Fetal Fraction Increase Linked to Birthweight: Expert Insights

Introduction The incidence of twin pregnancies has increased exponentially in recent years due to the growing demand for assisted reproductive technologies (ART). Twin pregnancies are more likely to be affected by intrauterine growth restriction (IUGR), preterm birth, and perinatal loss. Selective IUGR (sIUGR) is unique to twin pregnancies, whereby one twin appears to be compromised while the other grows normally. Twin pregnancies complicated by sIUGR are at increased risk of perinatal mortality and morbidity.1 Gestational age at delivery was found to be significantly earlier, and the rate of admission to the neonatal unit was higher when compared to those without sIUGR.2 The incidence of sIUGR is estimated as 10–20% of twin pregnancies.3 Monochronic (MC) twins are more likely to be affected by sIUGR than dichorionic (DC) twins, with an incidence of 19.7% compared to 10.5% for the latter.4 The recent International Society of Ultrasound in Obstetrics and Gynecology (ISUOG) guidance defines sIUGR as a condition in which the estimated fetal weight (EFW) of one twin is less than the 10th centile and an inter-twin EFW discordance greater than 25%. In addition, a discordance cut-off of 20% seems acceptable to distinguish pregnancies at increased risk of adverse outcome.5 Identifying those pregnancies likely to be affected by either sIUGR or birthweight discordant early can optimize counseling, the timing of interventions, and may, in the future, allow trials of preventative treatments. Several parameters identifiable in early pregnancy have been shown to be associated with a risk of sIUGR developing later during that pregnancy. Abnormal cord insertion has been found to be associated with sIUGR in twins. However, over 55% of twin pregnancies have been shown to show discordant cord insertions.6 A more specific screening test is urgently needed in clinical practice so as to provide the optimal treatment options for patients as well as health professionals. Current screening approaches for sIUGR primarily rely on ultrasound biometry, Doppler studies of umbilical and middle cerebral arteries, and assessment of cord insertion sites. However, these methods have limitations in early prediction. The onset and prognosis of sIUGR often correlate with differences in placental pathology. sIUGR in MC twins is thought to be caused by an unequal sharing of the placenta and distribution of blood through placental anastomoses,7 whereas in DC twins, this can result from placental insufficiency in one of the placentas.8 The villous trophoblast of the placenta undergoes continuous turnover throughout gestation, resulting in released apoptotic debris and cell-free fetal DNA (cffDNA) into the maternal circulation. It has been suggested that concentrations of cffDNA also may reflect placental pathology. The proportion of cffDNA in the maternal circulation is commonly termed fetal fraction (FF). The FF has been linked to adverse perinatal outcomes, such as IUGR in singleton pregnancies.9 Here, we tested the hypothesis that the FF in twin pregnancies would reflect birthweight discrepancies in the newborn. The biological basis for FF as a potential marker of placental function lies in the process of trophoblast turnover. The villous trophoblast undergoes continuous apoptosis throughout gestation, releasing cell-free fetal DNA into the maternal circulation. Placental hypoxia, inflammation, and oxidative stress—hallmarks of placental insufficiency—can accelerate trophoblast apoptosis and increase the release of cffDNA, thereby elevating FF. Moreover, variations in placental mass, vascular abnormalities, and differences in placental perfusion may all contribute to FF variation[ref]. Therefore, the primary objective of this study was to investigate whether FF in twin pregnancies can reflect birthweight discrepancies and serve as an early indicator for identifying pregnancies at risk of birthweight discordance and sIUGR. Materials and Methods Study Design We conducted a retrospective cohort study of all twin pregnancies undergoing cffDNA screening for aneuploidy and who delivered at International Peace Maternity and Child Health Hospital in Shanghai, China, between January 2018 and December 2021. Cases with chromosome abnormalities were excluded from this study. Other exclusion criteria were the presence of twin-to-twin transfusion syndrome (TTTS), monoamniotic twins, twin anemia-polycythemia sequence (TAPS), and congenital, structural and genetic malformations of the fetus. In order to ensure the placentas would be intact, only MC twin gestations delivered through cesarean section were included in this study. This study was approved by the Ethics Review Board of the International Peace Maternity and Child Health Hospital Affiliated to the Shanghai Jiao Tong University School of Medicine (No. GKLW2016-35) and was conducted in accordance with the Declaration of Helsinki. As this study is retrospective, it will not adversely affect the health of patients, nor will it involve the privacy and personal identity information of patients. The Ethics Committee of the International Peace Maternity and Child Health Hospital Affiliated to the Shanghai Jiao Tong University School of Medicine has waived the requirement of informed consent of patients. Following the established protocol [PMID: 25598039], 5 mL of peripheral blood was collected from