The Women's Health Initiative randomized trials showed a reduction in colorectal cancer risk with the use of estrogen plus progesterone (E + P), but not with estrogen alone (E-only), after intervention periods <7 years. Using data from the Cancer Prevention Study II Nutrition Cohort, we examined associations of colorectal cancer risk with E-only and E + P, including analyses by recency and duration of hormone use. During 13.2 years of follow-up, 776 cases of invasive colorectal cancer occurred among 67,412 postmenopausal women participants. Cox proportional hazards models were used to estimate multivariate-adjusted relative risks (RR) and 95% confidence intervals (95% CI) of colorectal cancer for current and former hormone users according to hormone type and duration of use. Relative to women who never used postmenopausal hormones, current, but not former, use of E-only was associated with a reduced risk of colorectal cancer (RR 0.76; 95% CI, 0.59-0.97). Among current E-only users, duration of use was inversely and linearly associated with risk (Ptrend = 0.01). Use of E-only for <5 years was not associated with reduced risk, whereas use for ≥20 years was associated with a 45% reduction in risk (RR, 0.55; 95% CI, 0.36-0.86). There were no statistically significant associations between E + P and colorectal cancer risk. Our results suggest a strong inverse association of long-term use of E-only with colorectal cancer risk, underscoring the importance of collecting data on duration of hormone use in epidemiologic studies of postmenopausal hormones and risk of disease. (Cancer Epidemiol Biomarkers Prev 2009;18(11):2835–41)

A large and diverse body of epidemiologic, basic, and clinical research suggests an important role of sex hormones in the etiology of colorectal neoplasia. This research includes >27 observational studies, conducted since the 1980s, on the association of postmenopausal hormone use with the risk of colorectal cancer (1-23) and adenomatous polyps (24-27). Based on findings from these observational studies, “ever” use of postmenopausal hormones is associated with a 20% to 40% reduced relative risk (RR) of colorectal cancer (1-3, 6, 8, 9, 11, 13-15, 18, 21, 22), whereas current or recent use is associated with a 30% to 40% reduced risk (1, 2, 7, 10, 11, 14, 15, 17, 18, 22). Importantly, several studies (11, 16, 22, 28) observed a statistically significant reduction in risk only after 5 years of hormone use.

A secondary goal of the Women's Health Initiative (WHI) randomized, placebo-controlled, primary prevention trials was to examine the effect of estrogen plus progesterone (E + P; 0.625 mg/d conjugated equine estrogens plus 2.5 mg/d medroxyprogesterone acetate) and estrogen alone (E-only; 0.625 mg/d conjugated equine estrogens) on risk of colorectal cancer. In the E + P trial, after 5.2 years of follow-up, risk of colorectal cancer was reduced among women in the intervention arm relative to placebo [RR, 0.56; 95% confidence interval (95% CI), 0.38-0.81; 112 cases; ref. 29]. In the E-only trial, however, there was no effect of the intervention on colorectal cancer risk (RR, 1.08; 95% CI, 0.75-1.55; 119 cases) during an average of 6.8 years of follow-up (30). In contrast to the randomized trials, results from the WHI observational study suggested a modest reduction in colorectal cancer risk with E-only use (RR, 0.80; 95% CI, 0.53-1.20; 168 cases) but no reduction in risk with E + P during 7.1 and 5.5 years of follow-up, respectively (31).

To date, a small number of observational studies have examined the associations of E + P and E-only with colorectal cancer risk, and results from these studies are inconsistent (2, 4, 13, 15, 16, 19, 22, 32, 33). The Breast Cancer Detection Demonstration Project, the largest prospective study to examine colorectal cancer risk by hormone type, found a nonstatistically significant reduction in risk of colorectal cancer with ever use of E + P and a significant reduction in risk associated with ever use of E-only (13). The reduced risk for E-only appeared to be attributable to current users and to users with ≥10 years of use, who incurred the strongest reduction in risk (RR, 0.74). However, statistical power to assess the effect of long duration of current hormone use was limited in the Breast Cancer Detection Demonstration Project, which accrued only 40 cases among current E-only users and 32 cases among current E + P users.

Based on these few studies, it seems plausible that risk of colorectal cancer may differ according to the type of hormonal preparation used and the recency and duration of hormone use. The Cancer Prevention Study II (CPS-II) Nutrition Cohort provides an excellent resource for the study of postmenopausal hormone use and risk of colorectal cancer because of its prospective design, duration of follow-up, large sample size, and regularly updated detailed information on hormone use. We therefore examined the associations between risk of incident colorectal cancer and use of postmenopausal hormones by type, recency, and duration of use.

Study Population

Study subjects were selected from the 97,786 female participants in the CPS-II Nutrition Cohort, a prospective study of cancer incidence, initiated in 1992 (34). The ∼184,000 participants in the Nutrition Cohort were selected from 1.2 million participants in CPS-II, a prospective mortality study established by the American Cancer Society in 1982 (35). Members of the 1982 Cohort, who were ages 50 to 74 years in 1992, were invited to participate in the Nutrition Cohort, as described elsewhere (34). The Study is approved by the Emory University Institutional Review Board.

At enrollment, participants completed a self-administered questionnaire that included information on demographic, medical, behavioral, environmental, occupational, and dietary factors. Follow-up questionnaires were sent to cohort members in 1997, 1999, 2001, 2003, and 2005 to update exposure information and to ascertain newly diagnosed cancers. The response rate for each of the follow-up questionnaires was ≥89%. For this analysis, the follow-up period ended June 30, 2005.

Analyses were restricted to postmenopausal women; women who were premenopausal/perimenopausal in 1992, but reported being postmenopausal in 1997 or 1999 (n = 2,658), began contributing person-time to the analysis at that time. Women were excluded based on the following criteria: premenopausal as of 1999, unknown menopausal status or unknown age at menopause (n = 2,520); lost to follow-up (n = 3,152); history of cancer, other than nonmelanoma skin cancer, at enrollment (n = 12,055); unknown type or duration of hormone use (n = 6,472); history of both E-only and E + P use (n = 3,975); use of only oral progesterone or vaginal cream (n = 1,798); reported current use of E + P after hysterectomy (n = 319); nonadenocarcinoma colorectal cancer (n = 15); and report of a colorectal cancer on a first follow-up survey that could not be verified (n = 68). After exclusions, the final analytic cohort consisted of 67,412 postmenopausal women.

