|Home | About | Journals | Submit | Contact Us | Français|
Lung cancer is the leading cause of cancer-related mortality among women. The role of hormone replacement therapy (HRT) in lung cancer development is unclear.
We evaluated a prospective cohort of 36,588 peri- and postmenopausal women aged 50 to 76 years from Washington State recruited in 2000 to 2002 (Vitamins and Lifestyle [VITAL] Study). Lung cancer cases (n = 344) were identified through the Seattle-Puget Sound Surveillance, Epidemiology, and End Results cancer registry during 6 years of follow-up. Hazard ratios (HRs) associated with use and duration of specific HRT formulations were calculated for total incident lung cancer, specific morphologies, and cancer by stage at diagnosis.
After adjusting for smoking, age, and other potential confounders, there was an increased risk of incident lung cancer associated with increasing duration of estrogen plus progestin (E+P) use (HR = 1.27 for E+P use 1 to 9 years, 95% CI, 0.91 to 1.78; and HR = 1.48 for E+P use ≥ 10 years, 95% CI, 1.03 to 2.12; P for trend = .03). There was no association with duration of unopposed estrogen use. Duration of E+P use was associated with an advanced stage at diagnosis (P for trend = .03).
Use of E+P increased the risk of incident lung cancer in a duration-dependent manner, with an approximate 50% increased risk for use of 10 years or longer. These findings may be helpful for informing women of their risk of developing lung cancer and delineating important pathways involved in hormone metabolism and lung cancer.
Lung cancer is the leading cause of cancer mortality for women in the United States.1 Tobacco causes 90% of lung cancers,2 and cessation is the only recommended method for prevention.3 However, the prevalence of tobacco use remains high,4 and lung cancer risk persists after smoking cessation.5 Accordingly, it is important to evaluate additional modalities that affect the risk of incident lung cancer.
Exogenous hormones may impact lung cancer development. The biologic mechanisms underlying hormone metabolism and lung cancer are not clear in direction or magnitude.6,7 Studies evaluating hormone replacement therapy (HRT) use and incident lung cancer have had conflicting results and many have had limitations such as not examining individual HRT formulations, duration of use, limited adjustment for the confounding effects of tobacco use, and/or limited to nonsmokers.8–21 There is concern that associations found in observational studies may not be replicated in placebo-controlled trials as shown in trials of both lung cancer chemoprevention and HRT,22–24 and several trials have suggested that estrogen plus progestin may be associated with an increased risk of lung cancer incidence25 and death.26–28
Given these conflicting findings, studies of HRT and lung cancer should examine the effects of duration and HRT formulation in the association between HRT and lung cancer and adjust for the confounding of smoking to decrease the chances of misconstruing the relationship between HRT and lung cancer.29 We used the prospective cohort study (Vitamins and Lifestyle [VITAL] study) to evaluate the association of HRT use with incident lung cancer.
The methods used in the VITAL cohort study have been described.30 From October 2000 to December 2002, we mailed questionnaires to 168,953 women aged 50 to 76 years who lived in the area covered by the Seattle-Puget Sound Surveillance, Epidemiology, and End Results (SEER) registry, using names from a commercial mailing company. A total of 41,157 women (24.4%) returned the questionnaire, and of those, 820 failed eligibility or quality control checks, leaving 40,337 women. Baseline data included items on medication use, diet, medical history, personal characteristics, and cancer risk factors. The responders were more educated than the general population and smoked less currently, but were similar in body mass index (BMI).30 The Fred Hutchinson Cancer Research Center's institutional review board approved the protocol.
The censored date was the date of withdrawal from the study, death, move out of the SEER catchment area, or last date of linkage to SEER for diagnosis of lung cancer. Deaths were ascertained by linkage to Washington State death files, and moves out of the area from the National Change of Address System, letters, and telephone calls.
For this report, we excluded participants with a previous diagnosis of lung cancer (n = 176) or for whom this datum was missing (n = 86). We excluded subjects whose lung cancer was identified on a death certificate only or whose lung cancer morphology was lymphoma (n = 4). We then excluded subjects who were not perimenopausal or postmenopausal (n = 1,995). Lastly, we excluded subjects with completely missing information regarding HRT use (n = 1,488), leaving 36,588 women.
Participants were observed for lung cancer occurring from baseline through December 31, 2007, through linkage to SEER. SEER has accurate and complete data collection31 and is reliable for lung cancer histology.32 If a subject had multiple diagnoses of lung cancer, we used the time to first diagnosis.
