Does body mass index impact assisted reproductive technology treatment outcomes in gestational carriers

Gestational carriers (GCs) are generally healthy women with proven fertility and a good obstetric history, who choose to carry a baby not genetically related to them for intended parent(s) [1]. Third-party reproduction is on the rise; in Canada, between 2001 and 2012, the number of births to GCs has increased six fold, and the proportion of ART births that involved a GC during the same years, doubled from 0.9 to 1.7% [2]. In the US, 2.4% of all IVF cycles performed between 2010 and 2014 involved GCs, resulting in 10,009 infants born [3].

Body Mass Index (BMI) measures body fat composite, based on height and weight. The World Health Organization (WHO) defines obesity as a BMI of 30 or greater. The abnormal fat accumulation in obese individuals can negatively impact their health in many ways, including an increased risk for vascular disease, diabetes and malignancies [4, 5]. As a result, obesity is an increasing public health concern and has been termed a ‘global pandemic’ [6].

Elevated maternal pre-pregnancy BMI can lead to a higher risk of maternal and fetal morbidities and mortality [7‐13]. It is controversial whether or not obesity impacts the response to assisted reproductive technology (ART) treatments [14‐18]. Some authors cite lower implantation rates (IR), clinical pregnancy rates (CPR) and live birth rates (LBR), as well as higher miscarriage rates among the obese population [17, 19‐25], whereas others do not report any differences [26‐29]. To better understand the impact obesity has on ART outcome and success rates, the effect on egg and embryo quality should be assessed separately from the effect on the uterine environment.

Measuring outcomes in GCs can enhance our understanding of the effect increasing BMI has on the uterine environment, while controlling for other confounding factors resulting from underlying causes of infertility. To further enhance our findings, and make them generalizable to other clinics, we chose to compare these outcomes to those achieved by our general infertile population at the clinic, during the same time period. Since, by design, surrogates have proven fertility and a favorable obstetric history, we hypothesized that per BMI they would achieve better ART outcomes, than infertile patients.

Importantly, studying the effect increased BMI has on outcomes among GCs, would assist in better defining screening policies for GCs; while some advocate using BMI as an exclusion criterion for GCs, this is not supported by the official American Society for Reproductive Medicine (ASRM) guidelines [30]. Previous studies, evaluating the impact obesity has on pregnancy outcomes in GCs have yielded conflicting results [31, 32].

BMI, per se, may not be the major determining factor influencing the outcome of infertility treatments; other metabolic and endocrinology factors may be at play [33‐35]. This study was designed to assess whether BMI is a predictor of pregnancy outcomes among GCs, and how these outcomes differ from outcomes among the general infertile population exposed to similar conditions.

This retrospective analysis received approval from the University of Toronto Research Ethics Board (#33894).

There were 336 GC cycles performed at CReATe Fertility Centre between 2003 and 2016. Figure 1 depicts a flow chart of the review process leading to the final patient population of this study; following application of exclusion criteria, 188 GC cycles were included in the analysis and matched by BMI and year of treatment with the general infertile population treated during the same years, after applying exclusion criteria on this group as well. Eight GCs and 12 controls contributed more than 1 cycle to the final sample.

Gestational surrogacy has been available since 1985 [1]. This has opened up options for women that are unable to carry a pregnancy (i.e. repeated implantation failure or repeated pregnancy loss), for women at risk for deterioration of their health if they conceive (i.e. malformations of the great vessels, kidney involvement of connective tissue diseases) and for patients wishing to conceive that do not have a uterus (agenesis of uterus, male patients). Cycles utilizing gestational carriers (GCs) are on the rise [2, 3]. GCs are usually younger than the general population seeking fertility solutions, and they have proven fertility, good obstetric history, and a relative scarcity of comorbidities, which was also the case in our study. A previous study concluded that previous proven fertility accounts for improved implantation rates in GC cycles, and is a positive prognostic factor of subsequent successful pregnancies [40]. Indeed, we were able to show improved implantation and clinical pregnancy rates as well as decreased miscarriage rates in our GC cohort compared with controls. It is important to note that the endometrial thickness prior to embryo transfer did not differ between cohorts, and that patients with RPL or RIF were excluded from the control cohort, thereby eliminating these factors as the driving forces for better outcomes among GCs. In this cohort, there was minimal utilization of PGT-A, and the oocytes utilized in GC cycles originated from younger women than those utilized in our control arm. However, when controlling for the age of the women contributing their oocytes to these cycles, BMI was still not a predictor for clinical pregnancy rates or live birth rates. Furthermore, it is well documented that the baseline aneuploidy rate in women younger than 35 is 30%, after which this rate increases significantly. Notably, in this study, the age of the oocyte source in both cohorts averaged below 35, thus cannot fully account for the remarkable differences in outcomes captured in this study [41]. It has been established that frozen embryo transfers yield superior results when compared with fresh embryo transfers [42]. Of note, in this cohort, the proportion of frozen embryo transfers was higher among the control population, which had worse outcomes per BMI strata. Furthermore, the cohorts also differed in prevalence of current recreational drug use, ‘social’ alcohol consumption and regular physical activity. These factors could not be controlled for in our analysis and thus represent a limitation of the retrospective design of this study.

