- Email: [email protected]
Cigarette smoking and cancer incidence risk in adult men: National Health Insurance Corporation Study
Cancer Detection and Prevention 29 (2005) 15–24 www.elsevier.com/locate/cdp
Cigarette smoking and cancer incidence risk in adult men: National Health Insurance Corporation Study Young Ho Yun MD, PhDa, Kyu Won Jung MPha, Jong-Myon Bae MD, PhDa, Jin Soo Lee MDa, Soon Ae Shin MPhb, Sang Min Park MD, MPhc, Taiwoo Yoo MD, PhDc,*, Bong Yul Huh MD, PhDc a
Research Institute and Hospital, National Cancer Center, Quality of Cancer Center, 809 Madu-dong, Ilsan-gu, Goyang-si, Gyeonggi-do 411-769, Republic of Korea b National Health Insurance Corporation, 168-9 Yeomri-dong, Mapo-gu, Seoul 121-749, Republic of Korea c Department of Family Medicine, Seoul National University College of Medicine, 28 Youngon-dong, Chongno-gu, Seoul 110-799, Republic of Korea Accepted 16 August 2004
Abstract We analyzed risk while adjusting for age, body mass index, frequency of moderate physical activity, alcohol consumption, preference for vegetables versus meats, and frequency of meat consumption in a multivariate analysis and based our findings on not mortality data but incidence data. 733,134 Korean men who were 30 years old or older, insured by the National Health Insurance Corporation, and had a medical evaluation in 1996 were included in the study and followed up through 2000. During the 4-year follow-up period of 3,590,872 person–years, we identified 7204 new cases. We used the Cox proportional hazards model to estimate adjusted relative risks (aRRs), 95% confidence intervals (CIs). The association of current cigarette smoking was significantly stronger as compared with never smokers; aRR was 1.49 (95% CI = 1.39–1.59) for all cancers, 4.46 (2.32–8.57) for esophageal, 3.83 (2.97–4.94) for lung, 3.01 (1.58–5.72) for laryngeal, 2.24 (1.48–3.39) for urinary bladder, 1.62 (1.42–1.84) for gastric, 1.75 (1.12–2.74) for oral and pharyngeal, 1.58 (0.97–2.27) for pancreatic, and 1.50 (1.29– 1.74) for liver cancer. Our findings, based on incidence data, confirmed that differences in smoking habit were responsible for most of the differences observed in smoking-related cancers. # 2004 International Society for Preventive Oncology. Published by Elsevier Ltd. All rights reserved. Keywords: Cigarette smoking; Cancer incidence; Risk; Cohort study
1. Introduction Epidemiologic studies have consistently shown an association between smoking and cancer of the lung, larynx, mouth, esophagus, urinary bladder, kidney, pancreas, and cervix [1–7]. Several reports also suggest an association between smoking and cancer of the stomach, liver, colon, and rectum [8–10]. Most large cohort studies collect detailed smoking histories, but variable information on other potential risk factors. Consultants for the tobacco industry have speculated that confounding may cause studies to exaggerate the risks attributed to smoking [11], although * Corresponding author. Tel.: +82 276 03332; fax: +82 276 63276. E-mail address: [email protected] (T. Yoo).
there is little evidence that this is occurs. Recently, US Cancer Prevention Study (CPS) II suggested that multiple covariates had little impact on the relative and attributable risk for cancer mortality after adjustments were made for age and sex [12]. Since that study was based on cancer mortality data, the effect of smoking on cancer incidence may have been confounded by its effect on other fatal diseases, such as cardiovascular disease [13]. Fewer large prospective studies have assessed the disease risks associated with smoking in Asia, where the widespread consumption of manufactured cigarettes began more recently than in North American and Northern Europe. Our study has the additional strengths of being able to measure cancer incidence, which is informative for sites with higher survival rates, and ability to control for several other major potential risk factors. It is
0361-090X/$30.00 # 2004 International Society for Preventive Oncology. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.cdp.2004.08.006
16
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
uncertain whether the CPS II results are applicable to groups with different genetic and environmental backgrounds [11,14]. In Korea, the National Health Insurance Corporation (NHIC) has been providing health insurance to government employees and teachers since 1980 along with biennial health examinations that include height, weight, and blood pressure measurements, chest radiography, blood counts, and blood chemistries. In addition, a self-administered questionnaire collects information regarding medical history, current health status, and individual life styles regarding tobacco and alcohol consumption, dietary preferences, and leisure-time physical activity. To analysis the association of smoking with the incidence of all cancers and the cancers of major organ sites for age, body mass index, frequency of moderate physical activity, alcohol consumption, preference for vegetables versus meats, and frequency of meat consumption, we performed National Health Insurance cohort study.
2. Material and methods 2.1. Cohort and follow-up The study subjects derived from 1,228,817 (980,841 men and 337,426 women) government employees and teachers who participated in a national health examination program begun in 1996, and they constituted the cohort for the National Health Insurance Cooperation Study. The NIHC permitted us to use their data for the present study. We excluded women from the present study because their smoking rate was too low (0.3% of the study population), and we excluded 165,915 men who were less than 30 years old. Among the eligible 814,926 male participants, 54,585 were excluded due to lack of information on life style, 23,835 due to lack of information on smoking status, and 3372 who were prevalent cases and documented to have received medical service in 1996. For the analysis of the impact of smoking duration, we excluded 208 current smokers and 10,753 former smokers for whom data on smoking duration was missing. The final cohort consisted of 733,134 Korean men aged 30 years or older. The incident cancer cases were identified from the Korean Central Cancer Registry (KCCR) and six regional cancer registries (RCRs). For the KCCR, a nationwide hospital-based cancer registry system, professionally trained and certified medical recorders collected all relevant data using a standardized manual based on the International Classification of Disease for Oncology (ICD-O) after reviewing the pathology reports and hospital discharge summaries [15]. The KCCR-affiliated hospitals included 94% of all university hospitals and 96% of all resident training hospitals in Korea [16]. Additionally, using a computerized search program that covered the period of 1997–2000, we investigated the RCR data from six major
cities (Seoul, Busan, Incheon, Daejeon, Kwangju, and Daegu) that collect cancer incidence data independent of the KCCR program. For those individuals with more than one event, we included the first diagnosis of cancer in the analyses. In general, microscopic examination (83.0%) was the leading methods of diagnosis, followed by clinical investigation (14.1%), specific biochemical/immunologic tests (1.0%), and clinical judgment (1.9%). We defined the follow-up period as the interval between the initial examination and the diagnosis of cancer. We followed subjects without cancer up to December 31, 2000. During the 4-year follow-up period of 3,590,872 person–years, we identified 7204 new cases. 2.2. Classification of smoking behavior Smoking status was classified as current, former, or never smoker on the basis of the response to the following questions on the initial questionnaire provided in 1996: ‘‘Do you smoke cigarettes now?’’, ‘‘How many cigarettes per day do you smoke?’’, and ‘‘How many years have you smoked in your lifetime?’’ Current smokers were further subgrouped by the average number of cigarettes smoked per day (1–9, 10–19, or 20 or more) and duration of smoking (1–19, 20– 29, or 30 or more years). We classified smoking behavior in order to compare our data with those of previous studies. We classified the smoking amount because previous studies showed that even light smoking increased cancer risk, and the smoking duration as increased cancer risk was observed among individuals smoking for 20 years or more [4,6,9,10,14]. 2.3. Statistical analysis We calculated the incidence rates per 100,000 person– years according to the smoking status at baseline by dividing the number of cancers by the total person–years of the follow-up. We used the never smokers as the reference group to calculate the relative risks of cancers due to smoking. We used Poisson regression to estimate the adjusted relative risks (aRRs) associated with current or former smoking and their 95% confidence intervals (CIs), first adjusting only for age at enrollment and then for other multiple covariates. We adjusted multivariate models for age at enrollment (30–39, 40–49, 50 or older), body mass index (BMI; <18.5, 18.5– 24.9, 25–29.9, 30 or more), frequency of moderate physical activity in leisure time (0, 1–2, 3–4, 5 or more times per week), alcohol consumption (number of drinks per week), preference for vegetables versus meats (vegetables, mixed, meats), and frequency of meat consumption (0–1, 2–3, 4 or more times per week). The BMI (kg/m2) calculated from height and weight at enrollment was used as a measure of relative body fat. The BMI score was divided into four groups according to the World Health Organization criteria for thinness or being overweight [17]. Moderate physical activity was defined as intense enough to produce sweat. For
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
17
Table 1 Sociodemographic characteristics of National Health Insurance Cooperation Study cohort by smoking status (n = 733,134) Characteristics Mean age (year)
Never (n = 164,447) 43.6
Former (n = 164,204) 44.7
Current (n = 404,483)a 42.4
Place of residence (%) Metropolis City Country
76,460 (46.5) 58,411 (35.5) 29,576 (18.0)
75,666 (46.1) 59,056 (36.0) 29,482 (18.0)
171,503 (42.4) 148,890 (36.8) 84,090 (20.8)
Frequency of alcohol consumption (%) Never 2–3 per month 1–2 per week 3–4 per week =5 per week
63,570 (38.8) 39,926 (24.4) 43,495 (26.5) 13,386 (8.2) 3,566 (2.2)
51,490 (31.6) 40,068 (24.6) 47,148 (29.0) 18,498 (11.4) 5,542 (3.4)
79,193 (19.7) 86,751 (21.6) 144,188 (35.8) 69,099 (17.2) 23,131 (5.8)
2,907 (1.8) 111,450 (67.8) 47,962 (29.2) 2,061 (1.3)
2,213 (1.4) 108,819 (66.3) 51,066 (31.1) 2,042 (1.2)
8,626 (2.1) 281,958 (69.7) 108,767 (26.9) 4,991 (1.2)
66,758 (41.1) 64,027 (39.5) 17,688 (10.9) 13,826 (8.5)
62,161 (38.4) 64,871 (40.1) 19,957 (12.3) 14,732 (9.1)
174,369 (43.6) 163,175 (40.8) 37,030 (9.3) 25,647 (6.4)
Preference of food saltiness (%) Low salt Normal Salty
34,351 (20.9) 108,389 (66.1) 21,283 (13.0)
31,432 (19.2) 106,095 (64.9) 25,956 (15.9)
54,636 (13.6) 254,231 (63.0) 94,467 (23.4)
Food preferenceb (%) Vegetables Mixed Meats
33,605 (20.5) 121,050 (73.9) 9,241 (5.6)
34,422 (21.1) 117,912 (72.1) 11,111 (6.8)
81,340 (20.2) 287,606 (71.4) 34,037 (8.5)
76,710 (47.0) 80,767 (49.5) 5,630 (3.5)
73,832 (45.3) 80,582 (49.5) 8,499 (5.2)
178,241 (44.4) 207,694 (51.7) 15,685 (3.9)
BMI (%) <18.5 kg/m2 18.5–24.9 kg/m2 25–29.9 kg/m2 =30 kg/m2 Physical activity (%) Never 1–2 per week 3–4 per week =5 per week
Frequency of meat consumption (%) =1 per week 2–3 per week =4 per week a b
Excluding the missing data. Preference for vegetables vs. meats.
stomach cancer, we adjusted also for preference of food saltiness (low salt, normal, salty) because some studies suggest that food saltiness might be associated with development of stomach cancer [18–20]. To calculate the population attributable risk (PAR) of cancer [21], we used the smoking prevalence data of the 1998 Korea Health Survey (performed by the Ministry of Health and Welfare), which reported 67.5% of men as being current smokers, 15.7% former smokers, and 16.7% as never smokers [22]. All confidence intervals are at 95%, and we considered a P-value of 5% as significant. All statistical tests were twosided. We used SAS statistical package version 8.1 [23].
