Tobacco Documents Online

h British journal of Urology (1994), 74, 753-756 Urothelial cancer and cigarette smoking: fmdings from a regional case-controlled study T. SORAHAN, R.J. LANCASHIRE and G. SOLE* Department of Public Health and Eplderniology, University of Birmingham, Edgbaston, Birmingham and *County Hospital, Hereford, UK Objective To quantify the risks of urothelial cancer associated with smoking cigarettes. Subjects and methods Histories were taken from 989 patients diagnosed with urothelial cancer, 2059 dec- teral register controls and 1599 General Practitioner controls. Histories of patients and controls were com- pared by means of logistic regression. Results Statistically significant trends were observed for the risk of urothelial cancer to increase with estimated lifetime consumption of cigarettes. Among ex-smokers, risks reduced with time since quitting, and ex-smokers Introduction An earlier report from a large, regional, case-controlled study, carried out primarily to study urothelial cancer and the use of dyed maggots by coarse fishermen, did not indicate that angling or the use of dyed maggots by anglers were risk factors for urothelial cancer [1]. The report emphasized the importance of the established risk factor of cigarette smoking [2]. This second report pro- vides further information on the quantification of risk, the effects of using different types of cigarettes and the effects of stopping smoking on risk reduction. who quit 20 or more years ago experienced risks similar to lifelong non-smokers. Type of cigarette Was not an important predictor of risk. The two control series provided estimates of 45% and 33% respectively for the percentage of cancers in the series attributable to cigarette smoking. Conclusion Cigarette smoking is the single most import- ant cause of urothelial cancer, but stopping smoking leads quickly to reductions in risk. Keywords Urothelial cancer, smoking, epidemiology, case-controlled study Subjects and methods i A description of the case series and the two control series has been given in an earlier report [1]. In this new report, the smoking histories of 989 white men, who were resident in the West Midlands Region and diagnosed with urothelial cancer (transitional cell) in the period 1985-87, are compared in turn with those of two control series. The first control series, an unmatched group of 2059 male residents, was selected from the electoral registers of the Region. The second series of 1599 patients, matched on sex and year of birth, was assembled with the assistance of the General Practitioners (GP) of the cases. All study subjects in these three series had returned a b Accepted for publication 18 February 1994 simple one-page postal questionnaire requiring yes/no answers on five occupations, four sports and three types of tobacco usage (pipe, cigars, cigarettes). For those subjects who had 'ever regularly smoked cigarettes', information was sought on age at starting, age stopped (if stopped), average number of cigarettes a day (five choices supplied: less than 10, about 10, about 20, about 30, 40 plus) and the type of cigarette usually smoked (filter, non-filter and handmade). All smoking histories for this analysis were calculated with reference to smoking status in 1985. Histories from all 989 patients were compared with those from the electoral register controls by means of unconditional logistic regression [3] after adjusting for age in 1985 (15-29, 30-39, 40-49, 50-54, 55-59, 60-64, 65-69, 70-74). A further analysis compared histories from 659 patients with one or more GP controls with histories from the GP controls by means of conditional logistic regression applied to matched sets [3]. Results Relative risks derived from case and electoral register control histories are shown in Table 1. Ex-smokers who quit 20 or more years ago experienced only about 10% of the excess risk experienced by current smokers (1.24-1)/(3.12-1), and a regular (monotonic) trend is shown for risk reduction with period since quitting. This trend remained after adjustment for lifetime consumption of cigarettes. Statistically significant positive trends 753 This arficlo is for individual use only and may not Im furth~ reproduced or storod electronically without wdtte~ pen~sslon from the copyright holder.

