Association of Physical Activity with Development of Uterine Leiomyoma

Abstract

The relation between physical activity and uterine leiomyomata (fibroids) has received little study, but exercise is protective for breast cancer, another hormonally mediated tumor. Participants in this study were randomly selected members of a health plan based in Washington, DC, aged 35–49 years (734 African Americans, 455 Whites) enrolled between 1996 and 1999. Fibroid status was based on ultrasound screening. Physical activity was based on detailed interview questions. Logistic regression with adjustment for body mass index and other risk factors showed that women in the highest category of physical activity were significantly less likely to have fibroids (odds ratio = 0.6, 95% confidence interval = 0.4, 0.9 for the highest vs. the lowest category (equivalent to approximately ≥7 hours/week vs. <2 hours/week)). There was a dose-response pattern; a significant trend was seen for both African-American and White women. A multistate Bayesian analysis indicated that exercise was associated with tumor onset more strongly than with tumor growth. When data for women who reported major fibroid-related symptoms were excluded, results remained essentially unchanged, suggesting that the observed association could not be attributed to reverse causation (fibroids preventing exercise). The authors concluded that regular exercise might help women prevent fibroids.

body mass index, causality, exercise, leiomyoma, ultrasonics

Uterine leiomyomata, commonly called fibroids, are the leading indication for hysterectomy in the United States (1). Fibroid-related gynecologic problems include uterine bleeding, pelvic pain/pressure, infertility, and pregnancy complications (2). The tumors are benign, are of smooth muscle origin, and develop as clonal growths (3). Incidence increases with age up to menopause, after which fibroids tend to regress (4, 5). Both estrogen and progesterone may stimulate their development, although the mechanisms by which these hormones influence tumor onset and tumor growth are not well understood (reviewed by Schwartz and Marshall (6) and Schwartz et al. (7)).

The epidemiology of uterine fibroids has received limited research attention (8, 9). It is a common condition, especially for African Americans, a group with estimated hysterectomy rates for fibroids of 20 percent (10). Consistently reported risk factors are age, African-American ethnicity, and early age at menarche, none of which are modifiable for women concerned about their own risk (9). One of the modifiable risk factors that has been reported for breast cancer and endometrial cancer, other hormonally mediated tumors, is physical activity (reviewed by Lee (11) and Matthews et al. (12)), but it has been studied very little in relation to uterine fibroids. The authors tested the hypothesis that physical activity reduces fibroid development by collecting detailed exercise information in an epidemiologic study of uterine fibroids.

MATERIALS AND METHODS

The National Institute of Environmental Health Sciences Uterine Fibroid Study was designed to estimate age-specific cumulative incidence of uterine fibroids and to assess risk factors for fibroids. Detailed methods have been described previously (13). Briefly, the computerized membership records of a prepaid health plan in Washington, DC, were used to randomly select women aged 35–49 years. Those selected were screened for eligibility by telephone interview. Eligibility criteria were as follows: 1) the computerized listing had correctly identified a current member aged 35–49 years, and 2) a telephone interview could be conducted in English. This analysis was limited to African Americans and non-Hispanic Whites because over 90 percent of participants self-identified in these two groups. Eighty-three percent of African-American women and White women who had been identified as eligible agreed to participate (figure 1). The research was approved by the National Institute of Environmental Health Sciences and George Washington University Human Subject Review Boards, and participants gave informed consent.

FIGURE 1

Flowchart showing the number of African-American and White women in the Uterine Fibroid Study (enrollment, 1996–1999) with data on fibroid status, Washington, DC. HMO, health maintenance organization.

To assess fibroid status, premenopausal women both with and without any prior diagnosis of fibroids were asked to undergo ultrasound screening for fibroids. Medical record review was conducted for women who had undergone a hysterectomy. If neither ultrasound nor medical record review could be conducted, self-report of a prior diagnosis of uterine fibroids was accepted. Self-report of “no fibroids” was not accepted because undiagnosed fibroids were common (about half of the undiagnosed women in our study were found to have fibroids at ultrasound screening) (13). The study obtained sonogram data for 76 percent of participants and medical record data for an additional 7 percent. Fibroid status relied on self-report for 9 percent of African-American women and 4 percent of White women. Fibroid status could not be ascertained for 10 percent of both ethnic groups, so data for these women were excluded from the analyses.