each participant into EDTA-primed tubes and processed within 8 hours via double centrifugation to extract plasma. Library construction was performed using standard protocols with barcode tracking, and 24 libraries were pooled for multiplex sequencing on the Illumina HiSeq2000 platform using 36-cycle single-end sequencing, generating 35-base reads after trimming. Low-coverage whole-genome sequencing was performed at approximately 0.1–0.2× genome coverage (4–10 million reads per sample), with reads aligned to the human reference genome (hg18, NCBI build 36). Aneuploidy classification was conducted using a binary hypothesis t-test and logarithmic likelihood ratio (L-score), while the FCAPS (Fetal Copy-number Analysis through Maternal Plasma Sequencing) algorithm based on binary segmentation was implemented for CNV detection. Quality control included a minimum fetal fraction (FF) threshold of 3.5%, with repeat blood sampling required in 2.18% of cases due to quality control failure, assay failure, or low FF, resulting in a test failure rate of 0.098%. FF was estimated using an SNP-based method with parent-specific homozygous loci (♀AA♂BB) calculated as FF = 2d(B)/[d(A)+d(B)], where d represents the sequencing depth of each allele. Data Collection and Definition We obtained the maternal demographic characteristics, FF, obstetric outcomes, and neonatal outcomes from the medical records. These included maternal age, gravidity, parity, maternal weight on the non-invasive prenatal test (NIPT) testing day, the method of conception, chorionicity, gestational hypertension, gestational diabetes, thyroid disease, gestational age at delivery, and neonatal birthweight. The variables required for this analysis were documented as part of the standard clinical protocol, resulting in no missing data for the final analyses. All women with twin pregnancies were offered a scan to determine chorionicity as well as first-trimester nuchal translucency (NT) at 11+0-13+6 weeks. DC pregnancies were confirmed when the ultrasound assessment clearly indicated two placentas. The twin peak sign was used to distinguish chorionicity if only one placenta was visualized. MC twin pregnancies were identified either in the first trimester or early second trimester of pregnancy using the following ultrasonographic criteria: (I) the presence of a single placenta, (II) the presence of a thin dividing membrane, and (III) the absence of a twin peak (lambda) sign. MC was also confirmed by obstetricians through postpartum examination of the placenta (ie the presence of a single placenta with inter-twin anastomoses).10 Following this, a thorough genetic consultation was offered to all future parents of the twin cohort. The characteristics of the tests and potential advantages and disadvantages of the different modalities of prenatal genetic testing were discussed. Parents were also informed that the NIPT included information regarding the detection of only T21, T18 and T13, and that the evidence for using NIPT in twin pregnancies is less than it is in singletons. Sensitivity for T21 is 99%. The performance metrics for T13 and T18 are lower than singletons. Therefore, this offered limited information compared with invasive diagnostic testing. We used the platform offered by cffDNA-BGI-Health (Shenzhen, China). The screening was performed specific to our clinical site. FF was defined as the proportion of cffDNA in the maternal circulation. Birthweight discordance was calculated as the difference between the newborns’ weights of the twins divided by the weight of the larger twin as follows: [(body weight of the heavier twin − body weight of the lighter twin)/ (body weight of the heavier twin)] × 100%.11 Selective fetal growth restriction (sFGR) was defined as the presence of (I) a birthweight discordance of >25% and (II) a twin with birthweight less than the 10th percentile. In our birth center, participants underwent a 75-gram oral glucose tolerance test (OGTT) between 24 and 28 weeks of gestation. Gestational diabetes mellitus (GDM) was diagnosed if any of the following cut-off values were exceeded: fasting plasma glucose of 5.1 mmol/L (92 mg/dL), 1-hour glucose of 10.0 mmol/L (180 mg/dL), or 2-hour glucose of 8.5 mmol/L (153 mg/dL) according to IADPSG guidelines. Statistical Analysis Categorical variables were described by percentage, and continuous variables were described by means and standard deviations. The Kolmogorov–Smirnov test was used to test the normality of the continuous variables. The group differences between DC and MC twins were examined using the chi-squared, Student t and Mann Whitney U-tests where appropriate. The linear regression models were used to test the association of FF with birthweight difference. Logistic regression was used to obtain the odds ratios (ORs) and 95% confidence intervals (CIs) and to examine the effects of FF on birthweight discordance of 20 and 25% and sFGR. The pROC package was used to perform a receiver operator characteristics (ROC) curve analysis for the FF and birth outcomes that had significance in the above regression analyses. This analysis aimed to determine the optimum cut-off points, along with the corresponding specificity, sensitivity, and area under the ROC curve (AUC). The Youden index was utilized to identify the optimal cut-off point, maximizing the sum of sensitivity and specificity. Then the FF was converted into a categorical variable according to