Objective
The objective of the investigation was to study the effect of body mass index (BMI) on in vitro fertilization (IVF) outcomes within a polycystic ovary syndrome (PCOS) population.
Study Design
This was a retrospective cohort study including 101 cycles from 79 women younger than 40 years old with a clinically documented diagnosis of PCOS by Rotterdam criteria undergoing IVF at a university-based infertility clinic from 2001 through 2010. All participants were stratified by BMI calculated from height and weight recorded within 3 months of cycle start: lean (18.7-24.9 kg/m 2 , n = 51), overweight (25-29.9 kg/m 2 , n = 19), and obese (≥30 kg/m 2 , n = 31). Linear, logistic, and Poisson regressions were used as appropriate to estimate the effect of a range of BMIs on IVF outcomes while adjusting for potential confounders.
Results
Obese PCOS women had 69% lower odds of clinical pregnancy per cycle start (odds ratio [OR], 0.31; 95% confidence interval [CI], 0.11–0.86; P = .02) and 77% lower odds of clinical pregnancy per embryo transfer (OR, 0.23; 95% CI, –0.08 to 0.68; P = .008) compared with lean PCOS women. Among obese PCOS women, the odds of live birth were 71% lower per cycle start (OR, 0.29; 95% CI, 0.10–0.84; P = .02) and 77% lower per embryo transfer (OR, 0.23; 95% CI, 0.07–0.71; P = .01) compared with lean PCOS women. There was a trend toward decreased ovarian hyperstimulation syndrome incidence with increasing BMI among women with PCOS: 19.6% in lean, 10.5% in overweight, and 3.2% in obese.
Conclusion
PCOS is a broad syndrome, with our results demonstrating 2 distinct populations, lean and obese, which have different IVF outcomes including ovarian hyperstimulation syndrome risk profiles. This information is important for clinicians because it informs treatment decisions.
Polycystic ovary syndrome (PCOS) affects 4-6% of women. It is the most common endocrinopathy among reproductive-age women and accounts for a large proportion of patients seeking assisted reproductive technologies. Of those affected by PCOS, 50% are obese with a body mass index (BMI) of 30 kg/m 2 or greater. Yet this disease is a multifactorial condition that includes different pathophysiological pathways. One possible pathway is via increased luteinizing hormone pulse frequency in both lean and obese women with PCOS compared with normally cycling women, which may have effects on the follicular cohort that is recruited. Other physiological and hormonal events contributing to the dysfunction differ across a spectrum of weights.
In a non-PCOS population, BMI is known to affect in vitro fertilization (IVF) outcomes. Although weight reduction improves fertility in PCOS patients attempting spontaneous conception, studies of PCOS patients to date do not fully evaluate the impact of a range of BMIs on IVF outcomes. One study quantified the negative impact of morbid obesity on pregnancy outcomes in a group of women with PCOS, although a lean-only PCOS population was not included for comparison.
In 2011, Marquard et al found that obesity and PCOS were both independently associated with smaller oocyte diameter among 8 obese and 5 nonobese PCOS patients, although only patients undergoing intracytoplasmic sperm injection (ICSI) were included. A 2001 study from Norway evaluated 100 cycles from 56 PCOS patients categorized by insulin-resistance status, which did not correlate completely with BMI. Those findings showed that insulin resistance was associated with a lower oocyte count and increased follicle stimulating hormone (FSH) requirement; however, the data analysis controlled for body weight. A 2008 study of lean and obese PCOS and non-PCOS patients found that more oocytes were retrieved from lean than from obese PCOS patients, although the findings were limited by low PCOS prevalence (19%), which may have contributed to the lack of significant differences in clinical outcomes (implantation, miscarriage, live birth, and multiple birth rates).
The increased incidence of ovarian hyperstimulation syndrome (OHSS) in PCOS patients is also well documented but not differentiated with regard to BMI-adjusted risk. So the full effects of a spectrum of body masses among reproductive-aged women with PCOS remain poorly studied.
Considering that 50% of PCOS patients are not obese, characterizing associations of a range of BMIs with ovarian response and clinical IVF outcomes could inform the management of PCOS patients. Specifically, characterizing any relationships between BMI and OHSS would lay the groundwork for investigation of how BMI-dependent dosing of gonadotropins might reduce morbidity because of OHSS, a serious disease for which PCOS patients are at elevated risk.
We hypothesized that PCOS is a broad syndrome comprised of unique BMI-determined subpopulations that exhibit differences in IVF outcomes. Our secondary hypothesis was that there is a higher incidence of OHSS in the lean PCOS group when gonadotropin dosing is not BMI adjusted. This information would be important for clinicians with respect to making treatment decisions.
