Outcomes of ovulation induction in obese women with infertility after bariatric surgery.

Miranda Blanco-Breindel: Writing – review & editing, Writing – original draft, Investigation, Formal analysis, Data curation, Conceptualization. Angela Hui-Chia Liu: Writing – review & editing, Writing – original draft, Supervision, Methodology, Investigation, Formal analysis, Conceptualization. Andrew Rezk: Writing – original draft, Data curation. Esther Wu: Writing – original draft. Harry J. Lieman: Writing – review & editing, Supervision, Investigation, Conceptualization.

The case group (group 1) consisted of 27 women with infertility who underwent bariatric surgery with a mean preoperative BMI of 45.94 kg/m 2 . After 4.38 years of surgical weight loss, these women reached a postoperative BMI of 34.67 kg/m 2 at the start of their OI cycles. A total of 64 OI cycles (OI) with and without intrauterine insemination were analyzed in the case group and showed a mean FPL of 11.13 days, mean peak pretrigger E2 level of 166.17 pg/mL, cumulative PR of 12.50%, cumulative CPR of 9.38%, biochemical pregnancy loss rate of 35.71%, and clinical pregnancy loss rate of 7.14%. Twenty-seven age-matched obese women with infertility in the control cohort of group 2 had pretreatment BMI similar to the preoperative BMI of the case group (41.51 kg/m 2 , P = .09). They underwent 80 OI cycles with and without intrauterine insemination, which had a mean FPL of 13.93 days and mean peak pretrigger E2 level of 278.36 pg/mL and resulted in a cumulative PR of 10.00%, cumulative CPR of 6.25%, biochemical pregnancy loss rate of 44.44%, and clinical pregnancy loss rate of 11.11%. The control cohort of group 3 consisted of 28 age-matched obese women with infertility with pretreatment BMI comparable to the postoperative BMI of the case group (35.83 kg/m 2 , P = .45). These women underwent a total of 68 OI cycles with and without intrauterine insemination and had a mean FPL of 12.14 days, mean peak pretrigger E2 level of 246.43 pg/mL, cumulative PR of 20.59%, cumulative CPR of 19.12%, biochemical pregnancy loss rate of 33.33%, and clinical pregnancy loss rate of 33.33%. Descriptive statistics of patient characteristics of the case and control cohorts noted similar ages at start of OI cycles ( P = .39), diverse but comparable breakdown of racial/ethnic backgrounds ( P = .11), and infertility diagnoses ( P = .52). Baseline menstrual cycle lengths were categorized into eumenorrhea (22–34 days) and oligomenorrhea (≥35 days), and again, no significant difference was found between the case and control groups ( P = .53). Precycle ovarian reserve testing did not differ significantly by a mean antimüllerian hormone level ( P = .24) or antral follicle count ( P = .08) between the groups. Treatment cycle characteristics also did not vary significantly by OI agents ( P = .06), mean FPL ( P = .05), or peak pretrigger E2 level ( P = .55; Table 1 ). Intrauterine insemination was significantly more common in the control group 2 ( P <.01). Comparison of the cycle outcome of PRs for the case and control groups found a nonsignificant P value of .25, although the CPRs were significantly different ( P = .04; Table 2 and Fig. 1 ). Among pregnancies resulting from the OI cycles, no statistically significant differences were noted for the rates of biochemical pregnancy loss ( P = .86) and clinical pregnancy loss ( P = .16) between the case and control groups ( Table 2 ). Of the 212 total OI cycles attempted, 29 cycles (13.67%) had missing data on luteal progesterone levels or subsequent spontaneous menses due to loss to follow-up after negative pregnancy test (data not shown), precluding the comparison of ovulation rates between cases and controls. Table 1 Descriptive statistics of patient demographic, clinical, and cycle characteristics in the case and control groups. Patient and