Tobacco Documents Online

Type of cigarettes and cancers of the upper digestive and respiratory tract Carlo La Vecchia, Ettore Bidoli, Salvatore Barra, Barbara D'Avanzo, Eva Negri, Renato Talamini and Silvia Franceschi (Received 23 April 1990," revised 29 May 1990; aqcepted 7 June 1990) " The relationship between type of dgarettes smoked and tim risk of cancer of upper digestive and respiratory sites was invest/gated in a case-control study conducted in Northern Italy on 291 males with cancer oF the oral caviry or phaq, nx, 288 with cancer of the esophagus, 162 with cancer of the larynx, and 1,272 control subjects in hospital for acute conditions unrelated to tobacco or alcohol consumption. Using a distinction based on rat-yield or the brand smoked for the longest time (<22 mg, low to medium tar; ~22 rag, high tar), the multivariate relative risks among ever-smokers were 8.5 for low/medium and 16.4 for high tar cigarettes for oral and pharyngeal neoplasms, 3.3 and 7.8 for esophageal, and 4.8 and 7.1" for laryngeal cancers. The differences according to ~pe ofdgarettes were similar in proportional terms, and hence larger in absolute terms, when analysis was restricted to current smokers only. Thus, these data provide further quantitative evidence ou the importance of ripe of cigarette smoked on the risk of upper-digestive and respiratory tract cancers and have important public health implications. Key words: Cigarettes, neoplasms of respiratory tract, neoplasms of upper digestive tract. Introduction Over the last decade, evidence has accumulated that the risk of lung cancer associated with modern, lighter cigarettes, is lower than that from older, non-filter cigarettes. Although a precise evaluation is complicated by socio-demographic aspects and temporal factors related to changes in brand types, large multi-center studies have provided estimates of a lower risk of lung cancer among smokers of low tar cigarettes of the order of 30 to 50%.I-6 Published data are less available for other tobacco- related sites than for lung cancer. The American Cancer Society One Million Cohort Study found lower relative mortality for oral, esophageal, laryngeal, and bladder cancer in smokers of medium and low tar cigarettes.th.an. in srfiol~ers of-high" tar dig.arett.es, but n~mbers of cases wer~ relatively small.2 Additional information on cancer ..~ ~'.,~ ',~,,., is loruwcleci t~y a large case-control study conducted in several areas of the United States, which showed lower relative risks (R_R) for filter cigarettes in each stratum of quantity smoked,i In the same study,7 as well as in several European case-control studies,s- I0 there was a lower risk of bladder cancer assodated with use of newer cigarettes, which was attributed not only to the introduction of filters but also to the type of tobacco used, nitrosamines being less concentrated in blond than in black tobacco,it Case-control studies from France~2 and Italy,~3 moreover, provided some evidence that low tar cigarettes were associated with lower esophageal cancer risk, but the data were relatively scanty. To further'.explore this i.ssue, we have considered in this.ar~i.'de the relat.ionshipl beacon, q;pe of..cigarett~ .