ABSTRACT
Background
Overweight and obesity among women have been associated with lower success of assisted reproduction technologies (ARTs). However, the relation of adolescent body weight and adult weight change to ART outcomes is not well understood.
Objective
Our objective was to evaluate the associations of female BMI (in kg/m2) at 18 y and weight change from 18 y to current age with ART outcomes.
Methods
We included 486 women in a prospective cohort at the Massachusetts General Hospital Fertility Center (2004–2018) who underwent 863 ART cycles. At study entry, height and weight were measured by research personnel. Women recalled their weight at 18 y. Restricted cubic splines were used to evaluate the associations between BMI at 18 y and weight change since 18 y and ART outcomes adjusting for age, race, education, smoking, and height and accounting for the correlated cycles within women.
Results
Women had a median (range) BMI of 20.6 (14.8 to 36.4) at 18 y and 23.3 (16.1 to 45.8) at study entry. The median (range) weight change since 18 y was 7.4 kg (−12.1 to 60.1 kg). There was no association between BMI at 18 y and clinical ART outcomes. Long-term weight change had a nonlinear association with live birth such that higher weight gain since 18 y (particularly ≥15 kg) and weight loss were both associated with lower odds of live birth. The negative association between weight change and live birth was stronger in women ≥22.5 kg/m2 at 18 y such that each 10-kg increase was associated with a 30% (6%, 48%) lower odds of live birth.
Conclusions
Weight gain in adulthood is negatively associated with ART success, particularly among women who were heavier at 18 y. These results add to the growing literature supporting the benefits of preventing weight gain in adulthood on female fertility.
Keywords: assisted reproductive technology, body weight, fertility, infertility, in vitro fertilization
Introduction
Excess body weight among women attempting to get pregnant unassisted or through assisted reproductive technologies (ARTs) has been linked to worse reproductive outcomes, including higher risks of anovulation, irregular menses, infertility, miscarriage, and stillbirth (1). Although overweight or obese women are often encouraged to lose weight as a means to increase fertility, the majority of evidence comes from studies comparing overweight or obese women with normal-weight women rather than studies directly evaluating the effects of weight loss. In fact, 2 large randomized controlled trials found no differences in the proportion of women achieving a clinical pregnancy or live birth comparing after a weight loss intervention prior to infertility treatment compared with prompt infertility treatment (2, 3). Therefore, although short-term weight loss does not appear to be beneficial for the success of infertility treatment, it remains unclear whether long-term weight loss or weight gain prevention may be advantageous.
A handful of studies have evaluated the association between long-term weight change and fertility among women attempting to conceive without medical assistance (4–8). Although weight loss was rare in these studies and not always associated with fertility improvements, the majority of studies found that weight gain between late adolescence and adulthood was associated with longer time to pregnancy and higher likelihood of pregnancy loss (4–6). To date, there has not been an investigation of how long-term weight change influences outcomes of infertility treatment. Due to the nature of ART, where much of reproduction is externalized, this population of women may provide insights into the biological mechanisms underlying any observed associations between long-term weight change and reproductive success. Moreover, it may help to further address whether the well-documented deleterious effects of current BMI (in kg/m2) on ART outcomes are due to recent weight status or lifetime patterns in weight trajectories. Given that excess body weight in adulthood generally results from a slow, steady weight gain during adulthood (as opposed to short periods of gain) (9), studying lifetime trajectories of weight change may be the more relevant exposure.
Methods
Study population
The Environment and Reproductive Health (EARTH) Study is an ongoing prospective cohort that began following couples seeking infertility treatment at the Massachusetts General Hospital (MGH) fertility center in 2004 to identify environmental and nutritional determinants of fertility (10). All women aged 18–46 y at enrollment were eligible, and ∼60% of eligible women contacted by the research staff participated in the study. Women were followed through each of their infertility treatment cycles until they either had a live birth or discontinued treatment at MGH. Women of all body sizes were eligible to enroll in the EARTH Study; however, due to restrictions in place at the MGH fertility center, women were only allowed to initiate an ART cycle once their current BMI was ≤40.
For the present analysis, we included women who had completed at least 1 ART cycle as of September 2018 (n = 493 eligible women, 879 cycles). From there, we excluded 7 women (16 ART cycles) with unknown weight at 18 y and height at study entry. Our final sample included 486 women and all their eligible ART cycles (n = 863). To assess outcomes of controlled ovarian stimulation, we further refined the study population to 439 women and 651 fresh autologous ART cycles by excluding donor egg cycles (n = 33 cycles), cryo-thaw cycles (n = 138), fresh autologous cycles that failed prior to oocyte retrieval (n = 39), and cycles in which the oocyte retrieval was performed outside of the MGH fertility center (n = 2). The study was approved by the Human Subject Committees of the Harvard T.H. Chan School of Public Health and MGH.