Women who developed cancer during the follow-up period, other than colorectal or nonmelanoma skin cancer, were censored at the time of their reported cancer diagnosis. Cohort members with missing hormone data in 1997, 1999, 2001, or 2003 were censored at the time of their missing data report. Subjects who reported an unverifiable colorectal cancer on a later survey (n = 103) were censored on the date of their most recent cancer-free survey. Women who reported changing the type of hormones used (E + P to E-only or vice versa) were censored at the time of the report.

Case Ascertainment

In the analytic cohort, we identified 776 incident colorectal cancer cases diagnosed between enrollment and June 30, 2005. The majority of cases (n = 641) were first identified by self-report on a follow-up questionnaire and then verified by medical record abstraction or through linkage with state cancer registries when complete medical records could not be obtained (34). Our respondents have been found to accurately report a cancer diagnosis with high sensitivity (0.93; ref. 36). Case verification of other reported cancers through state cancer registries identified an additional 14 colorectal cancer cases. One hundred twenty-one cases were identified by death certificates where colorectal cancer was listed as the underlying cause of death (37); 93 of these deaths were subsequently verified through state cancer registries. Colon subsite was obtained from medical records, registry linkage, or death certificates for 559 colon cancer cases, including 373 cases of proximal colon cancer (International Classification of Disease for Oncology code C18.0, C18.2-5) and 186 cases of distal colon cancer (International Classification of Disease for Oncology code C18.6-7). The remaining 217 cases of colorectal cancer included 176 rectal cancer cases (International Classification of Disease for Oncology code C19.9-20.9).

Ascertainment of Hormone Use

Information on postmenopausal hormone use was collected at enrollment and on all follow-up questionnaires. Subjects were asked about the type of hormones used, whether hormone use was previous or current, and the duration of use. For this analysis, women were categorized at baseline into mutually exclusive groups of never hormone users (n = 32,916), E-only users (n = 25,437), or E + P users (n = 9,059). Those who were premenopausal/perimenopausal in 1992 were categorized according to their hormone use after they became menopausal and began contributing person-time to the analysis.

Adjustment Variables

Covariates were chosen based on their potential to confound the association between hormone use and colorectal cancer as determined through univariate analyses. Variables included in the final models were body mass index (BMI; 18.5 to <25, 25 to <30, ≥30), smoking (nonsmoker, current, former, smoker with unknown current status), education (high school or less, some college or vocational school, college graduate), race (White, Black, other), physical activity (metabolic equivalents of energy expenditure, quintiles <3.5, 3.5 to <4.5, 4.5 to <14, 14 to <19, ≥19), current use of nonsteroidal anti-inflammatory drugs (NSAID; no, yes), current use of multivitamins (no, <1/d, daily use), family history of colorectal cancer (no, yes), hysterectomy (no, yes), type of menopause (natural, surgical or radiation), red and processed meat intake (g/wk; quintiles), and history of colorectal endoscopy (never received endoscopy test, received test). Age at menopause, age at menarche, age at first live birth, parity, oral contraceptive use, vegetable intake, total calcium intake, energy intake, and alcohol use were found to have no appreciable influence (<2% change) on the RRs of colorectal cancer associated with hormone use.

Statistical Analysis

Age-adjusted and multivariate-adjusted RRs and 95% CIs were estimated using Cox proportional hazards models, stratified on age at enrollment (38). RRs reported in the text are multivariate-adjusted. All covariates were modeled using status at enrollment, except colorectal endoscopy, which was modeled as a time-dependent variable.

Postmenopausal hormone use was modeled as a time-dependent variable. At the start of each interval, defined by completion of a new questionnaire, each woman's hormone use was updated. Women who reported never having used postmenopausal hormones were categorized as never users. Women who reported current or past use of either E-only or E + P were categorized as either current or former users, and their total lifetime duration of use as of the start of that follow-up interval was calculated. For current users of E-only, total duration of use was classified as <5, 5 to <10, 10 to <15, 15 to <20, or ≥20 years. For current users of E + P, total duration of use was classified as <5, 5 to <10, or ≥10 years. Former E-only users were classified as <5 or ≥5 years, and former E + P users were left in a single category due to small numbers of users with ≥5 years use. P values for linear trend were generated by the likelihood ratio test statistic from models of years of hormone use as a continuous variable. Women who reported changing from never use to use of E + P or E-only were recategorized into the appropriate hormone use category.

We examined whether the association between current postmenopausal hormone use and risk of colorectal cancer varied by BMI (18.5 to <25, 25 to <30, ≥30) and age (<70, ≥70 years) and evaluated effect modification by modeling multiplicative interaction terms for current use of each hormone type. P values for interactions were generated by the likelihood ratio test statistic that compared a multivariate-adjusted model with only the main effects to a model that contained the main effects and interaction terms.

To examine whether the association with risk of colorectal cancer was statistically different for current E-only than for current E + P, we compared our main multivariate-adjusted model containing separate variables for current E-only and current E + P to a model that contained a single variable for current hormone use without differentiation by type. A P value for heterogeneity of effects by hormone type was then generated by a likelihood ratio test with 1 df.

At baseline, 48.8% of women had never used postmenopausal hormones, 21.7% were current E-only users, 10.6% were current E + P users, 16.1% were former E-only users, and 2.8% were former E + P users (Table 1). E + P users reported higher education and lower red meat intake and were more often former smokers than never users or E-only users. E-only users were more likely to have had surgical or radiation-induced menopause. Average age at study entry in 1992 was lowest among current E + P users and highest among former E-only users, likely reflecting temporal trends in use of these different hormonal preparations (39).

Table 1.