754 T. SORAHAN et al. Table 1 Risks of urothelial cancer with respect to cigarette smoking habits Electoral General register Relative risl~f practitioner Relative Variable Patients controls (with 95% ¢I) Patients controls (with 95% Smoking status Non-smoker 135 765 1.0 102 359 1.0 Current smoker 375 535 3.12"** (2.38-4.09) 248 382 2.26*** (1.71-2.99) Quit 1-9 years ago 182 340 1.94"** (1.43-2.63) 114 272 1.52" (1,10-2.09) Quit 10-19 years ago 145 239 1.54"* (1.12-2.11) 86 253 1.20 (0.85-1.68) Quit 20 or more years ago 152 180 1.24 (0.90-1.70) 109 333 1.17 (0.85-1.61) P (trend): excluding non-smokers P<O.O01 P (trend); excluding non-smokers P<O.O01 Average cigarettes per day <10 74 147 1.46 (0,99-2.14:) 50 179 0.98 (0.66-1.46) About 10 202 259 ]..92*** (1.4:2-2.59) 130 270 1.73"** (1.26-2.37) About 20 389 588 2.28*** (1.76-2.96) 265 504 1.89"** (1.4:4:-2.49) About 30 118 202 2.00*** (1.42-2.80) 67 170 1.42 (0.99-2.0,t) 40 or more 71 98 2.20*** (1.46-3.3I) 45 117 1.35 (0.88-2.05) P (trend) <0.001; excluding non-smokers P=O.101 P (trend) <0.001: excluding non-smokers P=0.365 Type of cigarette Filter 427 827 2.15"** (1.67-2.77) 274 561 1.71"** (1.31-2.24) Non-filter 341 332 1.86"** (1.42-2.44) 225 558 1.50"* (1.13-2.00) Handmade 86 135 2.31"** (1.57-3.39) 58 121 1.59" (1.08-2.35) Lifetime consumption in pack-years 1-9 82 386 1.18 (0.83-1.66) 58 215 0.94: (0.65-1.36) 10-19 14:2 300 1.76"* (1.28-2.40) 88 225 1.46" (1.03-2.06) 20-29 168 216 2.34*** (1.71-3.21) 108 201 1.93"** (J..39-2.69) 30-39 119 150 2.33*** (1.67-3.28) 82 163 1.76"** (1.24:-2.50) >40 343 242 2.60*** (1.97-3.43) 221 436 1.87"** (1.40-2.49) P (trend) < 0.001: excluding non-smokers P < 0.001 P (trend) <0.001; excluding non-smokers P < 0.001 Duration of smoking in years 1-9 30 263 0.93 (0.58-1.50) 19 10-19 86 350 1.37 (0.98-1.93) 62 20-29 14:0 289 1.84"** (1.35-2.51) 91 30-39 197 194 2.47*** (1.82-3.34) 122 >4:0 401 198 2.85 *** (2.15-3.79) 263 P (trend) < 0.001; excluding non-smokers P < 0.001 P (trend Age at starting to smoke 7-16 403 682 2.25*** (1.73-2.91) 17-20 337 498 1.87"** (1.43-2.4:3) >21 114 114 1.98"** (1.39-2.83) P (trend); excluding non-smokers P--0.175 76 0.85 (0.49-1.4:9) 177 1.29 (0.89-1.86) 216 1.51" (1.08-2.13) 287 1.53"* (1.12-2.10) 4:84: 2.13"** (1.60-2.84) 0.001; excluding non-smokers P < 0.001 253 556 1.63"** (1.24-2.14) 227 496 1.64"** (1.24-2.17) 77 188 1.51" (1.06-2.15) P (trend); excluding non-smokers P--0.831 I ! I ! I * P<0.05. ** P<0.01. *** P<0.001. ~', Risks expressed relative to the baseline category of non-smokers, calculated from unconditional logistic regression with adlustment for age (eight levels). ¢, Risks expressed relative to the baseline category of non-smokers, calculated from conditional logistic regression applied to matched sets (P < 0,001) are shown for separate analyses of cigarettes per day, lifetime consumption and duration of smoking, although the trend for cigarettes per day was non- significant when non-smokers were excluded from the British Journal of Urology (1994). 74 calculation. Risks associated with the use of either falter or plain cigarettes were similar, and this finding was not materially changed by simultaneous adjustment for lifetime consumption of cigarettes. Age at starting to

!1 I I I I I I UROTHELIAL CANCER AND CIGARETTE SMOKING 755 smoke was not an important predictor of risk either with or without adjustment for lifetime consumption. On the basis of the relative risks shown for lifetime consumption, 45% of the bladder cancers in this series are attributable to cigarette smoking. Corresponding findings derived from case and GP control histories are also shown in Table 1. A regular trend is shown for risk reduction with period since quitting. Statistically significant positive trends were also found for cigarettes per day, lifetime consumption and duration of smoking, although the trend with cigarettes per day is clearly dependent on a smoker/non-smoker comparison rather than reflecting a trend across the smoking categories themselves. On the basis of the relative risks shown for lifetime consumption, 33% of the bladder cancers in this series are attributable to cigarette smoking. Relative risks (not shown) were calculated from simul- taneous analysis of duration of smoking and cigarettes per day. For patients and electoral register controls, trends were discernible both for duration of smoking within cigarettes per day categories and for cigarettes per day within duration of smoking categories. Corresponding findings derived from patients and GP controls were not clear-out. Table 2 shows relative risks for three smoking vari- ables calculated separately for current smokers and ex-smokers. Trend statistics were calculated after exclud- ing non-smokers; trends across the 'exposed' categories are being assessed. Clear-cut trends are shown only for ex-smokers when cases were compared with electoral register controls. The elevated risks experienced by cur- rent smokers were not significantly associated with cigarettes per day, duration of smoking or lifetime consumption of cigarettes. Discussion Principal findings from the available data were consistent with the conclusions of an International Agency for Research on Cancer working group; cigarette smoking increases the risk of bladder cancer [2]. Importantly, from the point of view of health promotion, stopping smoking led to a marked reduction in risk and risks among long-term ex-smokers were similar to those of non-smokers. After excluding non-smokers, a group known