Risk factors for lung cancer among nonsmoking women

Abstract

To evaluate risk factors for lung cancer in nonsmoking women, we used data of a case-control study conducted between 1991 and 1996 in Germany. A total of 234 female histologically confirmed lung cancer patients and 535 population controls who had never smoked more than 400 cigarettes in their lifetime were personally interviewed with respect to occupation, exposure to environmental tobacco smoke (ETS), family history of cancer, prior physician-diagnosed lung diseases or cancer and diet. One-year radon measurements in the last dwelling were performed. Odds ratios (OR) adjusted for age and region and 95% confidence intervals (CI) were calculated via logistic regression. When cumulative duration of exposure to ETS in hours was considered, the OR for high compared to not or low ETS exposed women was 2.62 (CI:1.35–5.06) for occupational exposure and OR=1.67 (CI:0.86–3.25) for spousal exposure, exhibiting a significant trend for ETS at work. Working more than 10 years in jobs or industries with known or suspected lung carcinogens was associated with OR=2.0 (CI:0.99–4.0). An elevated risk due to prior lung diseases was present for pneumonia (OR=1.6; CI:1.07–2.40) and tuberculosis (OR=1.6; CI:0.77–3.37). No significant increase in risk with increasing residential radon levels or with the presence of a family history of lung cancer was apparent. Protective effects were observed for high vs. low consumption of fresh vegetables (OR=0.5; CI:0.25–0.82) and cheese (OR=0.3, CI:0.21–0.55). ETS at work, occupational hazards and previous pneumonia may be risk factors for lung cancer in nonsmoking women, while a diet rich in fresh vegetables and cheese seems to be protective. © 2002 Wiley-Liss, Inc.

Lung cancer is the third most common cancer death among European women, showing steadily increasing mortality rates since the 1970s.1 Tobacco smoke is one of the major causes, accounting for a range of 14% to 85% of the lung cancer deaths among women in the different European countries.2 Other risk factors, including environmental tobacco smoke,3, 4 residential radon,5, 6 occupational hazards,7 previous non-malignant lung diseases,8, 9 prior cancer,10 family history of cancer,11, 12, 13 diet14, 15 and hormonal factors,16 have been suggested to contribute to lung cancer occurrence among women, although to a much lower extent than tobacco smoke. Most of these factors are not nearly as well quantified as cigarette smoking. In order to examine the separate influence of these risk factors and to avoid residual confounding by smoking, data on non-smokers are advantageous.

Therefore, the International Agency for Research on Cancer (IARC) has coordinated a European multi-center case-control study of lung cancer in male and female nonsmokers that focused on the effects of ETS,4 occupation7 and diet.15 A subset of 134 female and 136 female controls from a German case-control study of lung cancer had been included in this study. Since then, information on an additional 100 female nonsmoking cases and 399 female controls of this then ongoing German study has become available. Together this large sample size of nonsmoking women allowed us to examine a wide range of potential risk factors for lung cancer in the German study population. Data on the small subgroup of male nonsmokers have been analyzed and published separately.17 The purpose of our article was to examine lung cancer risk in relation to exposure to environmental tobacco smoke, occupational carcinogens, residential radon, previous non-malignant lung disease, prior cancer, nutrition and family history of cancer among nonsmoking women. Special attention was given to possible differences in risk by cell type.

MATERIAL AND METHODS

The methods for this case-control study have been reported in detail elsewhere.6, 17, 18 Briefly, cases were residents of defined areas in East and West Germany, 32 to 75 years of age and diagnosed between February 7, 1991 and March 16, 1996 with an incident, primary lung carcinoma. Case subjects were identified from 15 study clinics in the defined study area. Inclusion criteria for eligible cases were: less than 76 years of age; histological or cytological confirmation of primary lung cancer; no metastases or relapses of lung cancer; an interview within 3 months after diagnosis; current residence in the study area; living in Germany for more than 25 years; and in an adequately good health state to face a personal interview of 1.3 hours duration. The response rate of eligible cases was 76%. Control subjects were randomly selected from the general population from mandatory registries or by random digit dialing and were frequency-matched to cases on sex, age and region. The response rate of eligible controls was 41%.

