Early prognostic factors of outcomes in monochorionic twin pregnancy: systematic review and meta-analysis

BACKGROUND: Monochorionic twin pregnancies are high-risk, however at present, no screening test is available to predict which monochorionic twin pregnancy will develop complications. OBJECTIVE: We sought to assess ability of first-trimester pregnancy-related factors (ultrasound measurements, maternal characteristics, biomarkers) to predict complications in monochorionic twin pregnancies. DATA SOURCES: Data sources were MEDLINE, Embase, ISI Web of Science, CINAHL, the Cochrane Central Registration of Controlled Trials and Research Registers, and Google Scholar, from inception to May 12, 2017. Gray literature and bibliographies of articles were checked. STUDY ELIGIBILITY CRITERIA: Studies that reported ultrasound measurements, maternal characteristics, or potential biomarkers, measured in the first trimester in monochorionic-diamniotic twin pregnancies, where the potential prognostic ability between the variable and twin-twin transfusion syndrome, growth restriction, or intrauterine fetal death could be assessed, were included. STUDY APPRAISAL AND SYNTHESIS METHODS: Quality assessment was evaluated using the Strengthening the Reporting of Observational Studies in Epidemiology checklist by 2 reviewers independently. For meta-analysis, odds ratios using a random effects model, or standardized mean difference were calculated. If a moderate association was found, the prognostic ability was evaluated by calculating the sensitivity and specificity. Risk of heterogeneity was reported as I 2 and publication bias was visually assessed by funnel plots and quantitatively by Egger test. RESULTS: In all, 48 studies were eligible for inclusion. Twenty meta-analyses could be performed. A moderate association was demonstrated in 3 meta-analyses, between: nuchal translucency >95th centile in one/both fetuses and twin-twin transfusion syndrome (odds ratio, 2.29 [95% confidence interval, 1.05-4.96], I-2 = 6.6%, 4 studies, 615 pregnancies); crown-rump length discordance >= 10% and twin-twin transfusion syndrome (odds ratio, 2.43 [95% confidence interval, 1.13-5.21], I-2 = 14.1%, 3 studies, 708 pregnancies); and maternal ethnicity and twin-twin transfusion syndrome (odds ratio, 2.12 [95% confidence interval, 1.17-3.83], I-2 = 0.0%, 5 studies, 467 pregnancies), but none demonstrated a prognostic ability for any outcome under investigation. CONCLUSION: It is not currently possible to predict adverse outcomes in monochorionic twin pregnancies. We have revealed a lack of research investigating first-trimester biomarkers in monochorionic twin pregnancies. Different assessment methods and definitions of each variable and outcome were an issue and this highlights the need for a large cohort study to evaluate these factors.