Relationship between Cigarette Smoking and Histologic Type of Lung Cancer, with Special Reference to Sex Difference

Relationship between Cigarette Smoking and Histologic Type of Lung Cancer, with Special Reference to Sex Difference Tomotaka Sobue, M.D.* 1, Takaichiro Suzuki, M.D. 1, Takeshi Horai, M.D. 2, Minoru Matsuda, M.D. 3 and Isaburo Fujimoto, M.D. 1 ^Department of Field Research, ^Department of Biology, Research Institute, ^Department of Respiratory Diseases, The Center for Adult Diseases, Osaka, Nakamichi 1-3-3, Higashinari-ku, Osaka 530 Abstract In order to clarify sex differences associated with the relationship between smoking and each histologic type of lung cancer, the smoking histories of 2,083 lung cancer patients (1,660 men and 423 women) diagnosed at the Center for Adult Diseases, Osaka, from 1965 to 1983 were analyzed. The rates expected of smokers among these cases were calculated from the sex and age specific rates of smokers in the general population, reported annually by the Tobacco and Salt Public Corporation. The odds ratios for smoking were estimated to be 5.2, 3.1, 6.9 and 4.1 in men for squamous cell carcinoma, adenocarcinoma, small cell carcinoma and large cell carcinoma, and 7.2,, 14.4 and 3.8 in women respectively. Significant differences in the odds ratios for each histologic type between men and women were observed only for adenocarcinoma. In squamous cell carcinomas, the odds ratios were estimated by dividing them into two groups according to location. No difference was found between men and women in this respect. Key words: (Jpn. J. Clin. Oncol. 18: 3-13, 1988) Lung cancercigarette smokingodds ratio Introduction were, as for men, higher for squamous cell and small cell carcinoma than for adeno- It has been clearly shown from various carcinoma, although the values of the odds epidemiologic studies that cigarette smoking ratios were relatively lower than those in plays an important part in the carcinogenesis men. 3 " 7 ' of lung cancer. Through these studies, it is In, however, only few studies have known that, in men, different histologic types shown odds ratios for smoking by histologic of lung cancer are affected differently by type for women, because of the difficulties cigarette smoking, that is, squamous cell of collecting a sufficient number of female carcinoma and small cell carcinoma have lung cancer cases. 8 ' 0) stronger associations with smoking than ade- In the present study, we have investigated nocarcinoma. 1 ' 2) cigarette smoking habits among lung cancer In the U.S.A. and western Europe, similar patients, analyzing the difference between studies have been performed for women, men and women from the viewpoint of the These studies showed that the odds ratios for relationship between the histologic type of cigarette smoking on lung cancer for women lung cancer and cigarette smoking. Furthermore, in order to investigate wheth- ~ '. ~. I Z.., er or not the effect of cigarette smoking Received: January 6, 1987.. Accepted: December 19, 1987. dlffers according to the location of the lung *For reprints and all correspondence. cancer, we also studied the risks of cigarette 18(1) 1988 3

