Reproductive risk factors for incident bladder cancer: Iowa Women's Health Study

Abstract

We studied the association between reproductive factors and bladder cancer incidence in a prospective cohort study of 37,459 Iowa women aged 55–69 years and initially free from cancer in 1986. Women reported reproductive history and were followed prospectively through 2003. After adjusting for age and smoking, there was an inverse association between age at menopause and incident bladder cancer (n = 192). Compared with menopause at age ≥≥48, the hazard ratio (HR) of bladder cancer was 1.32 (95% CI; 0.90–1.94) for menopause at 43–47, and 1.60 (95% CI; 1.06–2.39) for ≤≤42 (p-trend = 0.02). The associations were similar for ages at natural and surgical menopause. In addition, women with a history of bilateral oophorectomy had an increased risk of bladder cancer compared with those who did not undergo bilateral oophorectomy: HR = 1.58 (95% CI; 1.12, 2.22). Finally, there was an indication of a positive association between bladder cancer and shorter lifetime years of ovulation (p-trend = 0.09). There were no associations between incident bladder cancer and age at first birth, number of births, age at menarche, use of hormone replacement therapy or any other reproductive characteristics. This study provides evidence that increased risk of bladder cancer is associated with earlier age at menopause in postmenopausal women. © 2006 Wiley-Liss, Inc.

Bladder cancer is one of the most common cancers—61,420 cases are expected to be diagnosed in the United Sates in 2006,1 constituting about 6% of all new cases among men and 2% of all cases among women.2 Established risk factors for bladder cancers include age, gender, smoking and chemical carcinogens. In most regions of the world, including the United States and northern Europe, bladder cancer is 2.5–4 times more common among men than women. It was shown3 that in the United States the excessive risk in men persists in the absence of exposure to smoking, occupational hazards or urinary tract infections. Other risk factors, including hormonal and reproductive characteristics, could explain the lower bladder cancer risk in women when compared with men. The hypothesis of a possible association between sex hormones and bladder carcinogenesis is supported by the following facts: estrogen receptors (ERs) are expressed in normal bladder and bladder carcinoma in animal and human models4, 5, 6, 7, 8; and estrogen inhibits and testosterone stimulates the growth of bladder tumors in rats.9, 10, 11, 12

There have been few studies of bladder cancer and reproductive factors. The largest cohort study reported that bladder cancer risk was inversely associated with age at menopause.13 A case-control study found that a greater number of live births was inversely associated with bladder cancer among never-smokers.14 Two case-control studies in Italy15, 16 observed an increased risk of bladder cancer associated with the use of hormone replacement therapy (HRT). Three other studies found no association17, 18, 19 between bladder cancer and reproductive factors.

We examined in great detail the association of bladder cancer incidence with reproductive factors and hormonal characteristics in a large cohort of postmenopausal women in Iowa, United States. Our study differs from others in that our cohort was population based and it was designed to examine risk factors related to cancer, so extensive information on various reproductive factors was available.

Material and methods

The Iowa Women's Health Study cohort

In 1986 a questionnaire was mailed to 98,030 randomly selected women between 55 and 69 years of age who had valid Iowa driver's licenses in 1985. Thus women living in the entire state of Iowa, both rural and urban areas, were recruited. Forty-two percent of women (n = 41,836) returned the questionnaire. Only women who at baseline did not self-report cancer other than nonmelanoma skin cancer and who reported being menopausal were included. As a result, 37,459 women constituted our analytical cohort.

Five follow-up questionnaires were mailed in 1987, 1989, 1992, 1997 and 2004 to update vital status, residence and exposure information. Data from follow-up surveys indicate that the migration rate from Iowa among cohort members is <1% annually, allowing near complete follow-up.20 The vital status and address of women who did not respond to the follow-up surveys were determined through the National Death Index. Nonrespondents were found to differ only slightly from respondents at baseline; they were 3 months younger than respondents, slightly heavier, had slightly higher rates of all-site cancer and smoking-related diseases.20

At baseline, women were asked to report the age of first menstruation, details of each pregnancy (up to 10) and fertility problems. They were also asked if they had menstrual periods within last year and, if not, the age and the reason for period cessation: natural, surgical or medical. Participants were queried whether or not they had their uterus or one or both ovaries surgically removed and whether they had been diagnosed with endometriosis, fibroid tumors of the uterus, polycystic ovaries or other noncancerous cysts/tumors of the ovaries. In addition, they were asked about history of oral contraceptive and HRT use (never, past and current) and duration of use.

