Hormonal factors and risk of lung cancer among ... - Semantic Scholar

13 downloads 0 Views 88KB Size Report
Dec 14, 2002 - influence of sex-specific hormones. This study examines endocrine factors and risk of lung cancer among women by smoking status and ...
© International Epidemiological Association 2003

Printed in Great Britain

International Journal of Epidemiology 2003;32:263–271 DOI: 10.1093/ije/dyg064

Hormonal factors and risk of lung cancer among women? Michaela Kreuzer,1 Michael Gerken,2 Joachim Heinrich,3 Lothar Kreienbrock4 and H-Erich Wichmann3

Accepted

14 December 2002

Background Gender differences in the histological distribution of lung carcinoma and a possibly greater susceptibility of women than men to tobacco carcinogens, suggest a possible influence of sex-specific hormones. This study examines endocrine factors and risk of lung cancer among women by smoking status and histology. Methods

We used data of a case-control study on lung cancer conducted from 1990 to 1996 in Germany, including 811 histologically confirmed female cases and 912 female population controls. Information on various menstrual and reproductive factors, use of oral contraceptives (OC), hormone replacement therapy (HRT), and smoking was gathered through personal interviews using a structured questionnaire. Odds ratios (OR) and 95% CI adjusted for age, region, smoking, and education were calculated via logistic regression.

Results

A reduction in lung cancer risk was observed with the use of OC (OR = 0.69; 95% CI: 0.51–0.92), but no trend in risk with increasing duration of use, age at first use, or calendar year of first use was present. A history of HRT was associated with a reduced risk (OR = 0.83; 95% CI: 0.64–1.09), particularly after long duration (7 years) (OR = 0.59; 95% CI: 0.37–0.93). No clear association was found with regard to age at menarche, length of menstrual cycle, number of livebirths, and age at menopause. Overall results did not differ much by histological cell subtype. The reduction in lung cancer risk associated with the use of exogenous hormones was primarily seen among smoking women.

Conclusions Our data provide evidence for a possible role of hormonal factors in the aetiology of lung cancer in women. Keywords

Case-control study, hormones, lung cancer, smoking, women

Differences in the histological distribution of lung carcinoma between men and women, which cannot be fully explained by differences in smoking habits,1 and a possibly greater susceptibility of women to tobacco carcinogens than men,2,3 indicate that sex-specific hormones may play a role in the genesis of lung cancer. Siegfried4 hypothesized that oestrogens may possibly influence lung cancer development, either

1 Institute of Radiation Hygiene, BfS—Federal Office for Radiation Protection,

Neuherberg, Germany. 2 Tumour Centre, University of Regensburg, Regensburg, Germany. 3 Institute of Epidemiology, GSF—National Research Center for Environment

and Health, Neuherberg, Germany. 4 Institute of Biometry, Epidemiology and Data Information Processing,

School of Veterinary Medicine, University of Hannover, Hannover. Correspondence: Dr Michaela Kreuzer, BfS-Institute of Radiation Hygiene, Ingolstädter Landstr. 1, 85764 Neuherberg, Germany. E-mail: [email protected]

through direct promotion by oestrogen of cell proliferation in the lung, or as a result of the influence of estrogen on lungcarcinogen metabolism or development of lung diseases that predispose to lung cancer. Evidence for an association between hormonal factors and lung cancer among women is available from several lines. First, possible biological evidence was given by some experimental studies indicating the presence of steroid receptors in some lung tumour types.5–8 Steroids such as glucocorticoids and oestrogens regulate the differentiation and metabolism of pulmonary epithelial cells. Second, a familial aggregation of cancers of the reproductive system among female lung cancer patients9,10 in addition to an increased risk of lung cancer in female survivors of a primary cancer of the reproductive organs11–17 have been described, indicating a possibly endocrine-related common aetiology of cancers of the lung and the reproductive organs. Third, some menstrual18–20 and reproductive factors18,21–23 have been associated with lung cancer risk in women. Women having ever used oral

263

264

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

contraceptives (OC) was suggestive of a reduced lung cancer risk.23 Inconsistent results have been provided with respect to oestrogen replacement therapy.21,24–26 A positive synergism between smoking and oestrogen replacement therapy was reported by Taioli and Wynder21 and was hypothesized as a possible explanation for a heightened smoking risk of women compared with men. The aim of the present analysis of the data of a case-control study on lung cancer in Germany was to examine the relation of various hormonal factors in the aetiology of lung cancer in women. Particular emphasis was given on estimating these effects separately in smokers and non-smokers and for different histological types. Findings of this study on other risk factors have been published previously.27–30

