0.03%, and the false-positive rate for T18 was 0.07%., both of which are relatively low rates. The corresponding false-negative rates for T21 were 0% and for T18 were 3.6%. Because the estimates for T18 are based on only 38 cases, they are probably less reliable than the T21 estimates. However, in an average-risk population with lower disease prevalence, the false-positive rate would be higher, and the PPV would be lower. With respect to the procedure itself, because it is a noninvasive blood test, the procedure-related harms are minimal.
Overall, cffDNA has the potential benefits of improved diagnostic accuracy for T21 and T18 over existing screening tests, although to date, its utility has only been evaluated in high-risk women. It has the potential to have a role as an advanced screening test and may ultimately lead to fewer unnecessary diagnostic procedures. The potential harms are related to the downstream effects of false positives and false negatives. Overall, the potential benefits of testing for T21 and T18 outweigh the potential harms in high-risk women. Less evidence is available for testing for T13. Because the test has not been evaluated in a large cohort of women at average risk, the potential benefits and harms cannot currently be weighed.
Studies of cffDNA to date have evaluated it in women already identified as high risk by age, personal or family history of Down's syndrome, and/or with initial positive screening tests. Given its significantly greater sensitivity and specificity than currently available noninvasive screening tests and combinations of noninvasive screening tests, one of the proposed uses of cffDNA has been as an ‘advanced’ screening test. Individuals with a positive test on one of the currently available screening tests could potentially have cffDNA as the next step. However, a positive cffDNA test would still require progression onto a definitive diagnostic test.
The potential impact of incorporating a cffDNA test into routine prenatal care for high-risk women was calculated using a theoretical model. This model was developed using the clinical results in the MELISSA study described earlier. The model evaluated the addition of cffDNA testing using verifi™ test for T21, T18, and T13 into routine clinical practice. By using this model, it was estimated that including the verifi™ prenatal test as a screening test for fetal trisomies in high-risk women would result in a 66% reduction in invasive diagnostic induced miscarriages and would lead to 38% more women receiving a T21 diagnosis, and that total costs for prenatal screening and diagnosis would be decreased by 1% annually.
Currently, no studies have evaluated incorporating cffDNA into prenatal clinical practice in a large cohort of average-risk women compared with the current standard of care. Similarly, no studies have compared using cffDNA as a primary screening test versus the current standard of care.
Real-world use of the test
For women at high risk for chromosomal abnormalities, an improvement when compared with the current standard of care of prenatal testing or when incorporated into existing noninvasive screening strategies has been shown on the basis of the aforementioned economic model based on the MELISSA study. The recent large cohort study by Norton et al. was performed in multiple settings in three different countries reflective of real-world settings. It is a simple blood test sent for analysis at centralized laboratories and can easily be performed in either investigational or noninvestigational settings.
For average-risk women, because an improvement has not been shown in the investigational setting, an improvement cannot be attainable outside the investigational setting.
The CTAF concluded that criterion 1 (approval by the appropriate governmental bodies) was met. CTAF concluded that criterion 2 (availability of scientific evidence to permit conclusions about effectiveness) was met.
The CTAF concluded that criterion 3 (improvement of net health outcomes) is met for cffDNA as a prenatal screening test for T21 and T18 in high-risk women. In addition, they concluded that criterion 3 was not met for cffDNA as a prenatal screening test for T13 in high-risk women nor was it met for average-risk women.
The CTAF concluded that criterion 4 (the test must be as beneficial as any of the established alternatives) was met for the detection of T21 and T18 for women at high risk for fetal aneuploidy when cffDNA is used as an advanced screening test. However, CTAF concluded that criterion 4 was not met for the detection of T13 for women at high risk for fetal aneuploidy when cffDNA is used as an advanced screening test, nor for women at high risk for fetal aneuploidy when cffDNA is used as a primary screening test, and nor for women at average risk for fetal aneuploidy.
The CTAF concluded that criterion 5 was met for women at high risk for fetal aneuploidy T21 and T18 when cffDNA is used as an advanced screening test. In addition, they concluded that TA criterion 5 was not met for women at high risk for fetal aneuploidy T13 when cffDNA is used as an advanced screening test nor was it met for women at high risk for fetal aneuploidy when cffDNA is used as a primary screening test. Finally, CTAF concluded that criterion 5 was not met for women at average risk for fetal aneuploidy.
Although CTAF does not determine health plan benefit coverage, CTAF assessments are used by insurers when making coverage decisions.
In conclusion, cffDNA is a promising new technology with high sensitivity and specificity for the prediction of fetal aneuploidy, in particular T21 and T18, when evaluated as an advanced screening test in high-risk women. It has the potential to reduce the number of invasive diagnostic procedures, with their associated risks of fetal loss.
To date, only one study with a total of 11 trisomy cases has evaluated cffDNA for use in women at average risk for fetal aneuploidy, and it is not ready for routine use in average-risk women.
The cffDNA as a prenatal advanced screening test in high-risk women for detection of T21 and T18 meets technology assessment criteria. Future studies must address its use as a primary screening test for T13 in high-risk women and its use for screening for fetal aneuploidy in average-risk women.
WHAT'S ALREADY KNOWN ABOUT THIS TOPIC?
- Noninvasive prenatal testing by analysis of cell-free DNA has the potential to accurately diagnose common fetal aneuploidies, although how best to use this test in the context of other available prenatal tests has not yet been determined.
WHAT DOES THIS STUDY ADD?
- This study critically assesses the published literature on the use of maternal plasma DNA sequencing for fetal aneuploidy detection and compares it with the established alternatives. Finally, it provides guidelines for evidence-based use of maternal plasma DNA sequencing for the detection of fetal aneuploidy.
- 1American Cancer Society. Cancer facts & figures 2012. 2012.
- 18Clinical and cost consequences of incorporating a novel non-invasive prenatal test into the diagnostic pathway for fetal trisomies. J Managed Care Med 2012;15(2):34–41., .