Both cases and controls were interviewed face to face by regularly trained interviewers at the bedside of the patients or at home for controls or if the patient had already been discharged. No data were obtained from next-of-kin or other surrogates. A standardized questionnaire was used to determine demographic parameters, detailed residential history, active and passive smoking history, occupational exposure, dietary habits, previous lung disease, prior cancer, hormonal factors and a family history of cancer. Overall, the study population consisted of 4,303 cases and 4,451 population controls. The present analyses were restricted to nonsmoking women, defined as having smoked less than 400 cigarettes in a lifetime. This cutoff point was chosen in accordance with previous publications.4, 17, 18

All women were asked if they had ever been told by a physician that they had prior non-malignant lung disease such as asthma, tuberculosis, emphysema, chronic bronchitis or pneumonia as well as prior cancer at any age at least 2 years before diagnosis of lung cancer (or date of interview for controls). For positive response of prior cancer, information on site of cancer and type of therapy was collected. Information on a family history of cancer among first degree relatives (parents, siblings and offspring) was gathered, including age at disease, site of cancer and relation to the subject. Subjects were defined as having cancer in the family if at least one relative with cancer was reported. This factor was defined for lung cancer and cancer of any site.

The questionnaire on environmental tobacco smoke (ETS) gathered information on ETS exposure at home, at work and at other public places. The corresponding results on lung cancer risk for nonsmoking women and men combined18 and men only17 have been published elsewhere. In accordance with previous analyses4, 17, 18, 19 the cumulative duration of exposure in hours was used as variable for exposure to ETS and the 75th and 90th centile for all nonzero values of cases and controls from each source of exposure was calculated. Never exposed women or women with exposures below the 75th centile comprised the baseline in risk analysis; those with exposures above the 90th centile were defined as highly exposed women.

For each job held for at least 6 months, women were asked to describe their occupation, industry and duties carried out, as well as the beginning and the end of each job period. A list of jobs, branches and industries in which a risk for lung cancer has been confirmed (A-list) or is suspected (B-list) (see reference 20) was applied to the job title and industry codes. Women were defined as exposed in an A-list (B-list) job if they had worked for at least 6 months in an occupation on A-list (B-list). Cumulative duration of employment in years in jobs of list A or B was calculated and categorized into less than 10 years or 10 or more years.

Dietary habits were assessed by a brief food frequency questionnaire. Women were asked to estimate their usual frequency of consumption of 15 different food items that are considered as major contributors to the intake of beta-carotene, retinol and cholesterol in Germany. The average consumption during the past 20 years should be reflected. The frequency intakes of these items were defined in 6 categories: i) never, ii) less than monthly, iii) less than weekly, iv) weekly, v) several times per week and vi) daily. For the analyses, for each individual food item, 3 approximately equal-sized groups of intake based on the distribution of consumption among the controls were established.

One-year measurements of the radon concentrations in the living room and bedroom of the participants' last dwellings were performed by alpha-track detectors.6 Radon exposure was quantified as the occupancy weighted average of the radon concentrations in the living room and bedroom of the present home, weighted by the individual's time spent in both rooms. Only women with complete one-year radon measurements were included in the present analysis for radon.

Statistical Methods

Unconditional logistic regression models were used to calculate odds ratios and 95% confidence intervals.21 Since frequency matching for age and region was performed for smokers and nonsmokers combined, the restriction to nonsmokers has led to a slightly unbalanced distribution of age and region between cases and controls. In order to control for possible confounding, all odds ratios were adjusted for age and region. We examined the effects of each of the above-mentioned risk factors separately with mutual adjustment for the remaining risk factors and educational level (less than 10 years or 10 or more years of school attendance). Since no major confounding was found, results were presented adjusted for age and region only.

To determine the relationship between each of these factors and lung cancer risk, we performed linear trend tests in accordance with previous analyses within our study by treating duration of exposure to occupational hazards in years and residential radon exposure as continuous variable,6, 17 and using the median exposure in each category of the duration of exposure to ETS in hours4, 18 Potential differences in lifestyle and exposure to risk factors for lung cancer between subjects from East and West Germany were considered by repeating all analyses separately for each study area and discussing differences if present. To investigate differences with respect to cell subtype 2 groups were formed: i) adenocarcinoma and ii) non-adenocarcinoma including small cell carcinoma, squamous cell carcinoma, large cell carcinoma and others.