Materials and Methods Data were derived from a case-control study of lung cancer and indoor radon conducted between 1990 and 1996 in Germany. The data collection methods have been described previously.29,30 In brief, newly diagnosed (3 months) cases with histologically or cytologically confirmed lung cancer as a primary tumour were recruited from 15 study hospitals. Inclusion criteria for cases were: (1) an age 76 years at the date of interview; (2) current residence in the study area; (3) residence in Germany 25 years; (4) an adequately good clinical state to face an interview of around 1.3 hours duration; and (5) no suspicion of pulmonary metastases from a different tumour. Population controls satisfying inclusion criteria (1)–(4) were randomly selected from either residency registries or by modified random digit dialling and were frequency matched to the cases on sex, age, and region. The response rate of cases and controls fulfilling the inclusion criteria was 76% and 41%, respectively. A total of 4276 cases (3465 men and 811 women) and 4629 controls (3717 men and 912 women) were personally interviewed by trained interviewers. A structured questionnaire was used to obtain information on demographic characteristics, residential history, active and passive (non-smokers only) smoking history, job history, personal and family medical history, dietary habits, and endocrine factors. Subjects were defined as smokers if they had ever smoked regularly (at least one cigarette per day) for 6 months. Active and former smokers were summarized as smokers and were compared with non-smokers. One-year radon measurements were performed in the current residence.30 Among women, detailed information on hormonal factors such as age at first and last menstrual period, average length of menstrual cycle, age at first pregnancy, and number of full-term pregnancies was collected. Menopausal status could be determined only indirectly by cessation of menstrual periods. Women reporting that they had still their menstrual periods at the date of interview were defined as premenopausal women. Those reporting cessation of menstruation were asked about age at last menstrual period defined as age at menopause in the analyses and about cause of menopause (natural, surgery, radiation therapy, hormonal therapy, or others) including information on age at genital surgery. To assess the role of exogenous hormones women were asked about having ever used OC, age at first use, and duration of use in years. In addition, they were asked whether

they had ever used hormones other than OC as, for example, oestrogen replacement therapy in menopause. For positive response, information was recorded on age at first use, and duration of use in years, but not on type of hormones or dosage. The statistical analyses were based on multivariate techniques for case-control studies. Conditional logistic regression models were used to estimate odds ratios (OR) and 95% CI with strata for age and region and adjusting for educational level (10 years, 10 years of school attendance), cumulative tobacco smoke (log(packyear+1)), and years since stopping smoking (current smoker/quit 5 years ago or quit 5 years ago). As additional potential confounders for the association of endocrine factors and lung cancer risk we quantified variables for occupational exposure to known or suspected lung carcinogens as described previously,28 exposure to environmental tobacco smoke in terms of the summary indicator of ETS-exposure (low, medium, or high) for non-smokers,29 and time-weighted average radon exposure in the current residence as continuous variable.30 Since risk estimates were little affected by these confounder variables we present OR adjusted for age, region, smoking, and education only. Trend tests were performed using the continuous variable in the regression model and presenting the corresponding twosided P-value. For categorization of variables tertiles of the distribution among controls was used. Subgroup analyses were performed after stratification for smokers and non-smokers and by histological type using unconditional logistic regression models. Potential differences in OC and hormone replacement therapy (HRT) use or other factors between women from East and West Germany were considered by repeating all analyses separately for each study area and discussing differences if present.

Results The present analysis included 811 female cases and 912 female controls. Demographic parameters and the distribution of some potential confounders were given in Table 1. The mean age was 60 years among the cases and 59 years among the controls. The educational level tended to be higher among controls as compared with cases. A substantially higher proportion of female controls reported to have never smoked (65%) than among cases (31%). Among smokers more cases were current smokers, smoking more cigarettes per day for a longer duration, than controls. The main histological type was adenocarcinoma (48%), followed by small cell lung cancer (26%), squamous cell carcinoma (21%), and other tumour types (5%).

Menstrual and reproductive variables Table 2 shows the OR for menstrual and reproductive factors. No significant differences between cases and controls in terms of age at first menarche, length of menstrual cycle, and number of full-term pregnancies were observed. Lung cancer risk significantly decreased with increasing age at first pregnancy. To see whether women without pregnancy behave like women with late age at pregnancy, as is known for breast cancer, we additionally included women without pregnancy in the model. The corresponding risk estimate, however, was similar to that of women with young age at pregnancy. The ratio between postmenopausal and premenopausal women in the case group was higher than that in the control group. Among women who

HORMONAL FACTORS AND RISK OF LUNG CANCER

Table 1 Characteristics of cases and controls, Germany, 1990–1996 Cases

Controls

No.

%

No.

%

Totals

811

100.0

912

100.0

Age (years) 50 50–54 55–59 60–64 65–69 70–74

120 110 147 148 153 133

14.8 13.6 18.1 18.3 18.9 16.4

132 142 162 153 162 161

14.5 15.6 17.8 16.8 17.8 17.8

Study area West Germany East Germany

565 246

69.7 30.3

634 278

69.5 30.5

Years of school attendance 9 9 10–11 12 Unknown

17 635 111 47 1

2.1 78.3 13.7 5.8 0.1

13 584 192 118 5

1.4 64.0 21.1 12.9 0.5

354 378

43.7 46.6

294 453

32.2 49.7

78 1

9.6 0.1

161 4

17.6 0.4

Smoking status Non-smokers Ex-smokers Current smokers

248 87 476

30.6 10.7 58.7

593 147 172

65.0 16.1 18.9

Cigarette pack-years Non-smokers 20 20–40 40+

248 210 237 116

30.6 25.9 29.2 14.3

593 237 67 15

65.0 26.0 7.4 1.6

Histological type Small cell lung cancer Squamous cell carcinoma Adenocarcinoma Other carcinoma

207 174 386 44

25.5 21.5 47.6 5.4

– – – –

– – – –

Highest professional degree Without any formal vocational training Apprenticeship College/polytechnic/ university Unknown

265

(OR = 1.59; 95% CI: 0.94–2.69). No clear association of lung cancer with either age at natural menopause or age at both natural and surgical menopause was observed in each of the subgroups.