Materials and Methods
We investigated these hypotheses in a retrospective cohort study within an urban academic medical center’s IVF practice that included 101 cycles from 79 women younger than 40 years old who started a fresh, autologous IVF cycle with or without ICSI between January 2001 and December 2010. This study received appropriate approval from the Partners Institutional Review Board (no. 2012P000196) prior to data collection.
Women eligible for inclusion were those with clinically measured height and weight within 3 months of their cycle start date, who also met strict diagnostic criteria for PCOS as defined by the 2003 Rotterdam European Society for Human Reproduction and Embryology/American Society for Reproductive Medicine-sponsored PCOS consensus workshop group ( Table 1 ).
| (2 of the 3 required for PCOS diagnosis) |
| Oligoovulation or anovulation |
| Clinical (acne, hirsutism) and/or biochemical signs of hyperandrogenism |
| Total testosterone >70 ng/dL |
| Androstenedione >245 ng/dL |
| Dehydroepiandrosterone sulfate >248 ug/dL |
| Polycystic ovaries |
| ≥12 follicles 2-9 mm in diameter in each ovary |
This set of diagnostic criteria was chosen to permit inclusion of a wide range of PCOS phenotypes because lean patients are more likely to have milder expressions of the disease including spontaneous ovulation. Each patient’s medical record was meticulously reviewed to ensure that they fit the Rotterdam criteria for PCOS diagnosis.
Women whose embryo transfers were cancelled because of the high risk of OHSS were still included in the clinical outcomes analyses. Those with a transfer were restricted to day 3 fresh embryo transfers to allow for standardization of embryological outcomes. Multiple eligible fresh cycles for each patient were included, and all stimulation protocols were included. Infertility diagnoses were defined by physician documentation. ICSI was performed only if indicated by semen parameters.
Individual record review was performed for all patients and all cycles. Exclusion criteria included the use of in vitro maturation, FSH greater than 10 mIU/mL, uncontrolled thyroid disease as defined by thyroid-stimulating hormone of 5.7 mIU/L or greater based on our laboratory cutoff for an abnormal value, a history of chemotherapy or radiation exposure, a recurrent pregnancy loss, uterine factor, balanced translocation (in either partner), surgically documented endometriosis or pelvic adhesions, a history of pelvic inflammatory disease, adenomyosis, and submucosal myoma.
Clinical IVF outcomes included OHSS, chemical pregnancy, miscarriage, clinical pregnancy, and live birth. For the purposes of this study, OHSS was defined by abdominal bloating and discomfort or decreased urine output significant enough to warrant outpatient evaluation. The OHSS cases were considered moderate if they required more than 1 paracentesis for symptomatology and severe if hospital admission was required based on persistent emesis or the laboratory values showing hemoconcentration or electrolyte abnormalities; all other OHSS cases were classified as mild. These definitions for diagnosis came from internal hospital protocols.
Chemical pregnancy was defined as a positive serum β human chorionic gonadotropin without progression of a pregnancy prior to sonographic evidence of fetal cardiac activity. A clinical pregnancy was documented by sonographic evidence of fetal cardiac activity. A miscarriage was denoted by the loss of a clinical pregnancy prior to 20 weeks’ gestational age. A live birth was defined as the delivery of a viable infant.
Patients were stratified by BMI calculated from height and weight recorded within 3 months of the cycle start: lean (BMI 18.7-24.9 kg/m 2 ), overweight (BMI 25-29.9 kg/m 2 ), and obese (BMI of ≥30 kg/m 2 ). Generalized estimating equations were used to account for the correlation between and within women in the instance of multiple cycles per woman. The binomial distribution with log link was applied for dichotomous outcomes (OHSS incidence, chemical pregnancy, miscarriage, clinical pregnancy, and live birth); the normal distribution with identity link was applied for continuous outcomes (number of days of stimulation medication, total FSH dose, and peak estradiol); and the Poisson distribution with log link was applied for count outcomes (number of oocytes retrieved, metaphase II oocytes, bipronuclear embryos on day 1, and 8 cell embryos on day 3). All models used robust standard errors.
Age was included in all models a priori. Potential additional confounders were added to the model one by one: gravidity, maximum day 3 FSH level, and a diagnosis of male factor infertility. All of these variables altered effect estimates by greater than 10% in the univariate model so were retained per the 10% confounding rule for more detailed analysis. In the final multivariable models when all covariates were included, however, there was no evidence of confounding by age, gravidity, maximum day 3 FSH level, and male factor infertility diagnosis when compared with the unadjusted model.
