Cigarette smoking and lung cancer risk according to histologic type in Japanese men and women
Although cigarette smoking is a well-known risk factor for lung cancer, histology-specific risk has not been fully clarified in Japan. This case-control study evaluated the associations between smoking and lung cancer risk according to sex and histologic type. From among patients aged 30 years and over admitted to a single hospital in Japan between 1997 and 2009, 1670 lung cancer cases and 5855 controls were selected. History of smoking, quantity and duration of smoking, and passive smoking from spouses were assessed using a self-administered questionnaire. Odds ratios (ORs) and 95% confidence intervals (CIs) for each exposure were estimated by unconditional logistic regression. Ever-smoking was significantly associated with a higher risk of squamous cell and small cell carcinoma. The OR for these two histologic types combined was larger in women (OR = 24.98, 95% CI: 13.50–46.23) than in men (OR = 9.43, 95% CI: 5.73–15.51). Analysis of the quantity and duration of smoking showed that the OR for each exposure level tended to be larger in women than in men. For adenocarcinoma, clear positive associations with quantity and duration-related factors were observed among men, and a significant positive association with passive smoking from spouses was found among non-smoking women (OR = 1.44, 95% CI: 1.06–1.95). These results suggest sex- and histologic type- differences in the association of smoking with lung cancer risk. Although smoking control should be continued to prevent lung cancers, further studies are required to better clarify differences in smoking-related lung cancer risk between the sexes and histologic types.
During the past several decades, trends in cancer incidence have changed in Japan. Lung cancer has been the leading site in male cancer mortality since 1993, although it has the second highest incidence after stomach cancer. For women, lung cancer is the fifth leading-site in cancer incidence and is expected to continuously increase. Additionally, the patterns of lung cancer incidence according to histologic type have also changed, the incidence of adenocarcinoma having increased and that of both squamous cell carcinoma and small cell carcinoma having decreased.
Cigarette smoking has been regarded as the most important risk factor for lung cancer. Smoking control would undoubtedly be the most effective method for prevention of lung cancer. However, risk estimates for cigarette smoking appear to vary according to sex and histologic type.[3-5] Thus, it seems difficult to precisely predict the overall effect of smoking control on lung cancer morbidity. In a previous case-control study, we analyzed lung cancer risk of cigarette smoking according to histologic type. However, that study was unable to clarify any histology-specific risk in relation to the quantity and duration of smoking. Also, the risk of lung cancer according to histologic type could not be fully evaluated in women, due to the small number of cases of female squamous cell and small cell carcinoma. In comparison with studies from Western countries, most Japanese studies have included few female cases; therefore, there is little evidence of the association between smoking and histology-specific risk among Japanese women. The association of cigarette smoking with the risk of female lung cancer has rarely been investigated for squamous cell and small cell carcinoma.[7-12] Only a few studies have evaluated histology-specific risk according to quantity and duration of smoking among women.[8, 11, 12]
On the other hand, recent studies from Western countries have discussed whether susceptibility to smoking-related lung cancer may differ between men and women. However, data for female lung cancer risk have been limited in Japan, as mentioned above. Few Japanese studies have addressed this issue.[14, 15] To explore the sex difference, it is essential to collect a sufficient number of female lung cancer cases and to clarify lung cancer risk according to histologic type among both men and women.
Therefore, we conducted an updated hospital-based case-control study. Data were obtained from patients aged 30 years and over who were admitted to a single hospital in Miyagi Prefecture, Japan, between 1997 and 2009. The analysis was performed according to sex and histologic type. In this study, we focused on the effect of quantity and duration of smoking. The risk for passive smoking was also investigated. Since a common protocol was used for men and women, this study enabled us to examine differences in smoking-related lung cancer risk according to sex and histologic type.
Materials and Methods
In January 1997, we began a questionnaire survey in connection with the present study. Information on lifestyle and personal history was collected from all patients at their first admission to the Miyagi Cancer Center Hospital (MCCH), using a self-administered questionnaire. Between January 1997 and December 2009, the questionnaire was distributed to 23 531 first-admitted patients, of whom 21 056 responded. Details of the questionnaire survey have already been described elsewhere.[6, 16]
Cases and controls were selected from among patients who responded to the above questionnaire survey. To identify incident cases of cancer, a list of the patients was linked with the hospital-based cancer registry files. This registry codes all cancer cases at the MCCH using the International Classification of Diseases for Oncology, Third Edition (ICD-O-3). Through linkage to the registry, 21 056 respondents were classified into 1812 with past history of cancer, 1734 with lung cancer (1195 men, 539 women), 10 381 with other cancers (5653 men, 4728 women) and 7129 non-cancer patients (3708 men, 3421 women). Among the 1734 patients with lung cancer, 1733 aged 30 years and over (1194 men, 539 women) were identified as cases. After the exclusion of 63 patients with incomplete information on history of smoking, 1670 cases (1159 men, 511 women) remained for analysis.
