吸烟与肠癌

REVIEW Smoking and Colorectal Cancer A Meta-analysis Edoardo Botteri, MSc Simona Iodice, MSc Vincenzo Bagnardi, PhD Sara Raimondi, MSc Albert B. Lowenfels, MD Patrick Maisonneuve, Eng ALTHOUGH TOBACCO HAS BEENresponsible for approxi-mately 100million deaths inthe past century and 5.4mil- lion in 2005 alone, there are still an es- timated 1.3 billion smokers in the world. Currently, smoking prevalence has decreased in the United States and some other countries but has in- creased in less-developed world re- gions.1 In the lung, where there is intimate exposure to inhaled tobacco smoke, about 80%of all primary cancers are at- tributable to smoking.2 Smoking- related cancers are also frequent in the oropharynx and larynx, where there is direct contact with tobacco-related car- cinogens. Smoking also increases the risk of cancer in organs such as the kid- ney, bladder, cervix, lower urinary tract, and pancreas, organs for which expo- sure to tobacco degradation products is indirect.3 With respect to the diges- tive tract, esophageal and gastric can- cers have been strongly associatedwith tobacco, whereas the smoking– colorectal cancer (CRC) link remains controversial. Several large cohort stud- ies linked smokingwithCRC,4-6 andour group has recently reported a meta- analysis that revealed that smoking doubles the risk of colorectal polyps, known precursors for CRC.7 How- ever, other studies have failed to de- tect a significant relationship between smoking and CRC.8,9 This inconsis- tency among studies could be partly ex- plained by different study designs, population characteristics, and the heterogeneous treatment of the most likely confounders, such as diet, alco- hol, physical activity, and bodymass in- dex (BMI, calculated as weight in ki- lograms divided by height in meters squared).10 Because smoking can potentially be controlled by individual and popula- tion-related measures, detecting a link See also Patient Page. Author Affiliations:Division of Epidemiology and Bio- statistics, European Institute of Oncology,Milan, Italy (Dr Bagnardi, Mss Iodice and Raimondi, and Messrs Botteri and Maisonneuve); Department of Statistics, University of Milan Bicocca, Milan, Italy (Dr Bag- nardi); and Departments of Surgery and Community and Preventive Medicine, New York Medical Col- lege, Valhalla (Dr Lowenfels). Corresponding Author: Edoardo Botteri, MSc, Divi- sion of Epidemiology and Biostatistics, European In- stitute of Oncology, Via Ripamonti 435, 20141, Mi- lano, Italy (edoardo.botteri@ieo.it). Context Colorectal cancer is the third most common form of cancer and the fourth most frequent cause of cancer deaths worldwide. The association between cigarette smoking and colorectal cancer has been inconsistent among studies. Objective To clarify the association of cigarette smoking and colorectal cancer, we performed a comprehensive literature search and ameta-analysis of observational stud- ies considering both incidence and mortality. Data Sources We performed a literature search using PubMed, ISI Web of Science (Science Citation Index Expanded), and EMBASE to May 2008, with no restrictions. We also reviewed references from all retrieved articles. Study Selection All articles that were independent and contained the minimum in- formation necessary to estimate the colorectal cancer risk associated with cigarette smoking and a corresponding measure of uncertainty. Data Extraction Articles were reviewed and data were extracted and cross- checked independently by 3 investigators, and any disagreement was resolved by con- sensus among all 3. Results One hundred six observational studies were included in the analysis of in- cidence. Twenty-six studies provided adjusted risk estimates for ever smokers vs never smokers, leading to a pooled relative risk of 1.18 (95% confidence interval [CI], 1.11- 1.25). Smoking was associated with an absolute risk increase of 10.8 cases per 100 000 person-years (95% CI, 7.9-13.6). We