Non‐invasive prenatal screening for fetal triploidy using single nucleotide polymorphism‐based testing: Differential diagnosis and clinical management in cases showing an extra haplotype

Abstract Objective An extra haplotype is infrequently encountered in single nucleotide polymorphism(SNP)‐based non‐invasive prenatal testing (NIPT) and is usually attributed to an undetected twin or triploidy. We reviewed a large series to establish relative frequencies of these outcomes and identify alternative causes. Methods In 515,804 women receiving NIPT from September 2017 through March 2019, all results with an extra haplotype were reviewed. Known viable and vanished twin pregnancies were excluded. For positive cases, pregnancy outcome information was sought. Results Of 1005 results with an extra haplotype (1 in 513), pregnancy outcome was available for 773 cases: 11% were confirmed or suspected triploidy; 65% to vanished twin; 10% with pregnancy loss. Rare explanations included complete mole, chimera, undisclosed donor egg pregnancy, maternal organ transplant and one instance of maternal neoplasm. Among triploid cases that were detected and independently confirmed, 23/27 (85%) were diandric. Conclusion SNP‐based NIPT, with detection of an extra haplotype, is 11% predictive of triploidy. For results with an extra haplotype, ultrasound is recommended to establish viability, evaluate for twins (viable or vanished), and detect findings consistent with triploidy. Review of patient history, serum screening, and ultrasound will reduce the number of CVS or amniocenteses necessary to confirm a diagnosis of triploidy.

� In cell-free DNA prenatal screening, the presence of an extra haplotype could be indicative of either twin pregnancy or triploidy

What does this study add?
� Outcomes for a large series of prenatal screening tests where an extra haplotype was identified in cell-free DNA � Complete moles, chimeras, transplantation, and donor egg are additional explanations for extra haplotypes � We discuss prenatal management when an extra haplotype is observed in cell-free DNA

| INTRODUCTION
Triploidy is defined as the presence of three copies of each chromosome in a cell, instead of the usual two sets of homologs. In triploid fetuses, the extra set of chromosomes can be maternal (digynic) or paternal (diandric) in origin. On ultrasound examination, digynic triploidy is typically associated with severe fetal growth restriction and a small non-cystic placenta. In contrast, diandric triploidy often demonstrates normal fetal growth with multiple placental cysts. 1,2 For pregnancies with a diagnosis of triploidy, serial human chorionic gonadotropin (HCG) tests are recommended to ensure complete removal of residual trophoblastic tissue. 3,4 Ultrasound visualized fetal abnormalities in triploidy include open neural tube defects, ventral wall defects, syndactyly, and other anatomic abnormalities. 5 The incidence of triploidy at approximately 11-13 weeks gestation has been estimated to be approximately 1 in 4800 pregnancies (Supplemental Table 1), decreasing to <1:27,000 in the second trimester. 6 Rare cases (possibly mosaic or chimeric) can survive to the third trimester, and these usually result in stillbirth or neonatal death. 7 Although reports exist regarding multiple recurrences of digynic triploid pregnancies, 8 the overall risk for recurrence has not been established. For diandric triploid pregnancies, recurrence risk is approximately twice the general population risk. 9 Because of the risk for maternal malignant trophoblastic disease, early detection of triploidy is advantageous. 10 Conventional first and second trimester maternal serum biochemical tests can help screen for triploid pregnancies, 11,12 but in some countries serum testing is less widely utilized because of the increasing use of cell-free DNA (cfDNA) based non-invasive prenatal testing (NIPT) for autosomal trisomies, sex chromosome abnormalities and some other imbalances. Counting-based NIPT methods are unable to routinely detect triploidy because there is no proportional change in the number of DNA fragments across the different autosomes. As ACOG noted in a 2020 Practice Bulletin, 13 SNP-based NIPT can identify triploidy by detecting the presence of the additional haploid chromosome set. 14 However, the SNP pattern seen with triploidy can sometimes be difficult to distinguish from the pattern seen with dizygotic twins, particularly when the fetal fraction is low. 15 Curnow et al. 14 reported the outcomes for an initial cohort of 76 instances where there was SNP-based evidence for an extra haplotype in cfDNA. Curnow et al., also provided initial estimates for the proportions of these results attributable to vanished twins, viable twins, or triploidy. Since Curnow et al.'s report, pregnancies with two viable fetuses can be analyzed using a SNP-based NIPT. In this report, we present follow-up information for 773 pregnancies with extra haplotypes, collected since the introduction of twin NIPT. We also provide additional causes for the presence of extra haplotypes in cell-free DNA.