Data on physical activity were collected by computer-assisted telephone interview (available at http://dir.niehs.nih.gov/direb/studies/ufs/question.htm). Women were asked for current number of hours per week spent in each of three categories of nonoccupational physical activity: 1) vigorous activity such as running, 2) moderate activity such as dancing, and 3) walking. Participants also estimated the current number of hours per week doing household chores. Time spent doing a specific activity (e.g., running) was not reported, only the estimated time spent doing each category of activity. Less detailed data were collected about nonoccupational physical activity at age 30 years and as a teenager. The data for each time period were combined into summary variables based on estimated metabolic rates for each category of activity (14) and were then standardized to vigorous activity (refer to the supplemental information posted on the Journal's website (http://aje.oupjournals.org/)). Two summary physical activity variables for time of enrollment were calculated: one without household chores, one with household chores. The physical activity variables were categorized as follows: low (lowest 33 percent of the combined African-American and White distribution), medium (middle 33 percent), high (those in the 67th–83rd percentile), and very high (those above the 83rd percentile).

Data on most potential confounders were collected during the telephone interview (reproductive history, age at menarche, cigarette smoking, and height) or by self-administered questionnaire (education and alcohol intake). Weight was measured at the clinic visit. For those not attending the clinic (postmenopausal women and 4 percent of premenopausal women), weight was self-reported at interview. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared (kg/m²).

The relation between physical activity and uterine fibroids was evaluated by using logistic regression and Bayesian analyses. Because African-American women have a significantly higher risk of fibroids and substantially different distributions of many covariates, analyses were carried out separately for the two ethnic groups. Logistic regression analysis was used to estimate the age-adjusted odds and then the multivariate-adjusted odds of fibroids associated with variation in physical activity and BMI. Estimates were generated for women whose largest fibroid was small (<2 cm diameter), medium (2 but <4 cm), or large (≥4 cm) versus those without fibroids (CATMOD procedure, SAS v.9.1; SAS Institute, Inc., Cary, North Carolina). All p values are two sided. In Bayesian analyses, the survivorship approach of Dunson and Baird (15) was used to evaluate the effects of physical activity on tumor onset and tumor growth separately. For this survivorship approach, a woman's data on age at any prior diagnosis and size of fibroids at study ultrasound were used to distinguish between onset and growth. Data were censored at natural menopause, hysterectomy, or ultrasound examination. Heuristically, after the model was adjusted for age and time since any previous diagnosis, women with large fibroids will contribute to tumor growth more than to tumor onset and women with small fibroids will contribute more to onset than to growth (details described in the supplemental information on the Journal's website).

The potential confounders included a priori in analyses because of their known or suspected relation to fibroids were age, BMI, age at menarche, and number of deliveries after age 24 years (earlier births were not significantly related to fibroid development (16, 17)). Education, alcohol intake, and smoking were evaluated as potential confounders but were found not to affect either the physical activity or BMI associations, so they were not included in analyses.

RESULTS

Characteristics of the sample are shown in table 1. Compared with Whites, African Americans reported less vigorous and moderate activity but more walking and chores (refer to the supplemental information). This finding resulted in the weighted summary measures being similar for the two groups. Both ethnic groups reported more activity as teenagers and at age 30 years than at age at enrollment.

TABLE 1

Characteristics of participants in the Uterine Fibroid Study (enrollment, 1996–1999), Washington, DC

