Abstract
Twin pregnancies are at increased relative risk of many obstetric complications compared to singletons, including aneuploidy. Antenatal testing for aneuploidy in twin pregnancies, both invasive and noninvasive, has additional complexities compared to singleton pregnancies, in both dizygotic and monozygotic twinning. Although the accuracy of cell-free DNA-based noninvasive prenatal testing (cfDNA NIPT) in twin pregnancies for aneuploidy and fetal Rhesus D status appears comparable to singleton pregnancies, substantially fewer overall tests (and papers published) have been performed in twin pregnancies. Trisomy 18 and trisomy 13, in particular, have not been adequately investigated. Thus, when couples are being counseled for cfDNA NIPT they must be informed that there are insufficient data to definitively demonstrate its accuracy in standard clinical practice at present. Although research in this field is rapidly advancing, certain biological scenarios will mean that cfDNA NIPT will never be 100% accurate, as with current screening and invasive testing.
Keywords
NIPT, Twin, Aneuploidy screening, cfDNA, Test accuracy, Zygosity, Chorionicity
Introduction
Due to an increasing trend of advancing maternal age at conception, and the increased use of assisted reproductive technologies (ART), the number of twin and higher order multiple pregnancies is increasing in many countries, despite the introduction of the single embryo transfer policy .
Compared to singletons, twin pregnancies are at increased relative risk of many obstetric complications, including aneuploidy . This is probably the indirect consequence of the fact that older women not only have a higher risk of aneuploidy pregnancies, but also a higher risk of naturally conceiving twin pregnancies. Additionally, older women are more likely to need ART to conceive, and ART is a risk factor for twin pregnancies as well as for fetal aneuploidy .
Risks of prenatal invasive testing are higher in twins than in singletons and accurate noninvasive prenatal testing (NIPT) would therefore be welcomed by mothers of twin pregnancies, particularly by those women who had difficulties conceiving. Additionally, in twin pregnancies, there is also a greater potential for sampling errors at chorionic villous sampling (CVS) as well as at amniocentesis . The currently used conventional prenatal screening strategies: the first trimester combined test and the second trimester serum screening test have a high false positive rate that surpasses that in singleton pregnancies and may be even higher in pregnancies conceived by ART .
Data from published studies informing the use of cell-free DNA-based noninvasive prenatal testing (cfDNA NIPT) in twins are relatively small compared to that of singleton pregnancies. There is debate as to whether the diagnostic accuracy of cfDNA NIPT in twin pregnancy is equivalent to that of singletons, and what role cfDNA NIPT should have in routine clinical care. This is against the backdrop of the increasing use in clinical practice of cfDNA testing both in state-run and private healthcare systems.
In this chapter, we will outline the evidence and the current opinions of the professional bodies, review the differences in cfDNA NIPT accuracy between twins and singletons, and stratify the available evidence on impact of chorionicity.
Zygosity, Chorionicity, and Amnionicity
Twin pregnancies can be classified according to zygosity, chorionicity, and amnionicity, all of which have implications on prenatal diagnosis and subsequent management. Zygosity refers to the number of zygotes. Monozygotic (MZ) twins originate from one zygote (one sperm and one oocyte), and dizygotic (DZ) twins originate from two zygotes (two sperm and two oocytes). Consequently, MZ twins are genetically “identical” although there are rare exceptions reported of heterokaryotypic MZ twins . It is assumed that a third of twins are monozygous. DZ twins are genetically nonidentical. In MZ twins, it can be assumed that the cfDNA NIPT result will reflect the genetic makeup of both twins, whereas in DZ twins it is more complicated as, in case of an anomaly, the twins are most likely to be discordant, that is, one twin will be affected but the other twin will not.
However, in clinical practice, the prenatal risk is allocated by the chorionicity (placental type) and amnionicity. Chorionicity refers to the number of placentas: monochorionic (MC) twins share one placenta and have interfetal vascular placental anastomoses, whereas dichorionic (DC) twins have two individual placental masses. Up to 15% of twins are MC and 85%–90% are DC. Amnionicity refers to the number of amniotic fluid sacs: monoamniotic (MA) twins have one sac; diamniotic (DA) twins have two sacs. The chorionicity and amnionicity should be determined in all twin pregnancies at the first trimester ultrasound scan , as each combination (DCDA, MCDA, MCMA) is managed differently antenatally due to different perinatal/fetal risks and complications including: twin-twin transfusion syndrome, selective intrauterine growth restriction, intrauterine death, and cord entanglement . The sensitivity and specificity of “correct” first trimester ultrasound chorionicity allocation is > 99%. Clinically, zygosity cannot be assumed from chorionicity and/or amnionicity. Although all DZ twins are DC, and all MC twins are MZ, not all DC twins are DZ, and not all MZ twins are MC. Fifteen percent of DC twins are MZ; of the MZ twins 33% are DCDA, 66% MCDA, and 1% MCMA. Consequently, this adds a layer of complexity to cfDNA NIPT in twins, particularly in cases of DC twins which although most likely to be DZ and thus discordant for aneuploidy, may be MZ. Table 1 outlines the theoretical genetic risks in twin pregnancies.