cycle characteristics Group 1 (post–bariatric surgery cases) Group 2 (controls matched by preoperative BMI of group 1) Group 3 (controls matched by postoperative BMI of group 1) P value Cohort size (N) 27 27 28 Age (y) 35.31 33.89 34.43 .39 BMI (kg/m 2 ) 34.67 41.51 35.83 <.01 Weight (lb) 210.64 243.11 213.93 <.01 Infertility diagnosis (N/%) .52  Ovulatory 4 (14.81) 2 (7.41) 0 (0.00)  Tubal factor 4 (14.81) 4 (14.81) 4 (14.29)  Male factor 5 (18.52) 7 (25.93) 6 (21.43)  Unexplained 6 (22.22) 3 (11.11) 4 (14.29)  Uterine factor 6 (22.22) 2 (7.41) 3 (10.71)  Diminished ovarian reserve 7 (25.93) 4 (14.81) 1 (3.57)  Endometriosis 1 (3.70) 1 (3.70) 1 (3.57)  PCOS 9 (33.33) 11 (40.74) 12 (42.86) Baseline menstrual cycle length (N/%) .53  Oligomenorrhea (cycle length, >35 d) 8 (29.36) 11 (42.31) 12 (42.86)  Eumenorrhea (cycle length, 22–34 d) 19 (70.37) 15 (57.69) 16 (57.14)  NA 0 (0.00) 1 (3.70) 0 (0.00) Race (N/%) .11  Non-Hispanic white 3 (11.11) 6 (22.22) 4 (14.29)  Non-Hispanic black 5 (18.52) 7 (25.93) 9 (32.14)  Hispanic 18 (66.67) 7 (25.93) 8 (28.57) Ovulation induction cycle characteristics  No. of attempted cycles 64 80 68  Mean precycle AMH (ng/mL) 3.22 3.96 4.81 .24  Mean precycle AFC 21.89 18.72 27.50 .08  Follicular phase length (d) 11.13 13.39 12.14 .05  Peak pretrigger estradiol (pg/mL) 166.17 278.36 246.43 .55  Ovulation induction agent (%) .06  Clomid 28 (43.75) 50 (62.50) 31 (45.58)  Letrozole 27 (42.19) 25 (31.25) 29 (42.65)  Gonadotropin 9 (14.06) 3 (3.75) 7 (10.29)  Insemination method (%) <.01  Timed intercourse 17 (26.56) 3 (3.75) 7 (10.29)  Intrauterine insemination 47 (73.44) 77 (96.25) 61 (89.71) Note: AFC = antral follicle count; AMH = antimüllerian hormone; NA = not available; PCOS = polycystic ovary syndrome. Table 2 Treatment outcomes of ovulation induction cycles in the case and control groups. Treatment outcomes Group 1 (post–bariatric surgery cases) Group 2 (controls matched by preoperative BMI of group 1) Group 3 (controls matched by postoperative BMI of group 1) P value Cumulative pregnancy rate (%) 12.50 10.00 20.59 .25 Cumulative clinical pregnancy rate (%) 9.38 6.25 19.12 .04 Biochemical pregnancy loss (%) 35.71 44.44 33.33 .86 Clinical pregnancy loss (%) 7.14 11.11 33.33 .16 Figure 1 Pregnancy outcomes in ovulation induction cycles undergone by the case and control groups. Values are rates of pregnancy, clinical pregnancy, biochemical pregnancy, and clinical pregnancy loss. Group 1 (gray bars) consisted of cases who underwent bariatric surgeries before ovulation induction treatments (n = 27); group 2 (blue bars) included nonoperated controls matched by age and preoperative body mass indices (BMIs) of group 1 (n = 27); group 3 (red bars) included nonoperated women matched by both age and postoperative BMIs of group 1 (n = 28). Statistical significance was defined by a P value of <.05 in the chi-square analyses (indicated by ∗). Descriptive statistics of patient demographic, clinical, and cycle characteristics in the case and control groups. Note: AFC = antral follicle count; AMH = antimüllerian hormone; NA = not available; PCOS = polycystic ovary syndrome. Treatment outcomes of ovulation induction cycles in the case and control groups. Pregnancy outcomes in ovulation induction cycles undergone by the case and control groups. Values are rates of pregnancy, clinical pregnancy, biochemical pregnancy, and clinical pregnancy loss. Group 1 (gray bars) consisted of cases who underwent bariatric surgeries before ovulation induction treatments (n = 27); group 2 (blue bars) included nonoperated controls matched by age and preoperative body mass indices (BMIs) of group 1 (n = 27); group 3 (red bars) included nonoperated women matched by both age and postoperative BMIs of group 1 (n = 28). Statistical significance was defined by a P value of <.05 in the chi-square analyses (indicated by ∗).