Clinical management and outcome assessment
For fresh, autologous ART cycles, participants underwent 1 of 3 treatment protocols: 1) luteal phase gonadotropin-releasing hormone (GnRH) agonist, 2) follicular phase GnRH agonist/flare, or 3) GnRH antagonist. For cryo-thaw or donor egg recipient cycles, patients underwent endometrial preparation protocols as clinically indicated. During ovarian stimulation, patients were monitored for serum estradiol (Elecsys Estradiol Chemiluminescent Assay; Roche Diagnostics), follicle size, and endometrial thickness. Human chorionic gonadotropin (hCG) was administered ∼36 h before scheduled oocyte retrieval to induce oocyte maturation. Following retrieval, oocytes were fertilized using conventional insemination or intracytoplasmic sperm injection as clinically indicated. Fertilization was determined 17–20 h after insemination as the number of oocytes with 2 pronuclei. The procedures regarding the timing and grading of embryo quality have varied during the follow-up of EARTH patients and thus are not included as an outcome for analysis. Clinical outcomes were assessed among women who underwent an embryo transfer. Successful implantation was defined as an elevation in plasma β-hCG concentrations >6 IU/L (β-hCG Chemiluminescent Assay; Roche Diagnostics). Clinical pregnancy was defined as an elevation in β-hCG with the confirmation of an intrauterine pregnancy by ultrasound. Live birth was defined as the birth of a neonate on or after 24 wk of gestation.
Anthropometry
Current height and weight were measured at study entry by trained research nurses. Women were asked to recall weight at 18 y through a self-administered questionnaire. In a validation study conducted in a separate cohort of women aged 25–42 y, self-reported weight at 18 y was shown to be highly correlated with weight recorded in medical records (r = 0.87) (11). BMI at 18 y was calculated as weight at 18 y (kg) divided by current height squared (m2) and divided into 5 categories based on the distribution of BMI at 18 y in our cohort and the World Health Organization guidelines: <18.5, 18.5–19.9, 20–22.5, 22.6–24.9, and ≥25 (12). Weight change since 18 y was calculated as the difference between weight at study entry and weight at 18 y and categorized based on its distribution in our study population: lost weight (<0 kg), weight stable (within 0–5 kg), gained 5–9.9 kg, gained 10–14.9 kg, and gained ≥15 kg.
Statistical analysis
Participants’ characteristics were summarized and compared across categories of adult weight change and BMI at 18 y. Kruskal-Wallis test was used to compare differences in continuous measures across categories of BMI at 18 y, whereas χ2 tests and Fisher exact test (when 1 or more cell counts were ≤5) were used for categorical variables. Multivariable generalized linear mixed models were used to evaluate the association between categories of BMI at 18 y and adult weight change and ART outcomes, with a random intercept to account for within-person correlations in outcomes and unbalanced design (e.g., different number of cycles per woman) (13). A normal distribution and identity link function were specified for peak estradiol and endometrial thickness, a Poisson distribution and log link function were specified for oocyte counts, and a binomial distribution and logit link function were specified for fertilization and clinical outcomes. Tests for linear trends were conducted using the median values of each category of BMI at 18 y or weight change since 18 y as continuous variables. For specific point comparisons, the reference group was consistently women who had a BMI between 20 and 22.4 at 18 y and women who gained between 0 and 5 kg (weight stable). Results are presented as population marginal means, adjusted for covariates at their mean level for continuous covariates and weighted by relative frequency for categorical covariates (14). We also used restricted cubic splines to test for nonlinear associations between weight change and BMI at 18 y and probability of live birth. These models used the likelihood ratio test to compare the model with the linear term to the model with the linear and the cubic spline terms (15).
Confounding was assessed using prior knowledge and descriptive statistics from our study population through the use of directed acyclic graphs (Supplemental Figure 1). Variables retained in the final multivariable models were age (years), smoking status (ever smoker compared with never), race (white compared with other), education (college or less compared with graduate school), and height at baseline (centimeters). Models in which the exposure of interest was weight change since 18 y were also adjusted for BMI at 18 y (continuous). Infertility diagnosis, infertility treatment history, parity, and gravidity were considered potential intermediates on the casual pathway between BMI at 18 y or adult weight change and ART outcomes and therefore were not included in any of the models.
We formally tested whether the associations of weight change differed according to weight status at 18 y (dichotomized at 22.5 kg/m2, the 75th percentile in our cohort) using cross-product terms in the final adjusted model. We also performed several sensitivity analyses restricting our study population to only embryo transfer cycles, fresh autologous cycles, and never smokers to evaluate the robustness of our results. All analyses were performed using SAS (version 9.4; SAS Institute, Inc.).
Results
The study population comprised 486 women who underwent 1 (53%), 2 (25%), 3 (15%), or ≥4 (7%) ART cycles. These women had a mean (SD) age of 35.4 (4.0) years, BMI of 24.3 (4.5), and BMI at 18 y of 20.9 (2.8). On average, women gained 9.2 (9.9) kg over a mean time period of 17.4 (4.0) years. Out of the 486 women, 58 (12%) lost weight, 129 (27%) maintained a stable weight (0–5 kg), 117 (24%) gained 5–9.9 kg, 75 (15%) gained 10–14.9 kg, and 107 (22%) gained ≥15 kg since 18 y. Only 8% of women were overweight or obese at 18 y (BMI ≥25); however, this percentage increased to 33% by the time of ART. Of the 863 initiated ART cycles, 692 (80%) were fresh cycles, 138 (16%) were autologous cryo-thaw cycles, and 33 (4%) were donor egg recipient cycles (Figure 1). A total of 762 (88%) of the initiated cycles had at least 1 embryo transferred, with 471 (55%) resulting in implantation, 411 (48%) in clinical pregnancy, and 332 (38%) in live birth.
FIGURE 1.
