Adverse outcome following selective termination of presenting twin vs non‐presenting twin

Data are lacking on the impact on pregnancy outcome of the position of the abnormal fetus in a discordant twin pregnancy undergoing selective termination (ST). Tissue maceration post ST of the presenting twin may lead to early rupture of membranes, amnionitis and preterm labor. The aim of this study was to evaluate pregnancy complications and outcome following ST of the presenting vs non‐presenting twin.


INTRODUCTION
Multifetal pregnancy is associated with increased maternal and neonatal morbidity and mortality, with elevated risk of preterm delivery and pregnancy loss [1][2][3] . In addition, the rate of chromosomal and structural fetal malformation is higher in multifetal pregnancy compared with singleton [4][5][6] . Approximately 1-2% of twin pregnancies face management dilemmas following the diagnosis of a discordant anomaly affecting only one fetus 7,8 . In the case of a major fetal anomaly, parents should be counseled on management options for the pregnancy, which include expectant management and, if local legislation permits, selective termination (ST) 9 . ST is an invasive procedure with an inherent risk of complications, including loss of the healthy cotwin 10,11 . While ST is usually performed without delay in early pregnancy, the timing of midtrimester ST is controversial. Some authors suggest delaying ST into the third trimester to increase the chance of survival of the unaffected twin, should preterm delivery occur 12 . The decision to postpone ST to late pregnancy is complex and should take into consideration, among other factors, the nature of the fetal anomaly (e.g. an anomaly causing polyhydramnios), the patients' obstetric history (e.g. early deliveries) and the course of the present pregnancy (e.g. premature contractions). Data are lacking on whether the position of the abnormal fetus in the uterine cavity affects the course of pregnancy and outcome post ST. Tissue maceration of the presenting twin following ST may lead to early rupture of membranes, amnionitis and preterm birth of the unaffected twin. This study aimed to compare outcomes following ST of the presenting vs non-presenting twin and to shed light on the optimal timing of ST.

METHODS
This was a multicenter retrospective study of patients attending one of seven maternity hospitals in Israel between 2007 and 2021. Approval for the study was obtained from the local ethics committee (decision number: 0085-21-WOMC). Dichorionic diamniotic twin pregnancies diagnosed with a major discordant anomaly (structural or genetic) that underwent ST of the affected fetus were included. Cases were classified according to the position of the affected fetus at the time of the procedure: presenting vs non-presenting. The presenting twin was defined as the twin most proximal to the cervix.
Dating and chorionicity were based on first-trimester ultrasound findings. The primary outcome was a composite of early complications following ST, including infection, preterm prelabor rupture of membranes (PPROM) and pregnancy loss. Secondary outcomes were gestational age (GA) at delivery, the interval between procedure and delivery and overall survival. Cases with missing data on the position of the affected fetus or neonatal outcome were excluded from the study.
Israeli legislation requires institutional committee approval for all terminations of pregnancy. When a severe fetal anomaly is detected after 23 weeks of gestation, permission to terminate the pregnancy is granted if the probability of disability is ≥ 30%, as assessed by a specialist counselor.
Following local guidelines in all hospitals, only fetal medicine experts with experience in ultrasound-guided procedures performed ST in an outpatient setting. The technique, risks and outcomes of ST were discussed with the patients before written informed consent was obtained. During the discussion, parents were counseled on the options regarding the timing of ST. They were informed that having the procedure as early as possible carried a risk of pregnancy loss (< 24 weeks) or preterm delivery (≥ 24 weeks), and that delaying the procedure, while mitigating against the aforementioned complications, could result in spontaneous delivery before ST was performed.
Feticide was performed by intracardiac administration of 15% potassium chloride solution via a needle until asystole was achieved. Prophylactic antibiotics were not administered routinely in the participating medical centers, with the exception of Chaim Sheba Medical Center at Tel Hashomer, Ramat Gan, in which amoxicillin was given 1 h prior to the procedure. Moreover, amnioreduction was not performed routinely in any of the centers. The absence of fetal cardiac activity was confirmed on ultrasound 1 h after the procedure.

