Body Mass Index and Uterine Fibroid Development: A Prospective Study.

Objective Fibroids are hormonally dependent uterine tumors. The literature on adiposity and fibroid prevalence is inconsistent. Previous work usually combined all those with a body mass index (BMI) ≥30 kg/m2 into a single category and relied on clinically diagnosed fibroids, which misclassifies the many women with undiagnosed fibroids. We used a prospective cohort design with periodic ultrasound screening to investigate associations between repeated measures of BMI and fibroid incidence and growth assessed at each follow-up ultrasound.

The Study of Environment, Lifestyle & Fibroids followed 1693 Black/African American women, ages 23 to 35 years from Detroit, Michigan, with ultrasound every 20 months for 5 years. Measured height and repeated weight measures were used to calculate BMI. Fibroid incidence was modeled using Cox models among those who were fibroid free at the enrollment ultrasound. Fibroid growth was estimated for individual fibroids matched across visits as the difference in log-volume between visits and was modeled using linear mixed models. All models used time-varying BMI and adjusted for time-varying covariates.

Compared with BMI <25 kg/m2, those with BMI 30 to <35 kg/m2 had increased fibroid incidence (adjusted hazard ratio, 1.37; 95% CI, 0.96-1.94), those with BMI ≥40 kg/m2 had reduced incidence (adjusted hazard ratio, 0.61; 95% CI, 0.41-0.90). Fibroid growth had mostly small magnitude associations with BMI.

BMI has a nonlinear association with fibroid incidence, which could be driven by effects of BMI on inflammation and reproductive hormones. More detailed measures of visceral and subcutaneous adiposity and their effects on hormones, DNA damage, and cell death are needed.

uterine fibroid, uterine leiomyoma, body mass index Uterine fibroids are common hormone-dependent noncancerous tumors of the myometrium that occur in most women by menopause (1). Fibroids can cause severe symptoms (eg, heavy bleeding, pelvic pain), especially for those with large tumors. Fibroids are the leading indication for hysterectomy in the United States, with annual health care costs in the billions. African American women develop fibroids at earlier ages than White women, which results in more premenopausal years of tumor growth and, at least partially, accounts for their markedly higher myomectomy and hysterectomy rates for fibroids compared with White women (2, 3). Fibroid development is promoted by both estrogen and progesterone (4). Increased adiposity has numerous impacts on hormonal and metabolic processes, many of which could affect fibroid initiation or growth. These impacts include alterations in menstrual cycle frequency, disruption in ovulation, and changes in concentrations of reproductive and other hormones including estrogen, progesterone, LH, insulin, and leptin, as well as changes in SHBG (5, 6). The majority of fibroids show mutations in the MED12 gene, which codes for 1 of the subunits of the mediator complex that regulates gene transcription (7). Thus, factors that increase DNA damage are likely to increase fibroid initiation. Excess adiposity may be such a factor (8). A recent systematic review of associations between measures of adiposity and the prevalence or incidence of fibroids (n = 24 studies) showed an increased risk of uterine fibroids for those with increased body mass index (BMI), weight gain since age 18 and increased waist-to-hip ratio (9). The association with BMI was nonlinear, with increasing risk until a BMI of approximately 30 kg/m2 and declining risk at higher BMIs. The systematic review, however, showed high heterogeneity (I2 = 76%), suggesting inconsistent associations across studies. The high heterogeneity in existing studies can arise from variation in BMI measurement and in study designs, as well as differences in study populations. Measurement error of BMI is more likely when ascertained through self-report, when a single measure of BMI is used or when the exposure (BMI) is measured after the outcome (fibroid incidence) as in cross-sectional or case-control studies. The range of BMI within a study population varied widely among the studies, and prior studies did not fully explore potential nonlinear effects, particularly at high BMIs (≥40 kg/m2). Detection of fibroids is also prone to misclassification if clinical records or self-report is used. Many people with fibroids have not received a clinical diagnosis (1), therefore many “noncases” will actually be true cases (false negatives, classification bias) (10). Given weight-related stigma, which may limit clinical care, those with higher BMI may be more likely to be misclassified as noncases. To address these concerns, we evaluated the association between measured height and repeated measures of weight (to calculate time-varying BMI) and fibroid incidence and growth in a prospective cohort of young African American women with a wide BMI range (50 kg/m2). Standardized ultrasound evaluations were conducted at approximately 20-month intervals over 5 years to detect and measure fibroids. This study design reduces classification bias and presents an opportunity to evaluate associations among those with higher BMIs.