Exogenous hormones

experienced natural menopause, no trend in risk with increasing age at menopause was observed. The same holds true when age at both natural and surgical menopause was considered (OR = 1.0, 1.1, 0.9, 1.0 for 45, 45–49, 50–52, 53 years of age at menopause, respectively). Mean age at natural menopause was 49.1 years among the cases and 50.4 years among the controls. When analyses were stratified by smoking status and histology (Table 3), no association of age at menarche, menstrual cycle length, or number of children with lung cancer was observed for smokers and non-smokers and for each of the cell subtypes. A significant decrease in risk of lung cancer with increasing age at first pregnancy, however, was present for all histological cell subtypes and for smokers, but not for nonsmokers. The proportion of premenopausal women was higher among smoking controls (26%) than among non-smoking controls (17%), partly due to younger ages at interview among smokers (mean age: 58 years among cases and 57 years among controls) compared with non-smokers (62 years and 61 years, respectively). Risk of surgical menopause was slightly higher in non-smokers (OR = 2.35; 95% CI: 1.23–4.49) than in smokers

A total of 35% of the cases and 39% of the controls reported having ever used OC (Table 4), demonstrating a significantly reduced lung cancer risk (OR = 0.69; 95% CI: 0.51–0.92), but no trend in risk with increasing duration of use was observed (P-value for trend: 0.13). Mean age at first use was on average 28.6 years among the cases and 29.0 years among the controls; mean duration of use 9.8 years and 9.3 years, respectively. Since the dosage and composition of hormones in OC dramatically changed over the years, with much higher levels of oestrogens in the 1960s as compared with the 1970s and after 1980,31 we considered calendar year of first use of OC in three categories. Most of the women started to use OC in the 1960s and 1970s and only a small proportion after 1980. There were no major differences in lung cancer risk between these groups. A higher proportion of controls (30%) than cases (25%) reported the use of hormones other than OC such as oestrogen replacement therapy, yielding a slightly decreased lung cancer risk (OR = 0.83; 95% CI: 0.64–1.09) after controlling for menopausal status. Mean age at first use was 48.9 years among cases and 47.8 years among controls; mean duration of use 4.9 years and 5.9 years, respectively. Long-term users (7 years) compared with non-users showed a significantly decreased lung cancer risk (OR = 0.59; 95% CI: 0.37–0.93), yet no trend in risk associated with duration of use was present (P-value for trend: 0.19). Women who were treated with HRT for the first time in the 1960s and 1970s, showed a reduction in lung cancer risk while those who started with HRT after 1980 showed no decrease or increase in risk in comparison with non-users. In Table 5, risk estimates concerning use of hormones stratified for smokers and non-smokers and for histological cell type are given. Substantially more women used OC among smokers (55% of the controls and 38% of the cases) than among non-smokers (31% and 27%, respectively). This may partly be due to the fact that smoking women belong to a younger generation than non-smoking women. Smoking seems to modify lung cancer risk associated with the use of OC. Among smokers, the OR was significantly reduced for users compared with lifetime non-users (OR = 0.50; 95% CI: 0.34–0.74); whereas no effect was present among non-smoking women (OR = 1.18; 95% CI: 0.78–1.79). For all major histological cell types a reduction in risk was observed, which was most pronounced for small cell lung cancer. In all groups, however, no decrease in lung cancer risk with increasing duration of use was apparent. In the group of smoking women, lung cancer risk by calendar year of first OC use was associated with similar risk estimates at each category, showing OR of 0.47, 0.51, and 0.47 for women who started to use OC in the 1960s, 1970s, and 1980s, respectively. The corresponding OR in the group of non-smoking women were 1.41, 0.88 and 1.63, respectively. A positive history of HRT was associated with a slightly reduced lung cancer risk that did not differ much between the various histological cell types (Table 5). Risk was slightly reduced in smokers (OR = 0.73; 95% CI: 0.51–1.03), while no clear effect

266

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Table 2 Odds ratios (OR) for lung cancer associated with menstrual and reproductive factors, Germany, 1990–1996 Cases

Controls

No.

No.

ORa

95% CI

P-value for trend

Age at first menarche (years) 13 13–14 15+ Missing values

195 351 250 16

211 433 256 12

1.00 1.05 1.30

Ref. 0.79–1.40 0.95–1.79

0.30

Length of menstrual cycle (days) 28 28 29+ Missing values

118 540 54 99

103 641 70 98

1.00 0.64 0.81

Ref. 0.45–0.89 0.48–1.38

0.10

Age at first pregnancy (years) 22 22–25 26+ Missing values

286 238 170 4

223 284 287 3

1.00 0.88 0.67

Ref. 0.65–1.18 0.49–0.93

0.007

No. of full-term pregnancies 0 1 2–3 4+ Missing values

136 220 242 211 2

133 228 325 226 –

1.00 0.79 0.84 0.86

Ref. 0.55–1.14 0.59–1.19 0.60–1.24

0.85

Menopausal status Pre-menopausal Natural menopause Surgical menopause Other menopause (radiation, etc.) Missing values

105 420 218 33 35

184 468 212 29 19

1.00 1.88 1.99 2.32

Ref. 1.20–2.92 1.28–3.10 1.16–4.65

Age at natural menopause (years) 49 50–52 53+ Missing values

185 135 81 19

134 196 122 16

1.00 0.67 0.96

Ref. 0.45–0.98 0.62–1.50

0.81

a Odds ratio adjusted for age, region, log(packyear+1), time since smoking cessation (5 years, 5+ years), and educational level (10 years, 10).

