Cell‐free DNA fetal fraction in twin gestations in single‐nucleotide polymorphism‐based noninvasive prenatal screening

Abstract Objectives The performance of noninvasive prenatal screening (NIPS) for fetal aneuploidy in twin pregnancies is dependent on the amount of placentally derived cell‐free DNA, the “fetal fraction (FF),” present in maternal plasma. We report FF values in monozygotic (MZ) and dizygotic (DZ) pregnancies. Methods We reviewed FF in pregnancies at 10 to 20 completed weeks gestational age based on single‐nucleotide polymorphism (SNP)‐based NIPS where zygosity was routinely established in twin pregnancies. The cohort included 121 446 (96.3%) singleton, 1454 (1.2%) MZ, and 3161 (2.5%) DZ pregnancies. For DZ twins, individual FFs were measured. Results Combined FF for DZ and MZ fetuses were 35% and 26% greater than singletons, respectively. The individual FF contributions from each fetus in DZ twins were, on average, 32% less than singletons. FF in DZ twin pairs were moderately correlated (Pearson correlation coefficient.66). When a threshold of 2.8% FF was applied to define uninterpretable results, 1.7% (2102/121 446) of singletons, 0.8% (11/1454) of MZ pairs, and 5.6% (178/3161) of DZ pairs were uninterpretable. Conclusion For optimal aneuploidy NIPS in twin pregnancies, zygosity should be established and in DZ twins FF for both fetuses should be determined to identify those cases where results can be reliably interpreted.

FF contributions from each pregnancy may not be equal. Canick et al estimated that the average combined amount of FF in twin pregnancies was 35% higher than that in singletons, implying that the average FF for each individual twin is about 2/3 as much as in singleton pregnancies. 3 Struble et al 4 found that monozygotic (MZ) twins had a median FF of 14% which can be compared with 13% for singletons measured using the same method. 5 The median for the twin with the lower FF in DZ twins was 7.9%. 4 To ensure that a chromosomally abnormal twin would nearly always be detected, setting a minimum requirement for the lower FF was proposed when determining whether the test could be interpreted. Using the same methodology, Sarno et al 6 reported that twin pregnancies (DZ and MZ) had a median FF of 8% compared with 11% for singletons. When 4% was used as the minimum FF required, the rate of noninformative results (no-call rate) was 9% to 11%. 6,7 Detailed protocols and no-call rates have not been reported for other counting-based NIPS methodologies where the individual FFs for each twin are not separately determined.
The recent availability of a single-nucleotide polymorphism (SNP)based NIPS to distinguish DZ and MZ twins and measure the individual FFs in DZ pregnancies 8 provides an opportunity to further characterize the distribution of FF in MZ and DZ twins. The objective of this study was to provide this information based on a large cohort of DZ and MZ twin pregnancies. Presence or absence of twins was based on information provided by the referring physician. In rare instances, the SNP analysis indicated an undocumented DZ twin pregnancy, and these cases were included only if confirmed by the referring physician. Testing was not offered when there was a known fetal death prior to the time of NIPS. Zygosity assessment was performed by identifying the number of genetically distinct individuals represented in the maternal plasma sample; two distinct sources indicated MZ twins (one pair of genetically identical twins and the mother) while three distinct sources indicated DZ twins (two genetically distinct twins and the mother). 8  Where appropriate, data evaluation included descriptive statistics, two-sided t-test, and one-way analysis of variance (ANOVA) at the 5% significance level. No Bonferroni correction for multiple tests was applied to minimize type II (false negative) associations.

| METHODS
The study was based on retrospective analysis of deidentified data and therefore fell under an IRB exemption protocol, as determined by Ethical & Independent Review Services, Corte Madera, CA (E&I) protocol #: 17113-02.
The demographics are presented in Table 1. There were small but statistically significant differences in gestational ages when blood was drawn for NIPS for MZ, DZ, and singleton pregnancies (P < .0001). As expected, maternal age was somewhat higher in DZ pregnancies compared with MZ and singleton pregnancies (P < .0001).
Women with DZ pregnancies were also significantly heavier than women with MZ or singleton pregnancies (P < .0001). Figure 1A shows the relationship between FF and gestational age in twin pregnancies; the increase in FF with gestational age was very similar in MZ and DZ pregnancies and parallels that seen in singletons. The

What's already known about this topic?
• Adequate cell-free fetal DNA (fetal fraction [FF]) is essential for noninvasive prenatal screening.
• FFs in twin pregnancies may be higher or lower than that found in singleton pregnancies.
What does this study add?
• In this large series of twin pregnancies, the average total FF was higher than for singletons but the per fetus FF was lower.
• There can be large differences in the two FFs in dizygotic twin pregnancies.
• Optimal prenatal aneuploidy screening in twin pregnancies requires information on both zygosity and the individual FFs.
relationship between FF and maternal weight was also proportionate in each of these three groups of pregnancies ( Figure 1B).  This would correspond to a reduction in the detection rate of approximately 3.6% and 1.6%, respectively.
We also considered whether the two FFs for DZ twin pairs were correlated. Figure S2 shows a plot of the paired FFs when each twin pair is randomly assigned as twin A or twin B. The Pearson linear correlation coefficient, R, was.66 (probability that the correlation was not significantly different from 0, P < .0001).  Figure S1). Previous studies have also noted that relatively high proportions of DZ pregnancies with FF below 4%, although the proportions were not as high as those reported here. 6,7   [13][14][15] In twin pregnancies, we observed associations between FF with gestational age and with maternal weight ( Figure 1A, B) that paralleled observations in singleton pregnancies. We also noted that maternal age in DZ pregnancies were significantly greater, which may be a result from multiple large developing follicles associated with advancing maternal age resulting in increasing DZ twinning in these women. 16 Increasing maternal weight and body mass index have been associated with DZ twinning, 17 consistent with our findings. We did not consider the extent to which assisted reproductive technology may have been used and that may further explain the finding of increased maternal weight for DZ cases because these women may be older. 14, 17 We also did not distinguish between test-positive and test-negative cases or evaluate detection rates and false-positive rates. It is also possible that some MZ pregnancies were not identified or reported to the laboratory and these would remain unrecognized by the laboratory testing.

| DISCUSSION
In summary, because (a) FF differ between MZ and DZ pregnancies, (b) the level of FF needed for reliable testing will differ depending on zygosity, and (c) the prior risk used in screening algorithms differs depending on zygosity, it is advantageous to carefully consider both FF and zygosity at the time of NIPS. Accurate measurement of the individual FFs for DZ pregnancies will identify those cases where a high confidence result can be provided.

CONFLICTS OF INTEREST
Zachary Demko and Paul Billings are employees of Natera, Inc and hold stock in the company. Kimberly Martin and Peter Benn are consultants to Natera and hold stock options. Herman Hedriana is a former employee of Natera and does not own stock or hold stock options. Daniel Salzman serves on the advisory board of Natera; he is not an employee and does not receive any financial compensation from Natera.

FUNDING SOURCES
This work was funded by Natera, Inc.