Study Design The Study of Environment, Lifestyle & Fibroids (SELF) is a prospective cohort designed to identify risk factors for fibroid incidence and growth and has been previously described (11). Given the early onset of fibroids in African American individuals (12), SELF recruitment was restricted to those who self-reported “Black or African American” from a list of racial and ethnic categories. Participants were recruited from the Detroit, Michigan, area from 2010 to 2012 in collaboration with Henry Ford Health. Recruitment details and full eligibility criteria are included in Supplementary Appendix 1 (13). Eligible participants were aged 23 to 34 years with no prior clinical diagnosis of fibroids. SELF enrolled 1693 participants who were then invited to attend 3 follow-up visits. The enrollment and 3 follow-up visits included completion of computer-assisted web and telephone questionnaires, ultrasound examination, anthropometric measures, and collection of biospecimen. Follow-up visits were scheduled every 20 months; participants who missed a visit were encouraged to attend the next visit. Participants who were pregnant at the time of a visit were rescheduled until 3 to 4 months postpartum to optimize ultrasound imaging. Study visits concluded in 2018. SELF had high study retention; 91% of the enrolled cohort attended the final visit, 95% attended at least 2 visits, and 79% attended all 4 study visits (Supplementary Fig. S1 (13)). Participants had a median length of study participation of 4.8 years (25th-75th percentiles: 4.7-5.0 years). SELF was approved by the institutional review boards of the National Institutes of Health and Henry Ford Health (NIH IRB 10EN044). All participants provided written informed consent and received compensation. Measurement of BMI Study staff measured height at the enrollment visit, and weight at every follow-up visit. Height was measured twice to a ¼ inch using a seca stadiometer, with a third measurement if the first 2 measures differed by more than half an inch. The nearest 2 height measures were averaged. When study staff noted difficulty in measuring height because of hairstyle or mobility issues, self-reported height was used instead. Weight was measured twice to the nearest 1∕10 pound, with a third measure if the first 2 measures differed by more than 1 pound. The nearest 2 values were averaged. For participants whose weight exceeded the scale maximum (440 pounds), 440 pounds was assigned. BMI was calculated using weight (kg)/height (m)2. For analysis, BMI was modeled as time-varying and was categorized based on clinical cut points or used continuously. With very few participants with a BMI < 18.5 kg/m2, we used the referent category of BMI <25 kg/m2 for both the categorical and continuous analyses. Uterine Fibroid Assessment Experienced ultrasonographers conducted transvaginal ultrasound examinations at every visit to localize and measure fibroids ≥0.5 cm in diameter. The initial and refresher study trainings of sonographers included a review of factors for distinguishing fibroids from other pathologic changes in the uterus including adenomyosis and polyps, training on the protocol for conducting the examination, and standards for recording the data. Sonographers made 3 separate passes through the uterus and recorded the 3 diameters of tumors during each pass. We used the ellipsoid formula to calculate each fibroid's volume for each pass and then averaged the 3 volumes for analysis. During the examination, the sonographers noted difficulty with visualization (eg, calcifications, shadowing). Sonographers archived video and still images and the lead sonographer reviewed an 8% sample for each sonographer monthly, oversampled for fibroid cases (additional detail in Supplementary Appendix 2 (13)). Among the 1693 participants enrolled in SELF, 1610 had at least 1 follow-up visit. Though women with a prior clinical diagnosis of fibroids were ineligible for study, undiagnosed fibroids were expected; 364 (23%) of the 1610 had a fibroid detected at enrollment. Those with a prevalent fibroid at enrollment were not eligible for the analysis of fibroid incidence but could be included in the analysis of fibroid growth. Fibroid incidence We defined incident fibroids as the first appearance of any fibroid among the participants confirmed to be fibroid free at enrollment (N = 1246). If the sonographer noted factors that could impede the visualization of fibroids (eg, calcification or shadowing, only transabdominal ultrasound), the data were excluded from analysis (∼0.5% of ultrasounds), resulting in incidence data for 1232 participants. Fibroid growth The senior author (D.D.B.) and the lead sonographer identified fibroids that could be matched across 2 successive visits using archived images and fibroid locations. Fibroids could only be matched when an individual had fibroids at an early visit and a later visit. Matched fibroids could include those that were prevalent at enrollment as well as incident fibroids followed over time. Sixty-three percent of all fibroids detected could be matched from 1 visit to a subsequent visit. These matched fibroids were from 434 participants (n = 1357 interval growth measurements from successive visits). We defined fibroid growth as the change in the natural logarithm of the tumor volume (ln-volume) between visits (14). To account for differences in the time between visits, the change in ln-volume was scaled to 18 months (median time between visits was 19 months; 25th-75th percentiles: 18-21) by calculating daily growth rates and multiplying by 540. Further details are provided in Supplementary Appendix 3 (13). Ultrasound data from 44 participants with surgical treatment for fibroids (myomectomy, uterine artery embolization, or hysterectomy) or hysterectomy for nonfibroid-related conditions was excluded following treatment. Other fibroid treatment modalities (ie, GnRH agonists or antagonists) were not reported by SELF participants. Covariate Assessment We identified covariates of interest based on the fibroid literature (15), previous work in this cohort (16, 17), and the availability of data in SELF. We collected covariate data using computer-assisted web and telephone questionnaires at every visit including data on demographic characteristics (educational attainment, household income, employment); reproductive and hormonal exposures (years since last birth, number of births before last, years since last use of depot medroxyprogesterone acetate [DMPA], current use of oral combined contraceptive pills [OCP], age at menarche); and other characteristics (current smoker, exercise category (quintiles based on metabolic equivalents). For analysis, we used time-varying covariates (except for age at menarche) and indicator variables to allow for nonlinear associations (except where noted in table footnotes). All time-varying covariates were updated at the beginning of an interval except for years since last birth and number of births before last, which were updated at the end of the interval to incorporate events during the interval (14). We explored all covariates in adjusted models; those that did not change the effect estimates between BMI and fibroid development were not included in the final adjusted models (18). Covariates and how they were parameterized varied somewhat between models; category cut points for covariates are noted in table footnotes. Missing values for education (N = 2 participants), household income (N = 18 participants), and smoking (N = 1 participant) were imputed using carryover from the surrounding visits. No other covariates used in analysis had missing values. Statistical Analysis Analysis of fibroid incidence We used Cox regression (PHREG in SAS) with age as the time scale and robust sandwich variance, to estimate hazard ratios and 95% 95% CI between categories of BMI (<25 [referent], 25-<30, 30-<35, 35-<40, ≥40 kg/m2) and fibroid incidence. Participants were included from the age at study entry until the age at fibroid detection or the final study visit with censoring for nonfibroid-related hysterectomy. Age-adjusted models were estimated without covariates, then fully adjusted models were estimated with further adjustment for age at menarche, years since last birth, number of births before last, use of DMPA within 2 years, current use of OCP, current smoking, and household income. There was no apparent violation of the proportional hazards assumption. Using the fully adjusted model, we also explored the possible nonlinear association nonparametrically with restricted cubic splines (19) (SAS macro lgtphcurve9). BMI values <25 kg/m2 were assigned a value of 25 and used as the referent. We chose the optimal number of knots (n = 5) by examining the Akaike information criterion. Knot location was chosen using the prespecified percentile values (noted in Figure 1 footnote). To avoid sparse data beyond the highest knot (95th percentile), we restricted the spline analysis to those with BMI values <50 kg/m2. Tests for nonlinearity used the likelihood ratio test, comparing the model with only the linear term to the model with the linear and cubic spline terms. Fibroid growth We used linear mixed models (GLIMMIX in SAS) to estimate associations between categories of BMI and fibroid growth rates while accounting for correlated growth of the same fibroid over time and among fibroids from the same participant (14, 17, 20). Minimally adjusted models included time-varying continuous age, and fibroid characteristics that were strongly associated with growth (16), fibroid size (<0.5, 0.5-<4.2, 4.2-<14.1, ≥14.1 cm3), and number of fibroids (ordinal, 1, 2, ≥3). Fully adjusted models also included adjustment for age at menarche, years since last birth, use of DMPA within 2 years, current use of OCP, employment, and household income. To account for measurement error that varies with fibroid size (21), we estimated the residual variance separately for each category of fibroid volume. To compare growth of fibroids from participants with higher BMI categories compared with the reference category (BMI < 25 kg/m2), we estimated percent difference in volume per 18 months by transforming the model-based estimate (β) using ([exp(β) – 1] × 100). Further details are provided in Supplementary Appendix 3 (13). Sensitivity analyses We conducted a number of sensitivity analyses to evaluate residual confounding and model assumptions. Those diagnosed with polycystic ovary syndrome (PCOS) have higher than average BMI, and the hormonal changes specific to PCOS may impact fibroid development. Similarly, hypertension may cooccur with higher BMI and has been reported to have both positive (22-24) and null (25) associations with fibroids. To address the potential residual confounding by PCOS or hypertension, we removed individuals with a self-reported clinical diagnosis of PCOS during the study and separately removed individuals with a diagnosis of hypertension requiring medication from both incidence and growth analyses. When fitting the Cox model, we assumed incident fibroids became incident at the end of an interval. To test the robustness of our results, we instead assigned incidence to the mid-point of the interval. Finally, we examined the influence of statistical outliers in the growth analysis by removing fibroids with positive or negative growth beyond 3 SDs (14). All statistical analyses were conducted using SAS 9.4 (Cary, NC) and did not include adjustment for multiple comparison.