found a statistically significant dose- relationship with an increasing number of pack-years and cigarettes per day. How- ever, the association was statistically significant only after 30 years of smoking. Sev- enteen cohort studies were included in the analysis ofmortality. The pooled risk estimate for ever vs never smokers was 1.25 (95%CI, 1.14-1.37). Smoking was associated with an absolute risk increase of 6.0 deaths per 100 000 person-years (95% CI, 4.2-7.6). For both incidence and mortality, the association was stronger for cancer of the rec- tum than of the colon. Conclusion Cigarette smoking is significantly associated with colorectal cancer in- cidence and mortality. JAMA. 2008;300(23):2765-2778 www.jama.com ©2008 American Medical Association. All rights reserved. (Reprinted) JAMA, December 17, 2008—Vol 300, No. 23 2765 Downloaded From: http://jama.jamanetwork.com/ by a Lanzhou University User on 11/01/2013 between CRC and smoking could help reduce the burden of the world’s third most common tumor, which cur- rently causes more than 500 000 an- nual deaths worldwide.11 In the United States alone, an estimate of approxi- mately 50 000 deaths fromCRCwould have occurred in 2008.12 We conducted a meta-analysis with the following aims: (1) to review and summarize published data examining the link between smoking and CRC in- cidence and mortality; (2) to measure the smoking-CRC relationship accord- ing to different characteristics of the study populations, study designs, and CRC subsites; and (3) to study dose- response patterns of tobacco exposure on the risk of CRC. METHODS Search Strategy, Inclusion Criteria, and Data Abstraction We performed the following literature search toMay 2008 using PubMed and EMBASE,without restrictions: ([Smoke or cigarette or tobacco or smoking] and Cancer and [Colon or Rectum or Colo- rectal or Colorectum or Colon rectum]) or (“Colorectal cancer” [MeSH Major Topic] and “epidemiologic studies” [Mesh Terms]) (Medical Subject Headings). We also identified the most cited ar- ticles on the topic using ISI Web of Knowledge (ScienceCitation Index Ex- panded; Journal Citation Report) and reviewed articles quoting them. In ad- dition, we reviewed the references of all articles of interest and of the Interna- tional Agency for Research on Cancer Monographs on tobacco smoke and in- voluntary smoking10 to identify addi- tional relevant studies.Only reports ful- filling the following inclusion criteria were included in the meta-analysis. 1. Studies that contained the mini- mum information necessary to esti- mate the relative risk (RR) associated with tobacco smoking and a corre- sponding measure of uncertainty (ie, 95% confidence interval [CI], stan- dard error, variance, or P value of the significance of the estimate). 2. Case-control and cohort studies, published as original articles; ecologi- cal and prevalence studies were ex- cluded. 3. Studies that were independent. In case of multiple reports on the same population or subpopulation, we con- sidered the estimates from themost re- cent or most informative report. 4. Studies inwhichpopulationswere representative of the general popula- tion and therefore were free from dis- eases, such as ulcerative colitis or dia- betes, potentially modifying the smoking-related CRC risk. When available, we used adjusted risk estimates and those based on popu- lation-based controls. Articles were re- viewed and data were extracted and cross-checked independently by 3 in- vestigators (E.B., S.I., and S.R.). Anydis- agreement was resolved by consensus among the 3. An important study based on theBrit- ish Doctors study13 was excluded from the meta-analysis because no measure of variability for the mortality rate was reported, but its results were used for other analyses. Data Analysis The RR was used as a measure of the association between cigarette smok- ing and CRC. For case-control stud- ies, the odds ratio (OR) was used as es- timates of the