Women reporting natural cessation of menstruation were defined as undergoing natural menopause. Those who reported surgical cessation of menstruation and bilateral oophorectomy were defined as having surgical menopause. Women who reported surgical cessation of menstruation due to hysterectomy without bilateral oophorectomy were excluded from analysis of age of menopause and years of ovulation because it was not possible to determine the age of true menopause for these women. Years of ovulation were calculated as a difference between age at menopause and age at menarche minus duration of pregnancies, minus duration of breastfeeding (on average, 3 months per each child) and minus duration of oral contraceptive use. The years of ovulation variable was categorized into quartiles (≤29.4, 29.5–33.5, 33.6–36.9, ≥37.0 years).

Data on other reproductive and menstrual factors were stratified into logical categories. Cut points for age at menopause (≤42, 43–47, ≥48) were based on empirical data to address whether early age at menopause is associated with bladder cancer risk. Less than 10% of women undergo a natural menopause at or before age 42, 25% of women are menopausal at or before age 45 and menopause at or later than 48 is not considered “early.”21 We also repeated analyses with age at menopause presented as tertiles (≤47, 48–51, ≥52) and as a continuous variable. For age at menarche, we were particularly interested in seeing whether the tails of the distribution were associated with bladder cancer. These are less likely to be subject to misclassification. Menarche at 12–13 years was used as the reference.

The baseline questionnaire also collected information on body size, sociodemographic characteristics (including race, occupation, education level, marital status and residence) as well as lifestyle behaviors including smoking status and number of pack-years, usual alcohol intake within the last year, diet and physical activity.22, 23 Body mass index (BMI) in kg/m² was calculated from self-reported height and weight. Women were asked whether they participated in any leisure exercise and, if so, the frequency of moderate- and heavy-intensity activities. These were combined to create a 3-level activity score (low, medium and high). Women were queried at baseline about whether they were currently married, separated/divorced, widowed or never married and were categorized as being married or unmarried. Three categories were created for education level: less than high school, high school and more than high school. Occupational status was also stratified into 3 categories: women who never worked or were homemakers, those involved in farming or crafts and professionals or clerical workers. The baseline questionnaire included information on prevalent diseases diagnosed by a physician, including heart disease, diabetes mellitus and cancer.

Identification of bladder cancer occurrence

Incident bladder cancer cases were identified through annual linkage to the State Health Registry of Iowa, which is part of the National Cancer Institute's Surveillance, Epidemiology and End Results Program, via an annual computer match of social security number, name, maiden name and date of birth. Topographic and morphological codes from the International Classification of Diseases for Oncology, Third Edition, were used to classify incident bladder cancer cases (code C67) in the cohort.

Data analysis

Person-years were calculated from baseline (January 1986) until the first of the following: date of bladder cancer diagnosis, known or estimated date of emigration from Iowa, date of death (if death occurred in Iowa) or the midpoint between the date of last contact and the date of death (for deaths occurring outside of Iowa). All other women were assumed to be living in Iowa, and they contributed follow-up time until December 31, 2003.

Age-adjusted and multivariate-adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) were computed by Cox proportional hazards regression using SAS statistical software, version 8 (program PHREG; SAS Institute, Cary, NC). We tested the assumption of proportional hazards and found no evidence of violation of this assumption. For multivariate model selection, we started with variables that had been previously found to be risk factors for bladder cancer in this cohort, including age, smoking (smoking status and pack-years), BMI, history of diabetes mellitus, alcohol use, physical activity, as well as marital, educational and occupational status.24 We found that only age and smoking changed HRs of bladder cancer and reproductive factors by more than 10%. All other potential confounders did not materially alter HRs for any variable of interest and were not included into the final model.