was seen in non-smokers (OR = 1.05; 95% CI: 0.74–1.49). Long-term users (7 years) of hormones compared with nonusers, however, showed consistently over all sub-groups a lowered lung cancer risk. The trend tests for duration of use in years, however, did not achieve statistical significance. Lung cancer risk by calendar year of first HRT was 0.62, 0.45, 0.83 for the years before 1970, 1970–1979, and after 1979 among smokers and 0.85, 0.65, and 1.34 for non-smokers, respectively. Due to the lack of information on composition of HRT, we analysed separately the group of women with surgical menopause, who are likely to have had hysterectomy and consequently their HRT therapy may have consisted of oestrogens only (Table 6). A positive history of HRT was associated with a reduced lung cancer risk of OR = 0.8 for all women and among both smoking and non-smoking women. Again long-term users (7 years) showed consistently lowered risk estimates over all sub-groups.

Discussion The aim of the present study was to examine the role of endogenous and exogenous hormonal factors in the genesis of lung cancer among women. In contrast to tumours of the female reproductive system, hormonal factors showed either no association with lung cancer risk or an inverse association. We

found no strong support for an association of age at first menarche, length of menstrual cycle, and number of children with lung cancer. The lack of such an association in the present study is in agreement with other lung cancer studies.18,21–23 Up to now only three case-control studies in Chinese females provided evidence that shorter menstrual cycle length increases the risk of adenocarcinoma.18–20 Our data suggest a significant decrease in lung cancer risk with increasing age at first pregnancy. Other studies examining this factor did not observe an association with lung cancer.18,21,23 Since early age at first pregnancy is related to lower social class and our cases tended to have lower levels of education than controls, our finding could be due to residual confounding through social class. When limiting analyses to highly educated women having a professional degree such as college, polytechnic or university, however, lung cancer risk remained reduced (OR = 0.55; 95% CI: 0.21–1.43 among women with age at first pregnancy 26 years compared with those aged 22 years). Overall we found no clear association between age at natural (or both natural and surgical) menopause and lung cancer risk, and only weak evidence for earlier ages at natural menopause among smoking cases compared with smoking controls. A significant reduction in risk associated with late age at natural and surgical menopause was reported by Wu et al.,23 while Taioli and Wynder21 noted a significant reduction in lung

HORMONAL FACTORS AND RISK OF LUNG CANCER

267

Table 3 Odds ratios (OR) for lung cancer associated with menstrual and reproductive factors by smoking status and histology ORa

95% CI

ORa

95% CI

Age at first menarche (years) Non-smokers Smokers Adenocarcinoma Small cell lung cancer Squamous cell carcinoma

13 1.00 1.00 1.00 1.00 1.00

0.83 1.22 1.08 0.99 1.09

13–14 0.55–1.24 0.84–1.78 0.77–1.50 0.64–1.55 0.68–1.77

1.15 1.21 1.27 0.98 1.40

15+ 0.75–1.77 0.80–1.83 0.89–1.82 0.60–1.62 0.84–2.34

P = 0.69 P = 0.70 P = 0.51 P = 0.52 P = 0.53

Length of menstrual cycle (days) Non-smokers Smokers Adenocarcinoma Small cell lung cancer Squamous cell carcinoma

28 1.00 1.00 1.00 1.00 1.00

0.68 0.66 0.84 0.56 0.48

28 0.43–1.07 0.41–1.06 0.56–1.24 0.33–0.94 0.28–0.82

0.79 0.78 0.93 0.46 0.95

29+ 0.40–1.57 0.37–1.64 0.51–1.69 0.18–1.16 0.44–2.09

P = 0.10 P = 0.54 P = 0.29 P = 0.13 P = 0.73

Age at first pregnancy (years) Non-smokers Smokers Adenocarcinoma Small cell lung cancer Squamous cell carcinoma

22 1.00 1.00 1.00 1.00 1.00

1.22 0.62 0.90 0.73 0.69

22–25 0.81–1.82 0.41–0.93 0.64–1.27 0.46–1.15 0.43–1.11

0.96 0.50 0.75 0.48 0.53

26+ 0.63–1.48 0.32–0.78 0.52–1.09 0.28–0.80 0.31–0.91

P = 0.20 P = 0.005 P = 0.04 P = 0.02 P = 0.02

Never 1.00 1.00 1.00 1.00 1.00

0.72 1.05 0.72 1.19 1.34

1 child 0.43–1.19 0.65–1.68 0.48–1.08 0.67–2.10 0.73–2.47

No. of full term pregnancies Non-smokers Smokers Adenocarcinoma Small cell lung cancer Squamous cell carcinoma Menopausal status Non-smokers Smokers Adenocarcinoma Small cell lung cancer Squamous cell carcinoma

Premeno-pausal 1.00 1.00 1.00 1.00 1.00

Age at natural menopause (years) Non-smokers Smokers Adenocarcinoma Small cell lung cancer Squamous cell carcinoma

49 1.00 1.00 1.00 1.00 1.00

Natural menopause 1.91 1.00–3.65 1.55 0.91–2.61 1.53 0.96–2.45 2.45 1.26–4.78 1.50 0.73–3.10 0.76 0.61 0.76 0.49 0.78