Controls were selected from among 7129 non-cancer patients. Patients with benign tumors have been classified as non-cancer patients. Diagnoses of the non-cancer patients were confirmed using the disease registration database at the MCCH. Accordingly, 6643 non-cancer patients aged 30 years and over (3483 men, 3160 women) were selected as possible controls, from whom 602 (321 men, 281 women) with smoking-related diseases were excluded. Smoking-related diseases, which are defined based on the procedure used in our previous study, include cardiovascular disease, respiratory disease such as chronic bronchitis and emphysema, and benign tumors of the upper respiratory tract, esophagus, and urinary tract. Finally, a total of 6041 non-cancer patients (3162 men, 2879 women) were identified as controls. From them, 186 non-cancer patients with incomplete information on history of smoking were further excluded, leaving 5855 controls (3098 men, 2757 women) for analysis. The diagnoses among the 5855 controls were as follows: benign tumor 3436 (58.7%), digestive tract disease 849 (14.5%), urologic-gynecologic disease 314 (5.4%), endocrine or metabolic disease 165 (2.8%), orthopedic disorder 83 (1.4%), other benign disease 306 (5.2%), and no abnormal findings 702 (12.0%). The sites of benign tumors were stomach in 298 subjects, colorectum in 1309, lung in 35, breast in 36, gynecologic organs in 363, bone or connective tissue in 963, and other in 432. There was not a large difference in the distribution of the diagnoses between males and females, although the sites of benign tumors were inconsistent between sexes (data not shown).
This study was approved by the ethical review board of the Miyagi Cancer Center. The final response rate for the questionnaire survey was 84.7% for the cases and 89.8% for the possible controls.
Assessment of cigarette smoking
Information on exposure, i.e., history of smoking, quantity and duration of smoking, and passive smoking, was collected from the questionnaire survey. Exposure variables related to active smoking included history of smoking (never-, past-, current-smoker), age at the start of smoking (never, ≤19, ≥20–≤24, ≥25 years), the mean number of cigarettes smoked per day during the smoking period (never, ≥1–≤10, ≥11–≤20, ≥21), duration of smoking (never, >0–<20, ≥20–<30, ≥30 years), and the pack-years of smoking (never, >0–<8.5, ≥8.5–<19, ≥19). Subjects who quit smoking within 1 year before the present admission were regarded as current smokers. Category boundaries for age at the start of smoking were determined arbitrarily, considering the importance of the category for subjects who started smoking as teenagers. The number of cigarettes smoked per day was categorized using points used in our previous study. For duration of smoking and pack-years of smoking, tertile points among smokers in the female control group were used as category boundaries. These tertile points were considered to be suitable for simultaneously evaluating the risks for men and women, because very few women had extremely high quantities and durations of smoking.
The only exposure variable related to passive smoking was the spouse's smoking status (never-, past-, current-smoker). This information was obtained from married subjects.
Data on the histologic type of lung cancer were retrieved from the abovementioned hospital-based cancer registry. Among 1670 cases, 1561 (93.5%) were microscopically confirmed as lung cancer: squamous cell carcinoma in 420, small cell carcinoma in 160, adenocarcinoma in 843, large cell carcinoma in 41, and others in 97. “Others” included cases with an undefined type or mixed type of cancer. For the remaining 109 cases (6.5%), histopathologic data were unavailable.
In the present study, the risks for the three major histologic types (squamous cell, small cell, and adenocarcinoma) were evaluated.