Overview of the ART cycles included in the analysis of BMI at 18 y and adult weight change since 18 y in relation to ART outcomes in the Environment and Reproductive Health Study. ART, assisted reproductive technology; IUI, intrauterine insemination; SAB, spontaneous abortion; SB, stillbirth; TAB, therapeutic abortion.
Demographic characteristics of the study population were similar across categories of weight change since 18 y, with the exception of current age, height, current BMI, and education (Table 1). On average, women who had gained more weight since 18 y tended to be older and taller, had a higher current BMI, and were less educated compared with women who were weight stable. Women who had lost weight since 18 y tended to be the heavier in adolescence compared with women who were weight stable. Similar differences were seen across categories of BMI at 18 y (Supplemental Table 1). The Spearman correlation between BMI at study entry and BMI at 18 y was moderate (ρ = 0.53).
TABLE 1.
Characteristics of women by categories of weight change since 18 y in the EARTH Study1
| Weight change since 18 y (kg) | ||||||
|---|---|---|---|---|---|---|
| Characteristic | <0 | 0–4.9 | 5–9.9 | 10–14.9 | ≥15 | P value2 |
| Median weight change, kg | −3.1 | 2.5 | 7.4 | 12.3 | 22.0 | |
| Number of women | 58 | 129 | 117 | 75 | 107 | |
| Personal characteristics | ||||||
| Current age, y | 34.7 ± 3.7 | 35.0 ± 4.0 | 35.1 ± 4.1 | 36.3 ± 3.7 | 36.2 ± 4.4 | 0.01 |
| Height, cm | 163.7 ± 5.9 | 164.9 ± 6.7 | 165.4 ± 6.5 | 164.6 ± 7.1 | 166.6 ± 6.8 | 0.09 |
| BMI at 18 y, kg/m2 | 22.5 ± 2.4 | 20.8 ± 2.1 | 20.2 ± 2.6 | 20.6 ± 3.7 | 21.2 ± 2.9 | <0.001 |
| Current BMI, kg/m2 | 21.0 ± 2.1 | 21.8 ± 2.1 | 23.0 ± 2.7 | 25.2 ± 4.0 | 29.8 ± 4.2 | <0.001 |
| Years since 18 y | 16.7 ± 3.7 | 17.0 ± 4.0 | 17.1 ± 4.1 | 18.3 ± 3.7 | 18.2 ± 4.4 | 0.01 |
| Ever smoker | 13 (22.4) | 32 (24.8) | 34 (29.1) | 18 (24.0) | 30 (28.0) | 0.84 |
| White | 49 (84.5) | 108 (83.7) | 96 (82.1) | 60 (80.0) | 86 (80.4) | 0.93 |
| Education | <0.001 | |||||
| <College | 1 (1.7) | 4 (3.1) | 4 (3.4) | 7 (9.3) | 16 (15.0) | |
| College degree | 23 (39.7) | 39 (30.2) | 55 (47.0) | 29 (38.7) | 45 (42.1) | |
| Graduate degree | 34 (58.6) | 86 (66.7) | 58 (49.6) | 39 (52.0) | 46 (43.0) | |
| Cycle characteristics | ||||||
| Parous | 6 (10.3) | 12 (9.3) | 17 (14.5) | 10 (13.3) | 17 (15.9) | 0.56 |
| Previous IUI | 26 (44.8) | 35 (27.1) | 40 (34.2) | 29 (38.7) | 38 (35.5) | 0.17 |
| Previous IVF | 9 (15.5) | 19 (14.7) | 24 (20.5) | 20 (26.7) | 26 (24.3) | 0.18 |
| Primary infertility diagnosis | 0.67 | |||||
| Male factor | 17 (29.3) | 34 (26.4) | 36 (30.8) | 28 (37.3) | 34 (31.8) | |
| Female factor | 15 (25.9) | 41 (31.8) | 35 (29.9) | 20 (26.7) | 38 (35.5) | |
| Diminished ovarian reserve | 5 (8.6) | 15 (11.6) | 14 (12.0) | 5 (6.7) | 10 (9.4) | |
| Endometriosis | 1 (1.7) | 6 (4.7) | 7 (6.0) | 3 (4.0) | 10 (9.4) | |
| Ovulation disorders | 8 (13.8) | 7 (5.4) | 7 (6.0) | 6 (8.0) | 7 (6.5) | |
| Tubal factor | 1 (1.7) | 11 (8.5) | 5 (4.3) | 5 (6.7) | 9 (8.4) | |
| Uterine disorders | 0 (0.0) | 2 (1.6) | 2 (1.7) | 1 (1.3) | 2 (1.9) | |
| Unexplained | 26 (44.8) | 54 (41.9) | 46 (39.3) | 27 (36.0) | 35 (32.7) | |
| PCOS | 5 (8.6) | 12 (9.3) | 10 (8.6) | 6 (8.0) | 17 (15.9) | 0.31 |
| Treatment protocol | 0.42 | |||||
| Antagonist | 8 (13.8) | 15 (11.6) | 19 (16.2) | 10 (13.3) | 14 (13.1) | |
| Flare | 2 (3.5) | 14 (10.9) | 14 (12.0) | 10 (13.3) | 20 (18.7) | |
| Luteal phase agonist | 44 (75.9) | 88 (68.2) | 74 (63.3) | 51 (68.0) | 62 (57.9) | |
| Egg donor or cryo | 4 (6.9) | 12 (9.3) | 10 (8.6) | 4 (5.3) | 11 (10.3) | |
| Day 3 FSH, IU/L | 7.3 ± 2.1 | 7.6 ± 3.5 | 7.5 ± 2.6 | 7.0 ± 1.7 | 7.5 ± 2.5 | 0.93 |
| Embryo transfer day3 | 0.62 | |||||
| Day 2 | 3 (6.8) | 8 (7.8) | 4 (4.2) | 2 (3.1) | 6 (7.2) | |
| Day 3 | 18 (40.9) | 53 (51.5) | 43 (45.3) | 36 (55.4) | 44 (53.0) | |
| Day 5 | 23 (52.3) | 42 (40.8) | 48 (50.5) | 27 (41.5) | 33 (39.8) | |
| Number of embryos transferred3 | 0.40 | |||||
| 1 embryo | 11 (25.6) | 30 (29.1) | 21 (22.1) | 14 (21.5) | 22 (26.5) | |
| 2 embryos | 27 (62.8) | 58 (56.3) | 56 (59.0) | 36 (55.4) | 39 (47.0) | |
| ≥3 embryos | 5 (11.6) | 15 (14.6) | 18 (19.0) | 15 (23.1) | 22 (26.5) | |
1Values are number (%) of women or means ± SDs unless otherwise indicated. FSH, follicle stimulation hormone; IUI, intrauterine insemination; IVF, in vitro fertilization; PCOS, polycystic ovarian syndrome.