To examine potential effect-modification by smoking, we conducted a test for interaction between age at menopause and smoking status. We performed a linear test for trend by including the median value for each category of a covariate under study as a continuous term into the regression model.

The Iowa Women's Health Study (IWHS) was conducted under a protocol approved for human subjects research by the University of Minnesota Institutional Review Board. The return of baseline and follow-up questionnaires were considered to indicate a consent.

Results

The mean baseline age of the 37,459 women at risk was 62 years, and 99% were white. We studied the baseline characteristics of postmenopausal women in IWHS according to their age at menopause: ≤42, 43–47 and ≥48 (Table I). Mean age at baseline and mean age at menarche did not vary across these categories. The distribution of BMI was also similar across categories. Women with later menopause (≥48 years) were more likely to be married, tended to be more educated and more physically active and tended to smoke less: the percent of current smokers and of those smoking ≥40 pack-years were lower for this category. Women with early menopause (≤42 years) were more often involved in farming and crafts and less often were homemakers or did not work. Furthermore, in the early menopause category, the percent of ever users of alcohol was slightly lower and percent of diabetics was slightly higher than in other categories. Women with early menopause had lower mean age at first live birth, were less likely to have more than 3 live births, were less likely to have ever used oral contraceptives and tended to have undergone hysterectomy and use HRT.

Table I. Prevalence of Various Baseline Characteristics According to Reported Age at Menopause,2 IWHS, 1986

Baseline variable	Age at menopause
	≤42 (n = 4,038)	43–47 (n = 5,749)	≥48 (n = 20,920)
1 Values given in parentheses indicate SEs. 2 Women with surgical cessation of menstruation due to hysterectomy but no total oophorectomy were excluded from analysis.
Mean age at baseline (years)	62 (0.07)1	62 (0.06)	62 (0.03)
Body mass index, kg/m^2 (%)
≤24.9	40	42	40
25–29.9	36	35	37
≥30	24	23	23
Education (%)
Less than high school	24	20	17
High school	42	43	42
More than high school	34	37	41
Married (yes, %)	73	74	78
Occupation (%)
Homemaker/never worked	30	37	40
Professional/clerical	44	41	40
Farmer/crafts	26	22	20
Smoking (%)
Never	57	62	68
Former	21	20	19
Current	22	18	13
≥40 pack-years of smoking (%)	14	11	8
Alcohol intake (ever, %)	41	44	44
Regular physical activity (yes, %)	39	40	42
Physical activity (%)
Low	49	49	46
Moderate	27	27	28
High	24	24	26
Diabetes (yes, %)	8	6	6
Mean age at menarche (years)	13 (0.02)	13 (0.02)	13 (0.01)
Mean age at first live birth (years)	22 (0.07)	23 (0.06)	23 (0.03)
≥3 live births (%)	47	57	60
Hysterectomy (yes) (%)	51	29	13
Oral contraceptive use (yes) (%)	12	18	20
Hormone replacement therapy use (yes) (%)	52	41	31
Years of ovulation2
≥37.0	0	0	37
33.6–36.9	0	5	35
29.5–33.5	2	48	23
≤29.4	98	46	5

During 18 years of follow-up, 192 women (0.5%) aged 55–86 were diagnosed with bladder cancer. The mean age at diagnosis of bladder cancer (n = 192) was 73 years (range 55–86 years). After adjustment for age and smoking (smoking status and number of pack-years), we observed no associations between bladder cancer incidence and age at menarche, age at first live birth, number of live births, history of oral contraceptive use or HRT use, history of infertility, fibroid tumors, ovary cysts or endometriosis (Table II).