~. d. cance?s of the upper digestive and respiratory, tract, using The possibility of comparison of the pattern of risk for Dr, La Vecchia, D'Avanzo and Negri are at the Istituto di Richerche Farmacologiche 'Matio Negri', Via EEtrea 62, 20157 ,~filano, Italy. Dr La 7ecchia is al, o at the Institute of Social and Preventive Medicine, University of Lausanne, 1005 Lausanne, Switzerland. Dr~ Bidoli, Barra, Talamini and Franees~hi are in tae Aviano Cancer Center, 3308 l Aviano, Pordenone, Italy. Repdnt requests saould be addressed to Dr La ~ecchia at the lstituto di Ricerche Farnacologiche. This work was conducted within the framework of the National Research Council (CNPO, Applied Projects "Oncology' (Contract No. ~7.0154~.44) and *Risk Factors for Disease ", with contributions from the Italian Association for Cancer Research and the Itah'an League against Turnouts, Milan. dd Communications of Oxford ktd 69

C. La Veechi~ ~ =l. various sites and the size of the data set make this analysis of interest. Some of the present data on esophageal cancer have been published previously. ~ blaterials and methods The data were obtained from a series of ongoing studies of cancers of the uppei" digestive and respiratory tracts, . based on a netw. 9rk.of t.each!ng' a.nd general hospitals in the'Greater Milan area'and in the Provinc.e of Pordenone, northeastern Italy. The general design of these studies has been describ.ed previously. ~'14 Recruitment of cases and controls for the various studies began between 1984 and 1987, arid the present report is based on data collected before December 1989. ~ The cases were male.patie.nts'belo~, the ageof 7-5, who had histologically confirmed cancers of the oral cavity, and pharynx (,/ = 291), larynx (n = 162), .and esophagus (n'= 288), diagnosed within the year preceding the interview. They were admitted to the National Cancer Institute, several University Clinics or the Ospedale Maggiore di Milan (which includes the f6ur major teaching and general hospitals in the area), or the General Hospital of Pordenone. The median ages were 57 years for cases of cancer of the oral cavity, 59 for cases of cancer of the pharynx, 62 for cases of larynx cancer, and 60 for esophageal cancers. The comparison group was comprised of 1,272 male patients admitted for acute conditions to several University Clinics, to the Ospedale Maggiore of Milan, or the General Hospital of Pordenone. The age range was 22 to 74, and the median age was 57 years. After exdusion of patients with any diagnosis related to tobacco use or alcohol consumption, the main diagnostic categories were: traumatic conditions (25%); non- traumatic orthopedic disorders (26 %); acute surgical con- ditions, including plastic surgery (19%); and other illnesses such as diseases of the ear, nose, and throat, or disorders of the skin or teeth (30 %). Over 85 % of cases and controls resided in the regions of Lombardy or Friuli- Venezia-Giulia. On the average, less than two percent of eligibl8 subjects (cases and controls) refused to be interviewed. Tar yields of cigarettes of 18 of the most common Italian brands were determined by the Laboratory of the British Government Chemist using the same standard smoking procedure and analytical techniques as in production of the tar, nicotine, and carbon monoxide tables issued by the British Health Departments.15"16 These tables were used for .foreign cigarette brands, together with more recent (u.npublished) estimates of tar- yield..pro.duced by the Italian State Monopoly. The classi- fication adopted was ba[s'ed.on i:ai'-yield bht, gi~.en, the cut-0ff point chosen (22 rag), allows an approximately accurate distinction between filter and non-filter cigarettes, or black and blond tobacco, italian plain cigarettes, commonly smoked up to the early 1970s, had more than 22 mg of tar and were made with black