Baseline characteristics of women according to hormone use in 1992

NeverE-onlyE + P
Current useFormer useCurrent useFormer use
No. women 32,916 14,617 10,820 7,163 1,896 
Average age at study entry in 1992 63 61 66 57 59 
 Age-adjusted percents* 
BMI  
    18.5 to <25 46.1 53.1 44.4 64.1 53.9 
    25 to <30 32.2 31.2 33.4 24.9 28.9 
    ≥30 18.3 12.3 18.8 7.9 14.0 
    Underweight or unknown 3.5 3.4 3.4 3.2 3.3 
Smoking status 
    Never smoked 56.6 54.6 53.2 51.2 50.8 
    Current smoker 9.8 6.7 10.6 6.5 7.3 
    Former smoker 32.0 37.4 34.7 40.9 40.2 
    Smoker, unknown if current or former 0.3 0.3 0.4 0.3 0.5 
    Unknown smoking status 1.2 1.1 1.1 1.1 1.2 
Education 
    Less than high school 6.0 3.7 6.3 1.3 3.2 
    High school graduate 35.8 29.9 34.8 19.7 24.9 
    Some college 28.8 34.7 33.1 32.3 31.3 
    College graduate 28.7 31.0 25.2 45.9 40.0 
    Unknown 0.7 0.7 0.6 0.8 0.6 
Race 
    White 97.1 97.5 96.7 98.6 98.4 
    Black 1.8 1.3 2.2 0.3 0.4 
    Other 1.1 1.2 1.1 1.1 1.2 
Physical activity defined as MET h/wk 
    <3.5  9.8 8.4 10.1 6.5 9.1 
    3.5 to <4.5 30.8 30.3 31.0 30.0 28.9 
    4.5 to <14 18.4 20.6 18.7 20.3 22.0 
    14 to <19 20.2 20.0 18.9 21.2 19.4 
    ≥19 19.3 19.5 19.8 21.1 20.1 
    Unknown 1.5 1.3 1.5 0.8 0.5 
Current use of NSAIDs 
    No 43.1 34.4 37.7 37.9 37.9 
    Yes 52.9 62.3 58.4 58.9 59.1 
    Unknown 4.0 3.3 4.0 3.2 3.1 
Current use of multivitamins 
    No 62.2 52.2 58.3 52.0 52.1 
    <1/d 8.6 9.7 9.6 9.9 10.2 
    Daily use 27.6 37.0 30.5 37.1 36.4 
    Unknown 1.7 1.1 1.6 1.0 1.4 
Family history of colorectal cancer 
    No 94.2 93.4 94.3 93.3 93.1 
    Yes 5.8 6.6 5.7 6.7 6.9 
Hysterectomy 
    Yes 18.4 87.7 53.6 0.0 7.0 
    No  79.5 10.4 44.3 97.1 90.9 
    Unknown 2.0 1.9 2.1 2.9 2.0 
Type of menopause 
    Natural 84.2 25.1 51.6 96.4 93.2 
    Surgical or radiation 15.2 74.4 48.0 0.4 6.0 
    Unknown 0.7 0.5 0.4 3.1 0.8 
Red meat intake by g/wk 
    1st quintile 17.1 19.1 17.1 21.8 21.4 
    2nd quintile 17.5 19.5 17.7 21.6 17.7 
    3rd quintile 17.9 19.0 18.4 19.1 21.1 
    4th quintile 18.5 18.5 18.9 17.1 17.1 
    5th quintile 19.7 16.9 19.4 14.6 15.5 
    Unknown 9.4 6.9 8.5 5.7 7.3 
Colorectal endoscopy as of 1997§ 
    Never received endoscopy 39.1 23.9 26.2 26.2 26.7 
    Received endoscopy 31.5 40.4 32.8 36.4 37.0 
    Endoscopy history unknown 9.8 7.3 8.3 6.1 7.5 
NeverE-onlyE + P
Current useFormer useCurrent useFormer use
No. women 32,916 14,617 10,820 7,163 1,896 
Average age at study entry in 1992 63 61 66 57 59 
 Age-adjusted percents* 
BMI  
    18.5 to <25 46.1 53.1 44.4 64.1 53.9 
    25 to <30 32.2 31.2 33.4 24.9 28.9 
    ≥30 18.3 12.3 18.8 7.9 14.0 
    Underweight or unknown 3.5 3.4 3.4 3.2 3.3 
Smoking status 
    Never smoked 56.6 54.6 53.2 51.2 50.8 
    Current smoker 9.8 6.7 10.6 6.5 7.3 
    Former smoker 32.0 37.4 34.7 40.9 40.2 
    Smoker, unknown if current or former 0.3 0.3 0.4 0.3 0.5 
    Unknown smoking status 1.2 1.1 1.1 1.1 1.2 
Education 
    Less than high school 6.0 3.7 6.3 1.3 3.2 
    High school graduate 35.8 29.9 34.8 19.7 24.9 
    Some college 28.8 34.7 33.1 32.3 31.3 
    College graduate 28.7 31.0 25.2 45.9 40.0 
    Unknown 0.7 0.7 0.6 0.8 0.6 
Race 
    White 97.1 97.5 96.7 98.6 98.4 
    Black 1.8 1.3 2.2 0.3 0.4 
    Other 1.1 1.2 1.1 1.1 1.2 
Physical activity defined as MET h/wk 
    <3.5  9.8 8.4 10.1 6.5 9.1 
    3.5 to <4.5 30.8 30.3 31.0 30.0 28.9 
    4.5 to <14 18.4 20.6 18.7 20.3 22.0 
    14 to <19 20.2 20.0 18.9 21.2 19.4 
    ≥19 19.3 19.5 19.8 21.1 20.1 
    Unknown 1.5 1.3 1.5 0.8 0.5 
Current use of NSAIDs 
    No 43.1 34.4 37.7 37.9 37.9 
    Yes 52.9 62.3 58.4 58.9 59.1 
    Unknown 4.0 3.3 4.0 3.2 3.1 
Current use of multivitamins 
    No 62.2 52.2 58.3 52.0 52.1 
    <1/d 8.6 9.7 9.6 9.9 10.2 
    Daily use 27.6 37.0 30.5 37.1 36.4 
    Unknown 1.7 1.1 1.6 1.0 1.4 
Family history of colorectal cancer 
    No 94.2 93.4 94.3 93.3 93.1 
    Yes 5.8 6.6 5.7 6.7 6.9 
Hysterectomy 
    Yes 18.4 87.7 53.6 0.0 7.0 
    No  79.5 10.4 44.3 97.1 90.9 
    Unknown 2.0 1.9 2.1 2.9 2.0 
Type of menopause 
    Natural 84.2 25.1 51.6 96.4 93.2 
    Surgical or radiation 15.2 74.4 48.0 0.4 6.0 
    Unknown 0.7 0.5 0.4 3.1 0.8 
Red meat intake by g/wk 
    1st quintile 17.1 19.1 17.1 21.8 21.4 
    2nd quintile 17.5 19.5 17.7 21.6 17.7 
    3rd quintile 17.9 19.0 18.4 19.1 21.1 
    4th quintile 18.5 18.5 18.9 17.1 17.1 
    5th quintile 19.7 16.9 19.4 14.6 15.5 
    Unknown 9.4 6.9 8.5 5.7 7.3 
Colorectal endoscopy as of 1997§ 
    Never received endoscopy 39.1 23.9 26.2 26.2 26.7 
    Received endoscopy 31.5 40.4 32.8 36.4 37.0 
    Endoscopy history unknown 9.8 7.3 8.3 6.1 7.5 