RESULTS

We included 234 nonsmoking female cases and 535 nonsmoking female controls in the analyses. Table I shows sociodemographic characteristics of the study population. The mean age at interview was 62 years among the cases and 60 years among the controls, respectively. Most women were married and had completed 9 years of school; yet controls tended to be slightly higher educated than cases. The main histological cell type was adenocarcinoma (59%), followed by squamous cell carcinoma (20%) and small cell lung cancer (14%) and others (7%).

Table I. Sociodemographic Characteristics of Nonsmoking Female Cases and Controls, Germany, 1991–1996

	Cases (n=234)		Controls (n=535)		Test for difference
	Number	%	Number	%	p-value
Age at interview in years
<50	9	3.9	56	10.5
50–54	25	10.7	84	15.7
55–59	59	25.3	99	18.5
60–64	36	15.5	87	16.3
65–69	51	21.9	96	17.9
70–74	53	22.8	113	21.1	p = 0.01
Current residence
East Germany	109	46.6	205	38.3
West Germany	125	53.4	330	61.7	p = 0.04
Marital status
Married	157	67.1	342	63.9
Widowed	54	23.1	113	21.1
Divorced	12	5.1	39	7.3
Never married	11	4.7	40	7.5
Unknown	0	0.0	1	0.2	p = 0.50
Years of school attendance
< 9 years	5	2.1	7	1.3
9 years	185	79.1	373	69.7
10–11 years	24	10.3	98	18.3
≥ 12 years	20	8.5	57	10.7	p = 0.21
Histologic type
Small cell lung cancer	32	13.7	—	—
Squamous cell carcinoma	46	19.7	—	—
Adenocarcinoma	139	59.4	—	—
Other carcinoma	16	6.8	—	—

Previous Lung Disease and Cancer

The cancer patients reported about the same frequency of any prior physician-diagnosed non-malignant lung disease (including tuberculosis, pneumonia, emphysema, asthma and chronic bronchitis) as the controls (30%), yielding a non-significantly increased lung cancer risk of OR=1.1 (CI: 0.79–1.56) (Table II). Women who had a prior diagnosis of pneumonia exhibited an increased lung cancer risk (OR=1.6; CI: 1.07–2.40); the same holds true when women with prior tuberculosis were considered (OR=1.6; CI: 0.77–3.37), although no statistical significance was achieved for the latter. No effect was present for prior emphysema, chronic bronchitis and asthma. After additional adjustment for education or passive smoking, risk estimates did not change substantially. A prior cancer was reported by 33 cases (14%) and 37 controls (7%), yielding a significantly increased lung cancer risk (OR=2.1; 95% 1.25–3.45), which was reduced to OR=1.6 (CI: 0.81–3.86) after adjustment for treatment of prior cancer with radiotherapy. The majority of reported previous cancers were cancers of the reproductive system, which were associated with a 1.7-fold (CI:0.8–3.7) increased lung cancer risk after accounting for radiotherapy. Among them, breast cancer was reported by 12 cases and 10 controls, showing a 2.2-fold risk after adjustment for radiotherapy. No major differences in lung cancer risk associated with prior lung disease or cancer were observed for women with adenocarcinoma as compared to non-adenocarcinoma.