All tests were 2 sided, and statistical significance was assumed for P < .05. Analyses were performed using Statistical Analysis Software (SAS) version 9.2 (SAS Institute, Inc, Cary, NC).
Results
Seventy-nine women with PCOS contributed 101 cycles (51 lean, 19 overweight, 31 obese); 2 patients each had a cycle in 2 different BMI categories. The mean BMI in the lean, overweight, and obese exposure groups was 22.2, 27.1, and 39.0 kg/m 2 , respectively; the average age was similar across groups (32.0, 32.6, and 32.4 years, respectively) ( Table 2 ). Because BMI increased among these PCOS patients, hyperandrogenism was more prevalent and polycystic ovarian morphology on ultrasound was less prevalent ( Table 2 ).
| Outcome | Lean (n = 51) | Overweight (n = 19) | Obese (n = 31) |
|---|---|---|---|
| Age, y a | 32.0 (3.5) 23.7–39.2 | 32.6 (2.9) 27.5–36.8 | 32.4 (3.2) 24.8–38.0 |
| BMI, kg/m 2 a | 22.2 (1.6) 18.7–24.9 | 27.1 (1.3) 25.1–29.4 | 39.0 (7.4) 30.2–59.2 |
| Male factor b | 15 (29.4%) | 4 (21.1%) | 12 (38.7%) |
| ICSI b | 10 (19.6%) | 2 (10.5%) | 10 (32.3%) |
| Anovulation/oligoovulation b | 51 (100.0%) | 19 (100.0%) | 31 (100.0%) |
| Hyperandrogenism b | 33 (64.7%) | 15 (78.9%) | 26 (83.9%) |
| Polycystic ovaries on ultrasound b | 19 (37.3%) | 4 (21.1%) | 5 (16.1%) |
There were very few chemical pregnancies that did not progress in this cohort ( Table 3 ). There were 3 previable (between 20 and 23 weeks’ gestational age) births: 2 in lean patients and 1 in an overweight patient. Compared with lean PCOS women, obese PCOS patients were 69% less likely to achieve clinical pregnancy per cycle start (35.5% vs 56.9%; odds ratio [OR], 0.31; 95% confidence interval [CI], 0.11–0.86; P = .02) and 77% less likely per embryo transfer (36.7% vs 64.4%; OR, 0.23; 95% CI, 0.08–0.68; P = .008) ( Table 3 ). Compared with lean PCOS women, obese PCOS patients were 71% less likely to have a live birth per cycle start (32.3% vs 49.0%; OR, 0.29; 95% CI, 0.10–0.84; P = .02) and 77% less likely per embryo transfer (33.3% vs 55.6%; OR, 0.23; 95% CI, 0.07–0.71; P = .01) ( Table 3 ).
| Outcome | Lean (n = 51) | Overweight (n = 19) | Obese (n = 31) |
|---|---|---|---|
| OHSS incidence a | 10 (19.6%) (Referent) | 2 (10.5%) 0.74 (0.16–3.50) P = .71 | 1 (3.2%) 0.17 (0.02–1.46) P = .11 |
| Mild b | 9 (17.6%) | 2 (10.5%) | 0 |
| Moderate b | 1 (2.0%) | 0 | 0 |
| Severe b | 0 | 0 | 1 (3.2%) |
| Chemical pregnancy b | 3 (6.7%) | 0 | 2 (6.7%) |
| Miscarriage b | 2 (4.4%) | 0 | 1 (3.3%) |
| Clinical pregnancy per cycle start a | 29 (56.9%) (Referent) | 14 (73.7%) 1.63 (0.49–5.46) P = .43 | 11 (35.5%) 0.31 (0.11–0.86) P = .02 |
| Clinical pregnancy per embryo transfer a | 29 (64.4%) (Referent) | 14 (73.7%) 1.11 (0.31–3.95) P = .88 | 11 (36.7%) 0.23 (0.08–0.68) P = .008 |
| Live birth per cycle start a | 25 (49.0%) (Referent) | 13 (68.4%) 1.63 (0.53–4.98) P = .39 | 10 (32.3%) 0.29 (0.10–0.84) P = .02 |
| Live birth per embryo transfer a | 25 (55.6%) (Referent) | 13 (68.4%) 1.23 (0.40–3.82) P = .72 | 10 (33.3%) 0.23 (0.07–0.71) P = .01 |
Stay updated, free articles. Join our Telegram channel
Full access? Get Clinical Tree