50–52 0.46–1.26 0.37–1.02 0.50–1.15 0.27–0.87 0.42–1.45

ORa

P-value for trend

Referent

95% CI

2–3 children 0.82 0.51–1.32 0.98 0.61–1.56 0.76 0.52–1.11 1.05 0.59–1.87 1.19 0.66–2.17

4 children 0.90 0.55–1.47 0.91 0.56–1.49 0.80 0.53–1.20 1.31 0.74–2.32 1.13 0.61–2.10

Surgical 2.35 1.59 1.71 2.03 1.86

Other menopause 2.35 0.89–6.22 2.48 0.96–6.45 2.04 0.94–4.43 3.59 1.25–10.34 1.91 0.60–6.07

0.98 0.63 0.80 0.90 0.92

menopause 1.23–4.49 0.94–2.69 1.06–2.74 1.03–4.02 0.90–3.85 53+ 0.58–1.66 0.33–1.23 0.49–1.32 0.47–1.73 0.45–1.91

P = 0.75 P = 0.94 P = 0.80 P = 0.24 P = 0.51

P = 0.81 P = 0.27 P = 0.32 P = 0.71 P = 0.86

a Odds ratio adjusted for age, region, smoking (current smokers or quitted smoking 5 years ago), log(packyear+1), and educational level.

cancer risk with early age at natural and surgical menopause (40 years). No relation was found in a case-control study among Chinese women.22 An anti-oestrogenic effect of smoking has been reported, leading to a 1- to 2-year earlier occurrence of natural menopause among smoking women compared with non-smoking women.32,33 This was confirmed in our study, showing a mean age at natural menopause of 50.7 years among non-smoking controls, and 49.8 years and 49.0 years among smokers of 20 pack-years and 20 pack-years, respectively. In the present study among postmenopausal women, type of menopause was associated with an 1.13-fold non-significantly increased lung cancer risk among women with surgical menopause (not shown). No information was available on type of surgery such as removal of ovaries. Two previous reports found an increase in lung cancer risk after hysterectomy,23,34 being highest among women whose ovaries remained intact.23 A hysterectomy had been suspected as risk factor for lung cancer, because of the potential of traumaassociated lung embolism.23 Few epidemiological studies examined the role of the use of OC in the aetiology of lung cancer. Wu-Williams et al.22 noted an OR of 0.8 (95% CI: 0.5–1.2) for lung cancer among women

who had used OC as compared with non-users. This finding is consistent with that reported by Taioli et al.,21 (OR = 0.8; 95% CI: 0.5–1.5). A significant trend of reduction in risk of adenocarcinoma with increasing duration of use was noted by Wu et al.,23 although based on short-term users only (averaging 45 months among cases and 61 months in controls). All these studies controlled for smoking and made no distinction in risk between smokers and non-smokers. Our overall smoking adjusted risk estimate is in accordance with previous studies (OR = 0.7; 95% CI: 0.5–0.9), yet we found a significant interaction between smoking status and risk of lung cancer associated with the use of OC, showing a reduction in risk by 50% in smokers only. It is unclear whether this finding is spurious, because no dose–response relationship regarding duration of use was present. Also, women who started to use OC in the 1960s, the years with the highest level of oestrogens in OC, showed a similar lung cancer risk than those who started later when oestrogen levels were drastically reduced. It cannot be excluded that part of the observed reduced risk among smokers may be explained by residual confounding through social class. Differences between East and West German women were not present, neither with respect to the proportion of OC

268

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Table 4 Odds ratios (OR) for lung cancer associated with the use of exogenous hormones, Germany, 1990–1996 Cases

Controls

No.

No.

ORa

95% CI

Use of oral contraceptives Never Ever Missing values

528 279 4

557 354 1

1.00 0.69

Ref. 0.51–0.92

Age at first use (years) Non-users 25 25–31 32+ Missing values

528 91 96 88 8

557 110 115 121 9

1.00 0.60 0.74 0.69

Ref. 0.35–1.02 0.49–1.12 0.48–1.00

Duration of use (years) Non-users 5 5–11 12+ Missing values

528 86 87 102 8

557 105 130 115 5

1.00 0.69 0.65 0.69

Ref. 0.46–1.03 0.44–0.95 0.47–1.02

Calendar year of first use Non-users Before 1970 1970–1979 1980+ Missing values

528 141 116 18 8

557 159 165 22 9

1.00 0.77 0.60 0.77

Ref. 0.55–1.08 0.41–0.87 0.35–1.71

Use of hormonesb,c Never Ever Missing values

604 196 11

632 274 6

1.00 0.83

Ref. 0.64–1.09

Age at first use (years) Non-users 47 47–51 52+ Missing values

604 62 58 70 17

632 89 91 91 9

1.00 0.63 0.81 1.04

Ref. 0.42–0.96 0.53–1.25 0.70–1.55

Duration of use (years) Non-users 3 3–6 7+ Missing values

604 86 58 43 19

632 108 74 87 10

1.00 0.89 0.96 0.59

Ref. 0.62–1.29 0.62–1.48 0.37–0.93

Calendar year of first use Non-users Before 1970 1970–1979 1980+ Missing values

604 28 29 133 17

632 32 61 178 9

1.00 0.66 0.52 0.99

Ref. 0.35–1.23 0.30–0.88 0.72–1.36

a Odds ratio adjusted for age, region, log(packyear+1), time since smoking

cessation, education.

b Use of hormones other than oral contraceptives including oestrogen

replacement therapy. ratios additionally postmenopausal).