*Adjusted to the age distribution of the Nutrition Cohort.

Columns may not add to 100% due to rounding and/or missing data.

Metabolic equivalents of energy expenditure.

§Data collection for endoscopy history began with 1997 survey; percents reflect distribution at that time.

Current use of E-only was associated with reduced risk of colorectal cancer compared with never use (RR, 0.76; 95% CI, 0.59-0.97), whereas former use was not (Table 2). No statistically significant associations were observed with current or former use of E + P. Although the RR estimates differed between current E-only and current E + P, statistical heterogeneity was not evident (P = 0.27).

Table 2.

RR of colorectal cancer according to hormone use

NeverE-onlyE + P
Current useFormer useCurrent useFormer use
All colorectal cancer 
    Age-adjusted RR (95% CI)* 1.00 0.65 (0.53-0.80) 0.99 (0.82-1.19) 0.86 (0.65-1.13) 0.83 (0.56-1.22) 
    Fully adjusted RR (95% CI) 1.00 0.76 (0.59-0.97) 1.05 (0.86-1.28) 0.93 (0.70-1.23) 0.88 (0.59-1.29) 
    No. cases 414 115 158 61 28 
Colon cancer 
    Age-adjusted RR (95% CI)* 1.00 0.68 (0.54-0.87) 1.03 (0.83-1.27) 0.84 (0.61-1.17) 0.89 (0.58-1.38) 
    Fully adjusted RR (95% CI) 1.00 0.81 (0.61-1.08) 1.10 (0.88-1.37) 0.90 (0.65-1.26) 0.93 (0.61-1.44) 
    No. cases 315 91 127 44 23 
Rectal cancer 
    Age-adjusted RR (95% CI)* 1.00 0.54 (0.35-0.85) 0.85 (0.56-1.28) 0.89 (0.52-1.52) 0.62 (0.25-1.54) 
    Fully adjusted RR (95% CI) 1.00 0.59 (0.34-1.01) 0.89 (0.58-1.38) 0.99 (0.57-1.71) 0.68 (0.27-1.69) 
    No. cases 99 24 31 17 
Proximal colon cancer 
    Age-adjusted RR (95% CI)* 1.00 0.70 (0.51-0.94) 1.01 (0.78-1.32) 1.08 (0.73-1.61) 1.01 (0.59-1.72) 
    Fully adjusted RR (95% CI) 1.00 0.76 (0.53-1.10) 1.05 (0.79-1.39) 1.16 (0.78-1.74) 1.05 (0.61-1.80) 
    No. cases 193 55 79 31 15 
Distal colon cancer 
    Age-adjusted RR (95% CI)* 1.00 0.67 (0.45-1.02) 1.13 (0.78-1.63) 0.50 (0.26-0.97) 0.80 (0.37-1.74) 
    Fully adjusted RR (95% CI) 1.00 0.88 (0.53-1.46) 1.25 (0.84-1.86) 0.53 (0.27-1.03) 0.82 (0.38-1.80) 
    No. cases 99 30 40 10 
NeverE-onlyE + P
Current useFormer useCurrent useFormer use
All colorectal cancer 
    Age-adjusted RR (95% CI)* 1.00 0.65 (0.53-0.80) 0.99 (0.82-1.19) 0.86 (0.65-1.13) 0.83 (0.56-1.22) 
    Fully adjusted RR (95% CI) 1.00 0.76 (0.59-0.97) 1.05 (0.86-1.28) 0.93 (0.70-1.23) 0.88 (0.59-1.29) 
    No. cases 414 115 158 61 28 
Colon cancer 
    Age-adjusted RR (95% CI)* 1.00 0.68 (0.54-0.87) 1.03 (0.83-1.27) 0.84 (0.61-1.17) 0.89 (0.58-1.38) 
    Fully adjusted RR (95% CI) 1.00 0.81 (0.61-1.08) 1.10 (0.88-1.37) 0.90 (0.65-1.26) 0.93 (0.61-1.44) 
    No. cases 315 91 127 44 23 
Rectal cancer 
    Age-adjusted RR (95% CI)* 1.00 0.54 (0.35-0.85) 0.85 (0.56-1.28) 0.89 (0.52-1.52) 0.62 (0.25-1.54) 
    Fully adjusted RR (95% CI) 1.00 0.59 (0.34-1.01) 0.89 (0.58-1.38) 0.99 (0.57-1.71) 0.68 (0.27-1.69) 
    No. cases 99 24 31 17 
Proximal colon cancer 
    Age-adjusted RR (95% CI)* 1.00 0.70 (0.51-0.94) 1.01 (0.78-1.32) 1.08 (0.73-1.61) 1.01 (0.59-1.72) 
    Fully adjusted RR (95% CI) 1.00 0.76 (0.53-1.10) 1.05 (0.79-1.39) 1.16 (0.78-1.74) 1.05 (0.61-1.80) 
    No. cases 193 55 79 31 15 
Distal colon cancer 
    Age-adjusted RR (95% CI)* 1.00 0.67 (0.45-1.02) 1.13 (0.78-1.63) 0.50 (0.26-0.97) 0.80 (0.37-1.74) 
    Fully adjusted RR (95% CI) 1.00 0.88 (0.53-1.46) 1.25 (0.84-1.86) 0.53 (0.27-1.03) 0.82 (0.38-1.80) 
    No. cases 99 30 40 10 

*Adjusted for age at 1992 survey.