We used unconditional logistic regression analysis to estimate the overall lung cancer risk and histology-specific risk among men and women. First, odds ratios (ORs) and 95% confidence intervals (CIs) for ever-smokers (current and past smokers) compared to never-smokers were calculated according to sex, respectively. Second, the associations with quantity and duration of smoking were evaluated according to sex and histologic type. In this evaluation, study subjects were categorized using the boundaries for each exposure and the OR was calculated for each category by treating never-smokers as reference. Dose-response relationships were tested by treating each exposure category as a continuous variable. Although some previous studies have evaluated the risks in relation to quantity and duration of smoking within current smokers, the present study analyzed all subjects including never-, past- and current-smokers. Third, the OR for passive smoking from spouses (never, ever) was estimated only among non-smoking married subjects.
In the analysis, the following variables were considered to be potential confounders: age, year of recruitment, referral status, area of residence, alcohol drinking, occupation, and family history of lung cancer among parents and all siblings. These variables had been considered as confounders in our previous study. Occupation was categorized into four groups based on the rate of ever-smokers according to occupation among controls. Missing values were treated as an additional variable category. All analyses were performed using sas version 9.3 (SAS Institute, Cary, NC, USA).
The characteristics of cases and controls are presented in Table 1. In both sexes, the mean age in cases with squamous cell carcinoma was high. Cases with adenocarcinoma were more likely to have been referred from screening. Female cases with squamous cell or small cell carcinoma tended to be alcohol drinkers. For all histologic types, the frequency of cases with a family history of lung cancer was higher than that of controls.
Table 1. Characteristics of cases and controls by sex and histologic type
|Number of subjects||1159||511||380||40||133||27||473||370||3098||2757|
|Age group (years old) (%)|
|30–39||0.5 ||0.8 ||0.5 ||0.0 ||0.0 ||0.0 ||0.6 ||1.1 ||5.9 ||9.1 |
|40–49||3.6 ||5.7 ||1.8 ||0.0 ||3.8 ||7.4||3.8 ||5.7 ||11.6 ||18.8 |
|50–59||15.8 ||17.2 ||9.0 ||12.5 ||22.6 ||11.1 ||19.5 ||17.8 ||22.5 ||22.7 |
|60–69||28.0 ||29.1 ||24.5 ||22.5 ||37.6 ||25.9 ||28.8 ||33.0 ||30.3 ||25.0 |
|70–79||42.9 ||38.0 ||52.9 ||47.5 ||29.3 ||48.1 ||40.6 ||35.1 ||24.1 ||18.6 |
|80+||9.2 ||9.2 ||11.3 ||17.5 ||6.8 ||7.4 ||6.8 ||7.3 ||5.5 ||5.7 |
|Mean (years old)||68.3 ||67.0 ||70.6 ||70.4 ||66.2 ||67.9 ||67.1 ||66.3 ||61.7 ||58.8 |
|SD||9.8 ||10.5 ||8.5 ||9.3 ||9.7 ||10.0 ||9.9 ||19.3 ||12.6 ||13.7 |
|Year of recruitment (%)|
|1997–2002||34.9 ||36.6 ||32.9 ||32.5 ||43.6 ||44.4 ||35.5 ||37.0 ||50.0 ||52.7 |
|2003–2009||65.1 ||63.4 ||67.1 ||67.5 ||56.4 ||55.6 ||64.5 ||63.0 ||50.0 ||47.3 |
|Area of residence (%)|
|Southern Miyagi prefecture||77.4 ||83.2 ||72.9||77.5||82.7||88.9||79.7||83.0||89.0||88.0|
|Other area||22.6 ||16.8 ||27.1||22.5||17.3||11.1||20.3||17.0||11.0||12.0|
|Referral status (%)|
|From screening||18.3 ||31.7 ||16.6||7.5||11.3||14.8||22.4||35.9||18.2||17.3|
|Other||81.7 ||68.3 ||83.4||92.5||88.7||85.2||77.6||64.1||81.8||82.7|
|Professional or clerical work||33.2 ||25.1 ||28.9 ||30.0 ||30.8 ||33.3 ||36.4 ||23.5 ||43.6 ||35.9 |
|Industrial work or fishery||35.5 ||16.4 ||31.3 ||2.5 ||33.8 ||7.4 ||38.7 ||18.7 ||34.5 ||14.6 |
|Agriculture or forestry||16.6 ||9.2 ||21.1 ||2.5 ||25.6 ||0.0 ||12.0 ||10.8 ||12.4 ||9.2 |
|Othera||2.1 ||27.4 ||2.6 ||27.5 ||1.5 ||33.3 ||2.1 ||27.8 ||1.5 ||24.6 |
|Missing||12.6 ||21.9 ||16.1 ||37.5 ||8.3 ||26.0 ||10.8 ||19.2 ||8.0 ||15.7 |
|Alcohol drinking (%)|
|Ever||72.6 ||21.7 ||70.5||30.0||75.2||33.3||75.1||20.5||76.9||24.2|
|Never||25.8 ||73.2 ||27.6||65.0||23.3||59.3||23.7||74.3||21.9||72.7|
|Missing||1.6 ||5.1 ||1.8||5.0||1.5||7.4||1.3||5.1||1.2||3.1|
|Family history of lung cancer in parents and siblings (%)|