SOBUE ET AL. smoking for squamous cell carcinoma, divided by location into two groups. Subjects and Methods The present study was based on 2,780 primary lung cancer patients (2,183 men and 597 women) diagnosed during the 19 years from 1965 to 1983 at the Center for Adult Diseases, Osaka. Histologic classification was determined according to the WHO lung carcinoma classification. The patients were grouped into four major types: (1) squamous cell carcinoma, (2) adenocarcinoma, (3) small cell carcinoma, (4) large cell carcinoma. Four hundred and one (14.4%) cases who were classified into other miscellaneous histologic types or whose histologic types were unknown were excluded. Histologies were classified by cytology (52.5%), biopsy (14.7%) or operation (32.8%). From an investigation of medical records, 2,087 cases with smoking habits were obtained. Smoking habits had been routinely established for all outpatients at the time of the first visit to the hospital from a given questionnaire. For patients already in hospital, smoking habits obtained from nurses' interviews were also reviewed. Four pipe smokers were excluded from the study. Thus, the numbers of subjects for further analysis were 1,660 men and 423 women. Smoking intensity was categorized from the Brinkman Index (no. of cigarettes smoked/day X no. of years spent smoking). The rates expected for smokers for each histologic type, adjusted by age and year of diagnosis, were calculated for both sexes, using the sex and age specific rates of smokers in the general population, reported annually by the Tobacco and Salt Public Corporation from 1965 to 1983 (Table I). 10) According to these data, the rates of smokers in men for all age groups were approximately 46 times higher than those in women throughout the study period. In men, the rates of smokers were higher in the younger age groups, but had been decreasing in all age groups during the period. In women, on the other hand, the rates of smokers Table I. Sex and Age Specific Smoking Rates in (1965-1983) 20 Years Old and Over Year Age 20-29 30-39 40-49 50-59 60- All 1965 80.5 84.7 86.7 81.4 74.6 8 1966 83.5 84.8 87.3 83.4 78.0 83.6 1967 83.2 84.1 85.8 82.3 73.3 82.3 1968 78.0 79.3 82.5 81.2 70.8 78.5 1969 78.5 80.5 83.7 80.3 71.0 79.1 1970 79.9 78.4 81.0 78.3 67.8 77.7 1971 79.2 77.3 79.7 78.8 69.8 77.5 1972 79.9 77.1 81.1 79.8 68.6 77.6 1973 80.1 78.7 82.2 77.7 70.1 78.3 1974 82.9 79.7 80.6 78.0 69.7 78.8 1975 8 77.0 76.3 78.6 65.8 76.2 1976 80.8 74.8 75.4 77.5 64.4 75.1 Year Age 1965 1966 1967 1968 1969 1970 1971 1972 1973 1974 1975 1976 20-29 30-39 40-49 50-59 60- All 6.6 13.5 19.0 19.7 23.0 15.0 10.6 14.3 22.0 24.1 24.1 17.7 11.0 16.4 20.9 23.1 20.3 17.4 8.1 13.6 17.8 21.1 20.4 15.1 9.9 13.1 16.4 20.7 19.8 15.0 9.8 13.0 23.3 2 15.3 10.2 13.7 17.9 19.4 14.7 12.7 13.3 14.9 20.6 18.6 15.5 11.0 12.4 15.5 18.0 21.2 15.1 12.9 14.1 17.6 21.1 20.5 16.7 12.7 13.5 15.7 17.9 16.8 15.1 14.3 14.4 14.6 17.4 17.5 15.4 Source: Tobacco and Salt Public Corporation. Jpn. J. Clin Oncol.

SEX DIFFERENCES IN SMOKING IN LUNG CANCER in the 20-29 age group had been increasing rapidly, while those in the 50-59 and 60 + age groups had been decreasing. As a result, the rate of smokers in women became higher in the younger age groups in 1983, while they had been higher in the older age groups in 1965. In order to estimate the odds ratios for smoking by histologic type, we calculated the odds ratios by dividing the observed by the expected odds. Confidence intervals (95%) for odds ratios were calculated by a testbased method, 111 using the numbers of subjects in each age group from the survey of the Tobacco and Salt Public Corporation. Furthermore, we divided squamous cell carcinomas into two groups according to their location, based on findings obtained from bronchoscopy, bronchogram and tomogram. The first, central type, was taken to be those located at segmental or more proximal bronchial areas, and the second, peripheral type, at subsegmental or more peripheral bronchial areas, assuming the odds 1977 79.9 76.0 