Data collection
Data were obtained from computerized medical records. Maternal demographic characteristics included age, pregravid body mass index, parity, use of assisted reproductive technology, smoking status and previous preterm birth. Procedure-related data included indication for ST (structural or genetic), GA at which ST was performed (early, < 18 weeks; intermediate, 18-24 weeks; late, > 24 weeks) 12 . Postprocedure complications included infection (defined as clinical chorioamnionitis), PPROM and pregnancy loss. Only complications occurring within 3 weeks after the procedure were considered likely to be related to ST 10 . Delivery and neonatal characteristics included GA at delivery, preterm delivery (defined as delivery at < 37 weeks' gestation), mode of delivery, neonatal birth weight, Apgar score and neonatal intensive care unit (NICU) admission. Overall neonatal survival was defined as survival at discharge.

Statistical analysis
Categorical variables are presented as n (%). The distribution of continuous variables was evaluated using histograms. Continuous variables are reported as mean ± SD or median (interquartile range). The chi-square test and Fisher's exact test were used to compare categorical variables, while the independent samples t-test and Mann-Whitney U-test were used to compare continuous variables. Variables that were associated significantly with the primary outcome were included in the multivariable analysis, for which logistic regression was used. All statistical tests were two-sided and P < 0.05 was considered statistically significant. SPSS Statistics v28 was used to perform the analysis (IBM Corp., Armonk, NY, USA).

RESULTS
Overall, 253 dichorionic diamniotic twin pregnancies underwent ST during the study period, of which 190 met the inclusion criteria. Of those, the non-presenting twin was terminated in 117 (61.6%) cases and the presenting twin in 73 (38.4%). There was no difference in demographic characteristics between pregnancies in which the presenting vs non-presenting twin was reduced ( Table 1). The indications for ST are described in Table 2, according to the location of the reduced fetus. Although the most common indication for ST in both groups was structural abnormality, the prevalence of structural defects was higher in presenting twins and genetic defects in non-presenting twins.
Procedure-related characteristics and early complications are presented in Table 3. GA at ST was similar between groups (21.8 weeks vs 22.0 weeks; P = 0.891). The primary outcome (a composite of early postprocedure complications) and pregnancy loss occurred at a significantly higher rate following ST of the presenting vs the non-presenting twin (19.2% vs 7.7% (P = 0.018) Data are given as mean ± SD, median (interquartile range) or n (%). ART, assisted reproductive technology; BMI, body mass index; PTB, preterm birth. and 11.0% vs 3.4% (P = 0.037), respectively). No cases of pregnancy loss were reported when ST was performed after 24 weeks of gestation. Pregnancy and neonatal outcomes are summarized in Table 4. GA at delivery was significantly earlier following ST of the presenting compared with the non-presenting twin (34.4 weeks vs 37.4 weeks; P < 0.001), with the former group showing a higher rate of preterm birth (P < 0.001). Moreover, the interval between ST and delivery was significantly shorter in those pregnancies in which the presenting twin was terminated (P < 0.001). Mean neonatal birth weight was also lower in these pregnancies (P < 0.001); however, birth-weight centile Data are given as n (%) or n/N (%). *Trisomies other than 21, deletions, monosomies. †Non-heart or central nervous system (CNS) defect. FGR, fetal growth restriction.

708
Miremberg et al. and the incidence of small-for-gestational age were similar between groups. The rates of Cesarean delivery and NICU admission were higher following ST of the presenting twin. The overall survival rate in the entire cohort was 92.1%. There was a non-significant trend towards higher neonatal survival after termination of the non-presenting twin (94.9% vs 87.7%; P = 0.073). GA at reduction was not significantly associated with the composite of early complications following the procedure (P = 0.318). On multivariable regression analysis, the presenting twin was a significant risk factor (odds ratio (OR), 2.6 (95% CI, 1.0-6.4); P = 0.045) in predicting composite early complications following the procedure. Smoking was found to be associated significantly with the composite outcome (OR, 3.9 (95% CI, 1.0-15.2); P = 0.049), while indication for the procedure was not associated significantly (OR, 2.48 (95% CI, 0.8-7.8); P = 0.120).