Table II. Odds Ratios for Lung Cancer and Previous Non-malignant Lung Disease Or Cancer (Diagnosed By a Physician at Least 2 Years Before Interview) and a Family History of Cancer (First Degree Relatives), Germany, 1991–1996

	Positive/negative history		OR11	95% Cl	OR22	95% Cl
	Number of cases	Number of controls
1 OR1, Odds ratio adjusted for age and region, 95% confidence interval. 2 OR2, Odds ratio adjusted for age, region and treatment of prior cancer with radiotherapy, 95% confidence interval. 3 Reproductive cancers included breast and genital cancer.
Previous non-malignant lung disease
Tuberculosis	13/218	18/513	1.61	0.77–3.37
Asthma	12/220	30/505	0.98	0.49–1.98
Chronic bronchitis	19/213	64/465	0.69	0.40–1.19
Emphysema	4/228	7/524	1.13	0.32–4.03
Pneumonia	49/177	79/452	1.60	1.07–2.40
Any lung disease	71/163	154/381	1.11	0.79–1.56
Previous cancer
Cancer of any site	33/200	37/496	2.08	1.25–3.45	1.6	0.81–3.86
Reproductive cancer3	25/208	26/509	2.30	1.29–4.12	1.7	0.80–3.72
Genital cancer	13/220	16/519	1.91	0.90–4.08	1.4	0.57–3.57
Breast cancer	12/221	10/525	2.89	1.23–7.03	2.2	0.81–6.02
Family history of cancer
Cancer	88/146	205/330	1.02	0.74–1.40
Lung cancer	11/223	28/507	0.89	0.43–1.83
Lung cancer (age ≤ 45 years)	4/230	3/532	3.28	0.71–15.1

Family History of Cancer

Approximately one third of both cases and controls reported a history of cancer in their first degree relatives (Table II). A positive family history of cancer or lung cancer was not associated with an increased lung cancer risk. A family history of early age of onset of lung cancer (≤ 45 years) was rare but associated with a 3-fold increased lung cancer risk, although confidence intervals were very wide (OR=3.3; CI:0.71–15.1).

Environmental Tobacco Smoke

Table III presents risk estimates associated with exposure to ETS by spouse or at work. When duration of exposure to spousal ETS was considered, highly exposed women compared to women with no or low exposure showed a non-significantly increased risk (OR=1.67; CI:0.86–3.25). The lung cancer risk associated with ETS exposure at work demonstrated an increase in risk with increasing duration of exposure with OR=2.62 (CI: 1.35–5.06) for the highest category of exposure. The patterns were generally similar for adenocarcinoma and non-adenocarcinoma.

Table III. Odds Ratios for Lung Cancer and Duration of Exposure to Environmental Tobacco Smoke in Hours by Spouse or at the Workplace, Germany, 1991–1996

Duration of ETS exposure in hours by	Cases number	Controls number	OR1	95% CI
1 OR, Odds ratio adjusted for age and region, 95% confidence interval.
Spouse
Never or < 53000	163	442	1.00	Referent
53,000–76,000	20	44	1.25	0.70–2.22
76,000 +	17	25	1.67	0.86–3.25
Unknown	34	24	/	/
p for trend test				p = 0.083
At work
Never or < 20,500	194	465	1.00	Referent
20,500–40,000	15	47	0.82	0.45–1.52
40,000 +	20	19	2.62	1.35–5.06
Unknown	5	4	/	/
p for trend test				p = 0.017

Diet

After classifying the case and control groups into approximate thirds for the distribution of various food items (Table IV), a reduced risk for the middle and upper third compared to the bottom third of fruit consumption was observed that was not statistically significant nor was a trend present. A statistically significant increasing protective effect was apparent with increasing levels of intake of fresh vegetables (p for trend: 0.03). No protective effect was seen for high consumption of lettuce or raw carrots. The same held true for high intake of canned vegetables, cooked carrots and kale (data not shown). High intake of meat was associated with a non-significantly 1.6-fold increased lung cancer risk. No relationship was found with frequent consumption of sausages, fish or eggs. Women drinking daily at least one cup of milk showed a reduced lung cancer risk compared to those drinking milk less than monthly. A similar pattern of risk reduction was found for high consumption of other dairy products such as curd or yogurt. Notable was the protective effect of high consumption of cheese. Daily compared to weekly or less frequent consumption of cheese was associated with a reduction in lung cancer risk by a factor of 3 (OR=0.34; CI: 0.21–0.55). Odds ratios were not altered when educational level, occupational exposure or exposure to ETS was included into the regression model. Separate analyses for the 2 cell type groups produced no differences in risk estimates.