3. Results Among the 733,134 subjects, 404,483 (55.2%) were current smokers, 164,204 (22.4%) former smokers, and 164,447 (22.4%) never smokers. Among the current
smokers, 16.1, 62.8, and 21.1% smoked 1–9, 10–19, and 20 or more cigarettes per day, respectively, and smoking duration was 1–19 years (58.6%), 20–29 years (28.4%), and 30 years or more (13.0%). Among former smokers, 82.2% reported that they had smoked for 1–19 years, 13.3% for 20– 29 years, and 4.5% for 30 years or more. Table 1 shows the smoking statuses and characteristics of individuals in the Korean National Health Insurance Cooperation study. Major cancer sites in descending order were stomach, liver, bronchus or lung, rectum, colon, kidney, urinary bladder, mouth or pharynx, pancreas, esophagus, gall bladder or bile duct, hematopoietic system, larynx, and others (Table 2). The age-adjusted RR for all types of cancer for former smokers as compared with never smokers was 1.32 (95% CI = 1.23–1.42) and 1.51 (95% CI = 1.42–1.61) for current smokers (Table 2). We observed a statistically significant age-adjusted RR in both former and current smokers for
18
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
Table 2 Estimated relative risk and 95% confidence intervals of cancers by smoking statusa (n = 733,134) Ptrendb
Cancer site
Never
Former
Current
Current, cigarettes smoked per day 1–9
10–19
=20
All (n = 7204) Agec Multivariated
(n = 1284) 1.0 1.0
(n = 1831) 1.32 (1.23, 1.42) 1.33 (1.24, 1.43)
(n = 4,089) 1.51 (1.42, 1.61) 1.49 (1.39, 1.59)
(n = 701) 1.46 (1.34, 1.61) 1.43 (1.30, 1.57)
(n = 2,496) 1.49 (1.39, 1.59) 1.46 (1.36, 1.57)
(n = 886) 1.64 (1.50, 1.78) 1.61 (1.47, 1.76)
NS NS
Mouth and pharynx (n = 172) Agec Multivariated
(n = 25) 1.0 1.0
(n = 41) 1.53 (0.93, 2.51) 1.53 (0.93, 2.52)
(n = 106) 1.91 (1.24, 2.96) 1.75 (1.12, 2.74)
(n = 13) 1.37 (0.70, 2.68) 1.24 (0.62, 2.48)
(n = 67) 1.94 (1.23, 3.08) 1.77 (1.10, 2.83)
(n = 26) 2.28 (1.31, 3.95) 2.15 (1.23, 3.78)
NS NS
Esophagus (n = 163) Agec Multivariated
(n = 10) 1.0 1.0
(n = 29) 2.64 (1.29, 5.42) 2.22 (1.07, 4.60)
(n = 124) 6.19 (3.25, 11.8) 4.46 (2.32, 8.57)
(n = 16) 4.34 (1.97, 9.57) 3.65 (1.65, 8.09)
(n = 81) 6.54 (3.39, 12.6) 4.71 (2.42, 9.18)
(n = 26) 6.65 (3.20, 13.8) 4.16 (1.96, 8.82)
NS NS
Stomach (n = 1949) Agec Multivariatee
(n = 319) 1.0 1.0
(n = 496) 1.44 (1.25, 1.66) 1.45 (1.25, 1.67)
(n = 1134) 1.68 (1.48, 1.90) 1.62 (1.42, 1.84)
(n = 202) 1.69 (1.42, 2.02) 1.60 (1.33, 1.93)
(n = 687) 1.63 (1.43, 1.87) 1.59 (1.38, 1.82)
(n = 244) 1.79 (1.52, 2.12) 1.72 (1.45, 2.05)
NS NS
Colon (n = 417) Agec Multivariated
(n = 99) 1.0 1.0
(n = 148) 1.39 (1.08, 1.79) 1.37 (1.06, 1.77)
(n = 170) 0.83 (0.65, 1.06) 0.81 (0.63, 1.05)
(n = 36) 0.98 (0.67, 1.43) 0.97 (0.66, 1.43)
(n = 102) 0.80 (0.61, 1.05) 0.78 (0.59, 1.04)
(n = 32) 0.78 (0.53, 1.17) 0.76 (0.51, 1.15)
NS NS
Rectum (n = 453) Agec Multivariated
(n = 106) 1.0 1.0
(n = 131) 1.15 (0.89, 1.48) 1.17 (0.91, 1.52)
(n = 216) 0.96 (0.76, 1.22) 0.97 (0.76, 1.24)
(n = 38) 0.96 (0.66, 1.39) 0.95 (0.65, 1.39)
(n = 131) 0.94 (0.73, 1.22) 0.95 (0.73, 1.24)
(n = 47) 1.04 (0.74, 1.48) 1.05 (0.74, 1.50)
NS NS
Liver (n = 1434) Agec Multivariated
(n = 255) 1.0 1.0
(n = 403) 1.45 (1.24, 1.70) 1.53 (1.30, 1.79)
(n = 776) 1.41 (1.22, 1.62) 1.50 (1.29, 1.74)
(n = 192) 2.01 (1.67, 2.42) 2.20 (1.81, 2.67)
(n = 472) 1.38 (1.18, 1.60) 1.48 (1.26, 1.73)
(n = 111) 0.98 (0.78, 1.23) 0.98 (0.77, 1.23)
< 0.01 < 0.01
GB and biliary system (n = 161) Agec Multivariated
(n = 39) 1.0 1.0
(n = 40) 0.94 (0.60, 1.46) 0.94 (0.60, 1.47)
(n = 82) 1.02 (0.69, 1.49) 0.94 (0.63, 1.41)
(n = 9) 0.62 (0.30, 1.29) 0.58 (0.27, 1.25)
(n = 56) 1.12 (0.74, 1.69) 1.08 (0.70, 1.66)
(n = 17) 1.06 (0.60, 1.88) 0.86 (0.46, 1.59)
NS NS
Pancreas (n = 172) Agec Multivariated
(n = 28) 1.0 1.0
(n = 52) 1.71 (1.08, 2.70) 1.68 (1.68, 1.68)f
(n = 92) 1.63 (1.07, 2.50) 1.58 (1.58, 1.58)f
(n = 16) 1.56 (0.84, 2.88) 1.48 (1.48, 1.48)f
(n = 52) 1.49 (0.94, 2.36) 1.44 (1.44, 1.44)f
(n = 23) 2.08 (1.20, 3.63) 2.12 (2.12, 2. 12)f
NS NS
Larynx (n = 103) Agec Multivariated
(n = 11) 1.0 1.0
(n = 14) 1.17 (0.53, 2.57) 1.12 (0.51, 2.46)
(n = 78) 3.55 (1.88, 6.68) 3.01 (1.58, 5.72)
(n = 5) 1.25 (0.43, 3.59) 1.14 (0.39, 3.29)
(n = 47) 3.47 (1.80, 6.70) 2.94 (1.50, 5.74)
(n = 26) 6.07 (2.99, 12.3) 5.20 (2.52, 10.7)
<0.01 <0.01
Lung (n = 787) Agec Multivariated
(n = 73) 1.0 1.0
(n = 150) 1.88 (1.42, 2.49) 1.85 (1.39, 2.48)
(n = 564) 3.94 (3.08, 5.02) 3.83 (2.97, 4.94)
(n = 50) 1.87 (1.31, 2.68) 1.82 (1.25, 2.64)
(n = 326) 3.69 (2.86, 4.76) 3.62 (2.78, 4.71)
(n = 187) 6.82 (5.20, 8.95) 6.66 (5.02, 8.84)
<0.01 <0.01
Hematopoietic system (n = 148) Agec Multivariated
(n = 39) 1.0 1.0
(n = 29) 0.72 (0.45, 1.17) 0.71 (0.44, 1.17)
(n = 80) 0.90 (0.61, 1.32) 0.93 (0.63, 1.39)
(n = 20) 1.33 (0.77, 2.27) 1.33 (0.77, 2.33)
(n = 41) 0.73 (0.47, 1.14) 0.77 (0.49, 1.21)
(n = 19) 1.04 (0.60, 1.80) 1.09 (0.62, 1.92)
NS NS
Kidney (n = 211) Agec Multivariated
(n = 53) 1.0 1.0
(n = 52) 0.91 (0.62, 1.34) 0.96 (0.65, 1.42)
(n = 106) 0.94 (0.67, 1.31) 0.94 (0.66, 1.32)
(n = 10) 0.50 (0.25, 1.05) 0.53 (0.27, 1.05)
(n = 77) 1.10 (0.77, 1.56) 1.08 (0.75, 1.56)
(n = 19) 0.84 (0.49, 1.41) 0.86 (0.50, 1.47)
NS NS
Urinary bladder (n = 208) Agec Multivariated
(n = 29) 1.0 1.0
(n = 42) 1.35 (0.84, 2.17) 1.31 (0.80, 2.13)
(n = 137) 2.30 (1.54, 3.44) 2.24 (1.48, 3.39)
(n = 21) 1.95 (1.11, 3.42) 2.05 (1.16, 3.62)
(n = 85) 2.30 (1.51, 3.51) 2.23 (1.44, 3.45)
(n = 31) 2.65 (1.59, 4.40) 2.47 (1.46, 4.18)
NS NS
Others (n = 826) Agec Multivariated
(n = 198) 1.0 1.0
(n = 204) 0.97 (0.80, 1.18) 0.98 (0.80, 1.19)
(n = 424) 1.01 (0.85, 1.19) 1.01 (0.85, 1.20)
(n = 73) 0.99 (0.75, 1.29) 1.01 (0.77, 1.33)
(n = 272) 1.04 (0.86, 1.25) 1.04 (0.86, 1.26)
(n = 78) 0.92 (0.71, 1.20) 0.90 (0.68, 1.18)
NS NS
(n=) indicates cancer incidence, NS: not significant, GB: gall bladder. a Relative risk with Poisson regression. b For trends in the current smokers. c Adjusted for age. d Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preferences for vegetables and meats, and frequency of meat consumption. e Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preference for food saltiness, preferences for vegetables and meats, and frequency of meat consumption. f Standard error is zero.
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
cancer of the esophagus, stomach, liver, pancreas, and lung, and for current smokers for cancer of the mouth and pharynx, larynx, and urinary bladder. In contrast, RR of colon cancer was statistically significant only in former smokers. There was no significant association of smoking with the risk of rectal, gall bladder, kidney, hematopoietic, and other cancers. Multivariate analysis revealed only minor decreases in the RR estimates obtained with age-adjusted analysis. In current smokers, there were statistically significant positive linear trends in aRRs with increased smoking duration for all cancers (Ptrend < 0.01) (Table 3). Even when adjusted for age and other related factors, there were significant dose-response relationships between smoking duration and incidence of all cancers in current smokers. We also observed statistically significant positive linear trends in aRRs for esophageal (Ptrend < 0.01), stomach (Ptrend < 0.01), laryngeal (P < 0.05), lung (P < 0.01), and urinary bladder cancer (Ptrend < 0.01) with increased smoking duration. We observed a statistically significant positive linear trend in RR of rectal cancer (P < 0.01), but the RR was not significantly increased over the never smoker. In former smokers, the age-adjusted RR of all cancers increased significantly with smoking duration (Ptrend < 0.01) (Table 4). Significant positive linear trends in aRRs were also observed for colon (Ptrend < 0.05), liver (Ptrend < 0.01), lung (Ptrend < 0.01), and urinary bladder cancer (Ptrend < 0.01), and a marginally significant positive linear trend in RR was observed for esophageal cancer (Ptrend = 0.07). The multivariate adjustment showed minor changes in the aRR estimates of smoking for all categories of cancer. The dose–response relationship between smoking and each category of cancer site remained unchanged when it was significant after adjustment for age. The age-adjusted PAR was highest for esophageal cancer (98.3%) followed by lung cancer (78.7%), laryngeal cancer (59.5%), and urinary bladder cancer (52.0%) (Table 5). In addition, more than one-third of the cases of oropharyngeal cancer (45.8%), pancreatic cancer (39.9%), and stomach cancer (36.0%) and one-quarter of liver cancer cases (28.3%) were attributable to smoking. Overall, 30.4% of all cancers were attributable to smoking. The multivariate adjusted PAR for all cancers changed negligibly to 29.8% from the age-adjusted PAR of 30.4%. For esophageal cancer however, multivariate adjustment significantly reduced the PAR estimate from 98.3% to 86.1%. For oral and pharyngeal, stomach, pancreatic, laryngeal, lung, and urinary bladder cancer, multivariate adjustment resulted in minor PAR changes. Nevertheless, lung, esophagus, larynx, and urinary bladder remained the four major sites of smoking-related cancers, in that order.
4. Discussion In this large cohort study based on incidence data, as in similar studies based on mortality data, lung, esophageal,
19
laryngeal, urinary bladder, oral and pharyngeal, stomach, pancreatic, and liver cancer in Korean men were associated with smoking. After adjusting for age, the proportion of all cancers that were attributable to smoking was 30.4%, and further adjustments for body mass index, frequency of moderate physical activity, alcohol consumption, preference for vegetables versus meats, and frequency of meat consumption had little impact on the risk. Our risk estimate was lower than the CPS II estimate, which attributed 45% of all cancer deaths in men to smoking [12], but similar to the 2002 Japan Public Heath Center report, which attributed 25% of cancer mortality to smoking [24]. Our study confirms the previous finding [25–27] that after controlling for diet and alcohol drinking, in a socially homogenous group, stomach cancer is weakly associated with smoking. In addition, the risk increases with smoking duration even after multivariate adjustments, which supports the association of smoking with stomach cancer. Our study also agrees with previous studies that showed no consistent association between cigarette smoking and colorectal cancer [28–33]. Recent studies, however, have proposed an induction period of 30–40 years between a genotoxic exposure and the diagnosis [34,35]. Thus, a longer follow-up period may be needed. We observed a weak but significant association between smoking and liver cancer, and the association remained after multivariate adjustment, which included alcohol consumption. Since we did not include hepatitis status in our analysis, we cannot draw any conclusions. In previous studies, the association between smoking and liver cancer risk was not consistent [7,14,36– 39]. One study suggested that in areas where liver cancer is not endemic, such as in Western countries, smoking was a more important cause of liver cancer [40], but confounding effects cannot be excluded. It is possible, for example, that patients with hepatitis B infections reduce their smoking as their disease progresses. We need to clarify the relationship between liver cancer and number of cigarettes smoked by investigating the association after adjusting for the smoking status of hepatitis B carriers. Our study did not show an increase in kidney cancer risk among current smokers. This is in contrast with urinary bladder cancer risk, which more than doubled. Since the incidence of kidney cancer is low in Asia [41] and its association with smoking has not been established in Asia, further studies are needed in Asian cohorts. Our results were similar to those of Japanese studies except that we found a lower risk for laryngeal cancer [24]. The relative risk of most cancers, however, especially oral and pharyngeal, laryngeal, and lung cancer were lower than the risks found in large Western cohort studies. Also, our study did not show a consistent association of smoking with colorectal cancer. Several explanations for our lower RRs are possible, especially for the US studies. First, even though the consumption of cigarettes per capita has increased in Korea since 1980, when 69.4% of men smoked, cigarette consumption was low in the 1960s and 1970s [42,43].