To identify women for the case and control groups, we queried the electronic medical records at a single academically affiliated fertility center from 2016 to 2022 using the International Classification of Diseases, 10th Revision, codes for infertility diagnoses and bariatric surgery and the Current Procedural Terminology codes for follicular scan and intrauterine insemination. Of the patients identified, 27 had undergone bariatric surgery before receiving OI treatments and met inclusion criteria for the case group (group 1). Aged-matched control cohorts were subsequently selected on the basis of the following schema. Patients were not excluded by infertility diagnoses: • Group 1: 27 women with infertility with a history of surgical weight loss • Group 2: 27 obese women with infertility with BMI matched on the preoperative BMI of the case group • Group 3: 28 women with infertility with BMI matched on the postoperative BMI of the case group Group 1: 27 women with infertility with a history of surgical weight loss Group 2: 27 obese women with infertility with BMI matched on the preoperative BMI of the case group Group 3: 28 women with infertility with BMI matched on the postoperative BMI of the case group The controls were matched on the group level. Besides ensuring that there were no statistically significant differences, controls were identified if the overall group mean yielded a difference of <3 years in age and <4.9 kg/m 2 in BMI. The prespecified range used for matching BMI was determined as classes of obesity were categorized in increments of 5 according to the International Classification of Diseases, 10th Revision, diagnoses (i.e., class 1 obesity, 30–34.9 kg/m 2 ; class 2 obesity, 35–39.9 kg/m 2 ; and class 3 obesity, ≥40 kg/m 2 ). To include a diverse range of infertility diagnoses, up to 3 diagnoses were recorded per patient, and cases and controls were again matched on the group level to avoid statistically significant differences. Retrospective chart review of case and control cohorts was performed to extract patient demographic and clinical characteristics (age, BMI and pretreatment weight, menstrual cycle lengths, duration and amount of postsurgical weight loss for the case group, race/ethnicity, up to 3 infertility diagnoses, and precycle ovarian reserve markers—antimüllerian hormone and antral follicle count). Cycle information of OI agents, follicular phase length (FPL; from day 1 of menses to day of ovulation trigger), peak pretrigger estradiol (E2) level, rates of ovulation (confirmed by a luteal progesterone level of ≥3 ng/mL, spontaneous menses, or subsequent pregnancy), and the use of intrauterine insemination or timed intercourse were recorded. Ovulation induction medications were chosen at the discretion of the treating physicians and included letrozole (2.5, 5, 7.5 mg), clomiphene citrate (50, 100, 150 mg), and gonadotropins. All cycles were extracted for women who underwent multiple treatment cycles during the study period. All except 8 treatment cycles (96.23%) were monitored with follicular ultrasound examinations, and the serum E2 levels and doses of induction medications were stair-stepped until at least 1 dominant follicle was achieved before triggering ovulation with subcutaneous injection of exogenous human chorionic gonadotropin (hCG). The primary treatment outcomes analyzed were cumulative PR, defined by a single serum hCG level of >5 mIU/mL measured 2 weeks after either the intrauterine insemination procedure or timed intercourse; cumulative CPR, defined as sonographic evidence of intrauterine gestational sac; biochemical pregnancy, defined as spontaneously downtrended serum hCG levels before progression to clinical pregnancy; and clinical pregnancy loss, defined as losses between 5 and 10 weeks after intrauterine gestation has been noted on ultrasound. Descriptive analyses were conducted for patient demographic, clinical, and cycle characteristics. Comparison of numerical descriptive statistics of the case and control groups was conducted using the analysis of variance test. The chi-square statistics was performed to compare distribution of categorical variables and cycle outcomes of cumulative PR and CPR, as well as the rates of biochemical pregnancy and clinical pregnancy losses. A P value of <.05 defined statistical significance. This study was approved by the Institutional Review Board at the Montefiore Medical Center/Albert Einstein College of Medicine.

Our study offers preliminary yet encouraging evidence of the trend toward improved CPR in women with infertility undergoing OI cycles after surgical weight loss, despite a persistently obese postoperative BMI. Larger prospective studies are needed to validate our results and further examine the effects of surgical weight loss by type of bariatric surgery, OI agents, and different ranges of postoperative BMIs.