2From Kruskal-Wallis test for continuous variables and χ2 tests or Fisher exact tests (when 1 or more cell counts were ≤5) for categorical variables.
3Embryo transfer day and number were assessed only among fresh cycles with embryo transfer.
Women with higher BMIs at 18 y, specifically those with a BMI 22.5–24.9 and ≥25, had significantly lower peak serum estradiol concentrations than women with a BMI of 20–22.4 at 18 y (P-trend = 0.02) (Table 2). BMI categories at 18 y were not associated with endometrial thickness (P-trend = 0.76), total oocyte yield (P-trend = 0.85), or mature oocyte yield (P-trend = 0.69). There was a slight inverse association between BMI at 18 y and fertilization rate (P-trend = 0.05). Higher adult weight gain was associated with lower peak estradiol concentrations (P-trend = 0.01), particularly among women who gained ≥15 kg since 18 y compared with those who remained weight stable. On the other hand, women with higher adult weight gain had slightly higher total and mature oocytes retrieved compared with women who were weight stable (P-trend = 0.05 and P-trend = 0.04, respectively). This association was attenuated when women with polycystic ovarian syndrome (PCOS) were excluded. Endometrial thickness and fertilization rate were not related to adult weight change (P-trend = 0.48 and 0.92, respectively).
TABLE 2.
Association between BMI at 18 y and weight change since 18 y and controlled ovarian stimulation outcomes in the EARTH Study1
| Adjusted means (95% CI)2 | ||||||
|---|---|---|---|---|---|---|
| Characteristic | Women/cycles | Peak serum estradiol, pg/mL | Endometrial thickness, mm | Total oocyte yield, n | Mature oocyte yield, n | Fertilization rate, % |
| BMI at 18 y, kg/m2 | ||||||
| <18.5 | 81/112 | 2140 (1960, 2330) | 9.9 (9.4, 10.4) | 10.3 (9.2, 11.4) | 8.6 (7.7, 9.6) | 74.8 (70.4, 78.8) |
| 18.5–19.9 | 81/114 | 2090 (1910, 2270) | 10.8 (10.3, 11.3) | 10.8 (9.8, 12.0) | 9.2 (8.3, 10.3) | 70.5 (65.9, 74.7) |
| 20–22.4 (reference) | 175/261 | 2240 (2120, 2360) | 10.5 (10.2, 10.9) | 11.5 (10.7, 12.3) | 9.5 (8.9, 10.2) | 71.7 (68.8, 74.5) |
| 22.5–24.9 | 63/103 | 1930 (1730, 2130)3 | 10.1 (9.5, 10.6) | 10.9 (9.7, 12.3) | 8.8 (7.8, 9.9) | 69.1 (63.9, 73.9) |
| ≥25 | 39/61 | 1840 (1590, 2090)3 | 10.2 (9.5, 10.9) | 10.4 (9.0, 12.0) | 8.4 (7.2, 9.8) | 67.4 (60.6, 73.5) |
| P-trend4 | 0.02 | 0.76 | 0.85 | 0.69 | 0.05 | |
| Weight change since 18 y | ||||||
| Lost (<0 kg) | 54/81 | 2150 (1930, 2370) | 10.3 (9.7, 10.9) | 10.3 (9.1, 11.7) | 8.5 (7.4, 9.7) | 68.1 (62.2, 73.4) |
| Stable (gained 0–4.9 kg) (reference) | 115/164 | 2180 (2030, 2330) | 10.2 (9.8, 10.7) | 10.1 (9.3, 11.1) | 8.2 (7.5, 9.0) | 70.5 (66.5, 74.2) |
| Gained 5–9.9 kg | 105/157 | 2190 (2040, 2350) | 10.3 (9.8, 10.7) | 11.2 (10.3, 12.3) | 9.4 (8.6, 10.3) | 74.1 (70.5, 77.4) |
| Gained 10–14.9 kg | 72/104 | 2180 (2000, 2360) | 10.6 (10.1, 11.1) | 11.6 (10.5, 12.9)3 | 10.0 (9.0, 11.0)3 | 72.7 (68.3, 76.7) |
| Gained ≥15 kg | 93/145 | 1870 (1700, 2030)3 | 10.5 (10.0, 10.9) | 11.5 (10.5, 12.7) | 9.4 (8.5, 10.4) | 69.6 (65.4, 73.5) |
| P-trend for weight change | 0.01 | 0.48 | 0.05 | 0.04 | 0.92 | |
| P-trend for weight gain5 | 0.005 | 0.36 | 0.07 | 0.07 | 0.48 | |
1This analysis included only the 651 fresh autologous cycles contributed by 439 women in the Environment and Reproductive Health (EARTH) Study. One cycle was missing data on peak serum estradiol and 3 cycles were missing data on endometrial thickness.