74.5 75.5 67.4 78.1 1977 16.0 13.2 14.5 16.0 17.0 15.1 1978 78.2 76.0 75.3 76.3 65.5 74.7 1978 14.9 15.7 16.6 16.8 17.3 16.2 1979 80.3 76.1 71.2 74.6 62.0 73.1 1979 16.4 14.0 15.5 16.3 15.4 15.4 1980 77.1 73.4 69.1 7 6 70.2 1980 16.2 14.2 14.4 12.8 14.6 14.4 1981 76.4 75.9 68.6 69.6 60.9 70.8 1981 17.4 14.9 16.5 13.5 14.1 15.3 1982 76.2 74.7 67.5 72.2 58.8 70.1 1982 17.4 16.2 15.7 14.1 13.3 15.4 1983 70.9 71.3 65.2 65.7 56.5 66.1 1983 15.0 14.8 13.4 1 12.4 13.5 ratio for smoking would be higher in the central than in the peripheral type. Fortytwo patients whose tumors were too large or too invasive for their sites of origin to be determined were excluded. This approach was limited to squamous cell carcinoma, since the number of central types in adenocarcinoma was extremely few. The method used in the analysis was the same as above. The population attributable risk (PAR was ) estimated using the following equation 12) PAR= Ji x 100 - P.x(RR-l) P,x where I o = incidence rate in non-smokers, I = overall incidence rate, RR = estimated odds ratio and P. = rate of smokers. In this calculation, the smoking rates for all ages in 1975 (male 76.2%, female 15.1%) were used. We also estimated the incidence rates in non-smokers (Io) and smokers (I. X RR) respectively, using the lung cancer incidence rates for all age in (male 25.5 per 100,000, female 9.4 per 100,000), as reported by the Research Group for Population-based Cancer Registry. 13 ' Results Table II shows the distribution of 2,083 lung cancer cases by sex and histologic type. The commonest lung cancers were squamous cell carcinoma in men (4%) and adenocarcinoma in women (64.5%). Table III shows the age distribution by sex and histologic type. There was a tendency for the women to be younger than the men in squamous cell carcinoma. Patients' smoking habits by sex and histologic type are presented in Table IV. The rates of smokers were 90.1% for men and 39.2% for women over all histologic types together. In both sexes, the rates of current smokers were relatively high for squamous cell carcinoma and small cell carcinoma, and 18(1) 1988

SOBUE ET AL. Table II. Distribution of Lung Cancer Cases by Histologic Type (1965-1983) Hijtologic Type Squamous cell Small cell Large cell (*) 694 (4) 68 () 595 (35.8) 273 (64.5) 254 (15.3) 62 (14.7) 117 (7.0) 20 (4.7) 1660 (10) 423 (10) Table III. Hijtologic Type Age <29 30-39 40-49 50-59 60-69 70-79 80+ Age Distribution of Lung Cancer Cases by Histologic Type Squamous cell 1.2 3.0 22.2 46.7 24.8 2.2 10 No. of cases 694 Histologic Type Age <29 30-39 40-49 50-59 60-69 70-79 80+ Squamous cell 7.4 7.4 17.6 44.1 20.6 2.9 10 (*) 0.2 3.2 8.7 20.3 4 23.7 2.4 10 595 0.4 2.6 10.6 30.4 36.6 17.6 10 0.4 0.4 9.8 25.2 34.6 27.6 2.0 10 254 1.6 1.6 25.8 48.4 22.6 10 2.6 10.3 26.5 45.3 12.0 3.4 10 117 2 2 4 2 10 0.1 1.9 6.6 22.3 42.9 23.9 2.3 10 1,660 0.2 3.1 9.2 27.2 39.7 18.9 1.7 10 No. of cases 68 273 62 20 423 relatively low for adenocarcinoma. Current smokers in men were distributed in a higher Brinkman Index group than women for all histologic types. Table V shows the odds ratios for smoking by sex and histologic type, after adjustment for age and year of diagnosis. All histologic types in both sexes were significantly associated with smoking. A comparison of odds ratios by histologic type within each sex gave relatively high values in small cell and squamous cell carcinoma, and relatively low values in adenocarcinoma. A statistically significant difference was observed in adenocarcinoma, between men and women, but no such significant differences were observed in Jpn. J. Clin. Oncol.