Adjusted for age at 1992 survey, time-dependent colorectal cancer endoscopy screening, BMI, smoking, education, race, physical activity, use of NSAIDs, use of multivitamins, red meat intake, family history of colorectal cancer, hysterectomy, and type of menopause.

Forty-one colon cancers of overlapping or unknown subsite were excluded from colon subsite analyses.

Among current E-only users, colorectal cancer risk was inversely and linearly associated with duration of use (Ptrend = 0.01; Table 3). Current use of E-only for <5 years was not associated with risk (RR, 1.13), whereas use of E-only for ≥20 years was associated with a substantial reduction in colorectal cancer risk (RR, 0.55). This trend of decreasing risk after 5 years of E-only use persisted in subsite analyses of colon cancer (Ptrend = 0.01), particularly for tumors in the proximal colon (Ptrend = 0.007). There was no clear trend with duration of E-only use for risk of cancer in the distal colon or rectum, although numbers were smaller. RRs for current E + P use were near unity, even for long-duration use. The RRs of former users did not vary by duration (data not presented).

Table 3.

RR of colorectal cancer according to duration of use among current hormone users

CasesAge-adjusted RR (95% CI)*Fully adjusted RR (95% CI)Ptrend
Never 414  
Current E-only (y) 
    <5 27 1.06 (0.71-1.57) 1.13 (0.75-1.70) 0.01 
    5 to <10 21 0.70 (0.45-1.09) 0.76 (0.48-1.21) 
    10 to <15 24 0.66 (0.44-1.00) 0.74 (0.48-1.15) 
    15 to <20 18 0.57 (0.36-0.92) 0.65 (0.40-1.08) 
    ≥20 25 0.47 (0.32-0.71) 0.55 (0.36-0.86) 
Current E + P (y) 
    <5 22 0.79 (0.51-1.23) 0.84 (0.54-1.30) 0.72 
    5 to <10 24 1.06 (0.70-1.62) 1.15 (0.75-1.76) 
    ≥10 15 0.75 (0.45-1.27) 0.85 (0.50-1.45) 
CasesAge-adjusted RR (95% CI)*Fully adjusted RR (95% CI)Ptrend
Never 414  
Current E-only (y) 
    <5 27 1.06 (0.71-1.57) 1.13 (0.75-1.70) 0.01 
    5 to <10 21 0.70 (0.45-1.09) 0.76 (0.48-1.21) 
    10 to <15 24 0.66 (0.44-1.00) 0.74 (0.48-1.15) 
    15 to <20 18 0.57 (0.36-0.92) 0.65 (0.40-1.08) 
    ≥20 25 0.47 (0.32-0.71) 0.55 (0.36-0.86) 
Current E + P (y) 
    <5 22 0.79 (0.51-1.23) 0.84 (0.54-1.30) 0.72 
    5 to <10 24 1.06 (0.70-1.62) 1.15 (0.75-1.76) 
    ≥10 15 0.75 (0.45-1.27) 0.85 (0.50-1.45) 

*Adjusted for age at 1992 survey.

Adjusted for age at 1992 survey, time-dependent colorectal cancer endoscopy, BMI, smoking, education, race, physical activity, use of NSAIDs, use of multivitamins, red meat intake, family history of colorectal cancer, hysterectomy, and type of menopause.

No significant effect modification was found when results were stratified by BMI or attained age (Table 4). The slightly stronger association of E-only use among older women likely reflects their longer average duration of hormone use, which was 18.3 years, as opposed to 11.3 years in the younger women. When results were adjusted for duration, the statistical evidence for an interaction between current E-only use and age was weakened (P = 0.56). The RRs for E + P use were similar between younger and older women.

Table 4.

RR of colorectal cancer according to current hormone use stratified by BMI and attained age

CasesAge adjusted RR* (95% CI)Fully adjusted RR (95% CI)Pinteraction
BMI 
    Current E-only 
        18.5 to <25 Never 180 0.34 
Current use 64 0.72 (0.54-0.96) 0.86 (0.61-1.22) 
        25 to <30 Never 134 
Current use 35 0.64 (0.44-0.93) 0.77 (0.51-1.18) 
        ≥30 Never 87 
Current use 12 0.44 (0.24-0.81) 0.53 (0.28-1.00) 
    Current E + P 
        18.5 to <25 Never 180 0.85 
Current use 32 0.77 (0.53-1.14) 0.81 (0.55-1.20) 
        25 to <30 Never 134 
Current use 17 0.91 (0.55-1.52) 0.94 (0.56-1.58) 
        ≥30 Never 87 
Current use 18 0.96 (0.46-1.99) 0.97 (0.47-2.03) 
Attained age 
    Current E-only 
        Age <70 y Never 182 0.26 
Current use 66 0.71 (0.54-0.95) 0.87 (0.62-1.22) 
        Age ≥70 y Never 232 
Current use 49 0.58 (0.42-0.79) 0.68 (0.48-0.98) 
    Current E + P 
        Age <70 y Never 182 0.80 
Current use 47 0.87 (0.62-1.20) 0.94 (0.67-1.32) 
        Age ≥70 y Never 232 
Current use 14 0.80 (0.46-1.38) 0.87 (0.50-1.51) 
CasesAge adjusted RR* (95% CI)Fully adjusted RR (95% CI)Pinteraction
BMI 
    Current E-only 
        18.5 to <25 Never 180 0.34 
Current use 64 0.72 (0.54-0.96) 0.86 (0.61-1.22) 
        25 to <30 Never 134 
Current use 35 0.64 (0.44-0.93) 0.77 (0.51-1.18) 
        ≥30 Never 87 
Current use 12 0.44 (0.24-0.81) 0.53 (0.28-1.00) 
    Current E + P 
        18.5 to <25 Never 180 0.85 
Current use 32 0.77 (0.53-1.14) 0.81 (0.55-1.20) 
        25 to <30 Never 134 
Current use 17 0.91 (0.55-1.52) 0.94 (0.56-1.58) 
        ≥30 Never 87 
Current use 18 0.96 (0.46-1.99) 0.97 (0.47-2.03) 
Attained age 
    Current E-only 
        Age <70 y Never 182 0.26 
Current use 66 0.71 (0.54-0.95) 0.87 (0.62-1.22) 
        Age ≥70 y Never 232 
Current use 49 0.58 (0.42-0.79) 0.68 (0.48-0.98) 
    Current E + P 
        Age <70 y Never 182 0.80 
Current use 47 0.87 (0.62-1.20) 0.94 (0.67-1.32) 
        Age ≥70 y Never 232 
Current use 14 0.80 (0.46-1.38) 0.87 (0.50-1.51) 