Table IV. Odds Ratios for Lung Cancer Associated with Consumption of Fruits and Vegetables or Meat, Fish and Dairy Products, Germany, 1991–19961

	Cases number	Controls number	OR2	95% CI	p for linear trend
1 Nev, never; lm, less than monthly; lw, less than weekly; w, weekly; sw, several time weekly. 2 OR, Odds ratio adjusted for age and region, 95% confidence interval.
Fruit
Nev/lm/lw/w	17	34	1.00	Referent
Sw	36	98	0.55	0.28–1.09
Daily	177	403	0.66	0.37–1.19	0.94
Fresh vegetables
Nev/lm/lw/w	22	43	1.00	Referent
Sw	147	326	0.57	0.33–0.99
Daily	60	165	0.45	0.25–0.82	0.03
Lettuce
Nev/lm/lw	88	204	1.00	Referent
W/sw	80	184	1.24	0.81–1.91
Daily	61	147	1.23	0.74–2.05	0.09
Raw carrots
Nev/lm	87	198	1.00	Referent
Lw/w	111	258	0.91	0.65–1.28
Sw/daily	31	75	0.91	0.55–1.48	0.32
Meat
Nev/lm/lw/w	16	63	1.00	Referent
Sw	142	311	1.57	0.90–2.75
Daily	72	157	1.61	0.90–2.89	0.44
Sausages
Nev/lm/lw/w	30	71	1.00	Referent
Sw	81	182	0.99	0.60–1.62
Daily	120	279	0.99	0.61–1.62	0.43
Liver
Nev	44	94	1.00	Referent
Lm	104	237	0.87	0.57–1.34
Lw/w/sw/daily	82	204	0.80	0.51–1.26	0.61
Fish
Nev/lm/lw	100	212	1.00	Referent
W	98	257	0.74	0.53–1.03
Sw/daily	30	65	0.86	0.52–1.42	0.33
Eggs
Nev/lm/lw	52	102	1.00	Referent
W	70	186	0.68	0.44–1.05
Sw/daily	107	245	0.69	0.46–1.05	0.22
Milk
Nev/lm	78	159	1.00	Referent
Lm/sm/w	81	160	0.98	0.67–1.45
Daily	72	215	0.65	0.44–0.95	0.16
Curd, yogurt
Nev/lm/lw/w	88	162	1.00	Referent
Sw	78	165	0.84	0.57–1.22
Daily	64	208	0.53	0.34–0.81	0.08
Cheese
Nev/lm/lw/w	40	65	1.00	Referent
Sw	105	203	0.62	0.39–0.99
Daily	82	267	0.34	0.21–0.55	< 0.001

Occupation

As shown in Table V, cases were more likely than the controls to report having worked in a job or industry that is associated with suspected or known lung carcinogens, yielding an odds ratio of 1.32 (CI: 0.78–2.23). Only a small proportion of women had been employed in jobs or industries associated with known lung carcinogens (so-called A-list), while most of them worked in jobs or industries with suspected elevated lung cancer risk (so-called B-list). Odds ratios tended to increase with increasing duration of employment, with women working 10 or more years in such jobs showing a 2-fold increased lung cancer risk (OR=1.99; CI:0.99–4.00), but the test for trend was not significant. This odds ratio was only slightly reduced (OR=1.88, CI:0.91–3.87) after accounting for educational level and ETS at work. Lung cancer risk associated with ever having worked in such jobs was non-significantly higher among women with adenocarcinoma (OR=1.56; CI:0.86–2.85) than among women with non-adenocarcinoma (OR=1.00, CI:0.45–2.21). Table V also presents the number of cases and controls according occupations and industries belonging to list A or B and gives risk estimates if at least 2 exposed cases were present. Mainly addressed jobs were laundry or dry cleaning workers, butcher and meat workers, carpenters and joiners and ceramic and pottery workers.