|No||89.6 ||91.6 ||89.7 ||92.5 ||91.0 ||92.6 ||87.7 ||91.6 ||93.0 ||93.8 |
|Yes||10.4 ||8.4 ||10.3 ||7.5 ||9.0 ||7.4 ||12.3 ||8.4 ||7.0 ||6.2 |
|Smoking status (%)|
|Never||7.5 ||75.1 ||3.4 ||37.5 ||3.0 ||14.8 ||12.7 ||83.0 ||23.1 ||83.4 |
|Past||37.5 ||8.8 ||42.4 ||27.5 ||30.1 ||18.5 ||37.0 ||7.0 ||29.8 ||3.7 |
|Current||55.0 ||16.1 ||54.2 ||35.0 ||66.9 ||66.7 ||50.3 ||10.0 ||47.1 ||12.9 |
|Passive smoking from spousesb (%)|
|Number of subjectsc||76 ||355||12 ||14 ||4 ||4 ||52 ||284 ||634 ||2115 |
|Never||88.2 ||27.9 ||83.3 ||21.4 ||75.0 ||75.0 ||90.4 ||26.4 ||91.2 ||30.3 |
|Past||1.3 ||21.7 ||0.0 ||35.7 ||0.0 ||25.0 ||1.9 ||22.5 ||0.6 ||19.2 |
|Current||2.6 ||32.7 ||0.0 ||35.7 ||0.0 ||0.0 ||1.9 ||34.2 ||2.8 ||36.1 |
|Missing||7.9 ||17.7 ||16.7 ||7.2 ||25.0 ||0.0 ||5.8 ||16.9 ||5.4 ||14.4 |
Table 2 shows the overall lung cancer risk and histology-specific risk in relation to smoking history among men and women. The OR for squamous cell and small cell carcinoma combined is also presented in the table. For both sexes, having ever smoked was associated with an increased risk of lung cancer overall. For all histologic types, smoking was positively associated with the risk. Odds ratios for ever-smokers tended to be greater for squamous cell and small cell carcinoma than for adenocarcinoma. A similar tendency was observed for past and current smokers, respectively. Furthermore, for squamous cell and small cell carcinoma, the ORs were much larger among women than among men. On the other hand, for adenocarcinoma, the difference in ORs between men and women was small. To investigate whether ORs for smoking history (ever versus never) were constant across both sexes, we introduced an interaction term (sex*smoking history) into the statistical model and tested its statistical significance. The interaction term for squamous cell and small cell carcinoma combined suggested that the effect of cigarette smoking on the risk of these histologic types might differ between men and women (P = 0.03).
Table 2. Odds ratio (OR) and 95% confidence interval (CI) according to history of smoking by histologic type
|Never smoked||87||1 ||(Reference)||384||1 ||(Reference) || |
|Ever smoked||1072||3.91 ||3.07–4.98 ||127||2.58 ||2.00–3.34 ||0.01|
|Past||434||3.25 ||2.51–4.21 ||45||2.81 ||1.88–4.20 || |
|Current||638||4.53 ||3.52–5.84 ||82||2.48 ||1.84–3.34 || |
|Never smoked||13||1 ||(Reference) ||15||1 ||(Reference) || |
|Never smoked||4||1 ||(Reference) ||4||1 ||(Reference) || |
|Squamous cell + small cell|
|Never smoked ||17||1 ||(Reference) ||19||1 ||(Reference) || |
|Never smoked||60||1 ||(Reference) ||307||1 ||(Reference) || |
Table 3 shows the associations between the quantity and duration of smoking and lung cancer risk according to histologic type. A younger age at the start of smoking was positively associated with the risk of squamous cell and small cell carcinoma in both sexes. For adenocarcinoma, a significant association for age at the start of smoking was observed only among men. The number of cigarettes smoked was positively associated with lung cancer risk for all histologic types in both sexes. A longer duration of smoking was significantly associated with an increased risk of squamous cell and small cell carcinoma in both sexes. For adenocarcinoma, a similar positive association was observed; however, the trend test showed that this was not significant among women (P for trend = 0.17). Pack-years of smoking were positively associated with lung cancer risk for all histologic types except female adenocarcinoma. When the ORs for quantity and duration of smoking listed in Table 3 were compared between men and women, the ORs for squamous cell and small cell carcinoma tended to be larger in women, although the confidence interval for each category was wide in women. The test for interaction between sex and quantity and duration of smoking showed statistical significance for most of the variables in these two histologic types. For adenocarcinoma, no large difference in OR was observed between the sexes; however, the ORs among women were insignificant for all exposure levels.