| Chromosome Abnormality | X-Linked (Fetal Sexing Only) | X-Linked (Specific Diagnostic Test) | Autosomal Recessive | |
|---|---|---|---|---|
| Risk for singleton pregnancy | Y a | 1/2 | 1/4 | 1/4 |
| Risk of at least one twin being affected | 5/3 Y | 2/3 | 3/8 | 3/8 |
| Risk of both twins being affected | ~ 1/3 Y | 1/3 | 1/8 | 1/8 |
Vanishing Twins
An increasingly recognized problem of twinning is the “vanishing twin,” whereby single twin demise is noted on ultrasound prior to 14 weeks’ gestation. In such cases, the “vanishing twin” may be “reabsorbed” and difficult to visualize on ultrasound later in pregnancy. A vanishing twin has little consequence to the surviving cotwin. It may though have consequences for false positive rates in cfDNA NIPT (of a presumed singleton pregnancy).
Higher Order Multiple Pregnancies
There is little data from cohort studies on cfDNA NIPT in higher order multiples (triplets, quadruplets, etc.), including the effect that selective reduction has on cfDNA NIPT results of the surviving fetuses . Therefore, conclusions on test accuracy in higher order multiple pregnancies cannot be confidently drawn and are not covered in this chapter.
Role of CELL-FREE DNA NIPT in Twin Pregnancies
The majority of research on cfDNA NIPT in twin pregnancies has investigated detecting aneuploidy, with few studies looking at fetal Rhesus status, or fetal sexing. The focus of this section will be on autosomal and sex chromosome aneuploidy testing, but we will briefly cover the other areas as well. Again, it is essential to emphasize that the clinical efficacy has to be assessed separately for both DC and MC twin pregnancies.
CELL-FREE DNA NIPT for Fetal Aneuploidy in Twin Pregnancies
Trisomy 21 Test Accuracy
A systematic review by Gil et al., performed in 2015, on cfDNA NIPT for detecting aneuploidy demonstrated a lower detection rate (DR) and higher false positive rate (FPR) for detecting trisomy 21 in twins overall (DR: 93.7% [95%CI 83.6–99.2], FPR: 0.23% [95%CI 0.00–0.92], 5 studies, 430 tests, I 2 = 0.0%) compared to singletons (DR: 99.2% [95%CI 98.5–99.6], FPR: 0.09% [95%CI 0.05–0.14], 24 studies, 22,659 tests, I 2 = 0.0%) . The same authors updated their systematic review 12 months later and revealed a higher DR and lower FPR in twins when testing for trisomy 21 (DR: 100% [95%CI 95.2–100], FPR: 0% [95%CI 0–0.003], 5 studies, 1135 tests, I 2 = 0%) compared to their previous systematic review results, despite the short time period between reviews . All studies performed cfDNA NIPT after 10 weeks gestation, some were performed in “high-risk populations” , some in mixed-risk populations . The updated review published in 2017 included 3 new studies, an additional 934 pregnancies . The improved accuracy over the 12 months may be due to the increased number of tests included in the updated meta-analysis (1135 vs 430 tests), or be the result of technological advances and improved sequencing platforms. Another important difference in the 2017 review is that the authors changed their study inclusion criteria to exclude the 3 case-control studies, equating to 231 tests . Case-control studies are less informative than cohort studies because they do not reflect the prevalence of the condition. Finally, the updated review also demonstrated an improved test accuracy in singleton pregnancies over the same time period: DR: 99.7% [95%CI 99.1–99.9], FPR: 0.04% [95%CI 0.02–0.08], 30 studies, 226,995 tests, I 2 = 1.2%. This suggests improvements in cfDNA NIPT technology or bioinformatics most likely play an important role.
Since the publication of Gil’s updated review, three additional cohort studies by Du (92 tests), Fosler (115 tests), and Le Conte (420 tests) and one meta-analysis by Liao have been published. The three new studies demonstrated high test accuracy in twin pregnancies, comparable to that of singleton pregnancies, with a sensitivity of 1.00 (95%CI 0.63, 1.00) and specificity of 1.00 (95%CI 0.99, 1.00) which equated to 1 false positive result in 627 tests. Fig. 1 displays the improvement in test accuracy seen across the updated systematic reviews by Gil, and the individual results of the three additional studies which we have included in “(3) Gil meta-analysis 2017 updated.” Seven tests reported by Benachi and included in Gil’s updated analysis were also included in the 492 twin pregnancies described by Le Conte, therefore we have not included Benachi in the “(3) Gil meta-analysis 2017 updated” to avoid double counting. This study included 2 true positive results.