c Odds

adjusted

for

menopausal

status

(pre/

users, duration of use, nor the risk estimates. The use of OC was inversely associated with age at first pregnancy among both smokers and non-smokers. Among smokers, at each level of age at first pregnancy a reduced lung cancer risk with the use of OC was found, indicating that OC use seems to be an independent risk factor. Inconsistent findings were available with respect to lung cancer risk associated with HRT. In a large population-based cohort of women who had received HRT,25 a 30% excess risk

for lung cancer was observed (95% CI: 0.9–1.7), which, however, could also have been due to a higher prevalence of smokers in the cohort than in the background population. Two previous case-control studies found no association between HRT and lung cancer.18,23 Ettinger et al.26 reported a lower risk of death from lung cancer in a female cohort of long-term postmenopausal oestrogen replacement hormone users compared with age-matched postmenopausal non-users (relative risk = 0.22; 95% CI: 0.04–1.15, adjusted for current smoking). Our results indicate a lowered lung cancer risk with HRT, particularly among long-term users and independent of cell type. It is difficult to evaluate this effect without data on dosage and composition of hormones. When we limited the analyses to women with surgical menopause, who are most likely to have received oestrogens only, we found a reduction in risk by 50% for longterm users over all sub-groups. The reduction in risk was more pronounced among women starting in the years before 1980 than among those starting later, where probably levels of oestrogens were reduced. It should be noted that use of HRT differed between East and West German women. East German women were treated much less with HRT (16% among controls) than West German women (37% among controls) and for a shorter time period. Lung cancer risk associated with a positive history of HRT differed slightly between East German (OR = 1.10; 95% CI: 0.61–1.96) and West German women (OR = 0.80; 95% CI: 0.59–1.08). A 70% significant excess risk of adenocarcinoma (95% CI: 1.0–2.8) for HRT users versus non-users including a nonsignificant increase in lung cancer risk with increasing duration of use was demonstrated by Taioli et al.21 Interestingly, the increase in risk was primarily seen among smokers, while no effect was present among non-smokers.21 Some studies which provided evidence for a higher susceptibility of lung cancer in women than men given a similar level of tobacco smoke3 hypothesized that this observed positive synergism between HRT and smoking might be some possible explanation. Our results contradict a positive synergism of HRT with smoking. Several factors may contribute to these differences. First our study subjects were more likely to report long-term use than those of Taioli;21 second, composition and dosage of hormones may differ in both studies (unfortunately both studies did not collect this information); third, differences in smoking habits or residual confounding due to social class may produce different risk estimates. There is increasing evidence that oestrogens may play a role in the genesis of lung cancer, yet the mechanism is unclear. Experimental studies suggest the presence of oestrogen and other steroid receptors in lung tissue.5–8,35,36 Steroids are known to be involved in the differentiation and metabolism of pulmonary epithelial cells. The presence of these receptors has been suggested as indicative of their common role in the regulation of cancer growth in various organs36 and of potential hormonal responsiveness of these tumours.20 Since adenocarcinomas showed a significantly higher incidence of steroid receptors compared with other lung cancer cell types,4 some investigators hypothesized that the greater proportion of adenocarcinoma among female compared with male cases1 may be in part due to endocrine-related factors. In addition, a greater proportion of oestrogen receptors has been detected in females than in males.37 In the present study, however, no major

HORMONAL FACTORS AND RISK OF LUNG CANCER

269

Table 5 Odds ratios (OR) for lung cancer associated with the use of exogenous hormones by smoking status and histology Referent

ORa

95% CI

Use of oral contraceptives Non-smokers Smokers Adenocarcinoma Small cell lung cancer Squamous cell carcinoma

Never 1.00 1.00 1.00 1.00 1.00

1.18 0.50 0.80 0.49 0.61

Ever 0.78–1.79 0.34–0.74 0.57–1.11 0.31–0.77 0.38–0.99

Use of other hormonesb Non-smokers Smokers Adenocarcinoma Small cell lung cancer Squamous cell carcinoma

Never 1.00 1.00 1.00 1.00 1.00

1.05 0.73 0.89 0.71 0.74

Ever 0.74–1.49 0.51–1.03 0.66–1.19 0.46–1.09 0.51–1.21

95% CI

ORa

95% CI

P for linear trend

5 years 1.24 0.70–2.20 0.44 0.26–0.73 0.92 0.59–1.42 0.31 0.15–0.63 0.57 0.29–1.13

5–11 years 1.22 0.71–2.09 0.41 0.24–0.69 0.65 0.41–1.03 0.59 0.32–1.09 0.51 0.26–1.02

0.99 0.60 0.81 0.56 0.58

12+ years 0.52–1.89 0.37–0.96 0.52–1.26 0.31–1.02 0.30–1.10

P = 0.87 P = 0.40 P = 0.42 P = 0.18 P = 0.31

3 years 1.40 0.87–2.26 0.67 0.41–1.08 1.06 0.71–1.57 0.79 0.44–1.44 0.66 0.35–1.26

3–6 years 0.98 0.55–1.75 1.11 0.60–2.04 0.95 0.57–1.59 1.12 0.56–2.25 1.11 0.55–2.24

0.71 0.48 0.58 0.30 0.60

7+ years 0.37–1.36 0.27–0.86 0.34–0.98 0.12–0.73 0.29–1.23

P = 0.60 P = 0.053 P = 0.45 P = 0.15 P = 0.21

ORa

95% CI

ORa

a Odds ratio adjusted for age, region, smoking (current smokers or quitted smoking 5 years ago), log(packyear+1), and educational level. b OR additionally adjusted for menopausal status (pre/postmenopausal).