Table 3. Odds ratio (OR) and 95% confidence interval (CI) according to quantity and duration of smoking by histologic type
|Age at the start of smoking|
|Never smoked||716||13||1||(Reference)||4||1 ||(Reference)||17||1 ||(Reference)||60||1 ||(Reference)|
|25≤||282||37||6.30 ||3.25–12.21 ||14||7.85 ||2.55–24.20 ||51||6.71 ||3.76–11.96||49||1.85 ||1.23–2.79 |
|20 ≤ 24||1447||217||9.10 ||5.10–16.23||76||10.15 ||3.68–27.97||293||9.42 ||5.67–15.64 ||251||2.11 ||1.56–2.86 |
|≤19||480||81||14.56 ||7.82–27.10 ||31||15.09 ||5.22–43.62 ||112||14.90 ||8.65–25.65 ||78||2.55 ||1.76–3.69 |
|Never smoked||2299||15||1 ||(Reference)||4||1 ||(Reference)||19||1||(Reference)||307||1 ||(Reference)|
|25≤||171||13||13.69 ||5.87–31.90||11||61.19 ||17.82–210.10 ||24||22.87 ||11.37–45.96 ||28||1.43 ||0.92–2.22 |
|20 ≤ 24||198||5||14.12 ||4.28–46.59 ||9||137.00 ||34.28–547.52 ||14||35.93 ||14.96–86.32 ||16||1.16 ||0.66–2.03 |
|≤19||48||2||30.23 ||5.48–166.91 ||2||225.62 ||30.85–b||4||59.41 ||15.79–223.50 ||4||1.44 ||0.49–4.22 |
|Number of cigarettes per day|
|0 (Never smoked)||716||13||1 ||(Reference)||4||1 ||(Reference)||17||1 ||(Reference)||60||1||(Reference)|
|0 ≤ 10||365||28||3.71 ||1.87–7.37 ||11||5.08 ||1.60–16.16 ||39||4.06 ||2.23–7.37 ||48||1.42 ||0.94–2.14 |
|11 ≤ ≤ 20||1273||207||10.03 ||5.61–17.94 ||68||10.20 ||3.69–28.25 ||275||10.09 ||6.06–16.79 ||214||2.06 ||1.51–2.81 |
|21≤||587||103||14.00 ||7.61–25.76 ||41||15.95 ||5.62–45.28 ||144||14.78 ||8.67–25.18 ||115||2.82 ||2.00–3.98 |
|P for trend|| ||<0.0001||<0.0001||<0.0001||<0.0001|
|0 (Never smoked)||2299||15||1 ||(Reference)||4||1 ||(Reference)||19||1 ||(Reference)||307||1 ||(Reference)|
|0 ≤ 10||222||8||8.24 ||3.16–21.50 ||7||40.79 ||10.88–152.88 ||15||14.47 ||6.68–31.34 ||24||1.23 ||0.77–1.96 |
|11 ≤ 20||170||10||29.50 ||11.07–78.64 ||11||151.05 ||39.06–584.13 ||21||48.99 ||22.20–108.08 ||22||1.51 ||0.92–2.49 |
|21≤||28||4||39.24 ||9.83–156.66 ||3||176.86 ||31.04−‡||7||64.77 ||20.62–203.45 ||5||1.99 ||0.73–5.44 |
|P for trend|| ||<0.0001||<0.0001||<0.0001||0.03 |
|P for interaction between sex and exposurec|| ||0.0004||0.0002||<0.0001||0.15|
|Duration of smoking (years)|
|0 (Never smoked)||716||13||1 ||(Reference)||4||1 ||(Reference)||17||1 ||(Reference)||60||1 ||(Reference)|
|0 < 20||246||6||3.15 ||1.15–8.62 ||0||–|| ||6||2.11 ||0.80–5.54 ||18||1.38 ||0.78–2.45 |