*Adjusted for age at 1992 survey.

Adjusted for age at 1992 survey, time-dependent colorectal cancer endoscopy screening, smoking, education, race, physical activity, use of NSAIDs, use of multivitamins, red meat intake, family history of colorectal cancer, hysterectomy, and type of menopause (models stratified by attained age also adjusted for BMI).

To evaluate the robustness of our E-only results, we performed a series of sensitivity analyses. Because the biological effects of dermal hormone patches may differ from those of oral hormones, we excluded or censored person-time from 2,958 women who reported patch use, and the results remained similar to those of the main analysis (current E-only use, RR, 0.76; 95% CI, 0.59-0.98). Results also remained similar when analyses were restricted to women who had undergone hysterectomy (current E-only use, RR, 0.79; 95% CI, 0.57-1.09). Furthermore, the association with current use of E-only persisted when covariates, BMI, smoking, physical activity, use of NSAIDs, multivitamins, family history of colorectal cancer, and hysterectomy, were modeled as time-dependent variables and updated with each follow-up survey (RR, 0.77; 95% CI, 0.60-0.99).

Finally, to compare our observational results for E-only with the E-only WHI trial, which had an average length of follow-up of 6.8 years, we examined the colorectal cancer risk associated with current E-only use for <7 years and found no association (RR, 0.99; 95% CI, 0.69-1.43).

In this large longitudinal cohort study, current E-only use was associated with a statistically significant lower risk of colorectal cancer, and the association was stronger with increasing duration of use. Although E + P was not associated with colorectal cancer in our study, a reduction in risk similar to that observed with E-only could not be ruled out. Results of other prospective studies that have examined E-only versus E + P are inconsistent. A statistically significant reduction in colorectal cancer risk with E-only, but not E + P, was observed in the Breast Cancer Detection Demonstration Project cohort (13), but the opposite result was observed in a population-based U.K. cohort (32, 33). Results of case-control studies are also conflicting (2, 15, 16, 22), with two supporting a stronger colorectal cancer risk reduction for E-only than E + P (2, 22). A nonsignificant 20% reduction in risk for E-only was found in the WHI observational study, but none was found for E + P (31).

Our findings for E-only appear to differ from those of the WHI randomized trials, which suggested that E-only use was not associated with risk of colorectal cancer (30). One possible reason for this apparent discrepancy is the relatively short intervention period of the WHI E-only trial. Consistent with the relatively short duration of the WHI E-only intervention, current E-only use for <7 years in this study was not associated with risk of colorectal cancer. The risk reduction from E-only use in this study was driven, quite largely, by long-duration use. Collectively, our results, and those of other studies (11, 16, 22, 28), suggest that ≥5 years of hormone use is required to reduce risk of colorectal cancer.

Several other studies noted stronger colorectal cancer risk reductions with hormone use among older than younger women (11, 14, 22, 24, 26). The RR estimates for E-only were lower among older than younger women in our study as well. However, this difference was not statistically significant and appeared to be at least partly attributable to longer duration of use among the older women.

Exogenous estrogen has been hypothesized to have a greater effect on modifying risk of colorectal cancer among women with a low BMI due to the lower endogenous levels of estrogen among lean women (40). Although studies that have evaluated this potential effect modification have been equivocal (14-16, 25, 26, 41), in general, and consistent with our findings, the majority of studies suggest that the influence of estrogens on colorectal cancer risk is not dependent on BMI status.

Our study has several notable strengths: it is one of the largest to date, allowing reasonable statistical power for subgroup analyses; its prospective design eliminates the bias due to differential recall of hormone use. Because of the availability of detailed, regularly updated information about hormone use, we were able to characterize mutually exclusive groups of E-only users, E + P users, and nonhormone users, thus reducing misclassification. In addition, we were able to control for colorectal endoscopy screening in a time-dependent manner, reducing the potential for confounding due to differential screening practices among hormone users and nonusers. A limitation of this study is that hormone use was self-reported, although self-reported hormone use has been shown to be in good agreement with physician records (42). Our ability to precisely estimate the risk associated with long-duration E + P use was also limited by the relatively small numbers of long-term E + P users in our cohort. Our study participants were predominantly White, middle-aged or elderly, and well-educated; therefore, results may not be generalizable to populations with different characteristics.

The biological mechanisms responsible for the inverse association with E-only are unclear, but there is evidence to suggest that estrogens may specifically reduce risk of tumors with microsatellite instability (43). Microsatellite instability may often occur as a result of methylation in a specific mismatch repair gene, MLH1. Studies (44, 45) suggest that preventing age-related methylation, perhaps of the estrogen receptor or one of the mismatch repair genes (e.g., MLH1), may be an important mechanism through which estrogens exert a protective effect in the colon and thereby decrease risk of microsatellite instability tumors. Because we do not have tumor samples, however, we cannot address this issue directly.