Table V. Odds Ratios for the Relation of Employment in Various Occupations and Industries Known (A-List) or Suspected (B-List) to Present an Excess Risk of Lung Cancer, Germany, 1991–1996

	Cases	Controls	OR2	95% Cl
1 Women working in specific occupations and industries were compared in their risk to all other women. 2 OR, Odds ratio adjusted for age and region, 95% confidence interval.
Working in a job or industry of list A or B
Never A- or B-list	209	492	1.00	Referent
Ever A- or B-list	25	43	1.32	0.78–2.23
Never A- or B-list	209	492	1.00	Referent
B, never A-list	21	32	1.51	0.84–2.71
A-list	4	11	0.77	0.29–2.50
Duration of exposure in years
Never A- or B-list	209	492	1.00	Referent
< 10 years	9	24	0.81	0.37–1.80
≥ 10 years	16	19	1.99	0.99–4.00
p value for test for trend				p = 0.09
Specific industries/occupations of A or B list1
A-list
Copper smelting, zinc smelters, cadmium alloy production, aluminum production etc.	2	8	0.55	0.12–2.66
Painters	2	3	1.19	0.20–7.22
B-list
Butcher and meat workers	3	6	1.14	0.27–4.73
Wood industry (carpenters, joiners)	6	1	13.35	1.59–112.2
Printing industry	3	2	4.60	0.73–28.95
Various occupations in rubber manufacture	2	5	0.83	0.16–4.34
Ceramic, pottery, glass workers	4	7	1.57	0.44–5.65
Motor vehicle manufacturing (welders etc.)	1	0	—	—
Laundry, dry cleaning	4	11	0.69	0.22–2.19

Residential Radon

One-year radon measurements could be completed for 58% of the patients and 84% of the controls. On average, the current residence had been occupied for 24 years. The radon concentrations did not differ much between cases and controls; with a median annual exposure of 45 Bq/m³ among the cases and 44 Bq/m³ among the controls. There was no significant trend in lung cancer risk with increasing radon levels (p=0.22).

DISCUSSION

Our study adds to the evidence that exposure to ETS by spouse or at work, working in certain occupations associated with potential lung carcinogens, prior pneumonia and tuberculosis and prior cancer treated with radiotherapy may be risk factors for lung cancer in lifetime nonsmoking women, while a high consumption of fresh vegetables seems to be protective. Residential radon exposure and a positive family history of lung cancer did not appear to be important risk factors in the present study population. As in several other studies among nonsmoking women, adenocarcinoma was the major histologic type, yet no clear relationship to specific risk factors was apparent.

One strength of our study is the relatively large number of lifelong nonsmoking lung cancer patients and controls. Detailed information on the main risk factors such as occupation, residential radon, ETS, previous lung disease, nutrition and family history of cancer was ascertained by closely supervised, trained interviewers in standardized face-to-face interviews. No surrogate interviews were performed. All cases had histologically or cytologically confirmed, primary lung tumors and histological subtypes were evaluated by one reference pathologist.

Nevertheless some limitations of our study should be considered before evaluating our findings. A possible source of bias was the low response rate among controls. As shown in an analysis of non-responders,6 the major reason for not participating in our study was the requirement to conduct, in addition to a 1.3 hr interview, a 1-year radon measurement in the living room and bedroom of the home. Generally, higher educated people tended to be more willing to participate. The effect of a potential selection bias cannot be ruled out completely. Additional adjustment for years of school attendance or the highest vocational training in the various models, however, did not lead to major changes in the risk estimates. Nevertheless it cannot be excluded that the low response among controls might have introduced bias. Another concern of bias is residual confounding by smoking through misclassification of non-smoking status. To be included as nonsmoker in our study, a cutoff of 400 cigarettes lifetime was set. Since misclassification may be more likely among light smokers than among never-smokers, we repeated the analyses by excluding light smokers (9 cases and 16 controls) and only minor changes were observed. In addition, a European validation study using cross-interviews with next of kin estimated the extent of misclassification of never-smoking status at 1.2%, based on 405 index subjects.22