We reviewed the laboratory results for all women in the United
States who had SNP-based NIPT performed at Natera, Inc., from September 2017 through March 2019. To meet research compliance requirements, the study was limited to testing performed for women residing in the United States. The test methodology and algorithms to assess aneuploidy in singleton and twin pregnancies have been described. [16][17][18][19] Tests were excluded if there was a known vanished twin prior to testing. 13 Testing was also not indicated when the gestational age was less than 9 weeks, for triplet or higher multiple pregnancies, or if the patient was known to have used a donor egg.
Tests were also excluded if the sample arrived at the laboratory more than 8 days after blood collection, if insufficient blood volume (less than 13 ml), an incorrect collection tube was used, or if the sample was damaged, had hemolysis, or DNA degradation.
Results where the test requisition indicated a twin pregnancy were evaluated for aneuploidy according to an algorithm specifically designed for twin pregnancies. 19,20 All other tests were evaluated according to a standard algorithm for singleton pregnancies. When evidence of an extra haplotype was seen in a sample processed as a singleton pregnancy, results were not released as a formal written report until after an attempt to contact the referring clinical provider to confirm the pregnancy was singleton. For results where this posttest checking established the presence of viable twins, the results were reanalyzed using the SNP-based twin algorithm. Therefore, the only results with extra haplotypes included in this study were those in which the extra alleles were not attributable to a known viable twin at the time of reporting. Analysis was limited to tests with sufficient fetal DNA for aneuploidy risk assessment (>2.8% fetal fraction); cases identified as high risk for triploidy based on a fetal fraction-based risk assessment (FFBR) were excluded. 21 Results showing the unexplained presence of an extra haplotype were reported as "consistent with vanishing twin, unrecognized multiple gestation, or fetal triploidy," and these results were the subject of KANTOR ET AL.
-995 this study. Pregnancy outcome information was based on either unsolicited information provided by providers or through outreach to referring provider's offices by facsimile, telephone, or both. Follow-up was performed after the patient's expected date of delivery. Pregnancy outcomes were classified as: confirmed triploidy; suspected triploidy; confirmed vanished twin; suspected vanished twin; pregnancy loss; normal singleton; normal viable twin, or 'other explanations or multiple factors (see Table 1). Cases of confirmed triplody were based on cytogenetic or cytogenomic testing performed by independent laboratories. For these cases, we also requested information on the parental origin of the extra set of chromosomes, based on molecular genetic analysis and/or evidence of molar changes in the placenta by pathologic examination of placental tissue.
In results with confirmed or suspected triploidy, the measured fetal fractions were reviewed to determine whether digynic and diandric triploidy could be determined. The methods used to determine fetal fraction (FF) in SNP-based NIPT are based on the relative proportion of polymorphic alleles in cfDNA contributed by the fetus, with a maximum likelihood value computed from informative loci.
This calculation focuses on those chromosomes with a low likelihood of aneuploidy. When triploidy is present, the algorithm will be fitting trisomy data to a disomy SNP model. The FF value returned by the algorithm for a singleton pregnancy with triploidy is therefore not an accurate measure of the true proportion of fetal DNA present. Therefore, we refer to the FF value for triploid pregnancies that is calculated from the singleton pregnancy algorithm as 'algFF'.

| DISCUSSION
The discovery of an extra haplotype is a rare finding in SNP-based NIPT. We observed an extra haplotype 1 in every 513 results. The association with vanished twins and triploidy was confirmed, 14 and in addition, we identified a number of other rare etiologies for the finding.
Approximately 11% of cases were attributable to a triploidy.
Data from cytogenetic studies of spontaneous fetal loss have indicated that triploidy is one of the most common cytogenetic abnormalities seen, 23 and therefore, early NIPT should identify a proportion of these pregnancies. We estimated that in the population of pregnancies with a mean gestational age of 13 weeks 3 days, the frequency of diandric triploidy was approximately 1 in 5543 cases, somewhat higher than estimates made using serum and ultrasound screening (Supplemental Table 1). Conversely, the observed frequency of digynic triploidy, 1 in 31,874, was substantially lower than expected from serum and ultrasound screening. Causes for under ascertainment could be the low FF associated with digynic triploid pregnancy, and also the increased difficulty in detecting digynic triploidy, as compared to diandric, due to the identical nature of the extra haplotype and the background maternal cfDNA in the sample.
McKanna et al. 21 reported that digynic triploidy was 90-fold more common than expected in those referrals that had a 'no result' from NIPT, mostly attributable to low FF. This group of referrals were excluded from this study.
In our reporting of results with an extra haplotype, we did not routinely provide information about the parental origin of the additional alleles. This could potentially distinguish between twins and triploidy and also establish whether the triploidy is digynic or diandric. The combinations of SNPs present in maternal plasma in dizygotic twin pregnancies are complex, 15 and it can be difficult to distinguish dizygotic twin patterns from diandric triploidy. Our observations also showed that the two types of triploidy cannot be distinguished solely on the overall algFF of the pregnancy; we observed overlap in the values for the two types of triploidy. Despite these complexities, we believe that the algorithm used for allele interpretation can potentially be refined to further assist in the early identification of each type of triploidy.
KANTOR ET AL. -997 Our data confirmed that the most common explanation for an extra set of alleles is the presence of a vanishing twin, either confirmed or suspected by ultrasound, in approximately 65% of cases. It has been suggested that the occurrence of a demise of one twin later in pregnancy can be associated with an increased risk for preterm birth. 24 Identification of a vanishing twin pregnancy could explain abnormal maternal serum markers. 25,26 In rare instances, A strength of our study is that it is based on a large number of results from a single laboratory. A limitation is the incomplete followup. Results with follow-up with microarray analysis could miss low level mosaicism, and those with follow-up karyotype could miss same-sex chimerism. Also, some women could have had undetected cancer, however, based on our data, this explanation appears to be infrequent. We speculate that, relative to cases with follow-up, cases with no follow-up include a higher proportion of pregnancies experiencing loss (without chromosome analysis), and some of these could be attributable to triploidy. Estimates for the triploidy predictive value are therefore expected to be conservative. Our study would also undercount digynic triploidy, where very low FF often precludes any type of NIPT result. 21 With improvements in testing protocols that allow more reliable interpretation at lower FF, better detection of digynic triploidy can be anticipated. 30 In summary, detection of triploidy and other conditions associated with an extra haplotype is a secondary benefit of prenatal screening using a SNP-based NIPT. In conjunction with ultrasound, this testing allows early identification of a small set of high-risk pregnancies, some of which are also associated with risk to maternal health.