Participants included in the analysis of fibroid incidence had a mean age of 28.9 (SD 3.4) years. The majority of participants had some education beyond high school (77%) and were employed (60%); however, household incomes were low (46% <$20 000) (Table 1). Mean BMI at enrollment was 33.7 (SD 9.9) with 40% above a BMI of 35 kg/m2 (Table 1). Compared with those included in the incidence analysis, those included in the growth analysis (Supplementary Table S1 (13)) had expected differences in factors associated with fibroids: they were older, more likely to be nulliparous, and to never have used DMPA. The participants included in incidence and those included in growth exhibited similar patterns of characteristics across categories of BMI. Compared with those with lower BMI, those with higher BMI had less educational attainment and lower household income and were less likely to have menarche after age 13 years (Table 1, Supplementary Table S1 (13)). Table 1. | BMI category (kg/m2) | |||||| |---|---|---|---|---|---|---| | Overall | <25 | 25-<30 | 30-<35 | 35-<40 | 40+ | | | Number of participants | 1232 (100%) | 248 (20.1%) | 260 (21.1%) | 231 (18.8%) | 205 (16.6%) | 288 (23.4%) | | BMI, mean (SD) | 33.7 (9.9) | 22.2 (2.1) | 27.6 (1.5) | 32.4 (1.5) | 37.3 (1.6) | 47.6 (7.3) | | Age, mean (SD) | 28.9 (3.4) | 28.3 (3.5) | 29.0 (3.5) | 29.0 (3.3) | 29.2 (3.4) | 29.0 (3.5) | | Educational attainment, N (%) | |||||| | High school/GED or less | 283 (23.0) | 48 (19.4) | 41 (15.8) | 54 (23.4) | 58 (28.3) | 82 (28.5) | | Some college/associates/technical | 635 (51.5) | 118 (47.6) | 131 (50.4) | 115 (49.8) | 108 (52.7) | 163 (56.6) | | Bachelors/masters/doctorate | 314 (25.5) | 82 (33.1) | 88 (33.9) | 62 (26.8) | 39 (19.0) | 43 (14.9) | | Household income, N (%) | |||||| | $50 000 | 188 (15.3) | 51 (20.6) | 48 (18.5) | 40 (17.3) | 21 (10.2) | 28 (9.7) | | Employed, N (%) | 743 (60.3) | 152 (61.3) | 162 (62.3) | 155 (67.1) | 115 (56.1) | 159 (55.2) | | Current smoker, N (%) | 240 (19.5) | 45 (18.2) | 45 (17.3) | 51 (22.1) | 42 (20.5) | 57 (19.8) | | Exercise category, N (%) | |||||| | Low | 199 (16.2) | 51 (20.6) | 50 (19.2) | 28 (12.1) | 28 (13.7) | 42 (14.6) | | Low to moderate | 295 (23.9) | 60 (24.2) | 56 (21.5) | 64 (27.7) | 53 (25.9) | 62 (21.5) | | Moderate | 313 (25.4) | 67 (27.0) | 68 (26.2) | 61 (26.4) | 54 (26.3) | 63 (21.9) | | High | 235 (19.1) | 41 (16.5) | 49 (18.9) | 45 (19.5) | 43 (21.0) | 57 (19.8) | | Very high | 187 (15.2) | 27 (10.9) | 37 (14.2) | 32 (13.9) | 27 (13.2) | 64 (22.2) | | Age at menarche, N (%), y | |||||| | Age <11 | 213 (17.3) | 20 (8.1) | 39 (15.0) | 39 (16.9) | 44 (21.5) | 71 (24.7) | | Age 11 | 252 (20.5) | 49 (19.8) | 44 (16.9) | 50 (21.7) | 40 (19.5) | 69 (24.0) | | Age 12 | 342 (27.8) | 61 (24.6) | 82 (31.5) | 71 (30.7) | 49 (23.9) | 79 (27.4) | | Age 13 | 201 (16.3) | 52 (21.0) | 42 (16.2) | 36 (15.6) | 36 (17.6) | 35 (12.2) | | Age >13 | 224 (18.2) | 66 (26.6) | 53 (20.4) | 35 (15.2) | 36 (17.6) | 34 (11.8) | | Years since last birth, N (%) | |||||| | Never pregnant | 315 (25.6) | 65 (26.2) | 49 (18.9) | 53 (22.9) | 49 (23.9) | 99 (34.4) | | 0 births | 129 (10.5) | 25 (10.1) | 28 (10.8) | 33 (14.3) | 24 (11.7) | 19 (6.6) | | Within 3 y | 50 (4.1) | 10 (4.0) | 14 (5.4) | 13 (5.6) | 4 (2.0) | 9 (3.1) | | 3-12 y | 310 (25.2) | 70 (28.2) | 67 (25.8) | 51 (22.1) | 59 (28.8) | 63 (21.9) | | >12 y | 428 (34.7) | 78 (31.5) | 102 (39.2) | 81 (35.1) | 69 (33.7) | 98 (34.0) | | Births before last,a N (%) | |||||| | 0 birthsb | 890 (72.2) | 181 (73.0) | 187 (71.9) | 167 (72.3) | 140 (68.3) | 215 (74.7) | | 1 birth | 225 (18.3) | 48 (19.4) | 44 (16.9) | 46 (19.9) | 41 (20.0) | 46 (16.0) | | 2 births | 81 (6.6) | 17 (6.9) | 21 (8.1) | 14 (6.1) | 12 (5.9) | 17 (5.9) | | 3 births | 21 (1.7) | 1 (0.4) | 6 (2.3) | 2 (0.9) | 6 (2.9) | 6 (2.1) | | ≥4 births | 15 (1.2) | 1 (0.4) | 2 (0.8) | 2 (0.9) | 6 (2.9) | 4 (1.4) | | Currently using birth control pills, N (%) | 138 (11.2) | 27 (10.9) | 38 (14.6) | 35 (15.2) | 17 (8.3) | 21 (7.3) | | Years since last use of DMPA, N (%) | |||||| | Never | 665 (54.0) | 128 (51.6) | 130 (50.0) | 137 (59.3) | 117 (57.1) | 153 (53.1) | | Within 2 y | 154 (12.5) | 52 (21.0) | 40 (15.4) | 16 (6.9) | 13 (6.3) | 33 (11.5) | | ≥2 y | 413 (33.5) | 68 (27.4) | 90 (34.6) | 78 (33.8) | 75 (36.6) | 102 (35.4) | Missing: Exercise N = 3 (N = 2 BMI < 25, N = 1 BMI 30-<35). Employment N = 2. Abbreviations: BMI, body mass index; DMPA, depot medroxyprogesterone acetate; GED, general educational development. a Most recent birth captured in years since last birth. Only prior births that occurred