Table 6 Lung cancer risk associated with hormone replacement therapy among women with surgical menopause, Germany, 1990–1996

All subjects Never Ever Duration of use (years) 3 3–6 7+ Smokers Never Ever Duration of use (years) 3 3–6 7+ Non-smokers Never Ever Duration of use (years) 3 3–6 7+

Cases

Controls

ORa

95% CI

144 72

123 88

1.00 0.78

Ref. 0.49–1.21

33 17 19

25 25 37

1.27 0.65 0.47

0.67–2.40 0.30–1.37 0.23–0.93

103 50

45 34

1.00 0.78

0.41–1.49

19 14 14

11 6 17

0.82 1.32 0.46

0.37–2.03 0.40–4.30 0.18–1.15

41 22

78 54

1.00 0.80

0.42–1.54

14 3 5

14 19 20

1.91 0.30 0.52

0.80–4.56 0.08–1.10 0.18–1.53

a Odds ratio adjusted for age, region, smoking (current smokers or quitted

smoking 5 years ago), log(packyear+1), and educational level.

differences in risk by histologic type associated with the various hormonal factors were observed. One of the main strengths of this study is the large sample size of lung cancer cases and controls that made the overall risk estimates relatively stable and allowed stratification by smoking status and histology. All interviews were conducted with the subjects themselves in face-to-face interviews, in contrast to studies in which surrogates were used. All cases were histologically or cytologically verified as tumours that originated in the lung. In addition, an independent review of about 75% of the diagnosed material was completed, in order to classify the cell type uniformly. Moreover, we controlled for most potential confounders. Nevertheless, there are several limitations in our study. Some potential methodological problems concern the low response rate among controls and recruitment of cases. In Germany, cancer

patients had to be recruited via study hospitals, since no overall cancer registry existed, thus the extent of representativeness of our cases is not measurable. In previous analyses based on the former East German cancer registry we estimated a coverage of about 50% for the eastern study area. A selection bias due to the poor response among controls cannot be completely ruled out. In a sub-sample of refusals, a non-response-analysis was conducted. Non-response was mainly due to refusal of longterm measurements of radon (1 year) required in the subjects’ homes (38%), no time for interview and organization of the measurement (13%), followed by illness (13%) and other reasons. As the controls tended to be more highly educated than the cases and there may be a positive correlation between low social class and some of the hormonal factors such as young age at first pregnancy and less frequent use of hormones, special attention has been given to a potential selection bias by controlling for educational level. However, uncontrolled residual confounding by differential socioeconomic status and life style cannot be excluded and could have resulted in somewhat lowered risk estimates. Another potential weakness of our study is the lack of information on type and dosage of hormones. Since this study was primarily conducted to investigate the effects of residential radon on lung cancer risk, no detailed information on hormones was collected. Some of the observed differences in risk between smokers and nonsmokers might be explained by differences in type and dosage of hormones.

Conclusion Our study provides evidence that some hormonal factors, such as use of OC or other hormones, may reduce lung cancer risk to some extent and that the reduction is more pronounced among smokers. Yet, no clear dose–response relationship with duration of use was apparent. The previously reported positive synergism between smoking and HRT could not be confirmed in our study. Risk of lung cancer associated with the various hormonal factors did not differ much between histological cell subtypes.

Acknowledgements The authors thank the following people for their contribution to the study: Gabi Wölke in Erfurt, Gerlinde Dingerkus in

270

INTERNATIONAL JOURNAL OF EPIDEMIOLOGY

Wuppertal, Dr Jürgen Wellmann in Münster, Prof. KlausMichael Müller in Bochum, Angelika Schaffrath-Rosario; Angelika Steinwachs, Birgit Proksch in Neuherberg and Drs Jenny Change-Claude in Heidelberg and Judith Müller in

Neuherberg for helpful comments to earlier versions of the paper. Financial support was provided by the Federal Office for Radiation Protection, Salzgitter, Germany (grant StSch 1066, 4074, 4074/1, 4006, 4112).

KEY MESSAGES •

Use of oral contraceptives is associated with a 50% reduction in lung cancer risk among smoking women, yet no trend with duration of use is apparent.



Long-term hormone replacement therapy (7 years) reduces the risk of lung cancer among both smoking and non-smoking women.



Other studies need to confirm these results since the present study had no information on dosage and composition of hormones.

References 1 Lubin JH, Blot WJ. Assessment of lung cancer risk factors by histologic

category. J Natl Cancer Inst 1984;73:383–89. 2 Risch HA, Howe GR, Jain M et al. Are female smokers at higher risk

for lung cancer than male smokers? A case-control study by histologic type. Am J Epidemiol 1993;138:281–93. 3 Zang EA, Wynder EL. Differences in lung cancer risk between men

and women: examination of the evidence. J Natl Cancer Inst 1996;88:183–92. 4 Siegfried JM. Women and lung cancer: does oestrogen play a role?

Lancet Oncol 2001;2:506–13. 5 Chaudhuri PK, Thomas PA, Walker MJ et al. Steroid receptors in

human lung cancer cytosols. Cancer Lett 1982;16:327–32. 6 Beattie CW, Hansen NW, Thomas PA. Steroid receptors in human lung

cancer. Cancer Res 1985;45:4206–14. 7 Kobayashi S, Mizuno T, Tobioka PA et al. Sex steroid receptors in

diverse human tumors. Gann 1982;73:439–45. 8 Cagle PT, Mody DR, Schwartz MR. Estrogen and progesterone

receptors in bronchogenic carcinoma. Cancer Res 1990;50:6632–35.