2Adjusted for age, smoking status, race, education, and height. Weight change models were further adjusted for BMI at 18 y.
3 P < 0.05 compared with the reference category.
4Test for trend using the median level of BMI or weight change in each category as a continuous variable in the model.
5Test for trend using median value for each weight change category among women who remained weight stable or gained weight (excludes women who lost weight).
There were no associations between BMI at 18 y and live birth following ART in both categorical exposure (P-trend = 0.20) and nonlinear spline models (P for nonlinearity = 0.34) (Table 3, Figure 2). Weight change since 18 y had a significant nonlinear association with live birth (P for nonlinearity = 0.03). Higher weight gain since 18 y (particularly ≥15 kg) and weight loss were both associated with lower odds of live birth; however, confidence intervals were imprecise for weight loss. Higher BMI at study entry, specifically being overweight or obese, was associated with lower probability of live birth (Supplemental Table 2). To evaluate whether BMI at study entry could explain the association between weight change since 18 y and live birth, we further adjusted our models with an indicator variable for BMI at entry of <25 or ≥25. Results from this model were very similar to the results before adjustment (Supplemental Figure 2), suggesting that attained adult BMI does not entirely explain the observed association between weight change and live birth.
TABLE 3.
Association between BMI at 18 y and weight change since 18 y and probability of implantation, clinical pregnancy, and live birth in the Environment and Reproductive Health Study
| Adjusted probability (95% CI)1 | |||||
|---|---|---|---|---|---|
| Characteristic | Women | Cycles | Successful implantation | Clinical pregnancy | Live birth |
| BMI at 18 y, kg/m2 | |||||
| <18.5 | 92 | 150 | 58.7 (50.1, 66.8) | 52.2 (44.0, 60.3) | 43.4 (35.2, 51.9) |
| 18.5–19.9 | 89 | 153 | 54.0 (45.7, 62.0) | 48.3 (40.5, 56.2) | 40.7 (32.9, 48.9) |
| 20–22.4 (reference) | 197 | 358 | 56.1 (50.7, 61.4) | 46.8 (41.6, 52.0) | 36.6 (31.6, 42.0) |
| 22.5–24.9 | 68 | 129 | 51.1 (42.1, 60.0) | 41.5 (33.3, 50.2) | 32.5 (24.7, 41.4) |
| ≥25 | 40 | 73 | 50.1 (38.4, 61.8) | 47.1 (36.1, 58.5) | 38.9 (28.1, 50.8) |
| P-trend2 | 0.20 | 0.20 | 0.20 | ||
| Weight change since 18 y | |||||
| Lost (<0 kg) | 58 | 104 | 57.7 (47.5, 67.3) | 46.2 (36.8, 56.0) | 38.7 (29.4, 48.9) |
| Stable (gained 0–4.9 kg) (reference) | 129 | 222 | 55.1 (48.2, 61.8) | 47.4 (40.9, 54.1) | 40.1 (33.6, 47.0) |
| Gained 5–9.9 kg | 117 | 213 | 58.9 (51.9, 65.7) | 50.1 (43.3, 56.8) | 37.1 (30.6, 44.1) |
| Gained 10–14.9 kg | 75 | 131 | 57.6 (48.7, 66.1) | 51.3 (42.7, 59.8) | 45.4 (36.6, 54.4) |
| Gained ≥15 kg | 107 | 193 | 47.0 (39.6, 54.5) | 41.9 (34.9, 49.1) | 31.6 (25.1, 38.9) |
| P-trend for weight change | 0.07 | 0.35 | 0.19 | ||
| P-trend for weight gain3 | 0.07 | 0.24 | 0.17 | ||
1Adjusted for age, smoking status, race, education, and height. Weight change models were further adjusted for BMI at 18 y.
2Test for trend was performed using the median level of BMI at 18 y or weight change since 18 y in each category as a continuous variable in the model.
3Test for trend using median value for each weight change category among women who remained weight stable or gained weight (excludes women who lost weight).
FIGURE 2.
Associations between BMI at 18 y (A) and weight change since 18 y (B) and live birth fit using restricted cubic splines in the Environment and Reproductive Health Study. The solid line represents the adjusted odds ratio and the dashed lines represent the 95% CI. The curve for weight change is adjusted for age, smoking status, race, education, and height. Weight change models were further adjusted for BMI at 18 y.