RR because CRCs are sufficiently rare.14 When cohort stud- ies reported only crude data and no in- formation on person-years, we treated it as a control study using noncases as controls. We used random rather than fixed- effects models to estimate pooled RRs in order to take into account the hetero- geneity, however small, of the risk es- timates and therefore to be more conservative. Homogeneity of effects across stud- ies was assessed using the �2 statistic and quantified by I2, which represents the percentage of total variation across studies that is attributable to hetero- geneity rather than chance.15 Sub- group analyses and meta-regression models were carried out to investigate potential sources of between-study heterogeneity. When several risk estimates were present in a single study (ie, separate estimates for proximal and distal co- lon), we adjusted the pooled esti- mates for intrastudy (or within-study) correlation.16 In the dose-response analysis, we considered cigarettes per day, pack- years, and duration of smoking as ex- planatory variables. Because for many studies continuous exposures were re- ported as categorical data with a range, we assigned themid-point of the range as the average exposure. For the high- est open categories, we considered 60 cigarettes per day, 60 pack-years, and 60 years of duration as the maximum. In pooling dose-response data, we took into account correlation between RRs within the same study, using the method described by Greenland and Longnecker.17 Both linear and nonlin- earmodels were fitted and evaluated on the logarithm of the RR. Nonlinear trends were evaluated using fractional polynomial curves of the secondorder.18 To consider differences among stud- ies as a further source of random vari- ability, an additional component of the variance was estimated and added in weighting each observation.19 The model minimizing Akaike informa- tion criterion (AIC), a penalized like- lihood that takes into account the num- ber of parameters estimated in the model, was used as a general rule in the model choice. Publication bias was evaluated by funnel plots and quantified by the tests developed by Egger et al20 and Begg et al.21 All analyses were performed with SAS software version 8.02 (SAS Insti- tute Inc, Cary, North Carolina). RESULTS Detailed search steps are described in FIGURE 1. Briefly, from the initial lit- erature search we identified and screened 1663 abstracts. Two hun- dreds forty-one articles were consid- ered of interest and full text was re- trieved for detailed evaluation. References of all 241 articles were re- viewed, and 6 additional relevant stud- ies were identified. One hundred SMOKING AND COLORECTAL CANCER 2766 JAMA, December 17, 2008—Vol 300, No. 23 (Reprinted) ©2008 American Medical Association. All rights reserved. Downloaded From: http://jama.jamanetwork.com/ by a Lanzhou University User on 11/01/2013 twenty-six of these 247 articles were subsequently excluded from themeta- analysis (80 did not satisfy the inclu- sion criteria, 46were based on the same study populations). Smoking and CRC Incidence One-hundred six independent observa- tional studies thatmet the inclusion cri- teria were included in the final analysis of incidence (TABLE 1 and TABLE 2 and eTable 1 available at http://www.jama .com).Overall, themeta-analysis is based on a total of 39 779 incident cases. Studieswere publishedbetween1969 and 2008; 31 were conducted in North America, 39 in Europe, 33 in Asia, and 3 in other areas.One hundred fourwere written in English, 1 in Korean, and 1 in French. Sixty-five studies (61%) pro- vided at least 1 adjusted risk estimate, and 43 (66%) of them reported an ad- justed estimate for at least 1 of themain lifestyle-related potential confound- ers (diet, BMI, alcohol consumption, and physical activity). Detailed infor- mation on adjustments is reported in Table 1 and Table 2. Quantitative Data Synthesis. Ini- tially, we considered all