20
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
Table 3 Estimated relative risk of cancer and 95% confidence intervals for current smokers by smoking durationa Ptrendb
Cancer site
Never
Years of smoking duration 1–19
20–29
30
All Agec Multivariated
(n = 1284) 1.0 1.0
(n = 1216) 1.37 (1.26, 1.48) 1.36 (1.25, 1.48)
(n = 1361) 1.42 (1.32, 1.54) 1.40 (1.29, 1.52)
(n = 1506) 1.72 (1.60, 1.86) 1.68 (1.55, 1.82)
<0.01 <0.01
Mouth and pharynx Agec Multivariated
(n = 25) 1.0 1.0
(n = 33) 1.59 (0.92, 2.74) 1.42 (0.81, 2.48)
(n = 45) 2.27 (1.38,3.74) 2.17 (1.31, 3.60)
(n = 28) 1.88 (1.08, 3.27) 1.67 (0.94, 2.96)
NS NS
Esophagus Agec Multivariated
(n = 10) 1.0 1.0
(n = 24) 5.09 (2.40, 10.8) 4.21 (1.96, 9.03)
(n = 36) 4.91 (2.42, 9.95) 3.64 (1.77, 7.46)
(n = 63) 7.63 (3.90, 14.9) 5.07 (2.56, 10.1)
<0.01 <0.05
Stomach Agec Multivariatee
(n = 319) 1.0 1.0
(n = 358) 1.57 (1.34, 1.84) 1.55 (1.32, 1.83)
(n = 371) 1.55 (1.33, 1.81) 1.52 (1.29, 1.77)
(n = 404) 1.89 (1.63, 2.20) 1.79 (1.53, 2.10)
<0.01 <0.01
Colon Agec Multivariated
(n = 99) 1.0 1.0
(n = 59) 0.89 (0.63, 1.25) 0.87 (0.62, 1.23)
(n = 45) 0.63 (0.44, 0.90) 0.61 (0.42, 0.88)
(n = 66) 0.96 (0.70, 1.32) 0.96 (0.69, 1.33)
NS NS
Rectum Agec Multivariated
(n = 106) 1.0 1.0
(n = 62) 0.81 (0.58, 1.12) 0.80 (0.57, 1.13)
(n = 76) 0.98 (0.72, 1.32) 1.00 (0.74, 1.36)
(n = 78) 1.10 (0.82, 1.49) 1.12 (0.82, 1.52)
<0.01 <0.01
Liver Agec Multivariated
(n = 255) 1.0 1.0
(n = 253) 1.49 (1.24, 1.78) 1.59 (1.32, 1.92)
(n = 282) 1.33 (1.12, 1.58) 1.42 (1.19, 1.70)
(n = 240) 1.42 (1.18, 1.70) 1.49 (1.23, 1.80)
NS NS
GB and biliary system Agec Multivariated
(n = 39) 1.0 1.0
(n = 19) 0.85 (0.48, 1.51) 0.83 (0.46, 1.50)
(n = 31) 1.03 (0.64, 1.67) 1.00 (0.60, 1.65)
(n = 32) 1.12 (0.70, 1.81) 0.94 (0.57, 1.56)
NS NS
Pancreas Agec Multivariated
(n = 28) 1.0 1.0
(n = 27) 1.79 (1.03, 3.12) 1.83 (1.83, 1.83)f
(n = 29) 1.49 (0.88, 2.53) 1.48 (1.48, 1.48)f
(n = 35) 1.61 (0.97, 2.66) 1.48 (1.48, 1.48)f
NS NS
Larynx Agec Multivariated
(n = 11) 1.0 1.0
(n = 14) 2.16 (0.94, 4.93) 1.95 (0.85, 4.49)
(n = 30) 3.79 (1.87, 7.66) 3.12 (1.52, 6.39)
(n = 34) 4.24 (2.12, 8.48) 3.67 (1.81, 7.44)
<0.05 <0.05
Lung Agec Multivariated
(n = 73) 1.0 1.0
(n = 98) 2.55 (1.86, 3.49) 2.60 (1.88, 3.60)
(n = 183) 3.65 (2.77, 4.80) 3.55 (2.67, 4.73)
(n = 282) 4.89 (3.77, 6.35) 4.78 (3.64, 6.27)
<0.01 <0.01
Hematopoietic system Agec Multivariated
(n = 39) 1.0 1.0
(n = 38) 0.88 (0.54, 1.42) 0.90 (0.55, 1.46)
(n = 23) 0.87 (0.51, 1.49) 0.93 (0.54, 1.59)
(n = 19) 0.96 (0.54, 1.70) 1.01 (0.55, 1.83)
NS NS
Kidney Agec Multivariated
(n = 53) 1.0 1.0
(n = 30) 0.76 (0.48, 1.23) 0.76 (0.47, 1.25)
(n = 37) 0.93 (0.61, 1.43) 0.89 (0.57, 1.40)
(n = 39) 1.12 (0.73, 1.71) 1.15 (0.74, 1.78)
NS NS
Urinary bladder Agec Multivariated
(n = 29) 1.0 1.0
(n = 36) 1.66 (0.99, 2.79) 1.68 (0.99, 2.86)
(n = 49) 2.50 (1.56, 3.99) 2.39 (1.47, 3.87)
(n = 52) 2.68 (1.68, 4.28) 2.63 (1.62, 4.27)
<0.01 <0.05
Others Agec Multivariated
(n = 198) 1.0 1.0
(n = 165) 1.05 (0.85, 1.32) 1.03 (0.82, 1.30)
(n = 124) 0.91 (0.72, 1.14) 0.89 (0.70, 1.13)
(n = 134) 1.05 (0.84, 1.32) 1.10 (0.88, 1.39)
NS NS
(n=) indicates cancer incidence, NS: not significant, GB: gall bladder. a Relative risk with Poisson regression. b For trends in the current smokers. c Adjusted for age. d Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preferences for vegetables and meats, and frequency of meat consumption. e Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preference for food saltiness, preferences for vegetables and meats, and frequency of meat consumption. f Standard error is zero.
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
21
Table 4 Estimated relative risk of cancer and 95% confidence intervals for former smokers by smoking durationa Ptrendb
Cancer site
Never
Years of smoking duration 1–19
20–29
=30
All Agec Multivariated
(n = 1284) 1.0 1.0
(n = 1149) 1.27 (1.17, 1.37) 1.28 (1.18, 1.39)
(n = 335) 1.36 (1.20, 1.53) 1.38 (1.22, 1.55)
(n = 195) 1.57 (1.35, 1.82) 1.54 (1.32, 1.80)
<0.01 <0.01
Mouth and pharynx Agec Multivariated
(n = 25) 1.0 1.0
(n = 27) 1.49 (0.86, 2.56) 1.49 (0.87, 2.57)
(n = 12) 2.64 (1.32, 5.27) 2.63 (1.31, 5.26)
(n = 2) 1.01 (0.24, 4.30) 1.00 (0.23, 4.27)
NS NS
Esophagus Agec Multivariated
(n = 10) 1.0 1.0
(n = 16) 2.35 (1.06, 5.17) 2.00 (0.89, 4.51)
(n = 7) 3.27 (1.24, 8.59) 2.97 (1.13, 7.82)
(n = 6) 5.36 (1.94, 14.8) 3.85 (1.31, 11.4)
0.07 0.07
Stomach Agec Multivariatee
(n = 319) 1.0 1.0
(n = 311) 1.38 (1.18, 1.61) 1.38 (1.17, 1.62)
(n = 89) 1.46 (1.16, 1.85) 1.50 (1.18, 1.90)
(n = 47) 1.55 (1.13, 2.11) 1.53 (1.11, 2.10)
NS NS
Colon Agec Multivariated
(n = 99) 1.0 1.0
(n = 95) 1.37 (1.04, 1.82) 1.36 (1.02, 1.80)
(n = 23) 1.21 (0.76, 1.90) 1.15 (0.72, 1.83)
(n = 21) 2.08 (1.29, 3.36) 2.08 (1.29, 3.37)
0.05 0.06
Rectum Agec Multivariated
(n = 106) 1.0 1.0
(n = 89) 1.19 (0.90, 1.57) 1.21 (0.91, 1.61)
(n = 24) 1.20 (0.77, 1.87) 1.23 (0.78, 1.92)
(n = 6) 0.59 (0.26, 1.36) 0.61 (0.27, 1.41)
NS NS
Liver Agec Multivariated
(n = 255) 1.0 1.0
(n = 262) 1.43 (1.21, 1.70) 1.49 (1.25, 1.78)
(n = 72) 1.43 (1.10, 1.85) 1.51 (1.16, 1.97)
(n = 29) 1.28 (0.87, 1.88) 1.31 (0.88, 1.95)
<0.01 NS
GB and biliary system Agec Multivariated
(n = 39) 1.0 1.0
(n = 24) 0.88 (0.53, 1.47) 0.89 (0.53, 1.49)
(n = 6) 0.75 (0.32, 1.78) 0.74 (0.31, 1.75)
(n = 6) 1.47 (0.62, 3.51) 1.38 (0.57, 3.30)
NS NS
Pancreas Agec Multivariated
(n = 28) 1.0 1.0
(n = 31) 1.63 (0.98, 2.72) 1.65 (1.65, 1.65)f
(n = 13) 2.28 (1.18, 4.40) 1.98 (1.98, 1.98)f
(n = 5) 1.54 (0.59, 4.01) 1.56 (1.56, 1.56)f
NS NS
Larynx Agec Multivariated
(n = 11) 1.0 1.0
(n = 6) 0.79 (0.29, 2.15) 0.76 (0.28, 2.06)
(n = 4) 1.84 (0.58, 5.78) 1.72 (0.55, 5.43)
(n = 2) 1.64 (0.36, 7.48) 1.55 (0.34, 7.09)
NS NS
Lung Agec Multivariated
(n = 73) 1.0 1.0
(n = 66) 1.33 (0.96, 1.86) 1.32 (0.93, 1.86)
(n = 34) 2.27 (1.51, 3.41) 2.39 (1.58, 3.61)
(n = 33) 3.87 (2.56, 5.86) 3.74 (2.42, 5.76)
<0.01 <0.01
Hematopoietic system Agec Multivariated
(n = 39) 1.0 1.0
(n = 20) 0.71 (0.41, 1.21) 0.68 (0.39, 1.19)
(n = 6) 0.99 (0.42, 2.37) 1.01 (0.42, 2.41)
(n = 3) 1.11 (0.34, 3.65) 1.16 (0.35, 3.82)
NS NS
Kidney Agec Multivariated
(n = 53) 1.0 1.0
(n = 34) 0.90 (0.59, 1.39) 0.96 (0.62, 1.49)
(n = 11) 1.11 (0.58, 2.13) 1.14 (0.59, 2.19)
(n = 6) 1.25 (0.53, 2.93) 1.28 (0.54, 3.00)
NS 0.07
Urinary bladder Agec Multivariated
(n = 29) 1.0 1.0
(n = 22) 1.09 (0.62, 1.89) 1.09 (0.62, 1.92)
(n = 12) 2.26 (1.15, 4.45) 2.32 (1.17, 4.59)
(n = 7) 2.38 (1.03, 5.49) 1.75 (0.67, 4.58)
<0.01 <0.05
Others Agec Multivariated
(n = 198) 1.0 1.0
(n = 146) 1.05 (0.85, 1.30) 1.05 (0.85, 1.31)
(n = 22) 0.61 (0.39, 0.95) 0.62 (0.40, 0.97)
(n = 22) 1.17 (0.75, 1.82) 1.21 (0.77, 1.90)
NS NS
(n=) indicates cancer incidence, NS: not significant, GB: gall bladder. a Relative risk with Poisson regression. b For trends in the current smokers. c Adjusted for age. d Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preferences for vegetables and meats, and frequency of meat consumption. e Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preference for food saltiness, preferences for vegetables and meats, and frequency of meat consumption. f Standard error is zero.
22
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
Table 5 Population attributable risk of major smoking-related cancers in men Cancer site
Mouth and pharynx Esophagus Stomach Liver Pancreas Larynx Lung Urinary bladder All
PARa (%) Ageb
Multivariatec
45.8 98.3 36.0 28.3 39.9 59.5 78.7 52.0 30.4
41.3 86.1 36.1 32.8 37.8. 59.5 78.3 50.2 29.8
a Population attributable risk (smoking prevalence: former, 15.7%; current, 67.6%). b Adjusted for age. c Adjusted for multivariate.