after age 20 years are included (very early births did not influence fibroid incidence and were not included). b 0 births includes those with no birth (n = 444) and those with no birth before most recent after age 20 years (n = 446). Most fibroids detected in SELF, including undiagnosed fibroids detected at enrollment and incident fibroids detected over follow-up, were small. At the time of detection, incident fibroids had a median volume of 0.6 cm3 (25th-75th percentiles: 0.2-1.4 cm3). Fibroids followed for growth had a median volume of 2.2 cm3 (25th-75th percentiles: 0.7-8.6 cm3) at the beginning of each interval. Average fibroid incidence over each study interval was 10.0%. We observed a nonlinear association between BMI and fibroid incidence. Compared with those with BMI < 25 kg/m2, those with BMI 30-<35 kg/m2 had increased fibroid incidence (adjusted hazard ratio [aHR], 1.37; 95% CI, 0.96-1.94), and participants with BMI ≥ 40 kg/m2 had reduced fibroid incidence (aHR, 0.61; 95% CI, 0.41-0.90) (Table 2). A similar nonlinear association was also apparent in the restricted cubic spline analysis using continuous BMI (P = .0008 indicating nonlinearity) (Fig. 1). Average fibroid growth in SELF was 75.9% (95% CI, 68.8-83.4) volume increase per 18 months. Those with higher BMI generally had reduced growth compared with the referent group (BMI < 25 kg/m2), although the estimates had small magnitude (adjusted difference in growth: −2% to −10%) and the 95% CIs all included the null value (Table 3). Table 2. | BMI Category (kg/m2) | Person-years | N incident | Unadjusteda HR (95% CI) | Adjustedb HR (95% CI) | |---|---|---|---|---| | <25 | 951 | 47 (8.6%) | Ref | Ref | | 25-<30 | 1161 | 61 (9.3%) | 0.95 (0.66-1.38) | 0.96 (0.66-1.40) | | 30-<35 | 1020 | 78 (13.4%) | 1.37 (0.96-1.94) | 1.37 (0.96-1.94) | | 35-<40 | 896 | 56 (10.9%) | 1.13 (0.78-1.65) | 1.10 (0.76-1.61) | | ≥40 | 1318 | 53 (6.9%) | 0.71 (0.49-1.05) | 0.61 (0.41-0.90) | Abbreviations: BMI, body mass index; DMPA, depot medroxyprogesterone acetate; HR, hazard ratio; Y/N, yes/no. a Cox model with age as time scale (starting at age of enrollment), with no further adjustment. b Cox model with age as time scale (starting at age of enrollment), further adjusted for years since last birth (within 3 years, within 3-12 years, >12 years or no birth), number of births before last if occurring after age 20 years (ordinal 0, 1, 2, 3, 4+ births), current smoker (Y/N), income >$20 000 (Y/N), use of DMPA within 2 years (Y/N), age at menarche (<11, 11, 12, 13, ≥14 years), current use of combined oral contraceptive pill (Y/N). Table 3. | BMI category (kg/m2) | N growth intervals | Estimated % difference in growth (95% CI) | | |---|---|---|---| | Minimallya adjusted | Fullyb adjusted | || | <25 | 272 | Reference | Reference | | 25-<30 | 263 | −7.9% (−19.8 to 5.8) | −2.1% (−14.1 to 11.6) | | 30-<35 | 280 | −16.2% (−27.1 to −3.7) | −10.3% (−21.4 to 2.4) | | 35-<40 | 203 | −14.1% (−26.1 to −0.2) | −5.2% (−18.2 to 9.9) | | ≥40 | 341 | −6.5% (−18.2 to 7.0) | −2.9% (−14.6 to 10.4) | Abbreviations: BMI, body mass index; DMPA, depot medroxyprogesterone acetate. a Minimally adjusted includes adjustment for fibroid volume (<0.5 cm3, 0.5-4.19 cm3, 4.2-14.0 cm3, ≥14.1 cm3), number of fibroids (ordinal 1, 2, 3, ≥4), age (continuous). b Fully adjusted model also included years since last birth (<5 years, ≥5 years ago including no birth), years since last use of DMPA (<2 years, ≥2 years/never), income (<$20,000, $20-50,000, ≥$50 000), employment (employed yes/no), age at menarche (ordinal <11, 11, 12, 13, ≥14 years), and current use of combined oral contraceptive pill (yes/no). The sensitivity analyses were consistent with the primary results (Supplementary Tables S2 and S3 (13)). Excluding those with PCOS, or those with hypertension, from the incidence or growth analyses resulted in only minor changes in estimates. In the incidence analysis, assuming fibroid incidence occurred at the mid-point of the interval slightly strengthened the observed increased incidence in those with BMI 30 to <35 kg/m2 (aHR, 1.5; 95% CI, 1.1-2.2) compared with aHR, 1.4 (95% CI, 1.0-1.9) when we assumed incidence at the end of the interval. Removal of statistical outliers in the growth analysis resulted in a strengthening of the reduction in growth for those with higher BMI (adjusted difference in growth: −6% to −12% compared with −2% to −10% with statistical outliers included); however, like the main analysis, the categorized BMI variable, as a whole, was not significant in the model as indicated by a type 3 P > .05.