Overweight and obesity are now dramatically on the rise, particularly in urban settings [43, 44]. In Canada, 25.8% of the population is obese [45]. Because of safety concerns and the perception that obesity might worsen the prognosis following ART treatments, some professional societies set BMI limits for eligibility for ART treatments [46‐51]. The question whether BMI affects pregnancy outcomes in general, and ART outcomes specifically, is unresolved. Three previous studies that have assessed the effect BMI has on the uterine environment, by utilizing an oocyte donation model, yielded conflicting results [20, 26, 36]. However, such a model cannot control for other infertility factors that may contribute to or be impacted by increased BMI. Hence, using a GC model can help tease out such infertility factors, and increase the detection rate of the effects of increasing BMI on the uterine environment. A previous study looking at outcomes of 163 GC cycles was not able to capture differences in outcomes based on BMI category [32]. In a larger-scale study looking at 349 GC cycles, a threshold of BMI > 35 was found to be a significant predictor of worse outcomes [31].

Our study is the first to use the GC model while comparing with outcomes in the general infertile population. Comparing outcomes between GCs of different BMIs allowed us to control for other infertility factors associated with increased BMI. Our study is limited to the experience of one center. To make our findings more generalizable, we chose to compare outcomes among GCs to those achieved by our general infertile population of similar BMIs. An ideal study design would be a multi-center comparison of outcomes achieved by GCs from different BMI strata.

While most studies analyze pregnancies and live births as binary outcomes, it has been shown that when applicable, treating outcomes as continuous variables, specifically in smaller sample size studies, increases the power of the study [52]. The fact that we treated our primary outcomes as both continuous and binary variables adds strength to our analysis and provides a more in-depth interrogation of the effect BMI has on the tested outcomes. Further, since our sub-analysis concerns even smaller groups, we chose to apply a Holm-Bonferroni correction in pairwise comparisons to add robustness to the results. Indeed, we were able to show lower risks for maternal, fetal and neonatal complications among our GC population when compared with the general infertile population, matched by BMI. Furthermore, our GC population performed better for all measured outcomes, across most BMI strata (normal weight, overweight and obese), when compared with the general infertile population treated at our clinic. Since the GC population is inherently different than the general infertile population, in terms of age and favorable obstetric history, we performed linear regression to account for these potential confounders. In this regression model we were able to show that even after controlling for the above, BMI was not predictive of GC outcomes. In our cohort, BMI was not predictive of miscarriages; the miscarriage rate among our GC population was 11.6% and among our general infertile population it was 30.4%. A previous study found that increased BMI increases the risk for early miscarriages, as well as for repeat miscarriages [53]. These differences could be at least partially attributed to the fact that the women in the aforementioned cohort used their own eggs, and increased BMI could have potentially resulted in reduced oocyte quality, further demonstrating the importance of our study design.

Lastly, we had too few GCs with a BMI over 40 to make a judgment in that BMI range, future investigations should focus on the effect morbid obesity has on outcomes in GC cycles.

In conclusion, in our study, BMI was not statistically or clinically predictive of ART outcomes or of pregnancy outcomes, among GCs. Due to the retrospective nature of our study and the limited sample size, these findings should be interpreted with caution and larger datasets are needed to further validate our findings. We suspect that other metabolic factors, which can be influenced by BMI, are at play, and may serve as better predictors for outcomes [31, 33, 35, 54, 55]. Currently the ASRM guidelines do not include increased BMI as an exclusion criterion for carrying a pregnancy as a surrogate [30]. Based on our study and previous literature, BMI should not be automatically applied as an exclusion criterion for GCs, when considering ART treatment. Consideration of previous obstetric history, and overall health are likely more predictive of adverse outcomes, and therefore, more important when screening a gestational carrier candidate. Future studies in this area should also look at other metabolic factors, such as lipid profiles and insulin resistance, which may be more helpful in identifying poor prognosis GC candidates, rather than BMI alone.