to be at low risk, statistically significant dose- response effects were observed both for lifetime consump- tion of cigarettes and for duration of smoking. Smoking may be involved in at least two stages of bladder carcinogenesis. The findings of a marked reduction in risk with time since quitting and of daily consumption not being a risk factor among current smokers are consistent with smoking affecting a late stage for which any amount of regular smoking is sufficient to encourage turnout progression. If this interpretation were to be correct, the overall trends of risk with daffy consumption and with duration of smok- ing would have to be explained by smoking affecting another stage, an earlier stage. Confident interpretation of these issues is not assisted, however, by the very different findings for ex-smokers obtained from the two control groups; a satisfactory explanation for these different findings is not available. Type of cigarette was not found to be an important predictor of risk. The use of filter and plain cigarettes has been analysed in a number of studies [4-7]. Differences in risk estimates have been both modest and contrasting. Taken together, the studies are consistent with type of cigarette being unimportant. Burch et al. have noted, however, that as filter cigarettes were intro- duced only in the late 1950s, insufficient time may have elapsed for some of the studies to have fully evaluated their effects [8]. Important bias may have been introduced into this study if the smoking histories of the non-responders were different from those of the responders. Unfortunately, data for the non-responders were not avaffable from other sources, and the importance of this potential bias cannot be assessed. Bias would have been introduced if some heavy smokers (patients and controls) dairned they were light or moderate smokers. Bias would also have been introduced if only patients or only controls made such claims. Both biases are likely to be present, at least to some extent. In addition, smoking histories detailing changes in frequency of smoking and type of cigarette smokext were not available. The removal of these limi- tations would be unlikely to alter the Public Health message of the study; stopping smoking leads quickly to a reduction in risk of urothelial cancer, and cigarette smoking is unquestionably the single most important cause of this disease. Acknowledgements We thank all those consultants and GPs throughout the region who enabled us to collect these data, the Director and staff of the West Midlands Regional Cancer Registry, Pat Boyd for carrying out the interviews, Alison Taylor for running the ot~ce, Ann O'Connor for word pro- cessing, Debbie Pope for consolidation of data files, the Cancer Research Campaign for financial support, and patients and members of the public for returning our questionnaire. British lournal of Uroloay (1994), 74

756 T. SORAHAN et al. Table 2 Risks of urothelial cancer in current smokers and ex-smokers Patients and electoral register controls Patients and general practitioner controls Current smokers Ex-smokers* Current smokers E~-smokers Average cigarettes per day Non-smoker 1.O <10 2.13 About 10 3.68 About 20 3.60 About 30 1.83 40 or more 3.36 P (trend)~', excluding non-smokers (-) 0.751 Duration of smoking 1-19 3.18 20-29 2.63 30-39 3.01 40 + 3,24 P (trend), excluding non-smokers (+) 0.418 Lifetime consumption in pack-years 1-9 1.51 10-19 3.64 20-29 3.69 30-39 2.67 >40 3.13 P (trend), excluding non-smokers (+) 0.755 1.0 1.0 1.0 1.11 1.68 0.84 1.31 " 2.13 1.45 1.58 2.56 1.52 2.12 1.22 1.33 1.87 . 4.46 1.22 (÷) 0.009 (+) 0.270 (+) 0.898 1.12 2.17 1.22 1.63 1.65 1.43 2.04 2.27 1.26 1.88 2.23 1.61 (+) 0.001 (--) 0.676 (+) 0.187 1.12 1.47 1.02 1.30 1.72 1.46 1.56 2.26 1.46 2.05 2.66 1.45 2.04 2.20 1.48 (+) <0.001 (+) 0.141 (+) 0.376 * Quit anytime before 1st January, 1985. "~ 0 indicates whether slope is positive (+) or negative (-); significance levels of individual relative risks are not shown. References 1 Sorahan T, Sole G. Coarse fishing and urothelial cancer: a regional case-control study. Br I Cancer 1990; 62:138-41 2 international Agency for Research on Cancer IARC. Monographs on the Evaluation of the Carcinogenic Risk of Chemicals to Humans, Vol 38. Lyon: International Agency for Research on Cancer, 1986 3 Breslow NE, Day NE. Statistical Methods in Cancer Research. Vol. 1 The Analysis of Case-control Studies. Lyon: IARC Scientific Publications, 1980 4 Cartwright RA, Adib R, Appleyard Iet al. Cigarette smoking and bladder cancer: an epidemiological inquiry in West Yorkshire. 1 Epidemiol Commun Filth 1983; 37:256-63 5 Hartge P, Silverman D, Hoover R et al. Changing cigarette habits and bladder cancer risk; a case-control study. 1 Nat Cancer Inst 1987; 78:1119-25 6 Morrison AS, Buring ]E, Verhoek WG et al. An international study of smoking and bladder cancer. 1 Urol 1984; 131:650-4 7 Vineis P, Esteve J, Hartge P, Hoover R, Silverman DT, Terracini B. Effects of timing and type of tobacco in cigarette- induced bladder cancer. Cancer Res 1980; 48:3849-52 8 Butch JD, Rohan TE, Howe GR et al. Risk of bladder cancer by source and type of tobacco exposure: A case-control study Int ] Cancer 1989; 44:622-8 Authors T. Sorahan, PhD, Senior Research Fellow. RJ. Lancashire, BSc, Computer Officer. G. Sole, MS, FRCS, Consultant Urologist. Correspondence: Dr T. Sorahan, Institute of Occupational Health, University of Birmingham, Edgbaston, Birmingham B15 2T~, IlK. British lournal of Urology (1994), 74