We found a significantly increased lung cancer risk for prior pneumonia (OR=1.6; CI:1.1–2.4), a suggestive risk for tuberculosis (OR=1.6; CI:0.8–3.4), while no association was observed with chronic bronchitis, emphysema or asthma. Generally there is increasing evidence that long-term effects of specific lung diseases may lead to a heightened susceptibility to lung cancer.23 The biological mechanism by which pulmonary diseases may enhance the risk of lung cancer among nonsmokers, however, is still unclear. Recent studies that report risk estimates for nonsmoking women associated with a history of any prior lung disease showed odds ratios ranging from 1.2 to 1.8,8, 9, 23, 24 which are somewhat higher than our overall risk estimate (OR=1.1; CI:0.8–1.6). Potential limitations in our study concern misclassification of pulmonary disease and recall bias, since we relied on self-reported data and no validation could be performed. Also, we could not exclude that some of the prior lung diseases may in fact reflect early manifestations of lung cancer. We do not have information on the exact year of diagnosis of lung disease, but participants were asked to report only diseases diagnosed at least 2 years before interview.

Our data suggest a possible relationship between previous reproductive cancer with subsequent lung cancer that is consistent with other studies of different sources. Cohort studies in the United States and Denmark showed an increased risk of lung cancer following a primary of breast cancer25, 26 and genital cancer.27, 28 Unfortunately, most of these studies did not present risk estimates for nonsmokers or adjusted for smoking. One cohort study that followed women with prior endometrial cancer reported that 5 out of 6 second primaries of the lung were nonsmokers with adenocarcinoma.29 Also, a case-control study among nonsmoking women showed a 4.9-fold increased lung cancer risk associated with a history of reproductive cancer, which decreased to OR=2.9 after controlling for radiotherapy.10 It is unclear to what extent the increase in lung cancer risk through previous reproductive cancer is due to prior radiation treatment or due to shared risk factors of the lung, breast and genital system or other unknown relations. Radiation therapy for breast cancer has been suggested to increase the risk of lung cancer after a latency period of 10 years in smokers and nonsmokers.30 The odds ratio for a previous breast cancer in the present study, however, was still elevated after controlling for radiotherapy (OR=2.2; CI: 0.8–6.0), although to a lesser extent. A potential limitation in our study is that self-reported previous primaries and radiotherapy could not be validated. Bias may also occur if the second cancer was not a primary lung cancer but a metastatic of the prior cancer. Women had been asked to report cancers diagnosed by a physician at least 2 years before interview. For a subsample of about 50% of the affected study subjects, we had some information on the age at diagnosis of the first primary cancer. Among these cases the mean age at diagnosis of the first reproductive cancer was 49 years (46 years among controls); the subsequent lung cancer followed on average 12 years (15 years among controls) later. Misclassification of the second primary should be small, since metastatic cancers as well as relapses of lung cancer had been excluded by study design.

Few and inconsistent findings were available with respect to familial clustering of lung cancer in nonsmoking women. No clear association with a family history of lung cancer was found in large case-control study in Missouri nonsmoking women.13 Among nonsmoking women, some suggestive evidence for a familial aggregation of lung cancer with sisters and mothers31 or with sisters32 was noted, whereas a 7-fold significantly increased lung cancer risk associated with lung cancer in first degree relatives that was limited to early age onset patients (40 to 59 years of age) was reported by Schwartz et al.12 In the present study, neither overall nor in young age groups, an elevated familial aggregation of lung cancer (or cancer) was observed, yet it was remarkable that overall the affected relatives of lung cancer patients were 3 times more likely to be aged less than 45 years at diagnose of lung cancer than those of controls.

Exposure to ETS is well established as risk factor for lung cancer and was classified as a human carcinogen by the United States Environmental Protection Agency already in 1992.33 Most of the evidence was based on studies that investigated ETS exposure by spouse. Meanwhile there also exist a number of studies examining the effects of occupational ETS exposure or combined lifetime ETS exposure.3, 4, 18, 19, 34, 35, 36, 37, 38 Most of these studies suggested that occupational exposure may be associated with higher risk than residential exposure3 due to a possibly higher intensity of exposure at work. In the present study, long-term spousal ETS exposure showed a 1.7-fold increased risk, with suggestive evidence for a trend as compared to a 2.6-fold increased risk for long-term occupational exposure demonstrating a significant trend in risk.