Compared to those with BMI <25 kg/m2, we found evidence of an almost 40% increased risk of incident fibroids for those with a BMI of 30 to 35 kg/m2 and a similar magnitude of decrease for those with a BMI ≥40 kg/m2. There was little evidence of increased growth of fibroids associated with categories of BMI. Our findings for fibroid incidence demonstrate similarities to work conducted in other large (>500 participants) studies that relied on prevalent or clinically incident fibroids (a new medical diagnosis of fibroids). Previous work found generally nonlinear associations between BMI and clinical incidence or prevalence with risks that increased with increasing BMI and then plateaued (23, 26-31) or decreased (32) for those with higher BMIs. Two studies, both using a dichotomous BMI variable, found null associations (33, 34), and 1 reported only a protective association with higher BMI (35). The magnitude of risk reported in prior studies varied greatly because of a wide range of referent values for BMI (as low as <18.5 kg/m2 and as high as 22 kg/m2 to as high as ≥35 kg/m2). The current analysis is the first able to separately examine BMI categories of 30 to <35, 35 to <40, and ≥40 kg/m2, and we found a significantly reduced in risk at BMI ≥40 kg/m2. This association would have been missed in earlier studies because they combined participants across the 3 obesity categories. The 1 study that found an increasing risk followed by a declining risk similar to SELF was conducted in the Black Women's Health Study and reported a maximal risk estimate of adjusted incidence rate ratio 1.47 (95% CI, 1.11-1.93) at BMI 27.5 to 29.9 kg/m2 with a reduction to adjusted incidence rate ratio 1.21 (95% CI, 0.93-1.58) at BMI ≥32.5 kg/m2 compared with a reference category of BMI < 20 kg/m2 (32). Had the Black Women's Health Study examined categories of BMI above 32.5 kg/m2, the estimated risk may have continued to fall. To our knowledge, we are the first to examine associations between BMI and fibroid growth. These findings increase the ability of clinicians to provide more personalized guidance about the risk of fibroid development within different BMI ranges. Additional research is needed to investigate the impact of weight loss, especially the marked changes that can follow bariatric surgery. The nonlinear association between BMI and fibroid incidence may be driven by plausible biological pathways. Fibroid incidence is a combination of fibroid initiation and subsequent growth of the tumor to a detectable size, but our growth findings suggest that BMI influences fibroid initiation more than fibroid growth. Genetic alteration of a myometrial cell is widely recognized as the initiation event for uterine fibroids, the great majority of which harbor specific mutations in MED12 and/or HMGA2 (7). Increased adiposity has been linked to increased DNA damage (36) and impaired DNA repair (37) through an accompanied increase in chronic inflammation (38) and increased production of reactive oxygen species (ROS), resulting in oxidative stress (39) and reduce DNA repair capacity (37). These ROS effects have been observed with in vitro experiments with uterine tissue; ROS-exposed myometrial tissue develops mutations in the most commonly mutated location in exon 2 of MED12 (40). The associations between increased adiposity and increased genomic instability could explain the observed increase in fibroid incidence for SELF participants with a BMI of 30 to 35 kg/m2. The higher inflammation seen with a BMI >40 kg/m2 may have opposing effects because the effects of ROS on cellular function can vary with the concentration of ROS. Whereas intermediate concentrations of ROS can drive tumorigenesis, high levels result in cell death (reviewed in Sena (41)). Such an increase in cell death with high concentrations of ROS has been demonstrated in fibroid tissue (42) and could help explain the reduction in incidence seen in women with a BMI >40 kg/m2. The reduction in incidence seen at the highest BMI category may also be influenced by decreased concentrations of progesterone, the hormone most critical for fibroid development (43). Among premenopausal women, high BMI can lead to both ovarian and hypothalamus/pituitary dysfunctions that are associated with anovulation, which results in decreased progesterone production (44, 45). Even among those with elevated BMI and regular menstrual cycles with confirmed ovulation, serum and urinary concentrations of progesterone or progesterone metabolites are reduced (6, 46, 47). Individuals with the highest BMI levels may have reduced hormonal support for maintenance of early fibroid lesions and more ROS-induced cell death, both acting to limit fibroid incidence. Finally, the reduction in fibroid incidence in the highest category of BMI could be due, in part, to difficulties detecting small fibroids in those with very high BMIs. Although transvaginal ultrasound has fewer technical issues in those with higher BMI compared with transabdominal ultrasound, the SELF ultrasound technicians did report more issues with visualizing the entire uterus in those with BMI ≥40 kg/m2 (13%) compared with those with a BMI < 25 kg/m2 (4%) (Supplementary Table S4 (13)). However, despite visualization issues, fibroid detection rates varied little by BMI for those with visualization issues. Among those with BMI ≥40 kg/m2, 26% had fibroids detected when there was a report of difficulty with visualization, compared with 28% with fibroids detected when there were no reported difficulties with visualization (Supplementary Table S4 (13)). Additionally, if incident fibroids were systematically missed in those with higher BMI, we would expect that detection would be delayed and the fibroids would be larger at the time of incidence, but this was not the case (Supplementary