There was evidence of an interaction between BMI at 18 y (BMI <22.5 compared with ≥22.5) and adult weight change on live birth (Supplemental Figure 3). Among women with a BMI <22.5, the curve was nonlinear, with a negative association between weight gain and live birth becoming apparent at ≥20 kg and all amounts of weight loss being associated with lower odds of live birth. In contrast, among women with a BMI ≥22.5 at 18 y, the association between weight change and live birth was linear, with weight loss being associated with higher odds of live birth (although confidence intervals were wide) and all amounts of weight gain being associated with lower odds of live birth. In this linear model, the odds of live birth (95% CI) were 30% (6%, 48%) lower for every 10-kg increase in weight among women who were ≥22.5 at 18 y. When we restricted this analysis to women who were not overweight at 18 y (e.g., BMI at 18 y of 22.5–24.9), the association became slightly stronger (OR: 0.60; 95% CI: 0.40, 0.92 for a 10-kg increase).
When analyses were restricted to embryo transfer cycles (n = 762), fresh autologous cycles (n = 692), and never smokers (n = 640), the inverse association between weight gain and probability of live birth strengthened compared with the original analyses (Supplemental Table 3), particularly comparing the adjusted probability of live birth among women who were weight stable with those who gained ≥15 kg during adulthood. There was no evidence of effect modification by age, infertility diagnosis, or history of ART (data not shown).
Discussion
Higher adult weight gain since 18 y was associated with lower probability of live birth following ART, particularly among women who were heavier in adolescence. The inverse association between weight gain and live birth was also stronger after excluding cryopreservation and donor egg cycles as well as cycles that failed prior to embryo transfer. Regarding the intermediate outcomes of ART outcomes, we found lower peak estradiol concentrations among women in the highest category of weight gain and BMI at 18 y. There were borderline associations between BMI at 18 y and lower fertilization rates as well as weight gain since 18 y and slightly higher total and mature oocytes retrieved. This latter association may in part be due to a higher prevalence of PCOS as weight gain increased.
Several previous studies have underscored the potential long-lasting effects that childhood and adolescent BMI could have on subsequent fertility (8, 16–20). In 2 separate longitudinal studies of children in the United States and Australia, childhood obesity before 12 y of age increased the risk of female infertility (8, 20). Moreover, in multiple large longitudinal studies of US women, higher weight in late adolescence was associated with increased risk of ovulatory infertility (16), lifetime nulliparity (19, 21), and having problems becoming pregnant (18). Jokela et al. (17) also found an inverted J-shaped relation between BMI in adolescence and number of children in adulthood, independent of adult BMI. However, a handful of other studies found no relation between early life adiposity and time to pregnancy (6), risk of infertility (22), or risk of pregnancy loss (5, 18). In concordance with some of our findings, an animal study found that early onset obesity negatively altered reproductive function in adult female rats by reducing hormone concentrations; however, we did not observe any effects on ovarian parameters as observed in their study (23).
There are several reasons why we may have observed no association between adolescent body weight and reproductive success following ART in our study compared with others. First, it may be that earlier onset obesity (e.g., prior to menarche) may exert a stronger influence on later fertility potential than body weight at 18 y either due to cumulative impacts over the life course or due to it being a more sensitive window of development. Second, it is possible that adolescent body weight may be more tightly linked to fertility in younger adult women and in spontaneous conceptions (as opposed to following assisted reproduction). As our cohort primarily consisted of women 35 y and older and all women were undergoing ART, our results may not be directly comparable to the other study populations.
Our findings that indicate a detrimental impact of long-term weight gain on probability of live birth following ART are in line with several other studies conducted among women conceiving without medical assistance. Two cohorts of women from Denmark and the United States found that women who gained weight, particularly ≥15 kg, since adolescence had longer time to pregnancy compared with women who remained weight stable (4, 6). Results from the Nurses’ Health Study II also showed that women who gained ≥20 kg since 18 y had a higher risk of pregnancy loss compared with women who maintained a stable weight (5). Yet a more recent cohort of women from North America did not find any significant differences in fecundability among women gaining ≥40 lb since age 17 y compared with those who were weight stable (7). The fact that the negative association between weight gain and live birth remained after adjustment for BMI at study entry suggests that this association is not completely mediated by attained adult BMI, which has been strongly associated with outcomes of ART (24). We also did not find a significant benefit of long-term weight loss on probability of live birth, which supports our previous report of no benefit of short-term weight loss among women in this cohort (25) as well as the results from the randomized controlled trials of short-term weight loss prior to ART that found no benefit (2, 3).
Given that we relied on the use of self-reported weight at 18 y, there is likely some error in our 2 main exposure measures. Yet prior research has demonstrated a high correlation between self-reported and measured weight in adolescence (11) and that the absolute error in recall of weight at younger ages is not influenced by age, current body weight, current BMI, or educational attainment (26). Therefore, although it is well documented that women on a whole tend to underestimate their weight (27), which may lead to misclassification, we do not expect this to be differential by BMI at 18 y. We also could have underestimated the association between weight status at 18 y and weight change since 18 y and clinical outcomes due to the fact that only women with a BMI ≤40 were allowed to initiate an ART cycle at the MGH fertility center. Based on prevalence estimates from National Health and Nutrition Examination Survey, we would expect this to exclude up to 8% of US women 20–39 y (28). Given the positive correlation between BMI at 18 y and current BMI, it is likely that this exclusion resulted in more overweight/obese adolescents at 18 y being excluded from our analysis. There is also possible concern about selection bias stemming from evaluation of an exposure that could be related to seeking ART. However, a previous study found no association between BMI at 18 y and likelihood of seeking an infertility evaluation and only weak associations between current BMI and likelihood of seeking an infertility evaluation (29). This gives us confidence that our exposures are not strongly associated with selection into our cohort. As this was an observational study, there remains the possibility of residual confounding by factors that were not measured or poorly measured in our study. For example, we did not collect information on a history of weight cycling or dieting. Despite these limitations, the strengths of our study include the standardized assessment of a wide variety of participant characteristics and the ability to investigate clinically relevant reproductive outcomes in a potentially vulnerable subpopulation.