DISCUSSION
This study found that ST of the presenting twin in a dichorionic diamniotic twin pregnancy was associated with an increased risk of short-term procedure-related complications, including pregnancy loss, as well as adverse perinatal outcome, including preterm birth, lower birth weight and NICU admission. In addition, the interval between ST and delivery was significantly shorter in pregnancies in which the presenting twin was terminated. Late ST performed after 24 weeks of gestation was shown to be a safe option, with no pregnancy losses occurring in our cohort, irrespective of fetal position.
This study highlights the importance of determining fetal position in twin pregnancy. A number of parameters are used to identify a particular fetus during dichorionic twin pregnancy: fetal sex, location of the placenta, position of the intervening membranes and location of the fetus in relation to the internal cervical os. The latter is an important marker, as both fetal sex and placental location may be identical, and the position of the separating membrane may not be constant. All markers, especially the spatial relationship between the fetus and internal cervical os, should be used to map fetal position during amniocentesis in multiple pregnancy. Accordingly, this information is used to identify a normal-appearing fetus with a severe genetic/chromosomal abnormality for termination at 3 weeks after amniocentesis or in late pregnancy, if the procedure is postponed.
When considering selective reduction from twin to singleton due to an obstetric, social or maternal indication, the fetus to be reduced is usually chosen based on technical considerations, such as which twin is more accessible to intervention. Most often, the non-presenting twin is chosen 10,13 . When faced with a diagnosis of a discordant anomaly and considering ST, the position of the affected fetus is predetermined and seems to affect the success of the procedure.
The reason for the unfavorable outcome following ST of the presenting twin is not apparent. A significant inverse association between GA at intrauterine demise and interval to delivery was found in a study on the effect of single fetal demise in twin pregnancy 14 . The authors showed a correlation between the amount of dead fetoplacental products and the interval to delivery. This study emphasizes the importance of the position of dead fetoplacental products. Tissue maceration with proximity to the cervix may be one of the reasons for increased risk of early complications, including preterm birth of the unaffected twin.
The timing of ST is still debated in the literature, with mixed results from different studies [15][16][17][18] . Recent literature emphasizes the safety of third-trimester ST, demonstrating a reduced risk of pregnancy loss. A retrospective study evaluated 147 dichorionic twin pairs undergoing ST due to a discordant anomaly and demonstrated a 100% survival rate when ST was performed at 30-32 weeks of gestation 12 . These data are in concordance with our findings, which showed a 100% survival rate following late ST in both study groups. In addition, a retrospective case-control study evaluated complications in twin pregnancies that had undergone late ST compared with matched singletons and twins 19 . ST in the late second trimester resulted in a higher rate of pregnancy loss compared with when the procedure was performed in the third trimester (0 cases of pregnancy loss). However, compared with singletons, third-trimester ST was associated with a higher rate of placental complications, while late second-trimester ST resulted in a cumulative incidence of placenta-related complications comparable to that of singletons. We support offering patients the option of late ST if a diagnosis of discordant anomaly is made during the second trimester. However, this decision should be discussed extensively with the parents. Premature contractions, ruptured membranes, sudden emergency delivery, technical difficulty in performing ST, increased maternal anxiety and the need for a team on call to perform urgent ST (and, sometimes, to be present during delivery) may result in both fetuses being delivered before ST is performed.
Several strengths of this study should be mentioned. First, to our knowledge, it is the first study with a relatively large sample size to address the question of fetal positioning during ST. Second, this was a multicenter study involving seven maternity hospitals, in which all procedures were performed by a few highly skilled experts trained in invasive fetal procedures. However, this study is not without limitations. First, the retrospective nature of this cohort study could decrease its accuracy. To keep the data as accurate as possible, we excluded cases with incomplete data and those for which the position of the reduced fetus was not mentioned clearly in the medical record. Moreover, the rate at which discordant anomalies were diagnosed in the presenting twin in our cohort was similar to that in previous reports 10,11 . Second, records of cervical length measurement were not available. Finally, this study lacks data on placenta-related complications and long-term neonatal outcome.
In conclusion, this study highlights the correlation between ST of the presenting twin and adverse pregnancy outcome. We believe that the position of the affected fetus should be considered when discussing management options for dichorionic diamniotic twin pregnancies with a discordant anomaly.