studies provid- ing either adjusted risk estimates or only crudedata fromwhichwecalculatedun- adjusted risk estimates. Although the pooledRR for ever vs never smokerswas similar for adjusted and nonadjusted studies (adjustedRR,1.18; 95%CI, 1.11- 1.25; TABLE 3 and FIGURE 2; unad- justed RR, 1.11, 95% CI, 1.05-1.31; P= .45), unadjusted risk estimates were highly heterogeneous (�2 P�.01; I2, 77%). Because adjusted risk esti- mates are generally more reliable and unbiased than unadjusted ones, we de- cided to limit themain analysis to stud- ies providing adjusted risk estimates, which did not show evidence of hetero- geneity (�2 P=.17; I2, 28%), and to re- port an extended analysis based on all studies as supplementary material. The pooled RRs for current vs never smokers and former vs never smokers were respectively 1.07 (95% CI, 0.99- 1.16) and 1.17 (95% CI, 1.11-1.22; Table3).Sexwasnotasignificant source ofheterogeneity,while theriskestimates werehigher for rectal thanforcoloncan- ceramongcurrent smokers (P=.02)and ever smokers (borderline significant, P=.08).Whenweconsideredcolonsub- sites,wefoundanonsignificantincreased risk for cancer of the proximal colon compared with the distal colon among everand formersmokers.Whenconsid- eringadjustedandunadjustedestimates together, cohort studies showedhigher risk estimates compared with case- control studies (P=.07; eTable 2). We found higher risks in studies in which controls underwent colonoscopy than instudies inwhichtheydidnot.Thiswas statistically significant for ever smokers (eTable 2 available at http://www.jama .com) and for former smokers (Table 3). Risk estimates were systematically higher in population-based than in hos- pital-based studies, but the difference was not statistically significant. After evaluating dose-response pat- terns for cigarettes per day, pack- years, and duration of smoking for ever vs never smokers, we observed a lin- ear increase in risk with increasing smoking consumption The risk increased by 7.8% (95% CI, 5.7%- 10.0%) for every additional 10 cigarettes per day or by 4.4% (95% CI, 1.7%- 7.2%) for every additional 10 pack- years (for example an individual who smoked 1 pack of cigarettes per day for 50 years or 2 packs per day for 25 years has a 24% increased risk of develop- ing CRC compared with someone who never smoked). We also observed a nonlinear increase in risk with increas- ing smoking duration (FIGURE 3). The risk starts to increase after approxi- mately 10 years of smoking and reaches statistical significance after 30 years. From 19 cohort studies that re- ported information on person-years in smokers andnonsmokers,we could cal- culate absolute annual rates of CRC cases: 65.5 cases per 100000 in smokers and 54.7 per 100 000 in nonsmokers, corresponding to an absolute risk Figure 1. Flowchart of Selection of Studies for Inclusion in Meta-analysis 121 Independent studies included in the meta-analysis of smoking and colorectal cancer 104 Incidence 15 Mortality 2 Both incidence and mortality 167 Eligible for inclusion in meta-analysis 247 Full-text articles considered for inclusion 6 Additional studies identified from retrieved articles 1663 Articles identified in MEDLINE, EMBASE, ISI Web of Science (Science Citation Index Expanded) search 241 Identified for possible inclusion 46 Excluded (duplicate report on the same study population) 80 Excluded 39 Not enough data to estimate relative risk and 95% confidence interval 15 No data on smoking 14 No colorectal cancer cases 6 No controls 6 Study population with particular disease 1422 Excluded (title and/or abstract were not relevant for the end point of the study) SMOKING AND COLORECTAL CANCER ©2008 American Medical Association. All rights reserved. (Reprinted) JAMA, December 17, 2008—Vol 300, No. 23 2767 Downloaded From: http://jama.jamanetwork.com/ by a Lanzhou University User on 11/01/2013 Table 1. Case-Control Studies Reporting Adjusted Incidence Risk Estimatesa Source (Country or Region) No. of Cases No. of Controls Sexb Site of Cancer Adjusting Variablesc Smoking Category Tuyns et al,68 1982 (Europe) 340 NA Both Colon, rectal Age, sex Ever Peters et al,69 1989 (United States) 147 147 M Colon, rectal, colorectal Age, other Former Kato et al,70 1990 (Japan) 221 578 Both Colon, rectal Age, sex, other Current, former Slattery et al,71 1990 (United States) 231 391 M, W Colon Age, BMI, diet Ever Choi and Kahyo,72 1991 (Korea) 130 390 M Colon, rectal, colorectal Age, diet, alcohol, other Current, former, ever Olsen and Kronborg,73 1993 (Europe) 49 362 Both Colorectal Age, sex, diet Current, former Baron et al,74 1994 (Europe) 569 512 Both Colon, rectal, colorectal Age, sex, BMI, diet, physical activity, other Current, former D’Avanzo et al,31 1995 (Europe) 1584 2879 M, W, Both Colon, rectal, colorectal Age, sex, diet, alcohol, family history, other Current, former Inoue et al,75 1995 (Japan) 432 31 782 M, W Colon, rectal Age Ever Kotake et al,76 1995 (Japan) 363 363 Both Colon, rectal Age, sex Current Newcomb et al,25 1995 (United States) 779 2315 W Colon, rectal Age, BMI, alcohol, family history, other Current, former, ever Freedman et al,8 1996 (United States) 163 326 Both Colorectal Age, sex, BMI, diet, alcohol, family history Current, former Slattery et al,22 1997 (United States)d 1993 2410 M, W Colon Age, diet, family history, physical activity, other Ever Slattery et al,77 2004 (United States) 1846 2221 M, W, Both Colon, rectal Age, BMI, alcohol, physical activity Current, former Yamada et al,78 1997 (Japan) 66 132 Both Colorectal Age, sex, BMI, alcohol Former Tavani et al,79 1998 (Europe) 1953 4154 Both Colon, rectal Age, sex, BMI, diet, alcohol, family history, physical activity, other Current, former Yoshioka et al,80 1999 (Japan) 106 100 Both Colorectal Age Ever Steindorf et al,81 2000 (Europe) 180 180 Both Colorectal Age, sex, other Current, former Chiu et al,82 2001 (United States) 1340 2434 M, W Colon, rectal Age, diet, family history, other Current, former, ever Ji et al,83 2002 (China) 1805 1552 M, W Colon, rectal Age, alcohol, other Current, former Sharpe et al,84 2002 (Canada) 585 905 M Colon, rectal Age, alcohol, family history, other Ever Tiemersma et al,85 2002 (Europe) 102 537 Both Colorectal Age, sex, BMI, alcohol, other Current, former Diergaarde et al,30 2003 (Europe) 176 249 Both Colorectal Age, sex, diet, alcohol Ever Kim et al,86 2003 (Korea) 125 247 Both Colorectal Age, sex, BMI, diet, alcohol, other Ever Minami and Tateno,87 2003 (Japan) 488 2444 M, W, Both Colon, rectal Age, sex, alcohol, family history, other Current, former, ever van der Hel et al,88 2003 (Europe) 258 871 W Colon, rectal, colorectal Age, BMI Ever Ho et al,89 2004 (China) 822 926 Both Colon, rectal, colorectal Age, sex, diet, alcohol, family history, physical activity, other Current Nkondjock and Ghadirian,90 2004 (Canada)d 402 688 M, W, Both Colon Ever Ghadirian et al,91 1998 (Canada) 402 688 Both Colon Age, sex, family history, other Ever Ates¸ et al,92 2005 (Europe) 181 204 Both Colorectal Age, sex Ever Jin et al,93 2005 (China) 140 280 Both Colorectal Age, sex, alcohol, family history Ever Chia et al,27 2006 (United States) 1792 1501 Both Colorectal Age, sex, BMI, family history, other Current, former, ever Verla-Tebit et al,26 2006 (Europe) 540 614 M, W, Both Colorectal Age, sex, BMI, diet, alcohol, family history, physical activity, other Current, former, ever Gao et al,94 2007 (China) 315 439 Both Colon, rectal, colorectal Age, sex, alcohol Ever Hu et al,95 2007 (Canada) 1723 3097 M, W Colon Age, BMI, physical activity, other Ever Le Marchand et al,96 1997 (United States)d 1192 1192 M, W Colon, rectal Age, BMI, diet, alcohol, family history, physical activity, other Current, former Lüchtenborg et al,23 2007 (United States) 1959 1959 M ,W Colorectal Age, BMI, diet, alcohol, family history, physical activity, other Current, former Steinmetz et al,97 2007 (Europe) 674 5456 M, W