Furthermore, the mean age of our study subjects at enrollment was about 43, relatively lower than in CPS II [10,12]. Therefore, our study had fewer long-term heavy smokers. Among the 404,511 current smokers in our study, only 1166 (0.29%) smoked 20 cigarettes or more a day for 30 years or more. The corresponding figure among 164,226 former smokers was 747 (0.45%). Because smoking is against the law for teenagers, Korean men start smoking at an average of 20.7 years of age [22], relatively later than the 17 years in US CPS II [10]. The older starting age for smoking may explain, in part, the lower relative risks in our study cohort. The difference in genetic susceptibility might have also contributed by modifying the smoking-related cancer risk [44]. For example, the frequency of *4 null alleles of CYP2A6, the enzyme that activates N-alkyl nitrosmaines such as NNK, is high in Koreans (11.0% compared to 1% in Caucasians) [45,46]. Inactive CYP2A6 alleles may decrease the risk of becoming tobaccodependent, and one study showed that the *4 null alleles of CYP2A6 significantly reduced the lung cancer risk [47]. Finally, while our study is based on incidence data, most other cohort studies are based on mortality data. Since after cancer diagnosis survival would have to be poorer among smokers than non-smokers [48–51], mortality-based studies are likely to show a stronger association with higher RR than incidence-based studies. Nevertheless, we believe the most important contributing factor to be the short follow-up period of the relatively healthy young male government employees and teachers. If we extend the follow-up period, we are likely to find a higher RR since more cases of smoking-related cancers will be diagnosed, as the participants get older. In prospective cohort and case-control studies, the combination of smoking with other factors increases the risk of smoking-related cancers [52,53]. In our study, however, socio-environmental factors such as place of residence and personal factors such as leisure-time physical activity, body mass index, alcohol consumption, and dietary factors had little impact on the relative and attributable risk
for the major smoking-related or overall cancer incidence. This is consistent with the recent CPS II [12], which suggests that despite differences in ethnicities and personal life style, smoking is responsible for the major portion of smokingrelated cancers. Our study had several limitations. The study population differed in general from other Koreans in terms of socioeconomic and educational status. Since they were government employees and teachers, they may have had behaviors that were conducive to better health. However, since multivariate analysis as compared with age-adjusted analysis revealed only minor decreases in cancer risk estimates from smoking, the differences of socioeconomic and educational status between the general population and our study population may show small difference in cancer risk. We did not apply the strict definition of never smokers used by CPS II, which defined current smokers as those who reported smoking at least one cigarette a day for 1 year or more [10,12]. Misclassification of never smokers as current smokers in our study could result in an underestimated cancer risk. Additionally, we could not adjust for the effects of environmental tobacco smoke and changes in smoking behavior during follow-up. In conclusion, our study, which was based on incidence data rather than mortality data, confirmed that smoking increases the risk of many cancers, including oral and pharyngeal, esophageal, stomach, liver, pancreatic, laryngeal, lung, and urinary bladder cancers. Overall, the PAR due to smoking was 30.4% of all cancers, and differences in smoking were responsible for most of the differences observed in smoking-related cancers.
Acknowledgements We appreciate the help provided by the Korean Central Cancer Registry (KCCR) and those of six regional cancer registries (Seoul, Busan, Incheon, Daejeon, Kwangju, and Daegu).
References [1] Wynder EL, Graham EA. Tobacco smoking as a possible etiologic factor in bronchogenic carcinoma. JAMA 1950;43:329–36. [2] Levin ML, Goldstein H, Gerhardt PR. Cancer and tobacco smoking. JAMA 1950;143:336–8. [3] Kahn HA. The Dorn study of smoking and mortality among US veterans. Report on eight and one-half years of observation. In: Haenszel W, editor. Epidemiological approaches to the study of cancer and other chronic diseases. National Cancer Institute Monograph, vol. 19. Rockville, MD: US Department of Health, Education, and Welfare; 1966. p. 1–125. [4] Doll R, Peto R. Mortality in relation to smoking: 20 years’ observations on male British doctors. BMJ 1976;2:1525–36. [5] Doll R, Gray R, Hafner B, Peto R. Mortality in relation to smoking: 22 years’ observations on female British doctors. BMJ 1980;5:967–71.
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24 [6] Hammond EC, Horn D. Smoking and death rates – report on forty-four months of follow-up of 187,783 men. II. Death rates by cause. JAMA 1958;166:1294–308. [7] Akiba S, Hirayama T. Cigarette smoking and cancer mortality risk in Japanese men and women-results form reanalysis of the six prefecture cohort study data. Environ Health Perspect 1990;87: 19–26. [8] US Department of Health, Education, and Welfare. Smoking and Health Report of the Advisory Committee to the Surgeon General of the Public Health Service. PHS Publication no. 1103. Washington: US Department of Health, Education, and Welfare, Public Health Service; 1964. [9] International Agency for Research on Cancer IARC Monographs on the evaluation of carcinogenic risks to humans, vol. 38. Tobacco smoking. Lyon, France: IARC; 1986. [10] Thun MJ, Day-Lally C, Myers DG, et al. Trends in tobacco smoking and mortality from cigarette use in Cancer Prevention Studies I (1959 through 1965) and II (1982 through 1988). In: National Cancer institute, smoking and tobacco control, monograph 8: changes in cigarette-related disease risks and their implication for prevention and control. Bethesda, NIH publication no. 97-4713. Washington, DC: NIH; 1997. [11] Sterling TD, Rosenbaum WL, Weinkam JJ. Risk attribution and tobacco-related deaths. Am J Epidemiol 1993;138:128–39. [12] Thun MJ, Apicella LF, Henley SJ. Smoking vs. other risk factors as the cause of smoking-attributable deaths: confounding in the courtroom. JAMA 2000;284:706–12. [13] Baron JA, Rohan TE. Tobacco. In: Schottenfeld D, Fraumeni JF, editors. Cancer epidemiology and prevention. 2nd ed. New York: Oxford University Press; 1996. [14] Doll R, Peto R, Wheatley K, et al. Mortality in relation to smoking: 40 years’ observations on male British doctors. Br Med J 1994;309:901– 11. [15] World Health Organization. International classification of disease for oncology. 2nd ed. Geneva, Switzerland: WHO; 1990. [16] Ministry of Health and Welfare, Korea. Annual report of Korea Central Cancer Registry program, January–December 2000. Seoul: Ministry of Health and Welfare; 2002. [17] World Health Organization. Physical status: the use and interpretation of anthropometry: report of a WHO Expert Committee. World Health Organization 1995;854:1–452. [18] Tsugane S, Sasazuki S, Kobayashi M, Sasaki S. Salt and salted food intake and subsequent risk of gastric cancer among middle-aged Japanese men and women. Br J Cancer 2004;12(90):128–34. [19] Van den Brandt PA, Botterweck AA, Goldbohm RA. Salt intake, cured meat consumption, refrigerator use and stomach cancer incidence: a prospective cohort study (Netherlands). Cancer Causes Control 2003;14:427–38. [20] Xibin S, Moller H, Evans HS, Dixing D, Wenjie D, Jianbang L. Residential environment, diet and risk of stomach cancer: a casecontrol study in Linzhou, China. Asian Pac J Cancer Prev 2002;3:167– 72. [21] Rothman KJ, Greenland S. Modern epidemiology. 2nd ed. Philadelphia: Lippincott-Raven Publishers; 1998. [22] Ministry of Health and Welfare, Korea. 1998 Korea Health Survey – Summary Report. Seoul: Ministry of Health and Welfare; 1999. [23] SAS. SAS/STAT software: changes and enhancement through release 8.1. Cary (NC): SAS Institute Inc.; 2002. [24] Hara M, Sobue T, Sasaki S, et al. Smoking and risk of premature death among middle-aged Japanese: ten-year follow-up of the Japan Public Heath Center-based prospective study on cancer and cardiovascular disease (JPHC Study) cohort I. Jpn J Cancer Res 2002;93:6– 14. [25] Tredaniel J, Boffetta P, Buiatti E, et al. Tobacco smoking and gastric cancer: review and meta-analysis. Int J Cancer 1999;83: 223–9.
23
[26] Shopland DR. Cigarette smoking as a cause of cancer. In: Harras A, Edwards BK, Blot WJ, Ries LA.., editors. Cancer rate and risks. 4th ed. National Cancer Institute: Bethesda, MD; 1996. [27] Chao A, Thun MJ, Henley SJ, et al. Cigarette smoking, use of other tobacco products and stomach cancer mortality in US adults: the Cancer Prevention Study II. Int J Cancer 2002;101:380–9. [28] Heineman EF, Zahm SH, et al. Increased risk of colorectal cancer among smokers: results of a 26-year follow-up of US veterans and a review. Int J Cancer 1995;59:728–38. [29] Newcomb PA, Storer BE, Marcus PM. Cigarette smoking in relation to risk of large bowel cancer in women. Cancer Res 1995;55: 4906–9. [30] Nyren O, Berstrom R, Nystrom L, et al. Smoking and colorectal cancer: a 20-year follow-up study of Swedish construction workers. J Natl Cancer Inst 1996;88:1302–7. [31] Sturmer T, Glynn RJ, Lee IM, et al. Lifetime cigarette smoking and colorectal cancer incidence in the Physician Health Study I. J Natl Cancer Inst 2000;92:1178–81. [32] Slattery ML, Curtin K, Anderson K, et al. Association between cigarette smoking, lifestyle factors, and microsatellite instability in colon tumors. J Natl Cancer Inst 2000;92:1831–6. [33] Terry PD, Miller AB, Rohan TE. Prospective cohort study of cigarette smoking and colorectal cancer risk in women. Int J Cancer 2002;99:480–3. [34] Giovannucci E, Rimm EB, Stampfer MJ, et al. A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S. men. J Natl Cancer Inst 1994;86:183–91. [35] Giovannucci E, Colditz GA, Stampfer MJ, et al. A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S. women. J Natl Cancer Inst 1994;86:192–9. [36] Trichopoulos D, MacMahon B, Sparros L, et al. Smoking and hepatitis B-negative primary hepatocellular carcinoma. J Natl Cancer Inst 1980;65:111–4. [37] Yu MC, Mack T, Hanisch R, et al. Hepatitis, alcohol consumption, cigarette smoking, and hepatocellular carcinoma in Los Angeles. Cancer Res 1983;43:6077–9. [38] Carstensen JM, Pershagen G, Eklund G. Mortality in relation to cigarette and pipe smoking: 16 years’ observation of 25,000 Swedish men. J Epidemiol Community Health 1987;41:166–72. [39] Mizoue T, Tokui N, Nishisaka K, et al. Prospective study on the relation of cigarette smoking with cancer of the liver and stomach in an endemic region. Int J Epidemiol 2000;29:232–7. [40] Hsing AW, McLaughlin JK, Hrubec Z, et al. Cigarette smoking and liver cancer among US veterans. Cancer Causes Control 1990;1:217– 21. [41] Kuper H, Boffetta P, Adami HO. Tobacco use and cancer causation: association by tumour type. J Int Med 2002;252:206–24. [42] Korea Institute of Tuberculosis. The survey on the smoking habits in Korea. Tubercul Resp Dis 1985;21:31–45. [43] Korean National Statistical Office. Report on the social statistics survey. Seoul: Korean National statistical Office; 2000. [44] Tyndale RF, Sellers EM. Variable CYP2A6-mediated nicotine metabolism alters smoking behavior and risk. Drug Metab Dispos 2001;29:548–52. [45] Kwon JT, Nakajima M, Cahi S, et al. Nicotine metabolism and CYP2A6 allele frequencies in Koreans. Pharmacogenetics 2001;11:317–23. [46] Oscarson M. Genetic polymorphism in the cytochrome P4502A6 (CYP2A6) gene: implications for interindividual differences in differences in nicotine metabolism. Drug Metab Dispos 2001;29: 91–5. [47] Miyamoto M, Umetsu Y, Dosaka-Akita H, et al. CYP2A6 gene deletion reduces susceptibility to lung cancer. Biochem Biophys Res Commun 1999;261:658–60. [48] Yu GP, Ostroff JS, Zhang ZF, et al. Smoking history and cancer patient survival: a hospital cancer registry study. Cancer Detect Prev 1997;21:497–509.
24
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
[49] Fujisawa T, Iizasa T, Saitoh Y, et al. Smoking before surgery predicts poor long-term survival in patients with stage I non-small-cell lung carcinomas. J Clin Oncol 1999;17:2086–91. [50] Browman GP, Mohide EA, Willan A, et al. Association between smoking during radiotherapy and prognosis in head and neck cancer: a follow-up study. Head Neck 2002;24:1031–7. [51] Videtic GM, Stitt LW, Dar AR, et al. Continued cigarette smoking by patients receiving concurrent chemoradiotherapy for limited-stage
small-cell lung cancer is associated with decreased survival. J Clin Oncol 2003;21:1544–9. [52] Sterling TD, Rosenbaum WL, Weinkam JJ. Risk attribution and tobacco-related deaths. Am J Epidemiol 1993;138: 128–39. [53] Midgette AS, Baron JA, Rohan TE. Do cigarette smokers have diets that increase their risks of coronary heart disease and cancer. Am J Epidemiol 1993;137:521–9.