Our results, based on a large prospective study of women with regularly updated exposure information, suggest that long-term use of E-only reduces risk of colorectal cancer. A reduction in risk for E + P was not observed in these data yet could not be ruled out. The prevalence of menopausal hormone use peaked in 1999 and then declined dramatically in the United States following the 2002 publication of results from the WHI trials (39). Given evidence from randomized trials that the risks associated with hormone use likely outweigh the benefits, our finding of a potential benefit for colorectal cancer risk should not appreciably influence current clinical practice. Nevertheless, our study underscores the importance of continued follow-up of the WHI and prospectively followed cohorts with detailed information on lifetime hormone use to gain a better understanding of the long-term effects of exogenous hormones on disease risks.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We acknowledge and continue to mourn the untimely and tragic loss this year of our beloved colleague and friend, Eugenia “Jeanne” Calle, distinguished researcher and Vice President of Epidemiology, American Cancer Society. We miss her inspiring leadership, intellect and insight, and especially and most especially, her passion, contagious sense of humor, and love of life.

1
Calle
EE
,
Miracle-McMahill
HL
,
Thun
MJ
,
Heath
CW
 Jr
. 
Estrogen replacement therapy and risk of fatal colon cancer in a prospective cohort of postmenopausal women
.
J Natl Cancer Inst
1995
;
87
:
517
23
.
2
Campbell
PT
,
Newcomb
P
,
Gallinger
S
,
Cotterchio
M
,
McLaughlin
JR
. 
Exogenous hormones and colorectal cancer risk in Canada: associations stratified by clinically defined familial risk of cancer
.
Cancer Causes Control
2007
;
18
:
723
33
.
3
Corrao
G
,
Zambon
A
,
Conti
V
, et al
. 
Menopause hormone replacement therapy and cancer risk: an Italian record linkage investigation
.
Ann Oncol
2008
;
19
:
150
5
.
4
Csizmadi
I
,
Collet
JP
,
Benedetti
A
,
Boivin
JF
,
Hanley
JA
. 
Defining hormone replacement therapy in longitudinal studies: impact on measures of effect
.
Pharmacoepidemiol Drug Saf
2004
;
13
:
215
25
.
5
Davis
FG
,
Furner
SE
,
Persky
V
,
Koch
M
. 
The influence of parity and exogenous female hormones on the risk of colorectal cancer
.
Int J Cancer
1989
;
43
:
587
90
.
6
Fernandez
E
,
La Vecchia
C
,
Braga
C
, et al
. 
Hormone replacement therapy and risk of colon and rectal cancer
.
Cancer Epidemiol Biomarkers Prev
1998
;
7
:
329
33
.
7
Folsom
AR
,
Mink
PJ
,
Sellers
TA
,
Hong
CP
,
Zheng
W
,
Potter
JD
. 
Hormonal replacement therapy and morbidity and mortality in a prospective study of postmenopausal women
.
Am J Public Health
1995
;
85
:
1128
32
.
8
Furner
SE
,
Davis
FG
,
Nelson
RL
,
Haenszel
W
. 
A case-control study of large bowel cancer and hormone exposure in women
.
Cancer Res
1989
;
49
:
4936
40
.
9
Gerhardsson de Verdier
M
,
London
S
. 
Reproductive factors, exogenous female hormones, and colorectal cancer by subsite
.
Cancer Causes Control
1992
;
3
:
355
60
.
10
Grodstein
F
,
Martinez
ME
,
Platz
EA
, et al
. 
Postmenopausal hormone use and risk for colorectal cancer and adenoma
.
Ann Intern Med
1998
;
128
:
705
12
.
11
Jacobs
EJ
,
White
E
,
Weiss
NS
. 
Exogenous hormones, reproductive history, and colon cancer (Seattle, Washington, USA)
.
Cancer Causes Control
1994
;
5
:
359
66
.
12
Jacobs
EJ
,
White
E
,
Weiss
NS
,
Heckbert
SR
,
LaCroix
A
,
Barlow
WE
. 
Hormone replacement therapy and colon cancer among members of a health maintenance organization
.
Epidemiology
1999
;
10
:
445
51
.
13
Johnson
JR
,
Lacey
JV
 Jr
,
Lazovich
D
, et al
. 
Menopausal hormone therapy and risk of colorectal cancer
.
Cancer Epidemiol Biomarkers Prev
2009
;
18
:
196
203
.
14
Kampman
E
,
Potter
JD
,
Slattery
ML
,
Caan
BJ
,
Edwards
S
. 
Hormone replacement therapy, reproductive history, and colon cancer: a multicenter, case-control study in the United States
.
Cancer Causes Control
1997
;
8
:
146
58
.
15
Newcomb
PA
,
Storer
BE
. 
Postmenopausal hormone use and risk of large-bowel cancer
.
J Natl Cancer Inst
1995
;
87
:
1067
71
.
16
Newcomb
PA
,
Zheng
Y
,
Chia
VM
, et al
. 
Estrogen plus progestin use, microsatellite instability, and the risk of colorectal cancer in women
.
Cancer Res
2007
;
67
:
7534
9
.
17
Nichols
HB
,
Trentham-Dietz
A
,
Hampton
JM
,
Newcomb
PA
. 
Oral contraceptive use, reproductive factors, and colorectal cancer risk: findings from Wisconsin
.
Cancer Epidemiol Biomarkers Prev
2005
;
14
:
1212
8
.
18
Paganini-Hill
A
. 
Estrogen replacement therapy and colorectal cancer risk in elderly women
.
Dis Colon Rectum
1999
;
42
:
1300
5
.
19
Persson
I
,
Yuen
J
,
Bergkvist
L
,
Schairer
C
. 
Cancer incidence and mortality in women receiving estrogen and estrogen-progestin replacement therapy—long-term follow-up of a Swedish cohort
.
Int J Cancer
1996
;
67
:
327
32
.
20
Peters
RK
,
Pike
MC
,
Chang
WW
,
Mack
TM
. 