tobacco.; Whereas. the newe.i" filtered, .blond-tobacco brands marketed since the 'mid 1970s, gerlerally ha~,e less than 20 mg of tar. ~5 Relative risks for each group of cancers, together with their 95 go comqdence intervals,17 according to various tobacco-related variables, were derived from two series of unconditional multiple-logistic regression equations, using the GLIM Statistical System (Release 3, Oxford Numerical Algorithms Group, 1978), including (a) quinquennia of age and area of residence only, and (b) education and alcohol consumption besides age and residence, since cases had less education than controls and drank significantly more alcohol. For the comparison of the RR of high vs low tar cigarettes, analyses were restricted to ever-smokers, to permit allowance for other smoking-related variables. Thus, further terms were included in the regression equations for smoking status (ex/current), duration of smoking and number of cigarettes smoked. Results Table 1 gives the basic information for various measures of tobacco smoking. There were strong associations for each of the three sites considered: the multivariate RRs were 4.9 among ex-smokers and 14.3 among current smokers for cancers of the oral cavity and pharynx, 2.6 and 5.0 for esophageal cancer, and 3.1 and 6.3 for cancer a standard questio'dniire on ~o~io-d&:n'ogtapl~ic" factors, smdker~, "especially for ~'ancers .of" the ~ral c~'~,ity and r~#rcr~n,~l rh.qracrerJ;rJcs, u~e of tobacco, alcobot, coffee. and other methylxanthine-containing beverages, frequency and consumption of selected drugs. The questions on tobacco included the forms of smoking, the amount smoked per day, the age at starting and (for ex- smokers) the time since stopping, and the cigarette brands smoked for the longest time. Subjects reporting more than one brand were assigned in the analysis to the highest-tar category reported. r~barvn×, and direcrlv associated with the number of cigarettes per day and the duration of smoking. The risks declined appreciably in ex-smokers but were still somewhat above unity, five or more years after stopping. The effect of tar-yield among current and ever-smokers is examined in Table 2. Although the risks were sub- stantially elevated in both strata, there was an appreciable difference in the RRs; among ever-smokers, the multivariate RRs for oral and pharyngeal cancer were 8.5 7O

Table 1. D, .tributi,,n of 741 cases of cancers of upper digestive and respiratory tract and 1272 controls according to smoking habits Smoking Oral cavity Esophagus l..arynx Contn~ls habit and pharynx Relative risk estimates (95 % confidence interval) Oral cavity and pharynx l~sophagus .. Larynx Multivariate* Multivariateb Multivariate* Multivarlateb Muhivar.iatea Multivariateb • Never smokel ; Ex-smokers Cu l'reiq r smok Cigarette smc -:er$d. (cigarettes pc* day) - <15 15 - 24 >-25 Pipe/cigar sin. ,kers (grams pet da. ) Ever i Total dutatio~i t smoking (year: I <30 30- 39 ->40 Time since srn'i ,kinga cessation (years ex-smokers onl ) <5 ->5 6 17 8 289 83 71 41 398 222 200 113 585 1c 1¢ 1c " 1c ic 1c 5.6 4.9 2.9 2,6 3.2 3.1 (2.4 - 13.1) (2.1 - 1.1.4) (1.7 5.1) (1.5 -4.6) (1.5 - 7.0) (1.4 -6.7) 17.4 14.3 5.9 5.0 6.8 6.3 (7.7 - 39.4) (6.3 - 32.6) (3.6 - 10.0) (3.0 - 8.5) "(3.3 - 14.2). ' .