SEX DIFFERENCES IN SMOKING IN LUNG CANCER Table IV. Distribution of Lung Cancer Cases by Smoking Habit and Histologic Type Histologic Type Smoking habit Non-smoker Ex-smoker Current smoker Current smoker: Histologic Type Smoking habit Non-smoker Ex-smoker Current smoker Current smoker: Brinkman Index. BI* -399 400-800- 1600- BI* -399 400-800- 1600- Squarnous cell 5.5 2.6 91.9 5.5 28.0 30.8 27.7 Squamous cell 38.2 60.3 14.7 20.6 23.5 1 86.4 9.7 23.0 34.5 19.2 72.9 25.6 9.9 10.3 3.7 4.3 2.0 93.7 5.9 24.0 33.5 30.3 25.8 74.2 33.9 8.1 8.5 0.9 90.6 8.5 32.5 26.5 23.1 55.0 45.0 10.1 25.0 5.0 5.0 Table V. Observed and Expected Rates of Smokers and Odds Ratios by Histologic Type Histologic Type Rate of smoker Observed Expected Odds ratio (955fc CI*) Squamous cell 91.9 68.5 5.2 4.2-6.5 86.4 67.5 3.1 2.4-3.7 93.7 68.3 6.9 4.6-10.3 90.6 69.9 4.1 2.4-7.2 7.8 2.1 90.1 7.3 25.9 32.2 24.7 59.6 1.2 39.2 11.6 8.7 2.8 Histologic Type Squamous cell Rate of smokers Observed Expected Odds ratio (95& CI*) 60.3 17.4 7.2 4.8-10.8 25.6 16.4 1.4-2.2 74.2 16.7 14.4 9.3-22.2 45.0 17.7 3.8-9.3 Confidence interval. 18(1) 1988

SOBUE ET AL. Table VI. Distribution of Squamous Cell Carcinoma Cases by Location and Smoking Habit Location Smoking habit Non-smoker Ex-smoker Current smoker Current smoker: Table VII. Location Smoking habit BI* -399 400-800- 1600- Non-smoker Ex-smoker Current smoker Current smoker: BI* -399 400-800- 1600- * Brinkmmn Index. Location Rate of smoken Observed Expected Odds ratio (95 CI*) Location Rate of smokers Observed Expected Odds ratio (95# CI*) Confidence interval. Observed and Expected Rates of Smokers and Odds Ratios by location in Squamous Cell Carcinoma Cases Central 93.7 68.3 6.9 4.9-9.8 Central 64.5 17.9 8.3 4.6-15.1 Peripheral 91.1 68.5 4.7 3.4-6.5 Peripheral 55.6 17.0 6.1 3.4-11.1 Central 4.6 1.7 93.7 5.7 28.9 32.0 27.1 Central 32.3 3.2 64.5 19.4 25.8 3.2 Peripheral 6.6 2.3 91.1 4.3 26.2 33.4 27.2 Peripheral 44.4 55.6 11.1 22.2 22.2 5.5 2.0 92.5 5.1 27.6 32.7 27.1 38.8 59.7 13.4 20.9 23.9 squamous cell carcinoma, small cell carcinoma and large cell carcinoma. The distribution of smoking habits for the central and peripheral types of squamous cell carcinoma are presented in Table VI. For men, there were no differences in the rate of smokers and smoking intensity among current smokers. For women, the rate of smokers for the central type was higher than for the peripheral type, and smoking intensity among current smokers was distributed into a higher Brinkman Index for the central type than for the peripheral type. The odds ratios for the central and peripheral types of squamous cell carcinomas are presented in Table VII. The odds ratios were slightly higher for the central type than for the peripheral type, in both sexes, however, the difference was not statistically significant. There was a tendency for the odds ratios in women to be higher than those in Jpn. J. Clan. Oncol.