Hormone replacement therapy use was ascertained by inquiring about use of prescription estrogen and progestin as pills or patches, excluding oral contraceptives. Information was obtained on hormone therapy use status (never, former, current) and years of use (categorized as ≤ 1, 1 to 4, 5 to 9, 10 to 14, or ≥ 15). Past HRT use for less than 1 year was combined with no use (never). Current HRT use less than 1 year was classified as 1 to 4 years of use. Years of estrogen plus progestin (E+P) use and years of estrogen-only use were computed separately. The reference group was users of neither type of HRT. Women with periods of exposure to both unopposed estrogen and E+P had the relevant years of use included in both analyses. Subjects with estrogen only use were excluded from the duration of E+P use analysis and users of E+P only were excluded from the duration of estrogen use analysis.
Smokers were defined as individuals who smoked at least one cigarette per day for at least a year. Smoking status was classified as never, current, quit 10 years or more, or quit less than 10 years ago, as of the date of questionnaire completion. Duration of smoking was estimated by the number of years smoked, intensity was estimated by the usual number of cigarettes smoked per day, and pack-years was computed as years smoked × cigarettes per day/20.
Perimenopausal women were defined as having menses in the past year that were not regular. Women were assumed to be postmenopausal if they had no periods in the year before baseline, had ever used hormone therapy, had had a bilateral oophorectomy, or were ≥ 60 years at baseline. Age at menopause was categorized in 5-year increments (≤ 39 to ≥ 55 years) to the age at which menstrual periods ended or age of first use of hormone therapy or were 60 years old at baseline, whichever came first. Women who reported a hysterectomy without bilateral oophorectomy were considered to be postmenopausal if they had ever used hormone therapy or were ≥ 55 years at baseline. For those women, age at menopause was set to the age at which they first used hormone therapy (if before age 55 years); otherwise, it was set to missing.
Subjects reported age, race/ethnicity, marital status, and education. Previous history of cancer and self-report of diagnosed chronic obstructive pulmonary disease (COPD) were recorded. We categorized family history as none or at least one first-degree relative with lung cancer. BMI was calculated from the respondent's self-reported current weight and height, measured as kilograms per square meters.
Daily servings of fruit were assessed by a food frequency questionnaire that was an adaptation from the Women's Health Initiative and other studies,33–35 with the addition of highly supplemented foods. Subjects were extensively queried about use of nonsteroidal anti-inflammatory drugs and supplemental vitamins.
All analyses were performed using STATA SE-9 (StataCorp, College Station, TX). For the main analysis, Cox regression using robust SEs36 was used to estimate the hazard ratios (HRs) for associations of HRT use categories and lung cancer risk. Age was the time variable with left truncation for age at baseline and censoring (right truncation) as described above. Subjects with missing data on HRT use or other covariates in the model were excluded from analysis. We analyzed four exposure variables: HRT use, analyzed by categories of use (never, former, and current); ever-use of specific formulations of HRT (none, estrogen only, E+P only, and both); duration of use of E+P; and duration of estrogen only. We treated each duration of use category as a continuous trend variable to assess for trends in lung cancer risk, comparing each category with no use.
On the basis previous work, we used a model that adjusts for confounding by cigarette smoking that included years smoked, pack-years, and a squared pack-years term.37 We decided a priori to include age, nonwhite race/ethnicity, history of cancer, family history of lung cancer, COPD, BMI (categoric, including missing as a category), age at menopause (categoric, including missing as a category), and hysterectomy/oophorectomy status. We evaluated whether other factors associated with lung cancer risk (nonsteroidal anti-inflammatory drug use, supplemental vitamin E, education, and daily servings of fruit) individually confounded the association of HRT use with lung cancer in the adjusted model. None changed the point estimates of the HRT variables by more than 10% or the 5% level of significance so were not included.
We examined whether the associations of HRT with incident lung cancer use differed by morphology by treating each morphology as a separate outcome, exclusive of the other morphologies, compared with subjects who did not develop lung cancer. We also looked for differences of the HRT–lung cancer associations by smoking status and BMI subgroups. Because few never smokers developed lung cancer, we did not include this group in the stratified smoking status analyses. Likelihood ratio tests were conducted to assess the interaction between HRT use, analyzed as trend variables, and the subgroups. P values for interaction were obtained to compare the fit of the models with the interaction terms and without them. Finally, we examined whether the associations of HRT with incident lung cancer use differed by local versus regional/advanced/unknown stage at diagnosis by treating these stages as separate outcomes, exclusive of the other stages, compared with subjects who did not develop lung cancer. Unknown stage was combined with the latter groups because staging information is often missing in advanced cases that are not candidates for surgery. P values less than .05 were considered statistically significant.