(3.0 - 13.2) 56 63 25 313 6,6 5.9 3.5 "3:2. (2.8L 15.6) (2,5 - 13.9) (2.0-6.2) 140 I10 68 396 14.8 12.5 4.9 4.0 (6.5-33.8) (5.4-28.6) (2.9-8.4) . (2.3-6.8) 83 91 59 258 15.7 12.7 5,8 4,9 (6.8-36.5) (5.4-29.7) (3.3-9,9). (2.8-8.6) 6 7 1 14 20.5 15,3 6.10 i 5,~~ (4.2 - 100.2) 2.8 - 84.0) (2.0 - 17.9) (I:.6 7 18.2) 62 66 20 414 6.0 5.0 3.1 2..7 (2,6-A4.1) (2.1 -11.8) (I.8- 5.5) (1.6-4.8) 8~ 73 49 255 16,7 IL4 %0 4,3 (7,1- 39.2) (5,7 - 31.5) (2.9-8,8) (2.5 - 7.7) 137 127 85 300 24.2 18.9 6.8 5,6 (10.3 - 57.1) (8.0-44.6) (3.9- 11.7) (3.2.- 9.7) 32 26 18 111 11.1 10.1 3.9 3.5 (4.5 -27.7) (4.0-25.4) (2.0- 7.~) (1.8-6.9) 29 44 23 284 3.2 2.7 2.4 2:2 . (1.3-8.1) (1.l -6.7) (1.3-4.4) (1.2-4.1) 2,6 ~ 2.4 5:7 5.2 (2,7 - 12.1) (2.4- 11.1) 7.9 7,8 (3.7 - "~7.0) 0.6- 16.8) 5.3 (0.2-20,0) (0.2-123.6) 2.(] 1.9 (0.9-4£) (0.8-4.5) 6,2 5,7 (2.9 - 1.3.5) (2.6- 12.4) 8.8 7.9 (4.1 - t8.8) : (3.7- 16.9) 5,,~ 5.2 (2.3 - 13,0) (2..2 - 12.6) ,2.5 2.5 (~.I -~.8) (1:1-~.8) *Estimates fron: multiple logistic regression: allowance was made for age in quinquennia and area of residence. bEsdmates for n ultiple l,gistic regression: allowance was made for age in quinquennia, area of residence, years of education and alcbhol consumption (grams per day). ~:Reference cateI ory. '~Totals do not tdd up h,ecause of missing values. ~¢696E9890~

Table (elative risk of cancers of upper digestive and respiratory tract according to typ'e of cigarette currently smoked Type of cigarette Oral cavity Esophagus Larynx Controls Relative risk estimates (95% tonfidence intcrval~ and pharynx Or.',l cavity ~nd pharynx Esopha~us . Larynx Multivariatea Multivariateb Multivariat~a Muhivariate1' Multivariate~ Muhivariateb Ncvcr sin( kers : Cigarette moket~s (ever)" Tar yield < 22 mg :> 22 mg Not defin d Cigarette rookerY. (current olly) Tar yield <22 mg > 22 mg! Not defill 'd " 6 17 8 289 I¢ 1~ 1¢ 1~ " 1¢ 1¢ 131 135 94 651 9.4 8.5 3.5 3.3 5.0 4.8 (4.1 -21.6) (3,7- 19.4) (2,1 -6.0) (1.9- 5.6) (2.4- i0.5"i (2.3- 10.1) 143 112 55 234 21.3 I6.4 11.1 7,8. 8~5 7.1 '(9.2- 49.4) (7,1 -38.2) (6.3- 19.3) (4.4- 13,8) (3.9- 16".7) " (3.2- 15.6) 11 24 5 98 ..... 108 102 71 426 12.2 10,9 4.1' 3,7 5,9 ' 5,5 (5.3-28.0) (zi,7 -,25.0) (2.4-7.0) i2.I-6,3) (2,8- 12.5) (2.6-11.6) 106 80 39 99 37.5 ' I~ 30,0ii" 19.2 13,6 . 14,~9 !2,3 (15.9=88,3) (12,7-71.2) (10,5-35.2) (7,3-25,2) (6.5-341.½) (5.3-2815) 8 18 3 60 ...... ~Estimate' from'multlple logistic regression: allowance was made for age in quinquennla and ~rca of residence. " ' l~timate,, from m~]tiple logistic regression: allowance was made for age in quinqucnnia, area of residence, years of education and alcohol c~msump~ion (grams per day). eReferenc, I catego.ry,,' Table 3, Rel~,~iv.e risk of cancers of upper digestive and respiratory tract according to tar yield of cigarettes srfioked for the long.est time, in separate strata of age and alcohol consumption "' Covariate Relative risk estimatesa.b (95% confidence inter~al) :. : Larynx Number of Oral cavity Number of Esophagus Number of cases and pharynx cases cases <22 mg~ ~22 mg¢ <22 mgc ___22 mgc <22 mg~ :>22 mg~ Age (yea ~) <60 ~ " 153 9.2 19.7 140 2.2 7.0 73 3.2 5.9 (2.4 41.4) (4.7-82.0) (1.1-4.5) (3.4-14.7) (1.3-7.6) (2,3-15.1) :>60 138 9.2 22.8 148 6.1 20.3 . 89 11.4 20.4 (3.4 - 25.3) (8.1 - 64,3) (2,7 - 13.7) (8..5 - 48,3) (2.9 - ~4.~2) (4.9 - 84.9) Tota/ ale ,holic ..beverage consumE ion (~l~Jnks per week) -<35 :>35 56 11.5 23.7 86 4.2 12.2 66 5,~" 12.1 (3.7 - 35.4) (7.7 - 73.3) (2.0 - 9.1) (5.5 - 26.8) . (1.9 - 18.4) (3.8 - 38.7) 235 5,9 11.5 202 2.6 7.5 .. 96 4.6. 6.9 (1,8- 19,1) (3.3-39.8) (1,3-5.4) (3.2-17,5) ' • (1.8-II,8) (2,3-21.1) *Estimat, s fr~'multiple logistic regression: allowance was made for area of residence and age. bReferer,e category: never-smokers. ~Tar yie', I (rag per cigarette).