SEX DIFFERENCES IN SMOKING IN LUNG CANCER Table VIII. Population Attributable Risks and Estimated Incidence Rates in Non-Smokers and Smokers by Histologic Type Histologic Type Squamous cell Population attributable risk Estimated incidence rate (per 100,000) In non-smokers In smokers 76.2 2.5 13.2 6 3.5 10.9 8 0.7 4.9 70.3 0.5 2.2 Histologic Type Squamous cell 1 a eoa u o a a a B 3 n c o u o 10 Population attributable risk 48.4 10.8 66.9 29.7 Estimated incidence rate (per 100,000) In non-smokers 0.8 5.4 0.5 0.3 In smokers 5.6 9.7 6.6 1.2 15 r' A / A / - I E.4 W! / 1900 1920 1940 1960 _^ U.S.A. YEAR E. &W.: EngUnd and W»lo. Sourca: Doll and Pcto (1981),' ' Shimizu (1979) 1 " and Fujimoto (1986).»> Fig. 1. Number of cigarettes consumed in three countries. 1980 18(1) 1988

SOBUE ET AL. Table IX. Summary of Relative Risks for Smoking by Histologic Type Reported in Case-Control Studies Auther Doll Lubin Segi Nakamura Shimizu Tsuchiya Sobue Author Doll Lubin Segi Nakamura Shimizu Tsuchiya Sobue Years of collection 1948-52 1976-80 1962-70 1978-82 1977-82 1974-81 1965-83 Years of collection 1948-52 1976-80 1962-70 1978-82 1977-82 1974-81 1965-83 Place U.K. Europe Place U.K. Europe Cases 872 6,920 296 498 603 183 1,660 Cases Sample size Controls 1,357 13,460 592 498 727 33 128,934* Sample size 61-884 82 84 148 61 423 Controls 108 1,747 164 84 746 51 123,500* Smoking rate in controls 95$ 805*; 90$ 695fc 69& 73# 685fc Smoking rate in controls 485*; 325S 215fc 215*; 75*; 165*; 16% SQ ( 20.9 t 3.6 I 2.1 6.0 4.3 27.8 5.2 SQ ( 6.8 1.7 6.4 7.2 Odds SM 13.1 10.5 ( 3 ( 1 10.3 3.9 7.5 6.9 Odds SM 2.2 7.4 ( 0. 4.5 14.4 ratios LA ) 8.0 6 ) 7 ) 4.4 3.4 5.6 4.1 ratios LA ) 7.2.51 ) * Smoking ratei only were used. t For those wbo smoke 20 or more cigarettes/day. X For those who smoke 19 or le» cigarettes/day. SQ, iquamous cell carcinoma; SM, small cell carcinoma; LA, large cell carcinoma; AD, adenocarcinoma;, not available. men for both types, but the differences were not statistically significant. Population attributable risks and incidence rates in non-smokers are shown in Table VIII. There were great differences in population attributable risks between men and women for all histologic types, mainly due to differences in the rates of smokers. In female adenocarcinoma, only 12.0% could be explained as having arisen from smoking. Incidences rates in non-smokers were estimated to be lower in women than in men for squamous cell carcinoma, but in contrast, higher in women than in men for adenocarcinoma. Discussion 4.0 3.8 AD 0.9 3.5 1.9 0.9 2.8 1.9 3.1 AD 1.9 1.0 1.7 2.9 1.3 The results of the present study have shown the odds ratios for cigarette smoking in each histologic type of lung cancer in women to be similar to those in men except for adenocarcinoma. Previously reported relative risks by histologic type have been summarized in Table IX. 1 ' 14-15 > According to a large-scale case-control study performed in western Europe, 6 ' the relationship between cigarette smoking and lung cancer for women was stronger in squamous cell and small cell carcinoma than in adenocarcinoma, the relationship in adenocarcinoma being the same as for men. For men, we obtained lower values for odds ratios than those in the western Euro- 10 Jpn. J. Clin. Oncol.

SEX DIFFERENCES IN SMOKING IN LUNG CANCER pean series, 0 ' especially for squamous cell carcinoma while, for women, they were approximately the same. Recently, in, Shimizu 9 * and Nakamura 14 ' reported almost the same odds ratios for both men and women as ours, although the value for male squamous cell carcinoma was still lower than that in the western European series. In the study conducted in the 1960s, Segi 8) reported much lower odds ratios for all histologic types in both sexes, in comparison with those from recent studies in. One of the main reasons of this seems to be a difference between the two in study period, that is, Segi's study was conducted from 1962 to 1970, while the cases for our study were mainly collected after 1970. During these periods, the rates of smokers in the general population have steadily decreased in the older age groups in both sexes, as shown in Table I. One of the reasons for the smoking rates around 1960 being higher than they are now in was that, during World War II, cigarettes were rationed equally to all ese, helping to make cigarette smoking generally prevalent. On the other hand, however, the number of cigarettes consumed/day/person has increased in since 1900, and especially since 1960, with the exception of a sudden decrease around the time of World War II, as shown in Fig. 1. In, therefore, it is thought that levels of exposure and the duration of smoking were relatively low and short in the 1960s, compared with those after the 1970s, making the odds ratios lower in the 1960s than in the 1970s, when smoking rate only is used to evaluate the risk from smoking, without taking into consideration smoking intensity or other related factors. In the U.S.A. and U.K. also, cigarette consumption has increased steadily since 1900, this trend being approximately 10-20 years ahead of. So, in the U.S.A. and U.K., exposure levels and duration of smoking were thought to be higher and longer than in for the same period, and the odds ratios to be higher in the U.S.A. and U.K. than in, when smoking rates only were taken into consideration. Although, in squamous cell carcinoma, the odds ratio for the central type was higher than that for the peripheral type, no statistical significance was observed between these types in either sex. Similar odds ratio values being observed between men and women for both central and peripheral types. In this regard, it was confirmed that there was no sex difference in the association between cigarette smoking and squamous cell carcinoma. Shimizu reported no difference between smokers and non-smokers with regard to the original site of distribution in the bronchus in Kreyberg type I lung cancer. 