Cigarette smoking and cancer incidence risk in adult men: National Health Insurance Corporation Study Young Ho Yun MD, PhDa, Kyu Won Jung MPha, Jong-Myon Bae MD, PhDa, Jin Soo Lee MDa, Soon Ae Shin MPhb, Sang Min Park MD, MPhc, Taiwoo Yoo MD, PhDc,*, Bong Yul Huh MD, PhDc a
Research Institute and Hospital, National Cancer Center, Quality of Cancer Center, 809 Madu-dong, Ilsan-gu, Goyang-si, Gyeonggi-do 411-769, Republic of Korea b National Health Insurance Corporation, 168-9 Yeomri-dong, Mapo-gu, Seoul 121-749, Republic of Korea c Department of Family Medicine, Seoul National University College of Medicine, 28 Youngon-dong, Chongno-gu, Seoul 110-799, Republic of Korea Accepted 16 August 2004
Abstract We analyzed risk while adjusting for age, body mass index, frequency of moderate physical activity, alcohol consumption, preference for vegetables versus meats, and frequency of meat consumption in a multivariate analysis and based our findings on not mortality data but incidence data. 733,134 Korean men who were 30 years old or older, insured by the National Health Insurance Corporation, and had a medical evaluation in 1996 were included in the study and followed up through 2000. During the 4-year follow-up period of 3,590,872 person–years, we identified 7204 new cases. We used the Cox proportional hazards model to estimate adjusted relative risks (aRRs), 95% confidence intervals (CIs). The association of current cigarette smoking was significantly stronger as compared with never smokers; aRR was 1.49 (95% CI = 1.39–1.59) for all cancers, 4.46 (2.32–8.57) for esophageal, 3.83 (2.97–4.94) for lung, 3.01 (1.58–5.72) for laryngeal, 2.24 (1.48–3.39) for urinary bladder, 1.62 (1.42–1.84) for gastric, 1.75 (1.12–2.74) for oral and pharyngeal, 1.58 (0.97–2.27) for pancreatic, and 1.50 (1.29– 1.74) for liver cancer. Our findings, based on incidence data, confirmed that differences in smoking habit were responsible for most of the differences observed in smoking-related cancers. # 2004 International Society for Preventive Oncology. Published by Elsevier Ltd. All rights reserved. Keywords: Cigarette smoking; Cancer incidence; Risk; Cohort study
1. Introduction Epidemiologic studies have consistently shown an association between smoking and cancer of the lung, larynx, mouth, esophagus, urinary bladder, kidney, pancreas, and cervix [1–7]. Several reports also suggest an association between smoking and cancer of the stomach, liver, colon, and rectum [8–10]. Most large cohort studies collect detailed smoking histories, but variable information on other potential risk factors. Consultants for the tobacco industry have speculated that confounding may cause studies to exaggerate the risks attributed to smoking [11], although * Corresponding author. Tel.: +82 276 03332; fax: +82 276 63276. E-mail address: [email protected] (T. Yoo).
there is little evidence that this is occurs. Recently, US Cancer Prevention Study (CPS) II suggested that multiple covariates had little impact on the relative and attributable risk for cancer mortality after adjustments were made for age and sex [12]. Since that study was based on cancer mortality data, the effect of smoking on cancer incidence may have been confounded by its effect on other fatal diseases, such as cardiovascular disease [13]. Fewer large prospective studies have assessed the disease risks associated with smoking in Asia, where the widespread consumption of manufactured cigarettes began more recently than in North American and Northern Europe. Our study has the additional strengths of being able to measure cancer incidence, which is informative for sites with higher survival rates, and ability to control for several other major potential risk factors. It is
0361-090X/$30.00 # 2004 International Society for Preventive Oncology. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.cdp.2004.08.006
16
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
uncertain whether the CPS II results are applicable to groups with different genetic and environmental backgrounds [11,14]. In Korea, the National Health Insurance Corporation (NHIC) has been providing health insurance to government employees and teachers since 1980 along with biennial health examinations that include height, weight, and blood pressure measurements, chest radiography, blood counts, and blood chemistries. In addition, a self-administered questionnaire collects information regarding medical history, current health status, and individual life styles regarding tobacco and alcohol consumption, dietary preferences, and leisure-time physical activity. To analysis the association of smoking with the incidence of all cancers and the cancers of major organ sites for age, body mass index, frequency of moderate physical activity, alcohol consumption, preference for vegetables versus meats, and frequency of meat consumption, we performed National Health Insurance cohort study.
2. Material and methods 2.1. Cohort and follow-up The study subjects derived from 1,228,817 (980,841 men and 337,426 women) government employees and teachers who participated in a national health examination program begun in 1996, and they constituted the cohort for the National Health Insurance Cooperation Study. The NIHC permitted us to use their data for the present study. We excluded women from the present study because their smoking rate was too low (0.3% of the study population), and we excluded 165,915 men who were less than 30 years old. Among the eligible 814,926 male participants, 54,585 were excluded due to lack of information on life style, 23,835 due to lack of information on smoking status, and 3372 who were prevalent cases and documented to have received medical service in 1996. For the analysis of the impact of smoking duration, we excluded 208 current smokers and 10,753 former smokers for whom data on smoking duration was missing. The final cohort consisted of 733,134 Korean men aged 30 years or older. The incident cancer cases were identified from the Korean Central Cancer Registry (KCCR) and six regional cancer registries (RCRs). For the KCCR, a nationwide hospital-based cancer registry system, professionally trained and certified medical recorders collected all relevant data using a standardized manual based on the International Classification of Disease for Oncology (ICD-O) after reviewing the pathology reports and hospital discharge summaries [15]. The KCCR-affiliated hospitals included 94% of all university hospitals and 96% of all resident training hospitals in Korea [16]. Additionally, using a computerized search program that covered the period of 1997–2000, we investigated the RCR data from six major
cities (Seoul, Busan, Incheon, Daejeon, Kwangju, and Daegu) that collect cancer incidence data independent of the KCCR program. For those individuals with more than one event, we included the first diagnosis of cancer in the analyses. In general, microscopic examination (83.0%) was the leading methods of diagnosis, followed by clinical investigation (14.1%), specific biochemical/immunologic tests (1.0%), and clinical judgment (1.9%). We defined the follow-up period as the interval between the initial examination and the diagnosis of cancer. We followed subjects without cancer up to December 31, 2000. During the 4-year follow-up period of 3,590,872 person–years, we identified 7204 new cases. 2.2. Classification of smoking behavior Smoking status was classified as current, former, or never smoker on the basis of the response to the following questions on the initial questionnaire provided in 1996: ‘‘Do you smoke cigarettes now?’’, ‘‘How many cigarettes per day do you smoke?’’, and ‘‘How many years have you smoked in your lifetime?’’ Current smokers were further subgrouped by the average number of cigarettes smoked per day (1–9, 10–19, or 20 or more) and duration of smoking (1–19, 20– 29, or 30 or more years). We classified smoking behavior in order to compare our data with those of previous studies. We classified the smoking amount because previous studies showed that even light smoking increased cancer risk, and the smoking duration as increased cancer risk was observed among individuals smoking for 20 years or more [4,6,9,10,14]. 2.3. Statistical analysis We calculated the incidence rates per 100,000 person– years according to the smoking status at baseline by dividing the number of cancers by the total person–years of the follow-up. We used the never smokers as the reference group to calculate the relative risks of cancers due to smoking. We used Poisson regression to estimate the adjusted relative risks (aRRs) associated with current or former smoking and their 95% confidence intervals (CIs), first adjusting only for age at enrollment and then for other multiple covariates. We adjusted multivariate models for age at enrollment (30–39, 40–49, 50 or older), body mass index (BMI; <18.5, 18.5– 24.9, 25–29.9, 30 or more), frequency of moderate physical activity in leisure time (0, 1–2, 3–4, 5 or more times per week), alcohol consumption (number of drinks per week), preference for vegetables versus meats (vegetables, mixed, meats), and frequency of meat consumption (0–1, 2–3, 4 or more times per week). The BMI (kg/m2) calculated from height and weight at enrollment was used as a measure of relative body fat. The BMI score was divided into four groups according to the World Health Organization criteria for thinness or being overweight [17]. Moderate physical activity was defined as intense enough to produce sweat. For
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
17
Table 1 Sociodemographic characteristics of National Health Insurance Cooperation Study cohort by smoking status (n = 733,134) Characteristics Mean age (year)
Never (n = 164,447) 43.6
Former (n = 164,204) 44.7
Current (n = 404,483)a 42.4
Place of residence (%) Metropolis City Country
76,460 (46.5) 58,411 (35.5) 29,576 (18.0)
75,666 (46.1) 59,056 (36.0) 29,482 (18.0)
171,503 (42.4) 148,890 (36.8) 84,090 (20.8)
Frequency of alcohol consumption (%) Never 2–3 per month 1–2 per week 3–4 per week =5 per week
63,570 (38.8) 39,926 (24.4) 43,495 (26.5) 13,386 (8.2) 3,566 (2.2)
51,490 (31.6) 40,068 (24.6) 47,148 (29.0) 18,498 (11.4) 5,542 (3.4)
79,193 (19.7) 86,751 (21.6) 144,188 (35.8) 69,099 (17.2) 23,131 (5.8)
2,907 (1.8) 111,450 (67.8) 47,962 (29.2) 2,061 (1.3)
2,213 (1.4) 108,819 (66.3) 51,066 (31.1) 2,042 (1.2)
8,626 (2.1) 281,958 (69.7) 108,767 (26.9) 4,991 (1.2)
66,758 (41.1) 64,027 (39.5) 17,688 (10.9) 13,826 (8.5)
62,161 (38.4) 64,871 (40.1) 19,957 (12.3) 14,732 (9.1)
174,369 (43.6) 163,175 (40.8) 37,030 (9.3) 25,647 (6.4)
Preference of food saltiness (%) Low salt Normal Salty
34,351 (20.9) 108,389 (66.1) 21,283 (13.0)
31,432 (19.2) 106,095 (64.9) 25,956 (15.9)
54,636 (13.6) 254,231 (63.0) 94,467 (23.4)
Food preferenceb (%) Vegetables Mixed Meats
33,605 (20.5) 121,050 (73.9) 9,241 (5.6)
34,422 (21.1) 117,912 (72.1) 11,111 (6.8)
81,340 (20.2) 287,606 (71.4) 34,037 (8.5)
76,710 (47.0) 80,767 (49.5) 5,630 (3.5)
73,832 (45.3) 80,582 (49.5) 8,499 (5.2)
178,241 (44.4) 207,694 (51.7) 15,685 (3.9)
BMI (%) <18.5 kg/m2 18.5–24.9 kg/m2 25–29.9 kg/m2 =30 kg/m2 Physical activity (%) Never 1–2 per week 3–4 per week =5 per week
Frequency of meat consumption (%) =1 per week 2–3 per week =4 per week a b
Excluding the missing data. Preference for vegetables vs. meats.
stomach cancer, we adjusted also for preference of food saltiness (low salt, normal, salty) because some studies suggest that food saltiness might be associated with development of stomach cancer [18–20]. To calculate the population attributable risk (PAR) of cancer [21], we used the smoking prevalence data of the 1998 Korea Health Survey (performed by the Ministry of Health and Welfare), which reported 67.5% of men as being current smokers, 15.7% former smokers, and 16.7% as never smokers [22]. All confidence intervals are at 95%, and we considered a P-value of 5% as significant. All statistical tests were twosided. We used SAS statistical package version 8.1 [23].