To our knowledge, this is the first study to investigate the association between adolescent BMI or weight change since 18 y with intermediate and clinical outcomes among women undergoing ART. Although we found no association between BMI at 18 y and ART outcomes, higher adult weight gain was associated with lower probability of live birth. These results add to the growing literature of research demonstrating a deleterious effect of adult weight gain on fertility. Future research focused on efforts to prevent weight gain during adulthood is needed to better understand whether these types of interventions may have beneficial effects on fertility.
Supplementary Material
ACKNOWLEDGEMENTS
We would like to thank the study participants and our devoted research staff, including Jennifer B. Ford and Myra G. Keller, whose continued dedication and commitment make this work possible. We would also like to thank Dr. Myriam Afeiche for her contributions to data analysis in the early stages of this project.
The authors’ contributions were as follows—AJG and JEC: formulated the concept and design; AJG, LM, RD, IS, and JEC: were involved in data collection; AJG and MA: analyzed the data; AJG: drafted the manuscript; AJG: had primary responsibility for final content; and all authors: critically revised the manuscript for important intellectual content and read and approved the final manuscript.
Notes
Sources of support: This work was supported by NIH grants R01-ES009718, K99-ES026648, and R00-ES026648 from the National Institute of Environmental Health Sciences and P30-DK046200 from the National Institute of Diabetes and Digestive and Kidney Diseases. The funding sources had no involvement in the study design, collection, analysis, or interpretation of the data; in the writing of the report; and in the decision to submit the article for publication.
Author disclosures: The authors report no conflicts of interest.
Supplemental Figures 1–3 and Supplemental Tables 1–3 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn/.
Abbreviations used: ART, assisted reproductive technology; EARTH, Environment and Reproductive Health; GnRH, gonadotropin-releasing hormone; hCG, human chorionic gonadotropin; MGH, Massachusetts General Hospital; PCOS, polycystic ovarian syndrome.
Contributor Information
Audrey J Gaskins, Department of Epidemiology, Rollins School of Public Health, Emory University, Atlanta, GA, USA.
Mariel Arvizu, Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, MA, USA.
Lidia Mínguez-Alarcón, Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, MA, USA.
Ramace Dadd, Department of Environmental Health, Harvard T.H. Chan School of Public Health, Boston, MA, USA.
Irene Souter, Vincent Obstetrics and Gynecology, Massachusetts General Hospital and Harvard Medical School, Boston, MA, USA.
Jorge E Chavarro, Department of Nutrition, Harvard T.H. Chan School of Public Health, Boston, MA, USA; Department of Epidemiology, Harvard T.H. Chan School of Public Health, Boston, MA, USA; Channing Laboratory, Department of Medicine, Brigham and Women's Hospital and Harvard Medical School, Boston, MA, USA.
for the EARTH Study Team:
Data Availability
The data that support the findings of this study are available from the corresponding author, upon reasonable request. N
References
- 1. Jungheim ES, Travieso JL, Hopeman MM. Weighing the impact of obesity on female reproductive function and fertility. Nutr Rev. 2013;71:S3–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Mutsaerts MA, van Oers AM, Groen H, Burggraaff JM, Kuchenbecker WK, Perquin DA, Koks CA, van Golde R, Kaaijk EM, Schierbeek JM et al. Randomized trial of a lifestyle program in obese infertile women. N Engl J Med. 2016;374:1942–53. [DOI] [PubMed] [Google Scholar]
- 3. Einarsson S, Bergh C, Friberg B, Pinborg A, Klajnbard A, Karlstrom PO, Kluge L, Larsson I, Loft A, Mikkelsen-Englund AL et al. Weight reduction intervention for obese infertile women prior to IVF: a randomized controlled trial. Hum Reprod. 2017;32:1621–30. [DOI] [PubMed] [Google Scholar]
- 4. Wise LA, Rothman KJ, Mikkelsen EM, Sorensen HT, Riis A, Hatch EE. An internet-based prospective study of body size and time-to-pregnancy. Hum Reprod. 2010;25:253–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Gaskins AJ, Rich-Edwards JW, Colaci DS, Afeiche MC, Toth TL, Gillman MW, Missmer SA, Chavarro JE. Prepregnancy and early adulthood body mass index and adult weight change in relation to fetal loss. Obstet Gynecol. 2014;124:662–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Gaskins AJ, Rich-Edwards JW, Missmer SA, Rosner B, Chavarro JE. Association of fecundity with changes in adult female weight. Obstet Gynecol. 2015;126:850–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. McKinnon CJ, Hatch EE, Rothman KJ, Mikkelsen EM, Wesselink AK, Hahn KA, Wise LA. Body mass index, physical activity and fecundability in a North American preconception cohort study. Fertil Steril. 2016;106:451–9. [DOI] [PubMed] [Google Scholar]