A total of 36,588 subjects met inclusion criteria and were observed for a mean of 5.9 years (standard deviation = 1.2 years). Three hundred forty-four patients developed lung cancer. Non–small-cell lung cancer (NSCLC) accounted for 77.0% of the total: adenocarcinoma (n = 141 [41.0%]), squamous cell (n = 47 [13.7%]), large cell (n = 9 [2.6%]) and NSCLC, not otherwise specified (n = 68 [19.8%]). Small-cell lung cancer (SCLC) accounted for 12.5% of the total lung cancers. Other lung cancers, mostly comprising carcinomas not otherwise specified and carcinoid/neuroendocrine tumors, accounted for 10.5% of the total. Seventy-three subjects (21.2%) were diagnosed with a local stage at diagnosis, 97 (28.2%) regional, 164 (47.7%) distant, and 10 (2.9%) had an unknown stage.
When stratified by the formulation of HRT used, the subjects were similar in most respects (Table 1). The E+P only users were somewhat younger with fewer pack-years and years of smoking. This group also had a smaller percentage with a history of cancer but a higher percentage without a hysterectomy.
After adjustment, current and former use of HRT was associated with a nonsignificantly elevated risk of incident lung cancer (HR = 1.23, 95% CI, 0.92 to 1.66; and HR = 1.18, 95% CI, 0.86 to 1.63, respectively; Table 2). Compared with no use of HRT, use of the estrogen plus progestin formulation only was associated with an increased risk of incident lung cancer (HR = 1.47; 95% CI, 1.06 to 2.04). Use of unopposed estrogen only and use of both formulations were not associated with lung cancer.
There was evidence of a dose response associated with the duration of E+P use. Use of E+P for ≥ 10 years was associated with an increased risk of lung cancer compared with no use of HRT (HR = 1.48; 95% CI, 1.03 to 2.12; P for trend = .03). Conversely, there was no association between duration of estrogen use and lung cancer.
When we stratified the analysis by NSCLC and SCLC, no status, formulation, or duration of HRT use was statistically significantly associated with either morphology, though all except estrogen use for ≥ 10 years were associated with increased risks (Table 3). No category of HRT use, formulation, or duration was significantly associated with adenocarcinoma when this morphology was analyzed separately (data not shown).
There was no clear evidence of effect modification by either smoking status or BMI. No category of HRT use, formulation, or duration was significantly associated with incident lung cancer when stratified by smoking status or BMI categories (data not shown).
No HRT use, formulation, or duration category was significantly associated with local stage at diagnosis (Table 4). However, E+P use (HR = 1.52; 95% CI, 1.06 to 2.19; P = .04) was associated with more advanced stages, and increasing duration of E+P use was associated as well (P for trend = .03). Unopposed estrogen use and duration were not associated with advanced stages at diagnosis.
The use of E+P was associated with an increased risk of incident lung cancer in this study. Although HRT use has declined38 and is not recommended except for short-term treatment of menopausal symptoms,39 our results indicate millions of women may remain at risk of developing lung cancer. We found that the association of HRT with lung cancer was duration dependent, with the highest risk for users of E+P ≥ 10 years. Use of E+P was also associated with an advanced stage at diagnosis in a duration-dependent manner. We do not have the power to estimate a safe length of HRT use.
There were not large differences in the associations between NSCLC and SCLC. Similarly, smoking status did not seem to modify the associations. A recent cohort study that also showed increased risks of lung cancer associated with HRT use did not see differences in the association with regard to histology and smoking status.11 Although obesity may modify the association between HRT and lung cancer,21 BMI was not an effect modifier in our study. The relatively small number of cases in each subcategory limits our ability to detect differences.
Previous studies of the association between HRT and lung cancer have been mixed.8–21 In the Cancer Prevention Study II, current use of any HRT was associated with a decreased risk of lung cancer (relative risk = 0.76; 95% CI, 0.62 to 0.92) that was not duration dependent and did not differ by formulation.9 A case-control study found a duration-dependent decreased risk for lung cancer with HRT use (odds ratio = 0.88; 95% CI, 0.78 to 1.00 for each duration quartile) and did not differ by formulation.20 Another case-control study reported protective associations with HRT use that did not differ between unopposed estrogen and E+P.18 Women in the Canadian National Breast Screening Study using HRT (formulation not specified) for 10 years or longer, however, had an elevated risk (HR = 1.51; 95% CI, 1.14 to 1.99).11 The Heart and Estrogen/Progestin Replacement Study (HERS) found a nonsignificantly increased risk of lung cancer for women randomly assigned to conjugated E+P (HR = 1.39; 95% CI, 0.84 to 2.28).25
Recent Women's Health Initiative analyses found that subjects randomly assigned to E+P had more lung cancer deaths than those taking placebo.26,27 A retrospective study also found decreased survival among HRT users with lung cancer.28 SEER collects mortality data, but we did not analyze lung cancer mortality because we do not have information on HRT use after diagnosis. However, to explore a possible explanation for the effect of HRT on lung cancer mortality, we analyzed the association of HRT use with stage at diagnosis and found an increased risk of advanced stages at diagnosis with E+P, consistent with the Women's Health Initiative results on lung cancer mortality.26,27 Although the mechanisms underlying this association are unknown, E+P may lead to more aggressive disease, may mask early symptoms, and/or users may be less likely to seek or receive medical care in a timely fashion.