3. Results Among the 733,134 subjects, 404,483 (55.2%) were current smokers, 164,204 (22.4%) former smokers, and 164,447 (22.4%) never smokers. Among the current
smokers, 16.1, 62.8, and 21.1% smoked 1–9, 10–19, and 20 or more cigarettes per day, respectively, and smoking duration was 1–19 years (58.6%), 20–29 years (28.4%), and 30 years or more (13.0%). Among former smokers, 82.2% reported that they had smoked for 1–19 years, 13.3% for 20– 29 years, and 4.5% for 30 years or more. Table 1 shows the smoking statuses and characteristics of individuals in the Korean National Health Insurance Cooperation study. Major cancer sites in descending order were stomach, liver, bronchus or lung, rectum, colon, kidney, urinary bladder, mouth or pharynx, pancreas, esophagus, gall bladder or bile duct, hematopoietic system, larynx, and others (Table 2). The age-adjusted RR for all types of cancer for former smokers as compared with never smokers was 1.32 (95% CI = 1.23–1.42) and 1.51 (95% CI = 1.42–1.61) for current smokers (Table 2). We observed a statistically significant age-adjusted RR in both former and current smokers for
18
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
Table 2 Estimated relative risk and 95% confidence intervals of cancers by smoking statusa (n = 733,134) Ptrendb
Cancer site
Never
Former
Current
Current, cigarettes smoked per day 1–9
10–19
=20
All (n = 7204) Agec Multivariated
(n = 1284) 1.0 1.0
(n = 1831) 1.32 (1.23, 1.42) 1.33 (1.24, 1.43)
(n = 4,089) 1.51 (1.42, 1.61) 1.49 (1.39, 1.59)
(n = 701) 1.46 (1.34, 1.61) 1.43 (1.30, 1.57)
(n = 2,496) 1.49 (1.39, 1.59) 1.46 (1.36, 1.57)
(n = 886) 1.64 (1.50, 1.78) 1.61 (1.47, 1.76)
NS NS
Mouth and pharynx (n = 172) Agec Multivariated
(n = 25) 1.0 1.0
(n = 41) 1.53 (0.93, 2.51) 1.53 (0.93, 2.52)
(n = 106) 1.91 (1.24, 2.96) 1.75 (1.12, 2.74)
(n = 13) 1.37 (0.70, 2.68) 1.24 (0.62, 2.48)
(n = 67) 1.94 (1.23, 3.08) 1.77 (1.10, 2.83)
(n = 26) 2.28 (1.31, 3.95) 2.15 (1.23, 3.78)
NS NS
Esophagus (n = 163) Agec Multivariated
(n = 10) 1.0 1.0
(n = 29) 2.64 (1.29, 5.42) 2.22 (1.07, 4.60)
(n = 124) 6.19 (3.25, 11.8) 4.46 (2.32, 8.57)
(n = 16) 4.34 (1.97, 9.57) 3.65 (1.65, 8.09)
(n = 81) 6.54 (3.39, 12.6) 4.71 (2.42, 9.18)
(n = 26) 6.65 (3.20, 13.8) 4.16 (1.96, 8.82)
NS NS
Stomach (n = 1949) Agec Multivariatee
(n = 319) 1.0 1.0
(n = 496) 1.44 (1.25, 1.66) 1.45 (1.25, 1.67)
(n = 1134) 1.68 (1.48, 1.90) 1.62 (1.42, 1.84)
(n = 202) 1.69 (1.42, 2.02) 1.60 (1.33, 1.93)
(n = 687) 1.63 (1.43, 1.87) 1.59 (1.38, 1.82)
(n = 244) 1.79 (1.52, 2.12) 1.72 (1.45, 2.05)
NS NS
Colon (n = 417) Agec Multivariated
(n = 99) 1.0 1.0
(n = 148) 1.39 (1.08, 1.79) 1.37 (1.06, 1.77)
(n = 170) 0.83 (0.65, 1.06) 0.81 (0.63, 1.05)
(n = 36) 0.98 (0.67, 1.43) 0.97 (0.66, 1.43)
(n = 102) 0.80 (0.61, 1.05) 0.78 (0.59, 1.04)
(n = 32) 0.78 (0.53, 1.17) 0.76 (0.51, 1.15)
NS NS
Rectum (n = 453) Agec Multivariated
(n = 106) 1.0 1.0
(n = 131) 1.15 (0.89, 1.48) 1.17 (0.91, 1.52)
(n = 216) 0.96 (0.76, 1.22) 0.97 (0.76, 1.24)
(n = 38) 0.96 (0.66, 1.39) 0.95 (0.65, 1.39)
(n = 131) 0.94 (0.73, 1.22) 0.95 (0.73, 1.24)
(n = 47) 1.04 (0.74, 1.48) 1.05 (0.74, 1.50)
NS NS
Liver (n = 1434) Agec Multivariated
(n = 255) 1.0 1.0
(n = 403) 1.45 (1.24, 1.70) 1.53 (1.30, 1.79)
(n = 776) 1.41 (1.22, 1.62) 1.50 (1.29, 1.74)
(n = 192) 2.01 (1.67, 2.42) 2.20 (1.81, 2.67)
(n = 472) 1.38 (1.18, 1.60) 1.48 (1.26, 1.73)
(n = 111) 0.98 (0.78, 1.23) 0.98 (0.77, 1.23)
< 0.01 < 0.01
GB and biliary system (n = 161) Agec Multivariated
(n = 39) 1.0 1.0
(n = 40) 0.94 (0.60, 1.46) 0.94 (0.60, 1.47)
(n = 82) 1.02 (0.69, 1.49) 0.94 (0.63, 1.41)
(n = 9) 0.62 (0.30, 1.29) 0.58 (0.27, 1.25)
(n = 56) 1.12 (0.74, 1.69) 1.08 (0.70, 1.66)
(n = 17) 1.06 (0.60, 1.88) 0.86 (0.46, 1.59)
NS NS
Pancreas (n = 172) Agec Multivariated
(n = 28) 1.0 1.0
(n = 52) 1.71 (1.08, 2.70) 1.68 (1.68, 1.68)f
(n = 92) 1.63 (1.07, 2.50) 1.58 (1.58, 1.58)f
(n = 16) 1.56 (0.84, 2.88) 1.48 (1.48, 1.48)f
(n = 52) 1.49 (0.94, 2.36) 1.44 (1.44, 1.44)f
(n = 23) 2.08 (1.20, 3.63) 2.12 (2.12, 2. 12)f
NS NS
Larynx (n = 103) Agec Multivariated
(n = 11) 1.0 1.0
(n = 14) 1.17 (0.53, 2.57) 1.12 (0.51, 2.46)
(n = 78) 3.55 (1.88, 6.68) 3.01 (1.58, 5.72)
(n = 5) 1.25 (0.43, 3.59) 1.14 (0.39, 3.29)
(n = 47) 3.47 (1.80, 6.70) 2.94 (1.50, 5.74)
(n = 26) 6.07 (2.99, 12.3) 5.20 (2.52, 10.7)
<0.01 <0.01
Lung (n = 787) Agec Multivariated
(n = 73) 1.0 1.0
(n = 150) 1.88 (1.42, 2.49) 1.85 (1.39, 2.48)
(n = 564) 3.94 (3.08, 5.02) 3.83 (2.97, 4.94)
(n = 50) 1.87 (1.31, 2.68) 1.82 (1.25, 2.64)
(n = 326) 3.69 (2.86, 4.76) 3.62 (2.78, 4.71)
(n = 187) 6.82 (5.20, 8.95) 6.66 (5.02, 8.84)
<0.01 <0.01
Hematopoietic system (n = 148) Agec Multivariated
(n = 39) 1.0 1.0
(n = 29) 0.72 (0.45, 1.17) 0.71 (0.44, 1.17)
(n = 80) 0.90 (0.61, 1.32) 0.93 (0.63, 1.39)
(n = 20) 1.33 (0.77, 2.27) 1.33 (0.77, 2.33)
(n = 41) 0.73 (0.47, 1.14) 0.77 (0.49, 1.21)
(n = 19) 1.04 (0.60, 1.80) 1.09 (0.62, 1.92)
NS NS
Kidney (n = 211) Agec Multivariated
(n = 53) 1.0 1.0
(n = 52) 0.91 (0.62, 1.34) 0.96 (0.65, 1.42)
(n = 106) 0.94 (0.67, 1.31) 0.94 (0.66, 1.32)
(n = 10) 0.50 (0.25, 1.05) 0.53 (0.27, 1.05)
(n = 77) 1.10 (0.77, 1.56) 1.08 (0.75, 1.56)
(n = 19) 0.84 (0.49, 1.41) 0.86 (0.50, 1.47)
NS NS
Urinary bladder (n = 208) Agec Multivariated
(n = 29) 1.0 1.0
(n = 42) 1.35 (0.84, 2.17) 1.31 (0.80, 2.13)
(n = 137) 2.30 (1.54, 3.44) 2.24 (1.48, 3.39)
(n = 21) 1.95 (1.11, 3.42) 2.05 (1.16, 3.62)
(n = 85) 2.30 (1.51, 3.51) 2.23 (1.44, 3.45)
(n = 31) 2.65 (1.59, 4.40) 2.47 (1.46, 4.18)
NS NS
Others (n = 826) Agec Multivariated
(n = 198) 1.0 1.0
(n = 204) 0.97 (0.80, 1.18) 0.98 (0.80, 1.19)
(n = 424) 1.01 (0.85, 1.19) 1.01 (0.85, 1.20)
(n = 73) 0.99 (0.75, 1.29) 1.01 (0.77, 1.33)
(n = 272) 1.04 (0.86, 1.25) 1.04 (0.86, 1.26)
(n = 78) 0.92 (0.71, 1.20) 0.90 (0.68, 1.18)
NS NS
(n=) indicates cancer incidence, NS: not significant, GB: gall bladder. a Relative risk with Poisson regression. b For trends in the current smokers. c Adjusted for age. d Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preferences for vegetables and meats, and frequency of meat consumption. e Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preference for food saltiness, preferences for vegetables and meats, and frequency of meat consumption. f Standard error is zero.
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
cancer of the esophagus, stomach, liver, pancreas, and lung, and for current smokers for cancer of the mouth and pharynx, larynx, and urinary bladder. In contrast, RR of colon cancer was statistically significant only in former smokers. There was no significant association of smoking with the risk of rectal, gall bladder, kidney, hematopoietic, and other cancers. Multivariate analysis revealed only minor decreases in the RR estimates obtained with age-adjusted analysis. In current smokers, there were statistically significant positive linear trends in aRRs with increased smoking duration for all cancers (Ptrend < 0.01) (Table 3). Even when adjusted for age and other related factors, there were significant dose-response relationships between smoking duration and incidence of all cancers in current smokers. We also observed statistically significant positive linear trends in aRRs for esophageal (Ptrend < 0.01), stomach (Ptrend < 0.01), laryngeal (P < 0.05), lung (P < 0.01), and urinary bladder cancer (Ptrend < 0.01) with increased smoking duration. We observed a statistically significant positive linear trend in RR of rectal cancer (P < 0.01), but the RR was not significantly increased over the never smoker. In former smokers, the age-adjusted RR of all cancers increased significantly with smoking duration (Ptrend < 0.01) (Table 4). Significant positive linear trends in aRRs were also observed for colon (Ptrend < 0.05), liver (Ptrend < 0.01), lung (Ptrend < 0.01), and urinary bladder cancer (Ptrend < 0.01), and a marginally significant positive linear trend in RR was observed for esophageal cancer (Ptrend = 0.07). The multivariate adjustment showed minor changes in the aRR estimates of smoking for all categories of cancer. The dose–response relationship between smoking and each category of cancer site remained unchanged when it was significant after adjustment for age. The age-adjusted PAR was highest for esophageal cancer (98.3%) followed by lung cancer (78.7%), laryngeal cancer (59.5%), and urinary bladder cancer (52.0%) (Table 5). In addition, more than one-third of the cases of oropharyngeal cancer (45.8%), pancreatic cancer (39.9%), and stomach cancer (36.0%) and one-quarter of liver cancer cases (28.3%) were attributable to smoking. Overall, 30.4% of all cancers were attributable to smoking. The multivariate adjusted PAR for all cancers changed negligibly to 29.8% from the age-adjusted PAR of 30.4%. For esophageal cancer however, multivariate adjustment significantly reduced the PAR estimate from 98.3% to 86.1%. For oral and pharyngeal, stomach, pancreatic, laryngeal, lung, and urinary bladder cancer, multivariate adjustment resulted in minor PAR changes. Nevertheless, lung, esophagus, larynx, and urinary bladder remained the four major sites of smoking-related cancers, in that order.
4. Discussion In this large cohort study based on incidence data, as in similar studies based on mortality data, lung, esophageal,
19
laryngeal, urinary bladder, oral and pharyngeal, stomach, pancreatic, and liver cancer in Korean men were associated with smoking. After adjusting for age, the proportion of all cancers that were attributable to smoking was 30.4%, and further adjustments for body mass index, frequency of moderate physical activity, alcohol consumption, preference for vegetables versus meats, and frequency of meat consumption had little impact on the risk. Our risk estimate was lower than the CPS II estimate, which attributed 45% of all cancer deaths in men to smoking [12], but similar to the 2002 Japan Public Heath Center report, which attributed 25% of cancer mortality to smoking [24]. Our study confirms the previous finding [25–27] that after controlling for diet and alcohol drinking, in a socially homogenous group, stomach cancer is weakly associated with smoking. In addition, the risk increases with smoking duration even after multivariate adjustments, which supports the association of smoking with stomach cancer. Our study also agrees with previous studies that showed no consistent association between cigarette smoking and colorectal cancer [28–33]. Recent studies, however, have proposed an induction period of 30–40 years between a genotoxic exposure and the diagnosis [34,35]. Thus, a longer follow-up period may be needed. We observed a weak but significant association between smoking and liver cancer, and the association remained after multivariate adjustment, which included alcohol consumption. Since we did not include hepatitis status in our analysis, we cannot draw any conclusions. In previous studies, the association between smoking and liver cancer risk was not consistent [7,14,36– 39]. One study suggested that in areas where liver cancer is not endemic, such as in Western countries, smoking was a more important cause of liver cancer [40], but confounding effects cannot be excluded. It is possible, for example, that patients with hepatitis B infections reduce their smoking as their disease progresses. We need to clarify the relationship between liver cancer and number of cigarettes smoked by investigating the association after adjusting for the smoking status of hepatitis B carriers. Our study did not show an increase in kidney cancer risk among current smokers. This is in contrast with urinary bladder cancer risk, which more than doubled. Since the incidence of kidney cancer is low in Asia [41] and its association with smoking has not been established in Asia, further studies are needed in Asian cohorts. Our results were similar to those of Japanese studies except that we found a lower risk for laryngeal cancer [24]. The relative risk of most cancers, however, especially oral and pharyngeal, laryngeal, and lung cancer were lower than the risks found in large Western cohort studies. Also, our study did not show a consistent association of smoking with colorectal cancer. Several explanations for our lower RRs are possible, especially for the US studies. First, even though the consumption of cigarettes per capita has increased in Korea since 1980, when 69.4% of men smoked, cigarette consumption was low in the 1960s and 1970s [42,43].