Characteristic	African Americans (n = 734)		Whites (n = 455)
	No.	%	No.	%
Fibroid status (size)
None	166	23	207	45
Small (<2 cm diameter)	112	15	77	17
Medium (2–<4 cm diameter)	233	32	99	22
Large (≥4 cm diameter)	223	30	72	16
Source of fibroid status data
Ultrasound	591	80	408	90
Surgical report	76	10	14	4
Self-report	67	9	40	6
Age (years)
35–39	236	32	140	31
40–44	257	35	150	33
45–49	241	33	165	36
Education
High school or less	152	22	14	3
Some beyond high school	337	48	38	9
College degree	82	12	74	17
Postbaccalaureate	131	19	310	71
Missing	32		19
Age at menarche (years)
<11	85	12	23	5
11	130	18	68	15
12	202	28	129	28
13	165	23	148	33
14	69	9	50	11
>14	80	11	35	8
Missing	3		2
Parous
No	157	21	263	58
Yes	577	79	192	42
No. of full-term pregnancies delivered after age 24 years
0	383	52	293	64
1	225	31	62	14
2	101	14	87	19
≥3	25	3	13	3
Menopausal status
Premenopausal	620	84	416	91
Naturally postmenopausal	6	1	4	1
Surgically postmenopausal*	105	14	34	7
Postmenopausal—other†	3	<1	1	<1
Smoking status
Never	349	48	263	58
Past	170	23	156	34
Current, <10 cigarettes/day	123	17	16	4
Current, 10–19 cigarettes/day	74	10	14	3
Current, ≥20 cigarettes/day	13	2	6	1
Missing	5		0
Alcohol (drinks/week)
<0.5	366	55	71	17
0.5–2	144	22	125	30
>2–<7	74	11	111	27
≥7	76	12	105	25
Missing	74		43
Activity level for type of work performed the longest
Mostly sit or stand	473	64	340	75
Walk at least half the time	196	27	90	20
Participate in some strenuous activity	65	9	25	5
Body mass index‡
<25	176	24	262	58
25–29.99	216	30	110	24
30–34.99	153	21	43	9
≥35	185	25	40	9
Missing	4		0
Physical activity, without chores (indexed to estimated hours/week of vigorous activity)
Low (<1.85)	276	38	121	26
Medium (1.85–4.14)	232	32	155	34
High (4.15–6.55)	111	15	94	21
Very high (>6.55)	111	15	84	19
Missing	4		1
Physical activity, with chores (indexed to estimated hours/week of vigorous activity)
Low (<3.15)	247	37	131	30
Medium (3.15–5.84)	201	31	162	37
High (5.85–8.45)	107	16	67	15
Very high (>8.45)	107	16	73	17
Missing	72		22

*

Most women defined as surgically menopausal had had a hysterectomy (with or without removal of ovaries); only two had had both ovaries removed but had an intact uterus.

†

Menopausal because of treatment (e.g., chemotherapy).

‡

Weight (kg)/height (m)².

On the basis of logistic regression analyses, current physical activity was inversely associated with fibroids (table 2). BMI was positively associated with fibroids, but only in African-American women. The age-adjusted and multivariate-adjusted estimates for the activity variables were very similar, but adjustment tended to reduce the estimates for BMI. Including household chores in the summary measure of physical activity strengthened its association with fibroids in African Americans but attenuated the association in Whites. Variables for physical activity and BMI at younger ages were not related to fibroids in either racial/ethnic group (refer to the supplemental data). Physical activity was inversely related to all size categories of fibroids (figure 2). The several sensitivity analyses that were performed showed that these results changed little with restriction to premenopausal women, minor changes in the fibroid outcome variable, control for physical activity at work, and exclusion of women who reported fibroid-related symptoms that might have interfered with their ability to exercise (refer to the supplemental information).

FIGURE 2

Adjusted odds ratios for fibroids (all sizes of fibroids as well as small, medium, and large fibroids) associated with physical activity (top panel) and body mass index (BMI; weight (kg)/height (m)²) (bottom panel) among African Americans and Whites in the Uterine Fibroid Study (enrollment, 1996–1999), Washington, DC. The physical activity variables are the summary variables for activity at time of enrollment. The variable for African Americans includes chores; the variable for Whites excludes chores. For the BMI panel, few White women had a BMI of ≥35, so the third sets of odds ratios and 95% confidence intervals indicate a BMI of ≥30. The odds ratios were adjusted for age, age at menarche, number of full-term pregnancies after age 24 years, BMI (for the physical activity analysis), and physical activity (for the BMI analysis). Horizontal lines, 95% confidence intervals. The values to the right and left (separated by a colon) show the numbers of African Americans and Whites, respectively, with and without fibroids in each of the exercise and BMI categories.