|20 < 30||407||20||5.73 ||2.75–11.93 ||4||2.25 ||0.55–9.25 ||24||4.70 ||2.44–9.03 ||34||1.48 ||0.94–2.34 |
|30≤||1407||269||9.74 ||5.49–17.29 ||102||12.74 ||4.65–34.91 ||371||10.48 ||6.34–17.31 ||296||2.33 ||1.73–3.15 |
|P for trend|| ||<0.0001||<0.0001||<0.0001||<0.0001|
|0 (Never smoked)||2299||15||1 ||(Reference)||4||1 ||(Reference)||19||1 ||(Reference)||307||1 ||(Reference)|
|0 < 20||179||2||5.38 ||1.08–26.68 ||0||–|| ||2||4.37 ||0.93–20.55 ||11||1.01 ||0.52–1.96 |
|20 < 30||125||2||5.00 ||1.01–24.85 ||3||37.04 ||7.15–191.92 ||5||11.56 ||3.80–35.21 ||13||1.27 ||0.68–2.34 |
|30≤||103||15||21.82 ||9.56–49.78 ||19||123.80 ||38.93–393.67 ||34||43.01 ||22.17–83.45 ||20||1.39 ||0.83–2.32 |
|P for trend|| ||<0.0001||<0.0001||<0.0001||0.17 |
|P for interaction between sex and exposure|| ||0.03||0.04||0.0004||0.04|
|Pack–years of smoking|
|0 (Never smoked)||716||13||1 ||(Reference)||4||1 ||(Reference)||17||1 ||(Reference)||60||1 ||(Reference)|
|0 < 8.5||93||3||3.21 ||0.86–11.99 ||1||2.99 ||0.33–27.43 ||4||3.25 ||1.03–10.20 ||10||1.76 ||0.85–3.65 |
|8.5< 19||279||3||0.87 ||0.24–3.12 ||4||3.35 ||0.83–13.64 ||7||1.58 ||0.64–3.91 ||26||1.44 ||0.88–2.36 |
|19≤||1676||287||10.24 ||5.76–18.19 ||100||11.51 ||4.19–31.60 ||387||10.65 ||6.43–17.61 ||308||2.24 ||1.66–3.01 |
|P for trend|| ||<0.0001||<0.0001||<0.0001||<0.0001|
|0 (Never smoked)||2299||15||1 ||(Reference)||4||1||(Reference)||19||1 ||(Reference)||307||1 ||(Reference)|
|0 < 8.5||137||1||3.08 ||0.38–25.09 ||0||–|| ||1||2.73 ||0.35–21.64 ||9||0.95 ||0.46–1.94 |
|8.5 ≤ 19||133||3||6.28 ||1.61–24.47 ||4||46.61 ||10.29–211.11 ||7||13.86 ||5.15–37.30 ||12||1.10 ||0.58–2.08 |
|19≤||132||15||28.19 ||11.88–66.89 ||17||176.50 ||50.32–619.04 ||32||53.95 ||26.40–110.23 ||23||1.57 ||0.97–2.55 |
|P for trend|| ||<0.0001||<0.0001||<0.0001||0.10 |
|P for interaction between sex and exposure|| ||0.11||0.0057||0.0021 ||0.08|
Table 4 shows the association of passive smoking from spouses with lung cancer risk. Overall analysis demonstrated a marginal association of passive smoking with female lung cancer risk (OR = 1.31, 95% CI: 0.99–1.72, P = 0.0544), whereas no significant association was observed for lung cancer in men. With regard to histologic type, a significant positive association with passive smoking was found for female adenocarcinoma (OR = 1.44, 95% CI: 1.06–1.95).