Given the limited data in the current literature, to our knowledge, our study is the first to report the effect of surgical weight loss on OI outcomes among obese women with infertility of diverse racial/ethnic backgrounds and infertility diagnoses beyond polycystic ovary syndrome. We used a retrospective case-control design to match the nonoperated control cohorts by age and BMIs on the basis of the preoperative and postoperative BMIs of the surgical weight loss group. Our results showed that although cumulative PRs do not differ, there is a trend toward improved cumulative CPRs after bariatric surgery, despite pretreatment weights in the obese range before OI cycles. The pathophysiology underlying menstrual irregularities in obesity is multifold, and our findings provide insights into the aspects affected by surgical weight loss pertinent in OI treatments. Gonadotropin release is dampened due to increased levels of nongonadal estrogen production by the aromatase activity of adipose tissue that feedback negatively onto the hypothalamic-pituitary-ovarian axis. Even eumenorrheic obese women exhibit lower total-cycle luteinizing hormone (LH) and follicle-stimulating hormone levels, decreased early follicular phase LH pulse amplitude, and prolonged folliculogenesis ( 3 , 5 ). A recent prospective cohort study of over 700 patients by Bartha et al. ( 24 ) confirmed that the first-trimester serum progesterone levels markedly decreased in obese patients. Another prospective observational study by Rochester et al. ( 25 ) noted an increase in the total-cycle LH level and less delayed increase in the luteal progesterone level in ovulatory obese women at 6 months of surgical weight loss, although these reproductive endocrine disturbances did not reverse completely compared with normal-weight controls. Follicle-stimulating hormone was not correlated with BMIs in the study. Together, these observations offer plausible mechanisms to support our findings. As reported by Rochester et al. ( 25 ), there is an earlier increase in the postovulatory progesterone level and that the total luteal progesterone level increased almost twofold after bariatric surgery. Along with the association of decreasing first-trimester progesterone level with higher BMIs noted in the study by Bartha et al. ( 24 ), corpus luteal function may be improved after surgical weight loss and aids the progression to clinical pregnancy. This may explain the results in our study that the cumulative CPR, but not the overall PR, was statistically significantly different between the case and control groups. Our findings would also be consistent with the hypothesis from the Study of Women’s Health Across the Nation—that a shortened luteal phase in women with increased BMIs and subsequent luteal phase deficiency may lead to worse early pregnancy outcomes ( 26 ). Currently, the available research demonstrates only a modest link between obesity and early pregnancy loss in both spontaneous and assisted conceptions. A recent meta-analysis on obesity and the risk of miscarriage by McLean and Boots ( 27 ) found the adjusted risk/odds ratios for the risk to range between 1.2 and 1.9. The limited sample size in our study was underpowered to confirm this. The pathophysiology behind the association between obesity and risk of early pregnancy loss is complex, and no causal relationships have been definitively established. At the level of the oocyte, the mitochondrial function is disturbed due to high oxidative stress from the obese environment, leading to more reactive oxygen species and subsequently aneuploidy in mouse models ( 28 ). Existing data on the effect of obesity at the level of the endometrium are more inconsistent. Setton et al. ( 29 ) found that in donor-recipient cycles when sibling oocytes from the same normal-weight donor were transferred into obese and normal-weight recipients, a higher BMI was not associated with impaired implantation or increased early pregnancy loss. However, other studies such as that by Bellver et al. ( 30 ) found that obesity can have a negative effect on endometrial receptivity by delaying the window of implantation in artificial frozen embryo transfer cycles. Retiming the embryo transfer on the basis of the displaced window of receptivity led to similar pregnancy loss rates as nonobese controls. Our study has several strengths. Our study cohorts represent diverse infertility diagnoses and racial/ethnic backgrounds that reflect the minorities most severely impacted by obesity. Despite the varied infertility diagnoses, the breakdown of eumenorrheic and oligomenorrheic women was comparable between the cases and controls, allowing us to compare the effect of surgical weight loss on OI in both ovulatory and anovulatory patients with infertility. Our center does not use a BMI cutoff for patients to undergo OI treatments, allowing a properly matched control group with similar BMI as the very increased preoperative BMI of the surgical weight loss group. The case-control design enables matching for the relevant clinical and patient characteristics between the study cohorts, reducing confounding. Yet, there are many limitations to consider as well. Our study is retrospective in nature; therefore, the data available are limited, and retrospective studies preclude conclusions of casualty. Our cohorts are matched at the group rather than the individual level, and we did not match our cohorts by insemination methods, which could have skewed the study results. The data on luteal progesterone level and spontaneous menses subsequent to unsuccessful OI cycles were not well captured. However, most study participants underwent ultrasound and serum E2 monitoring, and medication dosages were stair-stepped if inadequate follicular response was observed. Presence of at least 1 dominant follicle was noted in 197 (92.92%) of the cycles before ovulation was triggered, making anovulation an unlikely event in our study. The sample sizes of our cohorts are small and underpowered to detect differences in outcomes of biochemical and clinical pregnancy losses as well as perform subgroup analyses to further delineate the effects of surgical weight loss by type of bariatric surgery, OI agents, and timed intercourse vs. intrauterine insemination cycles. We also do not have live birth data to investigate the risks of obstetric and pregnancy complications in assisted conceptions from OI treatments after surgical weight loss. Furthermore, we only included patients with infertility undergoing OI treatments and cannot comment on the effect of surgical weight loss on spontaneous conception.

M.B.-B. has nothing to disclose. A.H.-C.L. has nothing to disclose. A.R. has nothing to disclose. E.W. has nothing to disclose. H.J.L. has nothing to disclose.