TABLE 2

Age-adjusted and multivariate-adjusted relative odds of fibroids for African Americans and Whites in different exercise and body mass index categories, Uterine Fibroid Study (enrollment, 1996–1999), Washington, DC

Category	African Americans							Whites
	No.*	OR†	95% CI†	No.	aOR‡	95% CI	p trend	No.	OR†	95% CI	No.	aOR‡	95% CI	p trend
Physical activity, without chores
Low	222:54	1.00	Reference	219:54	1.00	Reference		71:50	1.00	Reference	71:50	1.00	Reference
Medium	183:49	0.92	0.59, 1.41	181:49	0.96	0.62, 1.50		94:61	1.10	0.67, 1.80	94:61	1.25	0.75, 2.08
High	80:31	0.63	0.38, 1.05	79:31	0.70	0.42, 1.18		47:47	0.75	0.43, 1.30	46:47	0.76	0.42, 1.34
Very high	80:31	0.64	0.38, 1.07	79:31	0.66	0.39, 1.12	0.07	36:48	0.57	0.32, 1.01	35:48	0.55	0.30, 1.02	0.02
Physical activity, with chores
Low	199:48	1.00	Reference	195:48	1.00	Reference		71:60	1.00	Reference	71:60	1.00	Reference
Medium	159:42	0.90	0.57, 1.44	159:42	0.98	0.61, 1.56		96:66	1.18	0.75, 1.88	96:66	1.40	0.86, 2.29
High	74:33	0.54	0.32, 0.91	72:33	0.59	0.35, 1.00		39:28	1.13	0.63, 2.04	39:28	1.40	0.75, 2.63
Very high	75:32	0.58	0.34, 0.97	74:32	0.62	0.36, 1.05	0.02	30:43	0.66	0.37, 1.17	29:43	0.69	0.37, 1.28	0.37
Body mass index§ at enrollment
<25	122:54	1.00	Reference	121:53	1.00	Reference		138:124	1.00	Reference	137:124	1.00	Reference
25–29.99	168:48	1.52	0.96, 2.39	166:48	1.36	0.86, 2.17		60:50	1.03	0.66, 1.63	59:50	0.95	0.59, 1.53
30–34.99	123:30	1.85	1.11, 3.09	122:30	1.60	0.94, 2.71		26:17	1.40	0.71, 2.73	26:17	1.07	0.53, 2.15
≥35	151:34	1.95	1.19, 3.19	149:34	1.56	0.94, 2.59	0.07	24:16	1.25	0.62, 2.45	24:15	0.94	0.45, 1.98	0.94

*

Values are expressed as number with fibroids:number without fibroids.

†

OR, odds ratio adjusted for age; CI, confidence interval.

‡

OR adjusted for age, age at menarche, parity after age 24 years, physical activity (for body mass index), and body mass index (for physical activity variables).

§

Weight (kg)/height (m)².

Because the association of physical activity with fibroids appeared generally similar for African-American and White women, a combined logistic analysis was conducted, controlling for the same covariates. The adjusted odds ratios for the summary variable without chores for high and very high physical activity relative to low were 0.72 (95 percent confidence interval: 0.49, 1.06) and 0.61 (95 percent confidence interval: 0.41, 0.90), respectively. The respective odds ratios when the summary variable that included chores was used were 0.85 (95 percent confidence interval: 0.57, 1.28) and 0.64 (95 percent confidence interval: 0.43, 0.95).

Results for the Bayesian analysis are shown in table 3. These analyses also show an inverse association between physical activity and fibroids for both African-American and White women. Physical activity was more strongly related to tumor onset than to tumor growth.

TABLE 3

Bayes' factors* showing strength of associations of physical activity and body mass index with fibroid onset and tumor growth for African-American and White women in the Uterine Fibroid Study (enrollment, 1996–1999), Washington, DC

Variable	Bayes' factor
	African Americans		Whites
	Onset	Growth	Onset	Growth
Physical activity (excluding chores)	9	3	>1,000	8
Physical activity (including chores)	217	3	35	13
Body mass index	23	1	1	1

*

Bayes' factors: <1, no association; 1–2.9, a weak association; 3–11.9, a moderate association; 12–149.9, a strong association; and ≥150, a very strong association.

DISCUSSION

The National Institute of Environmental Health Sciences Uterine Fibroid Study was designed to identify fibroid cases in an unbiased manner by screening randomly selected women for fibroids using transvaginal and transabdominal ultrasound. Medical records were sought for women who had undergone hysterectomies. Our results showed decreased development of uterine fibroids in both African-American women and White women who were in the upper third of the physical activity distribution (estimated to be equivalent to at least 4 hours of vigorous activity per week). Analyses controlled for BMI, which was an important risk factor for African-American women but not White women. The Bayesian model results showed that physical activity was more strongly related to tumor onset than to tumor growth.