191 Several methodological problems are contained in this study, mainly because the study period was extended for as long as possible in order to collect a sufficient number of cases. First, we used the smoking rate in the general population, as reported by the Tobacco and Salt Public Corporation, for calculating the expected smoking rate for each histologic type. It is generally considered appropriate to collect a control series from the same hospital patients or from their neighborhoods. For this procedure, the most important thing to bear in mind is the question of whether or not the control series is able to represent the population from which patients were derived. In the present study, it was necessary to assume that the distribution of smoking habits amongst the patients was not too different from that amongst all the lung cancer patients in. Compared with the 4,735 cases collected by the Lung Cancer TNM Classification Committee in the 1975-77 period, 201 thought to be the best statistics representing all cases in, there were no great differences with regard to age distribution, histology or smoking habit, although in women, smoking rates tended to be higher among our patients. We thought, therefore, that the assumption mentioned above would not invalidate our analysis all 18(1) 1988 11

SOBUE ET AL. that much. There was a study in which the rates from a large population survey were used as controls in an early stage of lung cancer research. 21 * Second, we could not obtain any information on the controls other than the smoking rates, making it impossible to adjust for other factors. Moreover, because of the long study period, time changes for these factors have to be considered. In fact, the factors related to smoking habit have been changed, i.e. number of cigarettes smoked/day/person, age of onset of smoking, amount of tar contained in each cigarette, filter or no filter, etc. Also, confounding factors other than smoking itself, such as intake of beta-carotene and cholesterol or air pollution conditions, have been changing. These factors could affect each histologic type differently. The exact reason for the odds ratios for smoking being higher lately remains, therefore, unknown. Third, we used the smoking habits for patients and controls observed at the time of diagnosis. Although this is common practice in a case-control study, the time taken to compare smoking habits between cases and controls has to be considered when smoking rates have changed. There is the idea that the effect of smoking on carcinogenesis occurs 10-20 years before the actual detection of clinical lung cancer, so comparisons of smoking habits between cases and controls should be made 10-20 years before the time of diagnosis. In our study, however, we were not able to do this, because the information was not available specifically for patients. If smoking rates 10-20 years before the time of diagnosis had been used, the odds ratios would have become lower than those observed, since exposure levels and histories of smoking among smokers were low, as mentioned above. Fourth, with regard to the age groups used for estimating the smoking rate in the general population, we were unable to obtain a specific smoking rate in the 60s and 70s groups separately. Since the age distribution for lung cancer, especially squamous cell and small cell carcinoma, must be shifted to a higher age than that of the general population, this adjusting procedure could bias the result. Fifth, there was a problem regarding the reliability of the histologic classification. Only 32.8% cases were diagnosed histologically on the basis of material obtained at surgery. Neither could we carry out pathological reviews. Although pathological classifications were determined by one wellexperienced pathologist during the study period, it was unavoidable for some cases to be misclassified. These misclassifications will make risk differences smaller between different histologic types, if they occurred randomly. The same kind of problem arose in determining the bronchial site from which the tumor originated. These problems do not necessarily invalidate the study, although they do suggest caution in the interpretation of the results. Rather they encourage further investigations which can deal with detailed information on both smoking and the other factors regarding histologic type of lung cancer. Acknowledgmen t This work was presented at the 25th Annual Meeting of the Lung Cancer Society, October 30, 1984. References 1) Doll, R., Hill, A. B. and Kreyberg, L. The significance of cell type in relation to the aetiology of lung cancer. Br. J. Cancer 11, 43-48 (1957). 2) Stayner, L. T. and Wegman, D. H. Smoking, occupation and histopathology of lung cancer: A case-control study with the use of the third national cancer survey. JNCI 70, 421-426 (1983). 3) Haenszel, W., Shimkin, M. B. and Mantel, N. A. A retrospective study of lung cancer in women. JNCI 21, 825-842 (1958). 4) Wynder, E. L., Covey, L. S. and Mabuchi, K. Lung cancer in women: Present and future trends. JNCI 51, 391^01 (1973). 5) Hinds, M. W., Stemmerman, G. N., Yang, H.Y., Kolonel, L. N., Lee, J. and Wegner, 12 Jpn. J. Clin. Oncol.