Table S5 (13)). It was surprising that fibroid growth varied little across BMI categories. The hormonal changes with increased BMI might be expected to limit fibroid growth, whereas the ROS effects might be variable. The estimates for growth associated with all elevated BMI categories were negative, consistent with limited hormonal support, but the magnitude of the growth reduction was small and CIs broad. However, given the number of incident fibroids in this cohort, our growth data are mostly from small fibroids. Differences in growth patterns for smaller and larger fibroids have been noted (16, 48), but to our knowledge, there is no research on differences in their responses to ROS or on the degree of hormonal support required for growth of fibroids of different sizes. A limitation of our study is our reliance on BMI as a measure of adiposity. Individuals with similar BMIs will have a wide range of percent body fat (49) and the anatomic location (eg, visceral, subcutaneous) of the adiposity will also vary (50). Both the percent body fat and the anatomic location of the adiposity may independently influence the associated changes in reproductive hormones, adipokines, and the chronic inflammation that accompanies increased adiposity (50, 51). Increased BMI is also associated with numerous comorbid conditions. We adjusted for current smoking and explored the associations in those without hypertension or PCOS. These factors did not explain the observed associations. Another limitation is that the fibroid growth analysis can only be conducted among fibroids that can be matched across consecutive visits. Fibroids lost between visits (rapidly shrinking fibroids) therefore do not contribute to the analysis, making it more difficult to detect decreases in growth than increases in growth. We did not collect fibroid or myometrial tissue or measure serum concentrations of estradiol or progesterone. Further tissue or biomarker studies will be needed to elucidate possible biologic mechanisms. Our study has important, unique strengths. We conducted the study in African American women who, though at high risk for fibroids, are understudied in the United States. Our use of repeated ultrasound to identify newly developing fibroids in a large epidemiologic study eliminates the biases that are inherent in prior studies, which relied on self-reported fibroid diagnoses or fibroid cases ascertained through medical records. The multiple ultrasound measurements of fibroids also allowed us to assess the association of BMI and fibroid growth. In addition, we used measured height and repeated measures of weight, which are not subject to the same misclassification as self-reported height and weight. Finally, we had a large proportion of our population with high BMI, which provided adequate power to examine for the first time both categorical and continuous associations with fibroid development across this broad spectrum.

In conclusion, compared with those with BMI values <25 kg/m2, we find that those with BMIs between 30 and 35 kg/m2 have an increased risk of fibroid incidence, whereas those with BMIs ≥40 kg/m2 have a decreased risk of fibroid incidence. The association with incidence could plausibly be a result of adiposity-related increased DNA damage and reduced DNA repair that could initiate tumorigenesis. Given the high variability in percent and location of body fat within categories of BMI, studies with more detailed assessment of body fat are needed to further understand the nonlinear association between adiposity and fibroid incidence and to look further for evidence of adiposity-related associations with fibroid growth. Acknowledgments The authors thank J. Julie Kim, Hazel Nichols, and Che-Jung Chang for helpful comments on a prior draft of this paper. Abbreviations - aHR adjusted hazard ratio - BMI body mass index - DMPA depot medroxyprogesterone acetate - OCP oral combined contraceptive pill - PCOS polycystic ovary syndrome - ROS reactive oxygen species - SELF Study of Environment, Lifestyle & Fibroids Contributor Information Quaker E Harmon, Epidemiology Branch, National Institute of Environmental Health Sciences, Durham, NC 27709, USA. Stacy Patchel, Westat, Public Health Practice, Durham, NC 27713, USA. Sheri Denslow, Social & Scientific Systems Inc., a DLH Holdings Company, Durham, NC 27703, USA. Ganesa Wegienka, Department of Public Health Sciences, Henry Ford Health, Detroit, MI 48202, USA. Donna D Baird, Epidemiology Branch, National Institute of Environmental Health Sciences, Durham, NC 27709, USA. Funding This work was supported by the Intramural Research Program of the National Institutes of Health, National Institute of Environmental Health Sciences (ZIAES09013), and funds from the American Recovery and Reinvestment Act funds designated for National Institutes of Health Research. Disclosures The authors declare no conflict of interest. Data Availability Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request. Participant data that underlie the primary results reported in this article can be requested for purposes of replication or meta-analysis by emailing [email protected] and [email protected]. All data releases will require a data use proposal and a data use agreement and will comply with the IRB and consent of the SELF study, which may require omission of some data elements. The NIH IRB may be asked to review the request and study consent forms to approve a data transfer.