The mechanisms that contribute to the association between exogenous hormone therapy and lung cancer risk are complex, with likely genetic and environment interactions.6,40 Estrogen and progesterone receptors are found in NSCLC tumors41,42 and the quantity of estrogen receptors is associated with decreased recurrence-free survival.42 Estradiol promotes growth in NSCLC cell lines that is blocked by antiestrogens.43 Similarly, a mouse model showed that estradiol increased the proliferative index of cells that had been initiated by expression of oncogenic Kras and concurrent deletion of Tp53.44 A combination of an estrogen antagonist and epidermal growth factor receptor tyrosine kinase inhibitor decreased tumor volume more than either drug alone in vivo in xenograft models41,45 and in vitro.46 Evidence for a protective association of HRT has been shown. Low levels of insulin-like growth factor 1 are associated with reduced risk of lung cancer,47 and HRT has been shown to decrease insulin-like growth factor 1 levels.14 Finally, a recent study found several associations between single nucleotide polymorphisms in genes involved with estrogen metabolic pathways and an increased risk of lung cancer.48
Our study has several strengths. We used a large, prospective, population-based cohort study design. We were able to analyze the duration of HRT use in a dose-response manner over a long period of time that is likely necessary for biologic plausibility. We controlled for the strong confounding effect of tobacco and examined multiple other variables that affect the risk of incident lung cancer. Finally, the SEER database is complete and accurate, so there is minimal risk of outcome misclassification.
Our results are similar to those observed in the randomized, placebo-controlled HERS trial,25 but there are potential limitations of our study. First, residual confounding may be a factor. We could not adjust for environmental tobacco or occupational exposures that increase lung cancer risk. However, we did adjust for several smoking variables and did not find evidence of confounding by indication from processes such as oophorectomy for which HRT is commonly taken. Second, the measurement of long-term use of HRT is based on subject recall and was not measured repeatedly. The VITAL questionnaire was validated for long-term use of supplements,49 suggesting our measure of HRT is reasonably accurate. We expect that exposure misclassification would have attenuated our results in this prospective study. Generalizability may be limited because only 24% of respondents returned the initial survey, but it is unlikely that selection bias could have affected our results because in a prospective design, women cannot participate jointly based on exposure and future (unknown) disease status.
Lung cancer is the second most common malignancy among women and causes more deaths than breast cancer.1 Our results indicate that the use of E+P is associated with an increased risk of incident lung cancer and advanced stages at diagnosis in a duration-response manner. These findings represent an important contribution to current evidence against HRT use, which may be useful for counseling women about their risk of developing lung cancer, as well as prompting further study about the biologic mechanisms underlying this association.
Supported by the CHEST Foundation of the American College of Chest Physicians and the LUNGevity Foundation to C.G.S. This work was also supported by the National Institutes of Health (Grant No. CA130328 to C.G.S., CA74846 to E.W., and CA96556 and CA119683 to J.S.). C.G.S. and D.H.A. were supported by the Department of Veterans Affairs. This study is the result of work supported by resources from the Portland Veterans Affairs Medical Center, Portland, OR, and Veterans Affairs Puget Sound Health Care System, Seattle, WA.
The views expressed in this article are those of the authors and do not necessarily represent the views of the Department of Veterans Affairs.
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
The author(s) indicated no potential conflicts of interest.
Conception and design: Christopher G. Slatore, Jason W. Chien, Jessie A. Satia, Emily White
Financial support: Christopher G. Slatore, Emily White
Administrative support: David H. Au, Emily White
Collection and assembly of data: Christopher G. Slatore, Emily White
Data analysis and interpretation: Christopher G. Slatore, Jason W. Chien, David H. Au, Jessie A. Satia, Emily White
Manuscript writing: Christopher G. Slatore, Jason W. Chien, David H. Au, Emily White
Final approval of manuscript: Christopher G. Slatore, Jason W. Chien, David H. Au, Emily White