Fruit and/or vegetable consumption has been shown to be inversely associated with lung cancer risk in a number of previous study that reported on nonsmoking women.14, 15, 16, 39, 40, 41, 42, 43 In accordance with these studies, our data showed a reduced lung cancer risk associated with high intake of fruit and fresh vegetables, but the test for trend was statistically significant for consumption of vegetables only. We found weak evidence for an elevated lung cancer risk associated with frequent consumption of meat. An adverse effect of frequent consumption of particularly fried, well-done red meat has been reported.44 Unfortunately we do not have information on type of meat (red or white). In the present study, protective effects with high intake of cheese, milk and other dairy products were observed, showing a statistically significant trend with consumption of cheese. Two other nonsmoker studies reported specifically on consumption of cheese. Both the European multi-center study15 and a large study in the United States14 among nonsmoking women and men found a significant reduction in lung cancer risk with high levels of consumption of cheese. The mechanism for any protective effect of cheese is unclear, yet cheese contains conjugated dienoic derivatives of linoleic acid, which have been suggested to show anticarcinogenic effects.45 With regard to milk consumption, no effect was found in the European study,15 while an increased risk for greater consumption of whole milk, in contrast to a reduced risk for greater consumption of skim or low-fat milk was found in the U.S. study.14 Information on type of milk was not available in the present study. Potential limitations of our study arise from the brief list of food items in the questionnaire that did not allow the estimation of specific macro- and micro-nutrients and of the total calorie intake. Also, it cannot be excluded that misclassification results from measuring past dietary consumption.

Studies investigating the relation between lung cancer risk and occupation among nonsmokers and women are very rare.7, 46, 47 In a recent European multi-center study in nonsmokers7 including 509 female cases and 1,011 female controls, a significantly increased lung cancer risk for women engaged in jobs of list A (known risk) or B (suspected risk) as compared to those never having worked in such jobs (OR=1.7; 95% CI:1.1–2.5) was observed, which is in line with our corresponding risk estimate (OR=1.3; CI:0.8–2.2). In the present study, risk was elevated among women occupied in the wood industry (OR=13.4), printing industry (OR=4.6) and among ceramic, pottery or glass workers (OR=1.6) and painters (OR=1.2). Owing to the small numbers of exposed cases and controls, however, all ORs were associated with wide confidence limits. Other studies in nonsmoking women have provided evidence for an increased lung cancer risk among laundry workers and dry cleaners7, 47 ceramic, pottery, glass workers and workers in the rubber manufacture,7 workers in eating and drinking establishments46 and women occupationally exposed to pesticides and asbestos.47

There is increasing evidence that residential radon may increase the risk of lung cancer.48 Little and inconsistent knowledge is available with regard to risk among nonsmokers. Indoor radon studies in nonsmokers or those reporting risk estimates for nonsmokers separately provided either no49, 50, 51, 52, 53 or suggestive evidence54, 55 for an increased risk, yet most of these studies were hampered to detect a risk due to limited power. The so far largest study in nonsmokers was recently published by Lagarde et al.5 It was a pooled analysis of 5 Swedish case-control studies with additional never-smokers from a previous Swedish study on radon53 that included overall 436 never-smoking cases and 1,649 never-smoking controls. The odds ratio of lung cancer in relation to the time-weighted average radon concentrations exceeding 140 Bq/m³ compared to less than 50 Bq/m³ was 1.44 (95% CI:1.00–2.06), showing suggestive evidence for a trend (p<0.10). A first analysis of our data that was based on women with complete 1-year radon measurements failed to detect a relation between radon and lung cancer in nonsmoking women, yet it has to be kept in mind that 42% of the cases were excluded from the analysis due to missing radon measurements. Most of these women were widows or divorced who died within the 1-year measurement period. Another possible weakness of our study is the lack of estimates of past exposures in previous homes. Women, however, did not move frequently in our study and the current home had been occupied on average for 24 years.

CONCLUSION

The present study provides evidence that passive smoking, working in certain occupations, a history of pneumonia or cancer may increase the risk of lung cancer in lifetime nonsmoking women, while a diet rich in fresh vegetables and cheese might be protective. To reduce lung cancer risk in nonsmoking women public health efforts should concern the increase of consumption of fruits and vegetables and the reduction of passive smoke exposure at work and at home.