the optimal cut-off points. The association of the FF (categorical variable) on birth outcomes was then assessed using logistic regression. The multivariate analyses between the FF and birth outcomes were carried out using the multiple logistic regression and multiple linear regression models. The models were subsequently adjusted for maternal age, weight, primipara, history of abortion, chorionicity (DC and MC), pregnancy via ART, and physical conditions (gestational hypertension and gestational diabetes). Finally, we calculated the statistical power of linear and logistic regressions based on the sample size, a significance level of α=0.05, and corresponding regression parameters. Statistical power is generally recommended to reach the level of 0.8.12 All analyses were performed with the Statistical Package for the Social Sciences (version 24; SPSS Inc, Chicago, IL) and R statistical software (version 3.5.1; http://www.R-project.org). Significance was defined as a 2-tailed probability value of <0.05. Results Study Population A total of 237 twin pregnancies who had normal NIPT results and delivered ≥28 weeks were enrolled. Five cases of intrauterine death of one fetus and 8 structural abnormalities in one or both fetuses were excluded from this study. Finally, 224 twin pregnancies were included in our analyses. The proportions of DC and MC twins were 77.6% (174) and 22.3% (50), respectively. Compared to the MC group, the DC group had a higher maternal age (p=0.006), longer gestational weeks at delivery (p<0.001), larger birthweight of the fetuses (p=0.001) and higher percentage of pregnancies conceived by ART (p<0.001). The incidence of preterm delivery was higher in the MC group (p=0.004). There was no difference in other pregnancy complications such as gestational hypertension, gestational diabetes and thyroid diseases (Table 1). The Association of the FF with Birthweight Discordance and sIUGR As shown in Table 2, the adjusted model demonstrated that the FF was positively correlated with the difference of birthweight (β=0.365, 95% CI: 0.122~0.608). That is, for every 1% increase in the FF, the difference of the birthweight between twins increased by 0.365%. The results of logistic analysis are shown in Table 3, indicating that a higher FF was significantly associated with an increased risk of birthweight discordance of 20% (adjusted OR: 1.073, 95% CI: 1.009~1.142), 25% (adjusted OR: 1.092, 95% CI: 1.006~1.185) and sIUGR (adjusted OR: 1.130, 95% CI: 1.038~1.231). The optimum cut-off points for the FF in birthweight discordance and sIUGR were obtained with a ROC analysis (Table 4 and Figure 1). Using ROC analysis, we obtained the optimum cut-off points of the FF ≥ 11.790, ≥ 14.800 and ≥ 14.800 for birthweight discordance of 20% and 25% and sIUGR, respectively. For birthweight discordance of 20%, the AUC was 0.591, the sensitivity was 52.9% and the specificity was 66.3% (p=0.017). The AUC of sIUGR was 0.620, with 37.5% sensitivity and 86.5% specificity (p=0.024). The FF was further segmented at 11.79 and 14.8. As shown in Table 5, compared with women with a FF of < 11.790, there was a 1.091-fold higher risk of birthweight discordance of 20% (adjusted OR: 2.091, 95% CI: 1.218~3.591) among those with FFs ≥11.790, and significantly increased risks of birthweight discordance of 25% (adjusted OR: 3.045, 95% CI: 1.297~7.149) and sIUGR (adjusted OR: 3.526, 95% CI: 1.443~8.618) among these with FFs ≥14.800. Table 5 Fetal Fraction, Birthweight Discordance and sFGR: Logistic Regression Analysis Discussion In this study, we assessed the association between the FF and birthweight discordance and sIUGR. We demonstrated that FF was positively correlated with the difference of birthweight between twins within the pregnancy. For every 1% increase in the FF, the difference of birthweight between the twins increased by 0.365%. A higher FF was significantly associated with the increased risk of birthweight discordance of 20% and 25% as well as sIUGR. Using ROC analysis, we obtained the optimum cut-off points of FF ≥ 11.790, ≥ 14.800 and ≥ 14.800 for birthweight discordance of 20% and 25% and sIUGR, respectively. For a birthweight discordance of 20%, the AUC was 0.591, the sensitivity was 52.9% and the specificity was 66.3% (p=0.017). The AUC of sIUGR was 0.620, with 37.5% sensitivity and 86.5% specificity (p=0.024). When this was compared with women with an FF of < 11.790, there were a 1.091-fold higher risk of birthweight discordance of 20% among those with FFs of ≥11.790, and there significantly increased risks of birthweight discordance of 25% and sFGR among those with FFs of ≥14.800. Prenatal screening in trisomy 21 in twin pregnancies, based on measurements of nuchal translucency and first-trimester biomarkers (free b-subunit of the human chorionic gonadotrophin and placenta-associated pregnancy protein A), is less efficient.13 Due to its superior performance in singleton pregnancies, NIPT is now part of the prenatal screening scheme for twin pregnancies in many countries, either as first- or second-line tests.14 cffDNA is known to originate from placenta. During normal gestation, cffDNA undergoes a progressive rise, making up approximately 13% of all the cfDNA in the maternal plasma towards the end of pregnancy and then this rapidly drops to undetectable levels post-partum.15 The release of cff-DNA is related to the size of the placenta as well as the rate of trophoblast