19 Liao ML, Wang JH, Wang HM et al. A study of the association between

squamous cell carcinoma and adenocarcinoma in the lung, and history of menstruation in Shanghai women, China. Lung Cancer 1996;14(Suppl.1):S215–21. 20 Seow A, Poh WT, Teh M et al. Diet, reproductive factors and

lung cancer risk among Chinese women in Singapore: evidence for a protective effect of soy in nonsmokers. Int J Cancer 2002;97: 365–71. 21 Taioli E, Wynder EL. Endocrine factors and adenocarcinoma of the

lung in women. J Natl Cancer Inst 1994;86:869–70. 22 Wu-Williams AH, Dai XD, Blot W et al. Lung cancer among women in

north-east China. Br J Cancer 1990;62:982–87. 23 Wu AH, Yu MC, Thomas DC et al. Personal and family history of lung

disease as risk factors for adenocarcinoma of the lung. Cancer Res 1988;48:7279–84. 24 La Vecchia C, Brinton LA, McTiernan A. Menopause, hormone

replacement therapy and cancer. Maturitas 2001;39:97–115. 25 Adami HO, Persson I, Hoover R et al. Risk of cancer in women

9 Sellers TA, Potter JD, Folsom AR. Association of incident lung cancer

receiving hormone replacement therapy. Int J Cancer 1989;44: 833–39.

with family history of female reproductive cancers: the IOWA Women’s Health Study. Genet Epidemiol 1991;8:199–208.

26 Ettinger B, Friedman GD, Bush T et al. Reduced mortality associated

10 Schwartz AG, Siegfried JM, Weiss L. Familial aggregation of breast

with long-term postmenopausal estrogen therapy. Obstet Gynecol 1996;87:6–12.

cancer with early onset lung cancer. Genet Epidemiol 1999;17:274–84.

27 Kreuzer M, Kreienbrock L, Gerken M et al. Risk factors for lung cancer

11 Kabat GC. Aspects of the epidemiology of lung cancer in smokers and

nonsmokers in the United States. Lung Cancer 1996;15:1–20. 12 Curtis RE, Hoover RN, Kleinerman RA et al. Second cancer following

cancer of female genital system in Connecticut, 1935–82. Natl Cancer Inst Monogr 1985;68:113–37. 13 Harvey EB, Brinton LA. Second cancer following cancer of the breast

in Connecticut, 1935–82. Natl Cancer Inst Monogr 1985;68:99–112.

in young adults. Am J Epidemiol 1998;147:1028–37. 28 Kreuzer M, Pohlabeln H, Ahrens W et al. Occupational risk factors for

lung cancer in young males. Scand J Work Environ Health 1999;25:423–30. 29 Kreuzer M, Krauss M, Kreienbrock L et al. Environmental tobacco

smoke and lung cancer: A case-control study in Germany. Am J Epidemiol 2000;151:241–50.

14 Ewertz M, Mouridsen HT. Second cancer following cancer of the female

30 Kreienbrock L, Kreuzer M, Gerken M et al. Case-control study on lung

breast in Denmark, 1943–80. Natl Cancer Inst Monogr 1985;68:325–29.

cancer and residential radon in West Germany. Am J Epidemiol 2001;153:42–52.

15 Storm HH, Ewertz M. Second cancer following cancer of the female

genital system in Denmark, 1943–80. Natl Cancer Inst Monogr 1985;68:331–40.

31 Keep PA. The birth and development of the pill; a short history of oral

16 Kabat GC. Previous cancer and radiotherapy as risk factors for lung

32 Baron JA, La Vecchia C, Levi F. The antiestrogenic effect of cigarette

cancer in lifetime nonsmokers. Cancer Causes Control 1993;4:489–95. 17 Frisch M, Melbye M. Risk of lung cancer in pre- and post-menopausal

women with ano-genital malignancies. Int J Cancer 1995;62:508–11. 18 Gao YT, Blot WJ, Zheng W et al. Lung cancer among Chinese women.

Int J Cancer 1987;40:604–09.

contraception. Singapore I Obstet Gynaecol 1984;15:1–5. smoking in women. Am J Obstet Gynecol 1990;162:502–14. 33 Midgette AS, Baron JA. Cigarette smoking and the risk of natural

menopause. Epidemiology 1990;1:474–80. 34 Wu AH, Henderson BE, Pike MC et al. Smoking and other risk factors

for lung cancer in women. J Natl Cancer Inst 1985;74:747–51.

HORMONAL FACTORS AND RISK OF LUNG CANCER

271

35 Su JM, Hsu HK, Chang H et al. Expression of estrogen and

37 Kaiser U, Hofmann J, Schilli M et al. Steroid-hormone receptors in cell

progesterone receptors in non-small-cell lung cancer: immunohistochemical study. Anticancer Res 1996;16:3803–06.

lines and tumor biopsies of human lung cancer. Int J Cancer 1996;67:357–64.

36 Canver CC, Memoli VA, Vanderveer PL et al. Sex hormone receptors

in non-small-cell lung cancer in human beings. J Thorac Cardiovasc Surg 1994;108:153–57.