Table II. Age-Adjusted Hazard Ratios (HRs) of Bladder Cancer in Relation to Reproductive and Menstrual History Variables, IWHS, 1986–2003

Characteristics	Number of cases	Person-years	Age-adjusted HR (95% CI)	HR adjusted for age and smoking1(95% CI)
1 Adjusted for age, smoking status and pack-years.
Age at menarche (years)
≤10	4	18,302	0.70 (0.26–1.88)	0.70 (0.26–1.89)
11	26	71,241	1.14 (0.74–1.74)	1.05 (0.68–1.63)
12–13	110	330,853	1 (reference)	1 (reference)
14	23	91,547	0.75 (0.48–1.17)	0.73 (0.45–1.16)
≥15	26	70,416	1.10 (0.72–1.69)	1.08 (0.70–1.65)
p-trend			0.70	0.79
Number of live births
0	24	51,696	1 (reference)	1 (reference)
1–2	63	187,089	0.74 (0.46–1.18)	0.81 (0.50–1.32)
3–4	70	233,225	0.69 (0.43–1.10)	0.79 (0.48–1.27)
≥5	31	113,441	0.64 (0.38–1.10)	0.79 (0.46–1.37)
p-trend			0.10	0.38
Age at first live birth (years)
≤20	51	170,424	1 (reference)	1 (reference)
20–25	77	247,757	0.99 (0.69–1.40)	1.09 (0.76–1.57)
26–30	22	78,986	0.81 (0.49–1.35)	0.91 (0.55–1.51)
>30	10	25,038	1.13 (0.57–2.24)	1.31 (0.66–2.59)
Nulliparous	24	51,696	1.39 (0.85–2.26)	1.31 (0.79–2.17)
p-trend			0.20	0.25
Oral contraceptive use
Never	163	477,068	1 (reference)	1 (reference)
Ever	29	111,194	0.91 (0.60–1.37)	0.88 (0.58–1.32)
Estrogen use
Never	114	361,715	1 (reference)	1 (reference)
Current/past	78	225,385	1.09 (0.82–1.45)	1.04 (0.78–1.39)
Infertility
No	150	482,823	1 (reference)	1 (reference)
Yes	20	50,223	1.31 (0.82–2.09)	1.25 (0.78–1.99)
Fibroid tumors
No	143	442,657	1 (reference)	1 (reference)
Yes	46	136,491	1.06 (0.76–1.48)	1.09 (0.78–1.53)
Endometriosis
No	178	548,233	1 (reference)	1 (reference)
Yes	7	21,980	1.05 (0.54–3.93)	1.05 (0.49–2.24)
Ovary cysts/tumors
No	156	511,761	1 (reference)	1 (reference)
Yes	28	62,527	1.49 (1.00–2.23)	1.26 (0.82–1.92)

There was an inverse association between age at menopause and bladder cancer incidence. Compared with menopause above 48, the age-adjusted HR was 1.45 for 43–47 years and 1.93 for ≤42 (p-trend = 0.0007). After adjustment for smoking, the HR was 1.32 for 43–47 years, and 1.60 for ≤42 (p-trend = 0.02) (Table III). In the model adjusted for age and smoking, each 5-year decrease in age at menopause increased risk of bladder cancer by 21% (95% CI; 1.08, 1.36). When we regrouped age at menopause into tertiles (≤47, 48–51 and ≥52), the HR adjusted for age and smoking was 1.14 (95% CI; 0.76–1.73) for the middle and 1.54 (95% CI; 1.05–2.27) for the lower tertile compared with the upper tertile (p-trend = 0.03).

Table III. Adjusted Hazard Ratios (HRs) of Bladder Cancer in Relation to Age at Menopause (Natural and Surgical) and Oophorectomy Status, IWHS, 1986–2003