for low to medium tar ( <22 rag) and 16.4 for high tar (~>22 mg). Corresponding estimates were 3.3 and 7.8 for esophagus cancers, and 4.8 and 7.1 for larynx cancers. The differences between the two tar categories were similar in proportional terms (and hence greater in absolute terms) when only current smokers were considered. The role of type of c!garettes is further considered in Table 3, in separate stra.ta-of age and alcohol consumpd6n. In all the strata examined the .RRs were consistently higher for high- as compared to low/medium-tar cigarette smokers, and no appreciable modifying effect on tar-related estimates was introduced by these two major covariates. In Table 4, the Comparison between tow- and high tar • cigarettes is further, considered, for smokers only, after allowance for smoking status, duration and amoune smoked (plus other' potentially relevant covari~ites). The RRs for high. vs low tar cigarettes were 2.3 for oral cavity and pharynx cancer, 2.8 for esophagus cancer and 1.8 for larynx cancer. All these estimates were significantly greater than 1.0. Discussion The present findings provide further quantitative evidence on the importance of type of cigarettes smoked in relation to the risk of upper-digestive and respiratory tract cancers. Taking the three sites together, the RR for high tar cigarettes was more than twice that for low tar ones, after allowance for duration and quantity smoked, as well as major potential confounding factors. Although the point estimate was greatest for esophagus cancer, the RRs for the three sites were not heterogeneous by formal statistical test. Thus, the importance of type of cigarette on the risk of the three neoplasms is not only clearly established, but also quantified to an extent similar to, if not greater than, that observed in studies related to lung cancer. 1 - ~ Table 4. Relative risks*'associated with smoking of cigarette brands with tar-yields 22 mg or more per cigarette Cancer Relative Confidence site risk~ interval ~,o., ~.,~:~) ,,u~a pharynx Z.5 1.6 - 3.2 Esophagus 2.8 2.0 - 4.0 Larynx 1.8 1.2 - 2.8 ~Relative risk adjusted for smoking status (ex/curren0, duration of smoking and number ofcigarertes smoked per day, in addition to age, area of residence, years of education, and alcohol consumption (grams per day). Reference category is smokers of cigarette brands with less than 22 mg tar-yield per cigarette. 'Brand' is the brand smoked for the longest period. Cigaretses and cancers of digestive and respiratory tract Considering the smoking pattern of the population studied,*~ it is difficult to attribute confidently and completely the differences observed to tar-yield rather than to type of tobacco. The introduction of filters, and the consequent reduction in tar-yield, occurred, in fact, in the same period as change in type of tobacco from black to blond, which may also represent a relevant factor, or co-factor, in the differences in risk observed. A case for the importance of black tobacco has been made particularly in relation to bladder cancer,8- ~0 in view of its high nitrosamine content,11 but the specific role of nitrosamines in the upper-digestive and respiratory tract carcinogenesis is less clear.5a2 Thus, we decided to base our classification on the tar-yields, recognizing, nevertheless, that a satisfactory distinction is not possible, and hence, in s~ict terms, these data simplyindica~te that newer cigarettes are less carcinogenic for these sites than older ones. Among other possible sources of bias or misclassi- fication, information on brand currently smoked should be considered. We collected data only on the cigarette brand smoked for the longest time, although more than one brand could be reported in the interview in case of uncertainty. Further, and probably more important, it is known that information on the distant past is often imprecise, and influenced by current or more recent habits. Finally, medium- and low tar cigarettes have become widespread in Italy only over the last two decades.