BMI has been examined as a potential risk factor for fibroids in several studies, but findings have been mixed (reviewed by Schwartz et al. (7)). In both the Nurses' Health Study and the Black Women's Health Study, the risk of a new clinical diagnosis of fibroids was significantly elevated with increased BMI, but the risks were modest and peaked for overweight women rather than showing a continuous dose response (18, 19). This same pattern was found for the African-American women in our study. Although our results for White women showed no risk increase associated with BMI, the confidence intervals were relatively broad and were not statistically inconsistent with results from the Nurses' Health Study, the majority of whose participants are White women.

Few studies have examined physical activity and risk of fibroids. Wyshak et al. (20) found a reduced risk of self-reported fibroids among women who had been college athletes compared with college-educated women who did not participate in collegiate athletics. The Nurses' Health Study and the Black Women's Health Study collected data on physical activity as a potential confounder for analysis of BMI data (18, 19), but effect estimates for physical activity variables were not reported.

Our study has several limitations. The first is the cross-sectional nature of data collection, which affects both the quality of physical activity data and the interpretation of our results. Data for both postmenopausal and premenopausal women were included in primary analyses because limiting them to premenopausal women would selectively exclude the most severe fibroid cases, those who had a hysterectomy to treat fibroids. However, since the exercise and BMI data were measured at time of enrollment, not time at hysterectomy, there will be more misclassification in these variables for the postmenopausal women. When this bias was evaluated by limiting analyses to premenopausal women, results were similar (data shown in the supplemental information). Cross-sectional data also leave the causal direction open to question. The association observed could arise because symptomatic fibroids limit exercise. To assess this possibility, women who reported fibroid-related symptoms most likely to limit exercise (heavy bleeding or pelvic pain) were excluded; results were essentially unchanged (data shown in the supplemental information). Thus, reverse causation does not appear to explain the inverse relation between physical activity and fibroids.

Our findings are also limited by reliance on questionnaire measures of physical activity and potential confounding factors. Inclusion of household chores strengthened the association for African Americans but attenuated the effect for Whites. Compared with White women, African-American women reported a higher average number of hours per week doing chores, and it is possible that African Americans' chores tended to be more strenuous than those performed by Whites. Our data were not sufficiently detailed to evaluate this possibility, nor was much information collected about physical activity at work. Residual confounding by other factors is also possible.

Finally, this study was based on a single sample of urban women. The African Americans in the sample represent a broad socioeconomic range, but the Whites are predominantly highly educated professionals of low parity. Although representative of the White women in the study age range who received medical care at the study site, they are not representative of the US population of White women.

Strengths of the study include 1) ultrasound screening for fibroids, which enabled us to identify the large sample of subclinical fibroid cases; 2) analysis with Bayesian methods to evaluate tumor onset separately from tumor growth; and 3) analyses of two racial/ethnic groups, each of sufficient size to test the hypothesis separately. The similar findings for both groups strengthens the plausibility of a true protective effect of physical activity on fibroids.

Physical activity has been reported to be protective for breast cancer, another hormonally related tumor (11, 21). Several mechanisms have been proposed that might explain the protective effect and might also be applicable to fibroid development. Exercise can reduce circulating sex hormones and insulin levels (22–25), thus limiting proliferative effects that may be associated with these factors. The bioavailability of circulating estrogen might also be reduced by exercise-induced increases in sex hormone-binding globulin (26, 27). Exercise has also been hypothesized to influence estrogen metabolism such that fewer estrogenic metabolic products may be formed, but recent studies have not demonstrated such effects (28, 29). Further research will be required to evaluate these possible mechanisms for fibroids. Women have little control of other factors known to affect fibroids (i.e., age, age at menarche, and African-American ethnicity), but exercise is a modifiable practice and also has other positive effects on women's health.

Abbreviations

Abbreviations
 
• BMI

  body mass index

This research was funded by the intramural program at the National Institute of Environmental Health Sciences, with support from the Office of Research on Minority Health, National Institutes of Health, Department of Health and Human Services. Glenn Heartwell managed the study.

Drs. Stephanie London and Walter Rogan reviewed an earlier draft of the manuscript.

Conflict of interest: none declared.