apoptosis.16 It is reported that cffDNA induces release of type 1 interferons, IFN-β and IFN-α, and other pro-inflammatory mediators.17 Administration of CpG oligonucleotides was shown to cause gestational hypertension and increased vasoconstriction in pregnant rat.18 It is important to note that elevated FF likely reflects, rather than directly predicts, placental dysfunction. The increase in FF may act as a pro-inflammatory trigger with the unique capability of initiating inflammatory cascades. Studies shown that oxidative stress causes trophoblast apoptosis and increased release of cffDNA.19 Other inflammatory factors, such as doxorubicin and high-mobility group box protein-1, can also affect the release of cffDNA. However, lipopolysaccharides did not affect the release of cffDNA in either early or late pregnancy,20 which indicated that cffDNA played a role in sterile inflammation rather than during on-going infections progress. Sterile inflammation can contribute to placental dysfunction and this can also result in pregnancy complications.21 From a clinical perspective, FF measurements obtained during routine NIPT screening could potentially serve as an early risk indicator for birthweight discordance and sIUGR in twin pregnancies. This information could complement ultrasound monitoring and help identify pregnancies that may benefit from more intensive surveillance. However, the moderate predictive accuracy suggests that FF should be used in conjunction with, rather than as a replacement for, established ultrasound-based assessment methods. Numerous associations between pregnancy complications and cffDNA plasma concentrations have been made, with the goal of predicting negative outcomes earlier and non-invasively. Higher concentrations of cffDNA during second trimester were found in pregnancies complicated with pre-eclampsia, preterm delivery and IUGR. Peek cffDNA levels have been shown 3 weeks prior to the onset of symptoms. In twins with birthweight discordant, we were able to see an increase in cffDNA levels which was represented by the FF. Intrauterine surgeries such as laser ablation to treat TTTS can also result in increased cffDNA levels.22 Therefore, the increase of cffDNA in twins with birthweight discordance and sFGR may be a consequence of the repair process. The incidence of pre-eclampsia in DC pregnancies affected by discordant growth is as high as 37.5%, compared to 21.9% in MC sFGR pregnancies.23 Hence, sIUGR in DC pregnancies was known to be a result of placental insufficiency affecting only one fetus.24 After correcting for both pre-eclampsia and gestational diabetes, the increase in cffDNA levels is still associated with the incidence of birthweight discordance and sIUGR, which showed that there may be other pathological mechanisms involved. sIUGR in MC pregnancies maybe related to discordance in the placental share, and the greater the discrepancy in the placental mass in MC pregnancies, the greater the discordance that would be observed. Placentae affected by discordant sharing have more tightly connected fetal circulations with more and larger arterio-arterial anastomoses. These may be beneficial to the growth restricted fetuses, allowing their co-twin to compensate for a degree of their placental insufficiency. However, this close connection can also present special risks to the appropriately grown fetus, which is vulnerable to sudden changes in the smaller co-twin’s blood pressure. While unequal placental sharing is the underlying cause of most sFGR in MC pregnancies, their interdependent feto-placental circulations have an important role in the eventual prognosis and perinatal outcome. Glucose transporters, which are essential for glucose transport at maternal-fetal interface, have also been found to be elevated in sFGR, especially in cases with abnormal umbilical arterial Doppler measurements. This may be due to hypo-perfusion. We suspect that the increased levels of cffDNA in MC twins with sIUGR is a consequence of apoptosis in such a hypoxic environment. Further, hypermetabolism may also be a reason for elevated cffDNA since the placenta attempts to increase glucose transport to the fetal circulation as an adaptive response. However, this study has several limitations that warrant consideration. First, the retrospective design limits our ability to establish causality and may introduce selection bias. Second, this was a single-centre study conducted in a Chinese population, which may limit the generalizability of our findings to other ethnic groups and healthcare settings. Third, the relatively small sample size of monochorionic twin pregnancies may have limited our power to detect chronicity-specific associations. Fourth, we did not have information on placental pathology or detailed ultrasound parameters throughout pregnancy, which would have provided valuable insights into the mechanisms underlying the association between FF and the fetal growth. Conclusions A high FF is associated with poor placental development and dysfunction, which can be used as potential indicators to predict birthweight discordance and sIUGR. However, given the modest predictive accuracy, FF should be used in conjunction with, rather than as a replacement for, established ultrasound-based surveillance methods. Future multi-centre prospective studies are needed to validate these findings and establish standardized FF thresholds for clinical decision-making.