~ These limitations of the present data set, nonetheless, are unlikely to have applied differently to cases and controls and have thus produced a spurious relationship. If anything, they should have tended to reduce the strength of the real association. Further, the appreciable differences in RRs according to type of cigarette in different age-groups indicate that these findings cannot be explained simply in terms of a cohort effect. Other sources of bias have been discussed elsewhere,~7 but are unlikely to explain the association observed. Cases and controls, in fact, were drawn from comparable catchment areas and participation was almost complete. The results were not substantially modified by allowance for major, identified, potential confounding factors, including socio-es0nornic indicators,l~'t8 .as well as alcohol consumption.and duration and.quaritity of. . smoking. This suggests tha~ potential r~sidual" - - is unlikely to explain totiilly the association observed with type of cigarettes. - The results have important public health implications, particularly in the areas where the study was conducted--where mortality rates for upper-digestive and respiratory neoplasms are among the highest in Europe. 19,20 Still, significant and substantial associations for all sites considered were observed even in the lower 73 L

C. La VecchJ~ et al. tar category, thus giving unequivocal indications for stopping smoking as a priority for prevention and public health. Acknowle.dgements--The authors wish to thank Mrs. J. Baggott, Mrs. M. P. Bonifacino, and the G. A. Pfeiffer Memorial Library for editorial assistance. References 1. Wynder EL, Stellman SD. Impai:t of long-term filter cigarette usage on lung and larynx cancer Risk: A case- control studyJNCI 1979, 62:471 - 7. 2. Lee PN; Garfinkel L Mortality and type of cigarette smoked. J EpidemioI Comraun Healt/~ 1981; 35:16 - 22. 3. US Department of Health and Human Services. The Healt~ Conseqt4ences of Smoking. Tke Changing Cigarette: A Report of the Surgeon General. "Washington DC: Government Printing Office, 1981. 4. Lee PN. Lung cancer inddence and type of dgarette smoked. In: Mizell M, Correa P, ads. Lung Cancer: Causes and Prevention. Deer Field Beach: Verlag .Chemic International, 1984; 273 - 84. 5. Lubin JH, Blot, WJ, Berrino, F, et aL Patterns of lung cancer risk according to type of cigarette smoked. IntJ Cancer 1984; 33: 569- 76. 6. Benhamou S, Benhamou E, Tkmarche M, Flamant R. Lung cancer and use of cigarettes: A French case-control study. JNCI 1985; 74: 1169-75. 7. Wynder EL, Augustine A, Kabat GC, Hebert JR. Effect of the type of cigarette smoked on bladder cancer risk. Cancer 1988{ 61: 622- 7. 8. Vineis P, EsteveJ, Terracini B. Bladder cancer and smoking in males: Types of cigarettes, age at start, effect of stopping and interaction with occupation. IntJ Cancer 1984; 34: 165 - 70. 9. Butch JD, Rohan TE, Howe GR, et aL Risk of bladder cancer by source and type of tobacco exposure: A case- control study, lntJ Cancer 1989; 44: 622- 8. 10. ClavelJ, Cordier S, Boccon-Gibod L, Hereon D. Tobacco and bladder cancer in males: Increased risk for inhalers and smokers of black tobacco. IntJ Cancer 1989; 44:605 - I0. 11. Hoffman D, Adams JD, Piade JJ, Hecht SS. Chemical studies on tobacco smoke. LXVII. Analysis of volatile and tobacco-specific nitrosamines in tobacco products. IARC Sci Pub/1980; 31: 507- 16. 12. Esteve J, Tuyns AJ, Raymond L, Vineis P. Tobacco and the risk of cancer. Importance of kinds of tobacco.. IARC Sci Publ 1984; 57:867 - 76. 13. La Vecchia C, Liati P, Decarli A, Negrello I, Franceschi S. Tar yields of cigarettes and the risk of oesophageal cancer. IntJ Cancer I986; 38: 381- 5. 14. La Vecchia C, Negri E, D'Avanzo B, Franceschi S, Decarli A, Boyle P. Dietary indicators of laryngeal cancer risk. Cancer Res 1990; in press. i5. Lk Vecchia C. Patterns of cigarette smoking and trends in lung cancer mortality in Italy.JEpidemiol Commun Health 1985; 39: 157-64. 16. Fairweather FA, Carmicheal IA, Phillips Gtz, Copeland GK. Changes in the tar, nicotine and carbon monoxide yields of cigarettes sold in the United Kingdom. Health Trends 1981; 13: 77- 81. 17. Breslow NE, Day NE. Statistical Methods in Cancer Researcb, >'oL 1, The Analysis of Case Control Studies. Lyon: IARC, 1980. t8. Wynder EL, Goodman MT, Hoffmann D. Demographic aspects of the low-yield cigarette: Considerations in the evaluation of health risk. JNCI 1984; 72: 817- 22. 19. Cislaghi C, Decarli A, La Vecchia C, Laverda N, Mezzanotte G, Smans M. Dati, Indicatori e Mappe di ,14ortalit~ Tumora/e Italia 1973-1977 Data, Statisics and Maps on Cancer Mortality. Bologna: Pitagora Editrice, I986. 20. Levi F, Maisonneuve P, Filiberti R, La Vecchia C, Boyle P. Cancer incidence and mortality in Europe. Soz Praventivmed 1989; 34 $2:$3 - $83. 74