SEX DIFFERENCES IN SMOKING IN LUNG CANCER E. Difference in lung cancer risk from smoking among ese, Chinese and Hawaiian women in Hawaii. Int. J. Cancer 27, 297-302 (1981). 6) Lubin, J. H. and Blot, W. J. Assessment of lung cancer risk factors by histologic category. fnci 73, 383-389 (1984). 7) Wu, A.H., Henderson, B. E., Pike, M. C. and Yu, M. C. Smoking and other risk factors for lung cancer in women. JNCI 74, 747-751 (1985). 8) Segi, M., Kurihara, M., Ishikawa, S. and Haenszel, W. Epidemiological survey on lung cancer and smoking. Lung Cancer 19, 157-165 (1979) (in ese). 9) Shimizu, H., Hisamichi, S., Motomiya, M., Oizumi, K., Konno, K., Hashimoto, K. and Nakada, T. Risk of lung cancer by histologic type among smokers in Miyagi Prefecture. Jpn. J. Clin. Oncol. 16(2), 117-121 (1986). 10) Senbaikosaikai "Annual statistics of Tobacco and Salt Public Corporation," (1965-1983). Senbaikosaikai, Tokyo (in ese). 11) Miettinen, O. S. Estimability and estimation in case-referent studies. Am. J. Epidemiol. 103, 226-235 (1976). 12) Lilienfeld, A.M. and Lilienfeld, D. E. "Foundation of epidemiology 2nd ed.," p. 217 (1980). Oxford University Press, New York. 13) Hanai, A., Kitamura, H., Fukuma, S. and Fujimoto, I. "Cancer incidence in 1975-1979," pp. 18-21 (1984). The research group for population-based cancer registration. 14) Nakamura, M., Hanai, A., Fujimoto, I., Matsuda, M. and Tateishi, R. Relationship between smoking and four major histologic types of lung cancer. Lung Cancer 26(2), 137-148 (1986) (in ese). 15) Tsuchiya, E., Kitagawa, T., Matsubara, T., Nakagawa, K., Kinoshita, I. and Sugano, H. The Influence of Smoking on the Development of Various Subtypes of Pulmonary Carcinoma. Jap. J. Cancer Clin. 33(1), 27-34 (1987) (in ese). 16) Doll, R. and Peto, R. The causes of cancer, Appendix E: Trends in lung cancer death rates in relation to cigarette usage and tar yields. fnci 66, 1292-1305 (1981). 17) Shimizu, H. Smoking habits in. Smoking and Health 1(3), 7 (1979) (in ese). 18) Fujimoto, I. Trends of lung cancer incidence in in "Progress in cancer clinics. No. 6 Lung cancer," eds. K. Suemasu and T. Yoneyama pp. 168-175 (1986). Medical View, Co. Ltd., Tokyo. 19) Shimizu, H., Tominaga, S., Nishimura, M. and Urata, A. Comparison of Clinico- Epidemiological Features of Lung Cancer Patients with and without a History of Smoking. Jpn. J. Clin. Oncol. 14(4), 595-600 (1984). 20) Yoshimura, K. and Yamashita, N. Clinical Statistical Observation of 4,931 Cases with Lung Cancer by Histological Types. Result of Field Study in. Lung Cancer 22(1), 1-17 (1982). 21) Kreyberg, L. Lung cancer and tobacco smoking in Norway. Br. J. Cancer 9, 495-510 (1955). 18(1) 1988 13