Table 4. Odds ratio (OR) and 95% confidence interval (CI) according to passive smoking by histologic type among non–smoking married subjects
|Smoking from wives|
|No||578||67||1||(Reference) ||10||1||(Reference) ||3||1||(Reference) ||13||1||(Reference) ||47||1||(Reference) |
|Yesa||22||3||1.29||0.34–4.91||0||–|| ||0||–|| ||0||–|| ||2||1.30||0.27–6.14|
|Smoking from husbands|
|No||640||99||1||(Reference) ||3||1||(Reference) ||3||1||(Reference) ||6||1||(Reference) ||75||1||(Reference) |
The present study reconfirmed the significant association of cigarette smoking with overall lung cancer risk and clarified the risk of active and passive smoking according to histologic type among men and women. The value of the overall OR for ever-smokers in the present study was similar to that in our previous study including incident cases during the period of 1997–2001.
Regarding the risk for active smoking according to histologic type, previous studies, mainly from Western countries, have shown a significant positive association for all histologic types.[3, 4] In general, the magnitude of the association was greater for squamous cell and small cell carcinoma than for adenocarcinoma among both men and women. However, data for Japanese populations have been insufficient. Most analyses on the risk related to the quantity and duration of smoking have been limited to men,[9, 10, 19] and histology-specific risk among Japanese women has been unclear. To our knowledge, three studies have evaluated the histology-specific risk in relation to quantity and duration of smoking among Japanese women; however, their results were inconsistent.[8, 11, 12] Furthermore, the smoking-related risk for female squamous cell and small cell carcinoma has not been well investigated.[7-12] The present study included a relatively larger number of female cancer cases than previous Japanese studies, making it possible to conduct detailed analyses based on histologic type. The major findings were as follows. A higher risk for having ever smoked was observed for squamous cell and small cell carcinoma in both men and women. Women who are heavy and long-term smokers were at higher risk for these carcinomas. For adenocarcinoma, a weaker positive association with ever-smoking was observed in both men and women. Significant associations with quantity and duration-related factors were observed mainly among men. These findings indicate that active smoking is closely related to the risk of squamous cell and small cell carcinoma among women as well as men. However, the magnitude of smoking-related lung cancer risk is likely to differ between men and women.
Some studies from Western countries have suggested a sex difference in the association of cigarette smoking with lung cancer risk.[20-26] In the Asian region, a case-control study from Korea showed that the magnitude of association for active smoking was larger in women than in men for three major histologic types. Our results indicate that such a sex difference may also exist in the Japanese population. However, the issue of a sex difference still seems controversial. Recent large-scale studies from Western countries have provided conflicting conclusions.[25, 26] One study showed a higher smoking-related lung cancer risk in women. The other study according to histologic type did not support such a higher female susceptibility. To clarify any sex difference, it is necessary to resolve problems related to the study design (cohort versus case-control), the measure used to estimate risk (absolute versus relative), the model of interaction (additive versus multiplicative), the treatment of individuals who have never smoked (inclusion versus exclusion), and potential confounders, as pointed out by Matteis et al. The conflicting conclusions may partly depend on these problems. Additionally, it has not been fully investigated whether men and women differ in their biologic susceptibility to tobacco carcinogens.[13, 15, 28] We interpret our results, i.e., the difference in histology-specific risk between men and women, as follows: First, the absolute risk of squamous cell and small cell carcinoma might be higher among male non-smokers than among female non-smokers, which could contribute to the lower ORs, i.e., lower relative risks, for these histologic types in men. Although case-control studies like ours cannot evaluate absolute risk, a cohort study in Japan demonstrated a higher incidence of squamous cell and small cell carcinoma among male non-smokers than among female non-smokers. Second, some unique lifestyle factors of female smokers might have been responsible for the higher risk of squamous cell and small cell carcinoma in women, despite the fact that tobacco-specific carcinogens play a major role in the development of cancer.[29, 30] Unclear associations between the quantity and duration of smoking and the risk of adenocarcinoma in women may also be linked to lifestyle factors in female smokers. For example, the baseline data for the Miyagi cohort covering the catchment area of the present study showed that female heavy smokers tended to be alcohol drinkers, less educated and consume few green vegetables and oranges. The differences in these lifestyle factors between never- and ever-smokers were greater among women than among men. We therefore attempted additional adjustment for dietary intake, including that of green and yellow vegetables and fruit, as confounders. This adjustment did not change the ORs for any of the histologic types in either men or women (data not shown); however, the possibility that lifestyle characteristics of female smokers may mediate the risk of lung cancer in women cannot be completely ruled out. The presence of some unique characteristics in female smokers appears to be supported from a historical perspective. During and after World War II, there was an extreme tobacco shortage in Japan.