Reproductive factors and colon cancers
.
Br J Cancer
1990
;
61
:
741
8
.
21
Potter
JD
,
McMichael
AJ
. 
Large bowel cancer in women in relation to reproductive and hormonal factors: a case-control study
.
J Natl Cancer Inst
1983
;
71
:
703
9
.
22
Prihartono
N
,
Palmer
JR
,
Louik
C
,
Shapiro
S
,
Rosenberg
L
. 
A case-control study of use of postmenopausal female hormone supplements in relation to the risk of large bowel cancer
.
Cancer Epidemiol Biomarkers Prev
2000
;
9
:
443
7
.
23
Weiss
NS
,
Daling
JR
,
Chow
WH
. 
Incidence of cancer of the large bowel in women in relation to reproductive and hormonal factors
.
J Natl Cancer Inst
1981
;
67
:
57
60
.
24
Murff
HJ
,
Shrubsole
MJ
,
Smalley
WE
, et al
. 
The interaction of age and hormone replacement therapy on colon adenoma risk
.
Cancer Detect Prev
2007
;
31
:
161
5
.
25
Potter
JD
,
Bostick
RM
,
Grandits
GA
, et al
. 
Hormone replacement therapy is associated with lower risk of adenomatous polyps of the large bowel: the Minnesota Cancer Prevention Research Unit Case-Control Study
.
Cancer Epidemiol Biomarkers Prev
1996
;
5
:
779
84
.
26
Purdue
MP
,
Mink
PJ
,
Hartge
P
,
Huang
WY
,
Buys
S
,
Hayes
RB
. 
Hormone replacement therapy, reproductive history, and colorectal adenomas: data from the Prostate, Lung, Colorectal and Ovarian (PLCO) Cancer Screening Trial (United States)
.
Cancer Causes Control
2005
;
16
:
965
73
.
27
Terry
MB
,
Neugut
AI
,
Bostick
RM
, et al
. 
Risk factors for advanced colorectal adenomas: a pooled analysis
.
Cancer Epidemiol Biomarkers Prev
2002
;
11
:
622
9
.
28
Risch
HA
,
Howe
GR
. 
Menopausal hormone use and colorectal cancer in Saskatchewan: a record linkage cohort study
.
Cancer Epidemiol Biomarkers Prev
1995
;
4
:
21
8
.
29
Rossouw
JE
,
Anderson
GL
,
Prentice
RL
, et al
. 
Risks and benefits of estrogen plus progestin in healthy postmenopausal women: principal results from the Women's Health Initiative randomized controlled trial
.
JAMA
2002
;
288
:
321
33
.
30
Anderson
GL
,
Limacher
M
,
Assaf
AR
, et al
. 
Effects of conjugated equine estrogen in postmenopausal women with hysterectomy: the Women's Health Initiative randomized controlled trial
.
JAMA
2004
;
291
:
1701
12
.
31
Prentice
RL
,
Pettinger
M
,
Beresford
SA
, et al
. 
Colorectal cancer in relation to postmenopausal estrogen and estrogen plus progestin in the Women's Health Initiative clinical trial and observational study
.
Cancer Epidemiol Biomarkers Prev
2009
;
18
:
1531
7
.
32
Tannen
RL
,
Weiner
MG
,
Xie
D
,
Barnhart
K
. 
A simulation using data from a primary care practice database closely replicated the Women's Health Initiative trial
.
J Clin Epidemiol
2007
;
60
:
686
95
.
33
Tannen
RL
,
Weiner
MG
,
Xie
D
,
Barnhart
K
. 
Estrogen affects post-menopausal women differently than estrogen plus progestin replacement therapy
.
Hum Reprod
2007
;
22
:
1769
77
.
34
Calle
EE
,
Rodriguez
C
,
Jacobs
EJ
, et al
. 
The American Cancer Society Cancer Prevention Study II Nutrition Cohort: rationale, study design, and baseline characteristics
.
Cancer
2002
;
94
:
2490
501
.
35
Stellman
SD
,
Garfinkel
L
. 
Smoking habits and tar levels in a new American Cancer Society prospective study of 1.2 million men and women
.
J Natl Cancer Inst
1986
;
76
:
1057
63
.
36
Bergmann
MM
,
Calle
EE
,
Mervis
CA
,
Miracle-McMahill
HL
,
Thun
MJ
,
Heath
CW
. 
Validity of self-reported cancers in a prospective cohort study in comparison with data from state cancer registries
.
Am J Epidemiol
1998
;
147
:
556
62
.
37
Calle
EE
,
Terrell
DD
. 
Utility of the National Death Index for ascertainment of mortality among Cancer Prevention Study II participants
.
Am J Epidemiol
1993
;
137
:
235
41
.
38
Kleinbaum
DG
.
Survival analysis: a self-learning text
.
New York (NY): Springer
; 
1996
.
39
Hersh
AL
,
Stefanick
ML
,
Stafford
RS
. 
National use of postmenopausal hormone therapy: annual trends and response to recent evidence
.
JAMA
2004
;
291
:
47
53
.
40
Terry
PD
,
Miller
AB
,
Rohan
TE
. 
Obesity and colorectal cancer risk in women
.
Gut
2002
;
51
:
191
4
.
41
Troisi
R
,
Schairer
C
,
Chow
WH
,
Schatzkin
A
,
Brinton
LA
,
Fraumeni
JF
 Jr
. 
A prospective study of menopausal hormones and risk of colorectal cancer (United States)
.
Cancer Causes Control
1997
;
8
:
130
8
.
42
Jain
MG
,
Rohan
TE
,
Howe
GR
. 
Agreement of self-reported use of menopausal hormone replacement therapy with physician reports
.
Epidemiology
1999
;
10
:
260
3
.
43
Slattery
ML
,
Potter
JD
,
Curtin
K
, et al
. 
Estrogens reduce and withdrawal of estrogens increase risk of microsatellite instability-positive colon cancer
.
Cancer Res
2001
;
61
:
126
30
.
44
Di Leo
A
,
Messa
C
,
Cavallini
A
,
Linsalata
M
. 
Estrogens and colorectal cancer
.
Curr Drug Targets Immune Endocr Metabol Disord
2001
;
1
:
1
12
.
45
Issa
JP
,
Ottaviano
YL
,
Celano
P
,
Hamilton
SR
,
Davidson
NE
,
Baylin
SB
. 
Methylation of the oestrogen receptor CpG island links ageing and neoplasia in human colon
.
Nat Genet
1994
;
7
:
536
40
.