- 1. Stewart EA, Cookson CL, Gandolfo RA, Schulze-Rath R. Epidemiology of uterine fibroids: a systematic review. BJOG. 2017;124(10):1501‐1512. [DOI] [PubMed] [Google Scholar] - 2. Laughlin SK, Baird DD, Savitz DA, Herring AH, Hartmann KE. Prevalence of uterine leiomyomas in the first trimester of pregnancy: an ultrasound-screening study. Obstet Gynecol. 2009;113(3):630‐635. [DOI] [PMC free article] [PubMed] [Google Scholar] - 3. Laughlin-Tommaso SK, Jacoby VL, Myers ER. Disparities in fibroid incidence, prognosis, and management. Obstet Gynecol Clin North Am. 2017;44(1):81‐94. [DOI] [PubMed] [Google Scholar] - 4. Flake GP, Andersen J, Dixon D. Etiology and pathogenesis of uterine leiomyomas: a review. Environ Health Perspect. 2003;111(8):1037‐1054. [DOI] [PMC free article] [PubMed] [Google Scholar] - 5. Lash MM, Armstrong A. Impact of obesity on women's health. Fertil Steril. 2009;91(5):1712‐1716. [DOI] [PubMed] [Google Scholar] - 6. Yeung EH, Zhang C, Albert PS, et al. Adiposity and sex hormones across the menstrual cycle: the Biocycle Study. Int J Obes (Lond). 2013;37(2):237‐243. [DOI] [PMC free article] [PubMed] [Google Scholar] - 7. Mehine M, Mäkinen N, Heinonen HR, Aaltonen LA, Vahteristo P. Genomics of uterine leiomyomas: insights from high-throughput sequencing. Fertil Steril. 2014;102(3):621‐629. [DOI] [PubMed] [Google Scholar] - 8. Włodarczyk M, Nowicka G. Obesity, DNA damage, and development of obesity-related diseases. Int J Mol Sci. 2019;20(5):1146. [DOI] [PMC free article] [PubMed] [Google Scholar] - 9. Qin H, Lin Z, Vásquez E, Luan X, Guo F, Xu L. Association between obesity and the risk of uterine fibroids: a systematic review and meta-analysis. J Epidemiol Community Health. 2021;75(2):197‐204. [DOI] [PubMed] [Google Scholar] - 10. Baird DD, Dunson DB, Hill MC, Cousins D, Schectman JM. High cumulative incidence of uterine leiomyoma in Black and White women: ultrasound evidence. Am J Obstet Gynecol. 2003;188(1):100‐107. [DOI] [PubMed] [Google Scholar] - 11. Baird DD, Harmon QE, Upson K, et al. A prospective, ultrasound-based study to evaluate risk factors for uterine fibroid incidence and growth: methods and results of recruitment. J Womens Health (Larchmt). 2015;24(11):907‐915. [DOI] [PMC free article] [PubMed] [Google Scholar] - 12. Marsh EE, Ekpo GE, Cardozo ER, Brocks M, Dune T, Cohen LS. Racial differences in fibroid prevalence and ultrasound findings in asymptomatic young women (18-30 years old): a pilot study. Fertil Steril. 2013;99(7):1951‐1957. [DOI] [PMC free article] [PubMed] [Google Scholar] - 13. Harmon QE, Patchel S, Denslow S, Wegienka G, Baird DD. Supplemental material: Body mass index and uterine fibroid development: a prospective study. Deposited 4 January 2024. Figshare Repository. 10.6084/m9.figshare.24943713 [DOI] [PMC free article] [PubMed] - 14. Peddada SD, Laughlin SK, Miner K, et al. Growth of uterine leiomyomata among premenopausal Black and White women. Proc Natl Acad Sci U S A. 2008;105(50):19887‐19892. [DOI] [PMC free article] [PubMed] [Google Scholar] - 15. Wise LA, Laughlin-Tommaso SK. Epidemiology of uterine fibroids: from menarche to menopause. Clin Obstet Gynecol. 2016;59(1):2‐24. [DOI] [PMC free article] [PubMed] [Google Scholar] - 16. Baird DD, Patchel SA, Saldana TM, et al. Uterine fibroid incidence and growth in an ultrasound-based, prospective study of young African Americans. Am J Obstet Gynecol. 2020;223(3):402.e1‐402.e18. [DOI] [PMC free article] [PubMed] [Google Scholar] - 17. Harmon QE, Patchel SA, Denslow S, et al. Vitamin D and uterine fibroid growth, incidence, and loss: a prospective ultrasound study. Fertil Steril. 2022;118(6):1127‐1136. [DOI] [PMC free article] [PubMed] [Google Scholar] - 18. Mickey RM, Greenland S. The impact of confounder selection criteria on effect estimation. Am J Epidemiol. 1989;129(1):125‐137. [DOI] [PubMed] [Google Scholar] - 19. Durrleman S, Simon R. Flexible regression models with cubic splines. Stat Med. 1989;8(5):551‐561. [DOI] [PubMed] [Google Scholar] - 20. Harmon QE, Patchel SA, Zhao S, Umbach DM, Cooper TE, Baird DD. Depot medroxyprogesterone acetate use and the development and progression of uterine leiomyoma. Obstet Gynecol. 2022;139(5):797‐807. [DOI] [PMC free article] [PubMed] [Google Scholar] - 21. Moshesh M, Peddada SD, Cooper T, Baird D. Intraobserver variability in fibroid size measurements: estimated effects on assessing fibroid growth. J Ultrasound Med. 2014;33(7):1217‐1224. [DOI] [PMC free article] [PubMed] [Google Scholar] - 22. Faerstein E, Szklo M, Rosenshein NB. Risk factors for uterine leiomyoma: a practice-based case-control study. II. Atherogenic risk factors and potential sources of uterine irritation. Am J Epidemiol. 2001;153(1):11‐19. [DOI] [PubMed] [Google Scholar] - 23. Templeman C, Marshall SF, Clarke CA, et al. Risk factors for surgically removed fibroids in a large cohort of teachers. Fertil Steril. 2009;92(4):1436‐1446. [DOI] [PMC free article] [PubMed] [Google Scholar] - 24. Haan YC, Diemer FS, Van Der Woude L, Van Montfrans GA, Oehlers GP, Brewster LM. The risk of hypertension and cardiovascular disease in women with uterine fibroids. J Clin Hypertens (Greenwich). 2018;20(4):718‐726. [DOI] [PMC free article] [PubMed] [Google Scholar] - 25. Radin RG, Rosenberg L, Palmer JR, Cozier YC, Kumanyika SK, Wise LA. Hypertension and risk of uterine leiomyomata in US Black women. Hum Reprod. 2012;27(5):1504‐1509. [DOI] [PMC free article] [PubMed] [Google Scholar] - 26. Faerstein E, Szklo M, Rosenshein N. Risk factors for uterine leiomyoma: a practice-based case-control study. I. African-American heritage, reproductive history, body size, and smoking. Am J Epidemiol. 2001;153(1):1‐10. [DOI] [PubMed] [Google Scholar] - 27. Terry KL, De Vivo I, Hankinson SE, Missmer SA. Reproductive characteristics and risk of uterine leiomyomata. Fertil Steril. 2010;94(7):2703‐2707. [DOI] [PMC free article] [PubMed] [Google Scholar] - 28. Parazzini F. Risk factors for clinically diagnosed uterine fibroids in women around menopause. Maturitas. 2006;55(2):174‐179. [DOI] [PubMed] [Google Scholar] - 29. Yang Y, He Y, Zeng Q, Li S. Association of body size and body fat distribution with uterine fibroids among Chinese women. J Womens Health (Larchmt). 