Characteristics	Number of cases	Person-years	Age-adjusted HR (95% CI)	HR adjusted for age and smoking1 (95% CI)
1 Adjusted for age, smoking status and pack-years. 2 Women with surgical cessation of menstruation due to hysterectomy but no total oophorectomy were excluded from analysis. 3 Only women with natural cessation of menstruation were included. 4 Only women with surgical cessation of menstruation due to total oophorectomy were included.
Age at menopause (years)2
≥48	93	332,489	1 (reference)	1 (reference)
43–47	36	89,318	1.45 (0.98–2.12)	1.32 (0.90–1.94)
≤42	33	61,381	1.93 (1.30–2.88)	1.60 (1.06–2.39)
p-trend			0.0007	0.02
Age at menopause (years) as tertiles2
≥52	43	168,890	1 (reference)	1 (reference)
48–51	50	163,599	1.23 (0.81–1.84)	1.14 (0.76–1.73)
≤47	69	150,700	1.82 (1.25–2.67)	1.54 (1.05–2.27)
p-trend			0.002	0.03
Age at natural menopause (years)3
≥48	81	290,803	1 (reference)	1 (reference)
43–47	24	61,910	1.37 (0.87–2.16)	1.22 (0.77–1.93)
≤42	13	23,435	1.96 (1.09–3.52)	1.64 (0.91–2.96)
p-trend			0.02	0.09
Age at surgical menopause (years)4
≥48	10	31,985	1 (reference)	1 (reference)
43–47	12	24,210	1.66 (0.72–3.84)	1.53 (0.66–3.56)
≤42	19	33,727	1.82 (0.85–3.91)	1.41 (0.64–3.08)
p-trend			0.14	0.47
Oophorectomy
No surgery	122	426,444	1 (reference)	1 (reference)
Unilateral	20	49,051	1.41 (0.88–2.25)	1.18 (0.72–1.94)
Bilateral	47	103,767	1.62 (1.16–2.27)	1.58 (1.12–2.22)
Years of ovulation2
≥37.0	29	118,301	1 (reference)	1 (reference)
33.6–36.9	41	118,733	1.45 (0.90–2.34)	1.43 (0.88–2.32)
29.5–33.5	35	117,826	1.28 (0.78–2.09)	1.22 (0.74–2.01)
≤29.4	49	113,837	1.86 (1.17–2.95)	1.60 (1.00–2.56)
p-trend			0.03	0.09

The association between bladder cancer and age at natural menopause behaved in a parallel fashion (Table III). In the model adjusted for age and smoking, HRs were 1.00, 1.22 and 1.64 (p-trend = 0.09) for ages at natural menopause of ≥48, 43–47 and ≤42 years, respectively. Although power was limited for surgical menopause, HRs were 1.00, 1.53 and 1.41 (p-trend = 0.47) for ages at surgical menopause of ≥48, 43–47 and ≤42 years.

In addition, we observed a significant association between bilateral oophorectomy and bladder cancer risk (Table III). Women with a history of bilateral oophorectomy had an increased risk of bladder cancer when compared with those who did not undergo bilateral oophorectomy: the HR adjusted for age and smoking was 1.58 (95% CI; 1.12, 2.22). Finally, after age adjustment, there was an inverse association between bladder cancer incidence and years of ovulation. After further adjustment for smoking, the trend remained but was no longer statistically significant (p = 0.09) (Table III).

Since smoking is a strong risk factor for bladder cancer and is associated with earlier age at menopause,2, 25, 26, 27 we examined the interaction between smoking and age at menopause and found it to be not statistically significant. There were comparable inverse associations between risk of bladder cancer and age at menopause for ever smokers (p-trend = 0.04) and never smokers (p-trend = 0.07) (Table IV). In case of never smokers, small number of cases (n = 64) could have prevented statistical significance.

Table IV. Age-Adjusted Hazard Ratios (HRs) for Bladder Cancer in Relation to Age at Menopause1 Stratified by Smoking Status, IWHS, 1986–2003

Age at menopause (years)	Number of cases	Person-years	Age-adjusted HR (95% CI)
1 Women with surgical cessation of menstruation due to hysterectomy but without bilateral oophorectomy were excluded from analysis.
Never smokers
≥48	39	227,655	1 (reference)
43–47	14	56,347	1.43 (0.78–2.64)
≤42	11	36,092	1.74 (0.89–3.40)
p-trend			0.07
Ever smokers
≥48	53	99,974	1 (reference)
43–47	22	31,658	1.32 (0.80–2.16)
≤42	21	24,467	1.64 (0.99–2.72)
p-trend			0.04

Discussion

Our results indicate that younger age at menopause was associated with increased risk of bladder cancer in a large cohort of postmenopausal women in Iowa, US. The associations between bladder cancer risk and ages at natural and surgical menopause behaved similarly. The finding of an association between increased bladder cancer incidence and early age at menopause is consistent with the results reported in Nurses' Health Study cohort—the only cohort study so far that examined the risk of bladder cancer and age at menopause.13 In that study, postmenopausal women with early menopause (≤45 years) had a 1.63-fold (95% CI; 1.2, 2.23) greater risk than those with mid to late age at menopause (≥50 years).