20
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
Table 3 Estimated relative risk of cancer and 95% confidence intervals for current smokers by smoking durationa Ptrendb
Cancer site
Never
Years of smoking duration 1–19
20–29
30
All Agec Multivariated
(n = 1284) 1.0 1.0
(n = 1216) 1.37 (1.26, 1.48) 1.36 (1.25, 1.48)
(n = 1361) 1.42 (1.32, 1.54) 1.40 (1.29, 1.52)
(n = 1506) 1.72 (1.60, 1.86) 1.68 (1.55, 1.82)
<0.01 <0.01
Mouth and pharynx Agec Multivariated
(n = 25) 1.0 1.0
(n = 33) 1.59 (0.92, 2.74) 1.42 (0.81, 2.48)
(n = 45) 2.27 (1.38,3.74) 2.17 (1.31, 3.60)
(n = 28) 1.88 (1.08, 3.27) 1.67 (0.94, 2.96)
NS NS
Esophagus Agec Multivariated
(n = 10) 1.0 1.0
(n = 24) 5.09 (2.40, 10.8) 4.21 (1.96, 9.03)
(n = 36) 4.91 (2.42, 9.95) 3.64 (1.77, 7.46)
(n = 63) 7.63 (3.90, 14.9) 5.07 (2.56, 10.1)
<0.01 <0.05
Stomach Agec Multivariatee
(n = 319) 1.0 1.0
(n = 358) 1.57 (1.34, 1.84) 1.55 (1.32, 1.83)
(n = 371) 1.55 (1.33, 1.81) 1.52 (1.29, 1.77)
(n = 404) 1.89 (1.63, 2.20) 1.79 (1.53, 2.10)
<0.01 <0.01
Colon Agec Multivariated
(n = 99) 1.0 1.0
(n = 59) 0.89 (0.63, 1.25) 0.87 (0.62, 1.23)
(n = 45) 0.63 (0.44, 0.90) 0.61 (0.42, 0.88)
(n = 66) 0.96 (0.70, 1.32) 0.96 (0.69, 1.33)
NS NS
Rectum Agec Multivariated
(n = 106) 1.0 1.0
(n = 62) 0.81 (0.58, 1.12) 0.80 (0.57, 1.13)
(n = 76) 0.98 (0.72, 1.32) 1.00 (0.74, 1.36)
(n = 78) 1.10 (0.82, 1.49) 1.12 (0.82, 1.52)
<0.01 <0.01
Liver Agec Multivariated
(n = 255) 1.0 1.0
(n = 253) 1.49 (1.24, 1.78) 1.59 (1.32, 1.92)
(n = 282) 1.33 (1.12, 1.58) 1.42 (1.19, 1.70)
(n = 240) 1.42 (1.18, 1.70) 1.49 (1.23, 1.80)
NS NS
GB and biliary system Agec Multivariated
(n = 39) 1.0 1.0
(n = 19) 0.85 (0.48, 1.51) 0.83 (0.46, 1.50)
(n = 31) 1.03 (0.64, 1.67) 1.00 (0.60, 1.65)
(n = 32) 1.12 (0.70, 1.81) 0.94 (0.57, 1.56)
NS NS
Pancreas Agec Multivariated
(n = 28) 1.0 1.0
(n = 27) 1.79 (1.03, 3.12) 1.83 (1.83, 1.83)f
(n = 29) 1.49 (0.88, 2.53) 1.48 (1.48, 1.48)f
(n = 35) 1.61 (0.97, 2.66) 1.48 (1.48, 1.48)f
NS NS
Larynx Agec Multivariated
(n = 11) 1.0 1.0
(n = 14) 2.16 (0.94, 4.93) 1.95 (0.85, 4.49)
(n = 30) 3.79 (1.87, 7.66) 3.12 (1.52, 6.39)
(n = 34) 4.24 (2.12, 8.48) 3.67 (1.81, 7.44)
<0.05 <0.05
Lung Agec Multivariated
(n = 73) 1.0 1.0
(n = 98) 2.55 (1.86, 3.49) 2.60 (1.88, 3.60)
(n = 183) 3.65 (2.77, 4.80) 3.55 (2.67, 4.73)
(n = 282) 4.89 (3.77, 6.35) 4.78 (3.64, 6.27)
<0.01 <0.01
Hematopoietic system Agec Multivariated
(n = 39) 1.0 1.0
(n = 38) 0.88 (0.54, 1.42) 0.90 (0.55, 1.46)
(n = 23) 0.87 (0.51, 1.49) 0.93 (0.54, 1.59)
(n = 19) 0.96 (0.54, 1.70) 1.01 (0.55, 1.83)
NS NS
Kidney Agec Multivariated
(n = 53) 1.0 1.0
(n = 30) 0.76 (0.48, 1.23) 0.76 (0.47, 1.25)
(n = 37) 0.93 (0.61, 1.43) 0.89 (0.57, 1.40)
(n = 39) 1.12 (0.73, 1.71) 1.15 (0.74, 1.78)
NS NS
Urinary bladder Agec Multivariated
(n = 29) 1.0 1.0
(n = 36) 1.66 (0.99, 2.79) 1.68 (0.99, 2.86)
(n = 49) 2.50 (1.56, 3.99) 2.39 (1.47, 3.87)
(n = 52) 2.68 (1.68, 4.28) 2.63 (1.62, 4.27)
<0.01 <0.05
Others Agec Multivariated
(n = 198) 1.0 1.0
(n = 165) 1.05 (0.85, 1.32) 1.03 (0.82, 1.30)
(n = 124) 0.91 (0.72, 1.14) 0.89 (0.70, 1.13)
(n = 134) 1.05 (0.84, 1.32) 1.10 (0.88, 1.39)
NS NS
(n=) indicates cancer incidence, NS: not significant, GB: gall bladder. a Relative risk with Poisson regression. b For trends in the current smokers. c Adjusted for age. d Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preferences for vegetables and meats, and frequency of meat consumption. e Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preference for food saltiness, preferences for vegetables and meats, and frequency of meat consumption. f Standard error is zero.
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
21
Table 4 Estimated relative risk of cancer and 95% confidence intervals for former smokers by smoking durationa Ptrendb
Cancer site
Never
Years of smoking duration 1–19
20–29
=30
All Agec Multivariated
(n = 1284) 1.0 1.0
(n = 1149) 1.27 (1.17, 1.37) 1.28 (1.18, 1.39)
(n = 335) 1.36 (1.20, 1.53) 1.38 (1.22, 1.55)
(n = 195) 1.57 (1.35, 1.82) 1.54 (1.32, 1.80)
<0.01 <0.01
Mouth and pharynx Agec Multivariated
(n = 25) 1.0 1.0
(n = 27) 1.49 (0.86, 2.56) 1.49 (0.87, 2.57)
(n = 12) 2.64 (1.32, 5.27) 2.63 (1.31, 5.26)
(n = 2) 1.01 (0.24, 4.30) 1.00 (0.23, 4.27)
NS NS
Esophagus Agec Multivariated
(n = 10) 1.0 1.0
(n = 16) 2.35 (1.06, 5.17) 2.00 (0.89, 4.51)
(n = 7) 3.27 (1.24, 8.59) 2.97 (1.13, 7.82)
(n = 6) 5.36 (1.94, 14.8) 3.85 (1.31, 11.4)
0.07 0.07
Stomach Agec Multivariatee
(n = 319) 1.0 1.0
(n = 311) 1.38 (1.18, 1.61) 1.38 (1.17, 1.62)
(n = 89) 1.46 (1.16, 1.85) 1.50 (1.18, 1.90)
(n = 47) 1.55 (1.13, 2.11) 1.53 (1.11, 2.10)
NS NS
Colon Agec Multivariated
(n = 99) 1.0 1.0
(n = 95) 1.37 (1.04, 1.82) 1.36 (1.02, 1.80)
(n = 23) 1.21 (0.76, 1.90) 1.15 (0.72, 1.83)
(n = 21) 2.08 (1.29, 3.36) 2.08 (1.29, 3.37)
0.05 0.06
Rectum Agec Multivariated
(n = 106) 1.0 1.0
(n = 89) 1.19 (0.90, 1.57) 1.21 (0.91, 1.61)
(n = 24) 1.20 (0.77, 1.87) 1.23 (0.78, 1.92)
(n = 6) 0.59 (0.26, 1.36) 0.61 (0.27, 1.41)
NS NS
Liver Agec Multivariated
(n = 255) 1.0 1.0
(n = 262) 1.43 (1.21, 1.70) 1.49 (1.25, 1.78)
(n = 72) 1.43 (1.10, 1.85) 1.51 (1.16, 1.97)
(n = 29) 1.28 (0.87, 1.88) 1.31 (0.88, 1.95)
<0.01 NS
GB and biliary system Agec Multivariated
(n = 39) 1.0 1.0
(n = 24) 0.88 (0.53, 1.47) 0.89 (0.53, 1.49)
(n = 6) 0.75 (0.32, 1.78) 0.74 (0.31, 1.75)
(n = 6) 1.47 (0.62, 3.51) 1.38 (0.57, 3.30)
NS NS
Pancreas Agec Multivariated
(n = 28) 1.0 1.0
(n = 31) 1.63 (0.98, 2.72) 1.65 (1.65, 1.65)f
(n = 13) 2.28 (1.18, 4.40) 1.98 (1.98, 1.98)f
(n = 5) 1.54 (0.59, 4.01) 1.56 (1.56, 1.56)f
NS NS
Larynx Agec Multivariated
(n = 11) 1.0 1.0
(n = 6) 0.79 (0.29, 2.15) 0.76 (0.28, 2.06)
(n = 4) 1.84 (0.58, 5.78) 1.72 (0.55, 5.43)
(n = 2) 1.64 (0.36, 7.48) 1.55 (0.34, 7.09)
NS NS
Lung Agec Multivariated
(n = 73) 1.0 1.0
(n = 66) 1.33 (0.96, 1.86) 1.32 (0.93, 1.86)
(n = 34) 2.27 (1.51, 3.41) 2.39 (1.58, 3.61)
(n = 33) 3.87 (2.56, 5.86) 3.74 (2.42, 5.76)
<0.01 <0.01
Hematopoietic system Agec Multivariated
(n = 39) 1.0 1.0
(n = 20) 0.71 (0.41, 1.21) 0.68 (0.39, 1.19)
(n = 6) 0.99 (0.42, 2.37) 1.01 (0.42, 2.41)
(n = 3) 1.11 (0.34, 3.65) 1.16 (0.35, 3.82)
NS NS
Kidney Agec Multivariated
(n = 53) 1.0 1.0
(n = 34) 0.90 (0.59, 1.39) 0.96 (0.62, 1.49)
(n = 11) 1.11 (0.58, 2.13) 1.14 (0.59, 2.19)
(n = 6) 1.25 (0.53, 2.93) 1.28 (0.54, 3.00)
NS 0.07
Urinary bladder Agec Multivariated
(n = 29) 1.0 1.0
(n = 22) 1.09 (0.62, 1.89) 1.09 (0.62, 1.92)
(n = 12) 2.26 (1.15, 4.45) 2.32 (1.17, 4.59)
(n = 7) 2.38 (1.03, 5.49) 1.75 (0.67, 4.58)
<0.01 <0.05
Others Agec Multivariated
(n = 198) 1.0 1.0
(n = 146) 1.05 (0.85, 1.30) 1.05 (0.85, 1.31)
(n = 22) 0.61 (0.39, 0.95) 0.62 (0.40, 0.97)
(n = 22) 1.17 (0.75, 1.82) 1.21 (0.77, 1.90)
NS NS
(n=) indicates cancer incidence, NS: not significant, GB: gall bladder. a Relative risk with Poisson regression. b For trends in the current smokers. c Adjusted for age. d Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preferences for vegetables and meats, and frequency of meat consumption. e Adjusted for age, place of residence, alcohol consumption, BMI, leisure-time physical activity frequency, preference for food saltiness, preferences for vegetables and meats, and frequency of meat consumption. f Standard error is zero.
22
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
Table 5 Population attributable risk of major smoking-related cancers in men Cancer site
Mouth and pharynx Esophagus Stomach Liver Pancreas Larynx Lung Urinary bladder All
PARa (%) Ageb
Multivariatec
45.8 98.3 36.0 28.3 39.9 59.5 78.7 52.0 30.4
41.3 86.1 36.1 32.8 37.8. 59.5 78.3 50.2 29.8
a Population attributable risk (smoking prevalence: former, 15.7%; current, 67.6%). b Adjusted for age. c Adjusted for multivariate.