2014;23(7):619‐626. [DOI] [PubMed] [Google Scholar] - 30. Baird DD, Dunson DB, Hill MC, Cousins D, Schectman JM. Association of physical activity with development of uterine leiomyoma. Am J Epidemiol. 2007;165(2):157‐163. [DOI] [PubMed] [Google Scholar] - 31. Lee JE, Song S, Cho E, et al. Weight change and risk of uterine leiomyomas: Korea Nurses’ Health Study. Curr Med Res Opin. 2018;34(11):1913‐1919. [DOI] [PubMed] [Google Scholar] - 32. Wise LA, Palmer JR, Spiegelman D, et al. Influence of body size and body fat distribution on risk of uterine leiomyomata in U.S. Black women. Epidemiology. 2005;16(3):346‐354. [DOI] [PMC free article] [PubMed] [Google Scholar] - 33. Chen CR, Buck GM, Courey NG, Perez KM, Wactawski-Wende J. Risk factors for uterine fibroids among women undergoing tubal sterilization. Am J Epidemiol. 2001;153(1):20‐26. [DOI] [PubMed] [Google Scholar] - 34. Samadi AR, Lee NC, Flanders WD, Boring JR III, Parris EB. Risk factors for self-reported uterine fibroids: a case-control study. Am J Public Health. 1996;86(6):858‐862. [DOI] [PMC free article] [PubMed] [Google Scholar] - 35. Parazzini F, Chiaffarino F, Polverino G, Chiantera V, Surace M, La Vecchia C. Uterine fibroids risk and history of selected medical conditions linked with female hormones. Eur J Epidemiol. 2004;19(3):249‐253. [DOI] [PubMed] [Google Scholar] - 36. Włodarczyk M, Jabłonowska-Lietz B, Olejarz W, Nowicka G. Anthropometric and dietary factors as predictors of DNA damage in obese women. Nutrients. 2018;10(5):578. [DOI] [PMC free article] [PubMed] [Google Scholar] - 37. Tyson J, Caple F, Spiers A, et al. Inter-individual variation in nucleotide excision repair in young adults: effects of age, adiposity, micronutrient supplementation and genotype. Br J Nutr. 2009;101(9):1316‐1323. [DOI] [PubMed] [Google Scholar] - 38. Gregor MF, Hotamisligil GS. Inflammatory mechanisms in obesity. Annu Rev Immunol. 2011;29(1):415‐445. [DOI] [PubMed] [Google Scholar] - 39. Furukawa S, Fujita T, Shimabukuro M, et al. Increased oxidative stress in obesity and its impact on metabolic syndrome. J Clin Invest. 2004;114(12):1752‐1761. [DOI] [PMC free article] [PubMed] [Google Scholar] - 40. Li Y, Xu X, Asif H, et al. Myometrial oxidative stress drives MED12 mutations in leiomyoma. Cell Biosci. 2022;12(1):111. [DOI] [PMC free article] [PubMed] [Google Scholar] - 41. Sena LA, Chandel NS. Physiological roles of mitochondrial reactive oxygen species. Mol Cell. 2012/10/26/2012;48(2):158‐167. [DOI] [PMC free article] [PubMed] [Google Scholar] - 42. Vidimar V, Gius D, Chakravarti D, Bulun SE, Wei JJ, Kim JJ. Dysfunctional mnsod leads to redox dysregulation and activation of prosurvival AKT signaling in uterine leiomyomas. Sci Adv. 2016;2(11):e1601132. [DOI] [PMC free article] [PubMed] [Google Scholar] - 43. Bulun SE, Moravek MB, Yin P, et al. Uterine leiomyoma stem cells: linking progesterone to growth. Semin Reprod Med. 2015;33(5):357‐365. [DOI] [PubMed] [Google Scholar] - 44. Bloom MS, Perkins NJ, Sjaarda LA, et al. Adiposity is associated with anovulation independent of serum free testosterone: a prospective cohort study. Paediatr Perinat Epidemiol. 2021;35(2):174‐183. [DOI] [PMC free article] [PubMed] [Google Scholar] - 45. Santoro N, Polotsky AJ, Rieder J, Kondapalli LA. Chapter 18—nutrition and reproduction. In: Strauss JF, Barbieri RL, eds. Yen and Jaffe's Reproductive Endocrinology. 8th edi. Elsevier; 2019:447‐458.e446. [Google Scholar] - 46. Oldfield AL, Vanden Brink H, Carter FE, Jarrett BY, Lujan ME. Obesity is associated with alterations in antral follicle dynamics in eumenorrheic women. Hum Reprod. 2023;38(3):459‐470. [DOI] [PMC free article] [PubMed] [Google Scholar] - 47. Jain A, Polotsky AJ, Rochester D, et al. Pulsatile luteinizing hormone amplitude and progesterone metabolite excretion are reduced in obese women. J Clin Endocrinol Metab. 2007;92(7):2468‐2473. [DOI] [PubMed] [Google Scholar] - 48. Flake GP, Moore AB, Sutton D, et al. The life cycle of the uterine fibroid myocyte. Curr Obstet Gynecol Rep. 2018;7(2):97‐105. [DOI] [PMC free article] [PubMed] [Google Scholar] - 49. Flegal KM, Shepherd JA, Looker AC, et al. Comparisons of percentage body fat, body mass index, waist circumference, and waist-stature ratio in adults. Am J Clin Nutr. 2009;89(2):500‐508. [DOI] [PMC free article] [PubMed] [Google Scholar] - 50. Powell-Wiley TM, Poirier P, Burke LE, et al. Obesity and cardiovascular disease: a scientific statement from the American heart association. Circulation. 2021;143(21):e984‐e1010. [DOI] [PMC free article] [PubMed] [Google Scholar] - 51. Hetemäki N, Mikkola TS, Tikkanen MJ, et al. Adipose tissue estrogen production and metabolism in premenopausal women. J Steroid Biochem Mol Biol. 2021;209:105849. [DOI] [PubMed] [Google Scholar] Associated Data This section collects any data citations, data availability statements, or supplementary materials included in this article. Data Citations - Harmon QE, Patchel S, Denslow S, Wegienka G, Baird DD. Supplemental material: Body mass index and uterine fibroid development: a prospective study. Deposited 4 January 2024. Figshare Repository. 10.6084/m9.figshare.24943713 [DOI] [PMC free article] [PubMed] Data Availability Statement Some or all datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request. Participant data that underlie the primary results reported in this article can be requested for purposes of replication or meta-analysis by emailing [email protected] and [email protected]. All data releases will require a data use proposal and a data use agreement and will comply with the IRB and consent of the SELF study, which may require omission of some data elements. The NIH IRB may be asked to review the request and study consent forms to approve a data transfer.