It is possible that earlier age at menopause increases risk of bladder cancer by maintaining estrogen concentrations for a shorter period of life. The increased risk of bladder cancer for women with bilateral oophorectomy versus those with intact ovaries and suggestive increased risk for women with fewer years of ovulation are also consistent with lower levels of estrogen after menopause.

Laboratory studies on human and animal cell lines support the hypothesis that sex hormones can play a role in the development of bladder cancer. Firstly, estrogen receptors (ERβ) are expressed in normal rat bladder tissue.4, 28 Secondly, the incidence of spontaneous and chemically induced bladder tumor development is significantly greater in male than in female rats.9 Thirdly, surgical castration in male rodents significantly reduces the occurrence of bladder carcinomas.12 Finally, administration of testosterone increases the incidence of bladder cancer and stimulates the growth of bladder tumors in rats, while estrogens inhibit the carcinogenesis of bladder tumors.9, 10, 11, 12

The role of estrogen in the bladder cancer carcinogenesis in humans has not been investigated extensively. The trigone region of the bladder is formed from the same embryonic bud as the anterior vagina in women and is responsive to steroid hormones.7 Sex hormone receptors are found in normal human bladder epithelium and in transitional cell carcinoma of the bladder—androgen receptors are found most consistently, ERs were found less frequently.6, 7, 29, 30, 31 In addition, researchers have shown that raloxifene induces apoptosis32 and tamoxifen enhances chemosensitivity within in vitro human bladder carcinoma cell lines,33 but these results were not confirmed in clinical trials.34

The biological mechanism of any effect of estrogen on the risk of bladder cancer is not clear. Epidemiological studies suggest that estrogen deficiency after menopause is associated with increased urinary tract infections,35, 36 which could cause chronic irritation of the bladder epithelium and thus increase bladder cancer risk.13, 37, 38, 39 This mechanism is in agreement with the fact that, similarly to Nurses' Health Study cohort,13 Swedish cohort17 and 2 case-control studies,18, 19 we did not observe any associations between bladder cancer risk and reproductive factors not related to menopause. We found no associations between incident bladder cancer and age at menarche, age at first live birth, number of live births, history of oral contraceptive use, use of HRT or history of several gynecological disorders.

The strengths of our study lie in its large prospective population-based design with nearly complete follow-up over 18 years. The study was initiated to study risk factors for cancer and detailed information on many risk factors was collected.

One limitation of our study is that only 192 bladder cancers occurred, which limits statistical power to detect weak associations and prevents the study of bladder cancer subtypes. Another possible limitation is that data on reproductive factors were self-reported. However, other studies have suggested good agreement between self-report and medical records for type, reason and age of menopause, details on hysterectomy and extent of ovarian surgery.40, 41 Moreover, the associations between the main reproductive factors and breast, endometrial and ovarian cancers in the IWHS cohort have been studied earlier22, 42, 43, 44 and were consistent with associations reported in literature; this suggests these reproductive questions are valid.

We did not ask when oophorectomy and hysterectomy occurred, and the surgeries may have happened after menopause for some of the women. However, ∼95% of women with natural menopause reported having neither hysterectomy nor bilateral oophorectomy. The 99% of women reporting surgical cessation of menstruation answered that they had undergone hysterectomy and/or bilateral oophorectomy.

In conclusion, our findings suggest that earlier age at menopause is associated with an increased risk of bladder cancer, which might be explained by urinary tract infections related to deficiency of endogenous estrogen.