Furthermore, the mean age of our study subjects at enrollment was about 43, relatively lower than in CPS II [10,12]. Therefore, our study had fewer long-term heavy smokers. Among the 404,511 current smokers in our study, only 1166 (0.29%) smoked 20 cigarettes or more a day for 30 years or more. The corresponding figure among 164,226 former smokers was 747 (0.45%). Because smoking is against the law for teenagers, Korean men start smoking at an average of 20.7 years of age [22], relatively later than the 17 years in US CPS II [10]. The older starting age for smoking may explain, in part, the lower relative risks in our study cohort. The difference in genetic susceptibility might have also contributed by modifying the smoking-related cancer risk [44]. For example, the frequency of *4 null alleles of CYP2A6, the enzyme that activates N-alkyl nitrosmaines such as NNK, is high in Koreans (11.0% compared to 1% in Caucasians) [45,46]. Inactive CYP2A6 alleles may decrease the risk of becoming tobaccodependent, and one study showed that the *4 null alleles of CYP2A6 significantly reduced the lung cancer risk [47]. Finally, while our study is based on incidence data, most other cohort studies are based on mortality data. Since after cancer diagnosis survival would have to be poorer among smokers than non-smokers [48–51], mortality-based studies are likely to show a stronger association with higher RR than incidence-based studies. Nevertheless, we believe the most important contributing factor to be the short follow-up period of the relatively healthy young male government employees and teachers. If we extend the follow-up period, we are likely to find a higher RR since more cases of smoking-related cancers will be diagnosed, as the participants get older. In prospective cohort and case-control studies, the combination of smoking with other factors increases the risk of smoking-related cancers [52,53]. In our study, however, socio-environmental factors such as place of residence and personal factors such as leisure-time physical activity, body mass index, alcohol consumption, and dietary factors had little impact on the relative and attributable risk
for the major smoking-related or overall cancer incidence. This is consistent with the recent CPS II [12], which suggests that despite differences in ethnicities and personal life style, smoking is responsible for the major portion of smokingrelated cancers. Our study had several limitations. The study population differed in general from other Koreans in terms of socioeconomic and educational status. Since they were government employees and teachers, they may have had behaviors that were conducive to better health. However, since multivariate analysis as compared with age-adjusted analysis revealed only minor decreases in cancer risk estimates from smoking, the differences of socioeconomic and educational status between the general population and our study population may show small difference in cancer risk. We did not apply the strict definition of never smokers used by CPS II, which defined current smokers as those who reported smoking at least one cigarette a day for 1 year or more [10,12]. Misclassification of never smokers as current smokers in our study could result in an underestimated cancer risk. Additionally, we could not adjust for the effects of environmental tobacco smoke and changes in smoking behavior during follow-up. In conclusion, our study, which was based on incidence data rather than mortality data, confirmed that smoking increases the risk of many cancers, including oral and pharyngeal, esophageal, stomach, liver, pancreatic, laryngeal, lung, and urinary bladder cancers. Overall, the PAR due to smoking was 30.4% of all cancers, and differences in smoking were responsible for most of the differences observed in smoking-related cancers.
Acknowledgements We appreciate the help provided by the Korean Central Cancer Registry (KCCR) and those of six regional cancer registries (Seoul, Busan, Incheon, Daejeon, Kwangju, and Daegu).
References [1] Wynder EL, Graham EA. Tobacco smoking as a possible etiologic factor in bronchogenic carcinoma. JAMA 1950;43:329–36. [2] Levin ML, Goldstein H, Gerhardt PR. Cancer and tobacco smoking. JAMA 1950;143:336–8. [3] Kahn HA. The Dorn study of smoking and mortality among US veterans. Report on eight and one-half years of observation. In: Haenszel W, editor. Epidemiological approaches to the study of cancer and other chronic diseases. National Cancer Institute Monograph, vol. 19. Rockville, MD: US Department of Health, Education, and Welfare; 1966. p. 1–125. [4] Doll R, Peto R. Mortality in relation to smoking: 20 years’ observations on male British doctors. BMJ 1976;2:1525–36. [5] Doll R, Gray R, Hafner B, Peto R. Mortality in relation to smoking: 22 years’ observations on female British doctors. BMJ 1980;5:967–71.
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24 [6] Hammond EC, Horn D. Smoking and death rates – report on forty-four months of follow-up of 187,783 men. II. Death rates by cause. JAMA 1958;166:1294–308. [7] Akiba S, Hirayama T. Cigarette smoking and cancer mortality risk in Japanese men and women-results form reanalysis of the six prefecture cohort study data. Environ Health Perspect 1990;87: 19–26. [8] US Department of Health, Education, and Welfare. Smoking and Health Report of the Advisory Committee to the Surgeon General of the Public Health Service. PHS Publication no. 1103. Washington: US Department of Health, Education, and Welfare, Public Health Service; 1964. [9] International Agency for Research on Cancer IARC Monographs on the evaluation of carcinogenic risks to humans, vol. 38. Tobacco smoking. Lyon, France: IARC; 1986. [10] Thun MJ, Day-Lally C, Myers DG, et al. Trends in tobacco smoking and mortality from cigarette use in Cancer Prevention Studies I (1959 through 1965) and II (1982 through 1988). In: National Cancer institute, smoking and tobacco control, monograph 8: changes in cigarette-related disease risks and their implication for prevention and control. Bethesda, NIH publication no. 97-4713. Washington, DC: NIH; 1997. [11] Sterling TD, Rosenbaum WL, Weinkam JJ. Risk attribution and tobacco-related deaths. Am J Epidemiol 1993;138:128–39. [12] Thun MJ, Apicella LF, Henley SJ. Smoking vs. other risk factors as the cause of smoking-attributable deaths: confounding in the courtroom. JAMA 2000;284:706–12. [13] Baron JA, Rohan TE. Tobacco. In: Schottenfeld D, Fraumeni JF, editors. Cancer epidemiology and prevention. 2nd ed. New York: Oxford University Press; 1996. [14] Doll R, Peto R, Wheatley K, et al. Mortality in relation to smoking: 40 years’ observations on male British doctors. Br Med J 1994;309:901– 11. [15] World Health Organization. International classification of disease for oncology. 2nd ed. Geneva, Switzerland: WHO; 1990. [16] Ministry of Health and Welfare, Korea. Annual report of Korea Central Cancer Registry program, January–December 2000. Seoul: Ministry of Health and Welfare; 2002. [17] World Health Organization. Physical status: the use and interpretation of anthropometry: report of a WHO Expert Committee. World Health Organization 1995;854:1–452. [18] Tsugane S, Sasazuki S, Kobayashi M, Sasaki S. Salt and salted food intake and subsequent risk of gastric cancer among middle-aged Japanese men and women. Br J Cancer 2004;12(90):128–34. [19] Van den Brandt PA, Botterweck AA, Goldbohm RA. Salt intake, cured meat consumption, refrigerator use and stomach cancer incidence: a prospective cohort study (Netherlands). Cancer Causes Control 2003;14:427–38. [20] Xibin S, Moller H, Evans HS, Dixing D, Wenjie D, Jianbang L. Residential environment, diet and risk of stomach cancer: a casecontrol study in Linzhou, China. Asian Pac J Cancer Prev 2002;3:167– 72. [21] Rothman KJ, Greenland S. Modern epidemiology. 2nd ed. Philadelphia: Lippincott-Raven Publishers; 1998. [22] Ministry of Health and Welfare, Korea. 1998 Korea Health Survey – Summary Report. Seoul: Ministry of Health and Welfare; 1999. [23] SAS. SAS/STAT software: changes and enhancement through release 8.1. Cary (NC): SAS Institute Inc.; 2002. [24] Hara M, Sobue T, Sasaki S, et al. Smoking and risk of premature death among middle-aged Japanese: ten-year follow-up of the Japan Public Heath Center-based prospective study on cancer and cardiovascular disease (JPHC Study) cohort I. Jpn J Cancer Res 2002;93:6– 14. [25] Tredaniel J, Boffetta P, Buiatti E, et al. Tobacco smoking and gastric cancer: review and meta-analysis. Int J Cancer 1999;83: 223–9.
23
[26] Shopland DR. Cigarette smoking as a cause of cancer. In: Harras A, Edwards BK, Blot WJ, Ries LA.., editors. Cancer rate and risks. 4th ed. National Cancer Institute: Bethesda, MD; 1996. [27] Chao A, Thun MJ, Henley SJ, et al. Cigarette smoking, use of other tobacco products and stomach cancer mortality in US adults: the Cancer Prevention Study II. Int J Cancer 2002;101:380–9. [28] Heineman EF, Zahm SH, et al. Increased risk of colorectal cancer among smokers: results of a 26-year follow-up of US veterans and a review. Int J Cancer 1995;59:728–38. [29] Newcomb PA, Storer BE, Marcus PM. Cigarette smoking in relation to risk of large bowel cancer in women. Cancer Res 1995;55: 4906–9. [30] Nyren O, Berstrom R, Nystrom L, et al. Smoking and colorectal cancer: a 20-year follow-up study of Swedish construction workers. J Natl Cancer Inst 1996;88:1302–7. [31] Sturmer T, Glynn RJ, Lee IM, et al. Lifetime cigarette smoking and colorectal cancer incidence in the Physician Health Study I. J Natl Cancer Inst 2000;92:1178–81. [32] Slattery ML, Curtin K, Anderson K, et al. Association between cigarette smoking, lifestyle factors, and microsatellite instability in colon tumors. J Natl Cancer Inst 2000;92:1831–6. [33] Terry PD, Miller AB, Rohan TE. Prospective cohort study of cigarette smoking and colorectal cancer risk in women. Int J Cancer 2002;99:480–3. [34] Giovannucci E, Rimm EB, Stampfer MJ, et al. A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S. men. J Natl Cancer Inst 1994;86:183–91. [35] Giovannucci E, Colditz GA, Stampfer MJ, et al. A prospective study of cigarette smoking and risk of colorectal adenoma and colorectal cancer in U.S. women. J Natl Cancer Inst 1994;86:192–9. [36] Trichopoulos D, MacMahon B, Sparros L, et al. Smoking and hepatitis B-negative primary hepatocellular carcinoma. J Natl Cancer Inst 1980;65:111–4. [37] Yu MC, Mack T, Hanisch R, et al. Hepatitis, alcohol consumption, cigarette smoking, and hepatocellular carcinoma in Los Angeles. Cancer Res 1983;43:6077–9. [38] Carstensen JM, Pershagen G, Eklund G. Mortality in relation to cigarette and pipe smoking: 16 years’ observation of 25,000 Swedish men. J Epidemiol Community Health 1987;41:166–72. [39] Mizoue T, Tokui N, Nishisaka K, et al. Prospective study on the relation of cigarette smoking with cancer of the liver and stomach in an endemic region. Int J Epidemiol 2000;29:232–7. [40] Hsing AW, McLaughlin JK, Hrubec Z, et al. Cigarette smoking and liver cancer among US veterans. Cancer Causes Control 1990;1:217– 21. [41] Kuper H, Boffetta P, Adami HO. Tobacco use and cancer causation: association by tumour type. J Int Med 2002;252:206–24. [42] Korea Institute of Tuberculosis. The survey on the smoking habits in Korea. Tubercul Resp Dis 1985;21:31–45. [43] Korean National Statistical Office. Report on the social statistics survey. Seoul: Korean National statistical Office; 2000. [44] Tyndale RF, Sellers EM. Variable CYP2A6-mediated nicotine metabolism alters smoking behavior and risk. Drug Metab Dispos 2001;29:548–52. [45] Kwon JT, Nakajima M, Cahi S, et al. Nicotine metabolism and CYP2A6 allele frequencies in Koreans. Pharmacogenetics 2001;11:317–23. [46] Oscarson M. Genetic polymorphism in the cytochrome P4502A6 (CYP2A6) gene: implications for interindividual differences in differences in nicotine metabolism. Drug Metab Dispos 2001;29: 91–5. [47] Miyamoto M, Umetsu Y, Dosaka-Akita H, et al. CYP2A6 gene deletion reduces susceptibility to lung cancer. Biochem Biophys Res Commun 1999;261:658–60. [48] Yu GP, Ostroff JS, Zhang ZF, et al. Smoking history and cancer patient survival: a hospital cancer registry study. Cancer Detect Prev 1997;21:497–509.
24
Y.H. Yun et al. / Cancer Detection and Prevention 29 (2005) 15–24
[49] Fujisawa T, Iizasa T, Saitoh Y, et al. Smoking before surgery predicts poor long-term survival in patients with stage I non-small-cell lung carcinomas. J Clin Oncol 1999;17:2086–91. [50] Browman GP, Mohide EA, Willan A, et al. Association between smoking during radiotherapy and prognosis in head and neck cancer: a follow-up study. Head Neck 2002;24:1031–7. [51] Videtic GM, Stitt LW, Dar AR, et al. Continued cigarette smoking by patients receiving concurrent chemoradiotherapy for limited-stage
small-cell lung cancer is associated with decreased survival. J Clin Oncol 2003;21:1544–9. [52] Sterling TD, Rosenbaum WL, Weinkam JJ. Risk attribution and tobacco-related deaths. Am J Epidemiol 1993;138: 128–39. [53] Midgette AS, Baron JA, Rohan TE. Do cigarette smokers have diets that increase their risks of coronary heart disease and cancer. Am J Epidemiol 1993;137:521–9.