[33, 34] Most of present study subjects grew up during this period. Furthermore, smoking by women seems to have been stigmatized, since smoking prevalence among Japanese women has been consistently lower than those among Western populations.[33, 35, 36] Taking into account theses historical backgrounds, female heavy and long-term smokers are likely to have had some specific lifestyle factors that may have been responsible for the extremely high risk of squamous cell and small cell carcinoma. Third, cigarette smoking may affect the endocrine system, thus contributing to the sex difference in histology-specific lung cancer risk. Estrogen receptors are expressed in some types of lung cancer. It has been suggested that the reproductive and hormonal factors might be associated with female lung cancer risk.[38, 39] Clinical and laboratory data indicate estrogenic or antiestrogenic effects of cigarette smoking among women.[40, 41]
In relation to the lung cancer risk posed by passive smoking, we evaluated the risk for passive smoking from spouses. Lung cancer risk for smoking by spouses among non-smokers has been investigated in some studies both in Japan and Western countries.[3, 42] Our result for overall risk is supported by these studies. On the other hand, the risk according to histologic type has been inconsistent in previous studies.[42, 43] The present study found a significant positive association of passive smoking from husbands with the risk of adenocarcinoma among non-smoking married women. A recent cohort study in Japan has also demonstrated a similar positive association. The mechanisms accounting for the risk of passive smoking from spouses may be complicated. Sidestream smoke components may be involved in the development of lung cancer.[30, 45] However, lifestyle and dietary factors that are correlated with spouses' smoking status may confound the risk for lung cancer.[46, 47] In the present study, additional adjustment for intake of green and yellow vegetable and fruit attenuated the OR for adenocarcinoma in women, whereas such adjustment did not change the ORs for squamous cell and small cell carcinoma in women, or the ORs for lung cancer in men (data not shown). It is still possible that the risk of female adenocarcinoma associated with passive smoking from husbands may be influenced by other lifestyle factors.
The present study had both strengths and limitations. In hospital-based case-control studies like ours, some methodological problems are likely to influence the results. First, we considered comparability between the cases and controls. The controls were selected from among patients admitted to the same hospital as the cases. However, the distribution of risk factors among the control subjects might have differed from that in the general population. To improve comparability between cases and controls, we therefore excluded patients with disease believed to have been related to smoking from the controls in advance. Although any diseases associated with smoking might still remain in the control group, the distribution of diagnoses was relatively similar between the male and female control group. It is unlikely that the results according to males and females have been differently distorted by this control selection procedure. Additionally, the proportions of ever-smokers among the controls (men 76.9%, women 16.6%) were comparable to those in the general population of Miyagi Prefecture (men 78.8%, women 20.5%). Second, it is necessary to consider the possibility of information bias. Female subjects might have under-reported smoking history, since smoking by women seems to have been stigmatized in Japan. However, both cases and controls are hospitalized patients; therefore, it is unlikely that the magnitude of under-reporting bias is different between female cases and controls. Any misclassification of female smokers would have been non-differential. The effect of this bias does not seem very serious.
One of the strengths of the present study was that a relatively large number of female lung cancer cases were included. Consequently, we were able to analyze the association of smoking with the risk of female squamous cell and small cell carcinoma, although statistical power might still be limited. These histologic types are rarely diagnosed in women. Therefore, case-control studies like this one seem suitable for investigating risk factors for female lung cancer. Another strength was that confounders such as occupation, alcohol drinking, and family history of lung cancer were controlled in the analysis. Previous studies had not considered the effects of these confounders.[3-5]
In conclusion, this case-control study clarified the association of cigarette smoking with histology-specific lung cancer risk among Japanese men and women. A higher risk for having ever smoked was observed for squamous cell and small cell carcinoma in both men and women. Women who are heavy and long-term smokers were at higher risk for these carcinomas. For adenocarcinoma, clear positive associations with quantity and duration-related factors were observed among men, and a significant positive association with passive smoking from husbands was found among non-smoking women. These results suggest that there are sex- and histologic type-differences in the association of smoking with lung cancer risk. Although smoking control should be continued to prevent lung cancers, further studies are required to better understand differences in smoking-related lung cancer risk in relation to sex and histologic type.
This work was supported by a Grant-in-Aid for the Second Term Comprehensive 10-Year Strategy for Cancer Control from the Ministry of Health, Labour and Welfare, Japan.
The authors have no conflict of interest.