- 8. He Y, Tian J, Oddy WH, Dwyer T, Venn AJ. Association of childhood obesity with female infertility in adulthood: a 25-year follow-up study. Fertil Steril. 2018;110:596. [DOI] [PubMed] [Google Scholar]
- 9. Malhotra R, Ostbye T, Riley CM, Finkelstein EA. Young adult weight trajectories through midlife by body mass category. Obesity (Silver Spring). 2013;21:1923–34. [DOI] [PubMed] [Google Scholar]
- 10. Messerlian C, Williams PL, Ford JB, Chavarro JE, Minguez-Alarcon L, Dadd R, Braun JM, Gaskins AJ, Meeker JD, James-Todd T et al. The Environment and Reproductive Health (EARTH) Study: a prospective preconception cohort. Hum Reprod Open. 2018;2018:hoy001. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Troy LM, Hunter DJ, Manson JE, Colditz GA, Stampfer MJ, Willett WC. The validity of recalled weight among younger women. Int J Obes Relat Metab Disord. 1995;19:570–2. [PubMed] [Google Scholar]
- 12. Physical status: the use and interpretation of anthropometry. Report of a WHO Expert Committee. World Health Organ Tech Rep Ser. 1995;854:1–452. [PubMed] [Google Scholar]
- 13. Fitzmaurice GM, Laird NM, Ware JH. Missing data and dropout. In: Applied longitudinal analysis. Hoboken (NJ): John Wiley; 2004.pp. 489–514. [Google Scholar]
- 14. Searle SR, Speed FM, Milliken GA. Population marginal means in the linear model: an alternative to least square means. Am Statist. 1980;34:216–21. [Google Scholar]
- 15. Durrleman S, Simon R. Flexible regression models with cubic splines. Statist Med. 1989;8:551–61. [DOI] [PubMed] [Google Scholar]
- 16. Rich-Edwards JW, Goldman MB, Willett WC, Hunter DJ, Stampfer MJ, Colditz GA, Manson JE. Adolescent body mass index and infertility caused by ovulatory disorder. Am J Obstet Gynecol. 1994;171:171–7. [DOI] [PubMed] [Google Scholar]
- 17. Jokela M, Kivimaki M, Elovainio M, Viikari J, Raitakari OT, Keltikangas-Jarvinen L. Body mass index in adolescence and number of children in adulthood. Epidemiology. 2007;18:599–606. [DOI] [PubMed] [Google Scholar]
- 18. Jacobsen BK, Knutsen SF, Oda K, Fraser GE. Obesity at age 20 and the risk of miscarriages, irregular periods and reported problems of becoming pregnant: the Adventist Health Study-2. Eur J Epidemiol. 2012;27:923–31. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Jacobsen BK, Knutsen SF, Oda K, Fraser GE. Body mass index at age 20 and subsequent childbearing: the Adventist Health Study-2. J Womens Health. 2013;22:460–6. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Jacobs MB, Bazzano LA, Pridjian G, Harville EW. Childhood adiposity and fertility difficulties: the Bogalusa Heart Study. Pediatr Obes. 2017;12:477–84. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Polotsky AJ, Hailpern SM, Skurnick JH, Lo JC, Sternfeld B, Santoro N. Association of adolescent obesity and lifetime nulliparity—the Study of Women's Health Across the Nation (SWAN). Fertil Steril. 2010;93:2004–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22. Lake JK, Power C, Cole TJ. Women's reproductive health: the role of body mass index in early and adult life. Int J Obes. 1997;21:432–8. [DOI] [PubMed] [Google Scholar]
- 23. Sagae SC, Menezes EF, Bonfleur ML, Vanzela EC, Zacharias P, Lubaczeuski C, Franci CR, Sanvitto GL. Early onset of obesity induces reproductive deficits in female rats. Physiol Behav. 2012;105:1104–11. [DOI] [PubMed] [Google Scholar]
- 24. Rittenberg V, Seshadri S, Sunkara SK, Sobaleva S, Oteng-Ntim E, El-Toukhy T. Effect of body mass index on IVF treatment outcome: an updated systematic review and meta-analysis. Reprod Biomed Online. 2011;23:421–39. [DOI] [PubMed] [Google Scholar]
- 25. Chavarro JE, Ehrlich S, Colaci DS, Wright DL, Toth TL, Petrozza JC, Hauser R. Body mass index and short-term weight change in relation to treatment outcomes in women undergoing assisted reproduction. Fertil Steril. 2012;98:109–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Klipstein-Grobusch K, Kroke A, Boeing H. Reproducibility of self-reported past body weight. Eur J Clin Nutr. 1998;52:525–8. [DOI] [PubMed] [Google Scholar]
- 27. Engstrom JL, Paterson SA, Doherty A, Trabulsi M, Speer KL. Accuracy of self-reported height and weight in women: an integrative review of the literature. J Midwifery Womens Health. 2003;48:338–45. [DOI] [PubMed] [Google Scholar]
- 28. Ogden CL, Carroll MD, Curtin LR, McDowell MA, Tabak CJ, Flegal KM. Prevalence of overweight and obesity in the United States, 1999–2004. JAMA. 2006;295:1549–55. [DOI] [PubMed] [Google Scholar]
- 29. Farland LV, Collier AY, Correia KF, Grodstein F, Chavarro JE, Rich-Edwards J, Missmer SA. Who receives a medical evaluation for infertility in the United States?. Fertil Steril. 2016;105:1274–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The data that support the findings of this study are available from the corresponding author, upon reasonable request. N