Predictors of spontaneous reduction in multiple pregnancies conceived following assisted reproductive technology.

OBJECTIVE: To identify patient and cycle parameters that relate to spontaneous reduction in multiple pregnancies following in vitro fertilization (IVF) and embryo transfer (ET).

STUDY DESIGN: A retrospective cohort study was conducted in an academic infertility center. All IVF cycles between January 2007 and June 2008 were evaluated and 69 infertile women were diagnosed with multiple gestation following IVF. Multiple pregnancy was diagnosed by transvaginal sonography at 6-7 weeks gestation following embryo transfer (ET). Repeat ultrasonography undertaken in late first trimester assessed ongoing multiple pregnancy versus occurrence of spontaneous reduction. Patient and IVF cycle parameters were compared between patients experiencing spontaneous reduction (n=26, cases) compared to those where the multiple pregnancy proceeded uneventfully (n=43, controls).

RESULTS: Spontaneous reduction (SR) was observed in almost 38% of the studied multiple gestations (n=69). Women experiencing SR were significantly older (p=.047), of a leaner body mass (p=.049), and exhibited significantly higher serum estradiol (E(2)) levels in the early follicular phase (p=.016) compared to the controls. Endometrial thickness (EMT) on the day of hCG administration related inversely (p=.017), whereas the number of embryos transferred (ET) related positively with the likelihood of SR (p=.027). On multivariate analyses, EMT and the number of ET were identified as independent predictors of SR following IVF-ET; EMT of less than 10mm was associated with a fourfold increased likelihood of SR (OR 4.18; 95% CI 1.02-17.01) whereas each additional embryo transferred doubled the risk of SR in multiple pregnancies resulting from IVF (OR 2.39; CI 1.02-5.58).

CONCLUSIONS: In multiple pregnancies conceived following IVF, occurrence of SR is relatively common. Increasing number of ET and EMT measuring <10mm are identified as independent predictors of likelihood of SR. While advancing age, body mass and baseline E(2) levels were associated with likelihood of SR, these associations disappeared on adjusted analyses. The observed relationship between EMT and SR is novel: the underlying mechanisms are unclear and merit further investigation.