The meta-analysis by Liao included only one of the new studies due to the search strategy being curtailed at July 1, 2016. For trisomy 21 it reported a sensitivity of 0.99 (95%CI 0.92, 1.00) and a specificity of 1.00 (95%CI 0.99, 1.00) based on 10 studies, equating to 2093 tests. However, it included case-control studies, and we cannot exclude the possibility of overlapping cohorts in 2 studies, potentially leading to double counting 323 participants, hence the results should be interpreted with caution.
The reviews by Gil and Liao demonstrate that the number of tests evaluated in twins is still substantially lower than in singletons (1135 and 2093 vs 226,995 tests, respectively). Even if the new 627 tests are added to results by Gil, the total number of trisomy 21 affected cases evaluated is relatively small (30/1762) and therefore the results from the twin cohort studies need to be interpreted with caution.
Trisomy 18 and Trisomy 13 Test Accuracy ( Tables 2 and 3 )
Studies using cfDNA NIPT to screen for trisomy 18 and trisomy 13 involved even smaller cohorts. Only 3 studies in the review by Gil 2017 evaluated trisomy 18 , and 2 studies assessed trisomy 13 , therefore, a meta-analysis was not performed. Unfortunately, placental tissue was not available for analysis in the 2 false-negative trisomy 18 cases . In the study by Huang the false negative result occurred in a discordant MZ twin whereby karyotyping confirmed 1 normal twin and 1 with trisomy 18, whereas NIPT reported a normal pregnancy. The authors state that discordant MZ twins are rare, and that this may be due to “trisomic rescue” or a postzygotic event which will be discussed in the section on “Technical Issues With cfDNA NIPT for Aneuploidy Screening in Twins.” The false negative result in Sarno was also in a discordant twin pregnancy, although the authors state the pregnancy was DC, they do not report the zygosity. When Canick et al. examined fetal fraction, they found that the lowest fetal fraction of a twin cohort ( n = 24 tests) was in a DZ pregnancy discordant for trisomy 13. The three studies published later reported lower test accuracy for detecting trisomy 18 and trisomy 13, as compared to trisomy 21, but the results are difficult to interpret due to the low number of cases examined. Two of the three studies each reported 1 case of trisomy 18 both of which were accurately detected by cfDNA NIPT. The study by Du reported 1 false positive result for trisomy 13 and had no trisomy 13 cases in the cohort, and Le Conte , who also included the samples from Benachi reported one true positive result for trisomy 13 and no false positive or false negative results. The review by Liao finally also reported lower test accuracy for trisomy 18 with a sensitivity of 0.85 (95%CI 0.55, 0.98) and specificity of 1.00 (95%CI 0.99, 1.00) based on 5 studies, equating to 1167 tests, 13 cases of trisomy 18. The results of their trisomy 13 meta-analysis reported a sensitivity of 1.00 (95%CI 0.30, 1.00) and specificity of 1.00 (95%CI 0.99, 1.00) based on 3 studies, equating to 605 tests, 3 cases of trisomy 13.
| Author | Total Tests in Cohort | T18 Cases | Sensitivity (95%CI) | Specificity (95%CI) |
|---|---|---|---|---|
| Huang | 187 | 2 | 0.50 (0.01, 0.99) | 1.00 (0.98, 1.00) |
| Benachi a | 7 | 0 | Not estimable | 1.00 (0.59, 1.00) |
| Sarno | 417 | 4 | 0.75 (0.19, 0.99) | 1.00 (0.99, 1.00) |
| Fosler | 115 | 1 | 1.00 (0.03, 1.00) | 1.00 (0.97, 1.00) |
| Le Conte | 420 | 1 | 1.00 (0.03, 1.00) | 1.00 (0.99, 1.00) |
| Du | 92 | 0 | Not estimable | 1.00 (0.96, 1.00) |
| Author | Total Tests in Cohort | T13 Cases | Sensitivity (95%CI) | Specificity (95%CI) |
|---|---|---|---|---|
| Benachi a | 7 | 0 | Not estimable | 1.00 (0.59, 1.00) |
| Sarno | 417 | 1 | 0.00 (0.00, 0.97) | 1.00 (0.99, 1.00) |
| Fosler | 115 | 0 | Not estimable | 1.00 (0.97, 1.00) |
| Le Conte | 420 | 1 | 1.00 (0.03, 1.00) | 1.00 (0.99, 1.00) |
| Du | 92 | 0 | Not estimable | 0.99 (0.94, 1.00) |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
