Document gD8oJOJ6a52zx9GoddwYnR0Qa

1" l 6-! British Journalof Haernatobgy. 1995. 91, 154-161 I Hairy cell leukaemia, occupation, and smoking I I J. CLAVEL, L. MANDEREAU, S . CORDIERC,. LE GOASTERD, . HEMON,F . CONSO*AND G . F L A N D R I N ~ lnstitut National de la Santt et de la Recherche Midicale, INSERM U170,VillejuiJ *Servicede Pathologie Professionnelle, H6pital Cochin, Paris, and thboratoire Central d'Htmatologie, H6pital Necker, Paris, France Received 12 December 1994; accepted for publication 30 May 1995 I I I I I Summary. The roles of farm practices, occupational exposures to organic solvents, and ionizing radiation in the risk of hairy cell leukaemia (HCL)were examined in a French hospital-based multicentre case-control study including 291 cases (229 men and 62 women) and 541 controls (425 men and 116 women). No positive association was observed with occupations involving exposure to organic solvents or with selfdeclared exposures to solvents, but a significant association with self-reported exposure to petrol or diesel was found for men (OR= 1.5 CI95%[1.0-2.1]). No association with ionizingradiation was detected. Agriculture employment gave an odds ratio of 1.7 (CI95%[1.1-2-4]) for men and 2.7 (CI95% [1*1-6.7]) for women. Among men, the association seems to affect farmers rather than agricultural workers. Self-declared exposure to pesticides or bovine cattle breeding was related to HCL risk in both genders. Finally, a signfficant negative association with smoking was observed in men, with an inverse exposurerisk relationship odds ratios of 0.6, 0.5 and 0.2,respectively, for cumulative consumptions of < l o . 10-23 and 224 pack-years), contrasting with an odds ratios clearly >1 in women. Keywords: hairy cell leukaemia, occupation, smoking, epidemiology, case-control study. I I I I ! I I I I Hairy cell leukaemia (HCL)is a rare B-cell chronic lymphoid leukaemia, well defined by the specificmicroscopic pattern of lymphocytes, 6rst described by Bouroncle et al(l958). Very few epidemiological studies of HCL have been conducted. The incidence of the disease is low, and HCL is estimated to constitute about 2%of all leukaemias (Bouroncleet al. 1958: Bernstein et al, 1990). HCL affects exclusively adults, with a mean age of about 55 years. A striking epidemiological feature of this malignancy is the strong male predominance (with a male :female ratio of about 4 : 1) consistently observed in all series, whereas the sex ratios of other leukaemias do not usually go beyond 2 : 1. Flandrin & Collado (1987).having interviewed a large series of cases in the context of a clinical trial on interferon-alpha, suggested that this imbalance could be related to environmental factors, especially to benzene and ionizing radiation. This suggestion was the starting point of the present case-control study. To our knowledge, two analytical case-control studies have investigated satisfactorily the role of environmental factors in HCL (Oleske et ai, 1985; Staines & Cartwright, 1993). These studies yielded some convergent 6ndmgs with respect to occupational exposures and smoking. Long-term exposure to organic chemicals gave Correspondence:Dr J. Clavel, INSEF3.4 U170, 16 av. Paul VaillantCouturier, 94807 Villejuif Cedex. France. 154 odds ratio of 3.1 in the study by Oleske et al (1985) and exposure to solvents, petroleum products and benzene were associated with odds ratios of 1-5, 2.1 and 2.0, respectively, in the study by Staines & Cartwright (1993).Odds ratios >2 were also found with woodworking (Oleske et al, 1985) or wood dust (Staines & Cartwright. 1993). No clear effect of ionizing radiation was observed in either study. Odds ratios associated with tobacco smoking were <1 in both studies, and even with a significant inverse dose-response relationship in the study by Staines & Cartwright (1993). Finally, elevated odds ratios were observed in farmers and foresten in the study by Oleske et aZ(1985),whereas in the study by Staines & Cartwright (1993) the overall odds ratio for farmers was not reported. Slightly elevated odds ratios were observed in the latter study for use of fertilizers and for exposure to' farm animals, but not for pesticide exposure. which concerned only three cases. However, the s m d size of these two studies hampers the interpretation of their findings. This paper reports the initial results of a larger case-control study investigating the role of organic solvents, pesticides and ionizing radiation in HCL risk. MATERIAL AND METHODS Selection of cases and controls. Cases and controls were retrospectively recruited from 1990 to 1992 from 18 IQ 1995 Blackwell Science Ltd Hairy Cell Leukaemia, Occupation, and Smoking 155 Table I. Causes Of hospitalization of the controls. &eumatdogical diseases Injuries urological diseases Osteopathies & n i p turnours (lipomas) Neurological diseases cardiovascular diseases Infectious diseases Respiratory diseases others unknown Totd Men 166 (39.1%) 124 (29.2%) 25 (5.9%) 21 (4.9%) 15 (3.5%) 13 (3.1%) 12 (2.8%) 8 (1.9%) 7 (1.6%) 22 (5.2%) 12 (2.8%) 425 (lGO%) Women 47 (405%) 22 (19.0%) 3 (2.6%) 10 (8.6%) 7 (6.0%) 12 (10.3%) 3 (2.6%) 2 (1.7%) 1(0.9%) 6 (5.2%) 3 (2.6%) 116 (100%) hospitals located throughout France (Angers, Bayonne, Bordeaux. Caen, Chambkry, Idle, Limoges, Lyon, Nancy, Nantes, Nice, Nimes. Paris, Poitiers, Rennes, Rouen, Strasbourg, Todouse). AU cases diagnosed from January 1980 to January 1990. including those deceased, were identified and recorded, but only living cases were included in this study. Diagnoses were confirmed histologically. Controls were chosen from among persons hospitalized at the same time in the same hospital (mostly in orthopaedic and rheumatology departments), using the admission records for the 10-year period. The choice of controls was guided both by the necessity of targeting the same age group as cases without selecting for particular diseases, habits or socio-economic status, and by the need to concentrate retrospective sampling on a few admission records in each city. Subjects admitted for cancers or for work-related diseases or accidents were not eligible as controls. For practical reasons, only those subjects whose addresses and telephone numbers correlated positively in the current telephone directory at the time of recruitment were retained as eligible. However, the cases were also checked in the telephone directory at the same time as the controls in order to verify further that no bias was introduced by this selection. Two controls were individually matched with each case according to gender, date of birth (f3 years), date of admission (f3 years). and residence (inside/outside the hospital catchment area). Cases and controls were sent by mail a self-administered questionnaire. Non-responders were sent the questionnaire at most twicemore. If a control had still not answered after three attempts he was replaced by another control from the list using the same selection criteria. In all,474 cases (381 men and 93 women: sex ratio 4.1) were identi6ed ftom the 18 haematology departments. Among them, 291 living cases (61%).229 men and 62 women, were recruited. Among the 183 cases excluded, 112 Table 11. Socio-demographiccharacteristics of cases and controls. Men Cases (229) Age Mean Standard error of the mean 55.6 07 Residence inside/outside catchment area of hospital Inside 76% Outside 24% Socio-economic status (last job held) Professional. technical workers Administrators and managers: Clerical,sales and service workers Production and agricultural workers Never employed 31.6% 23.7% 44.3% 0.4% Education level (highest diploma) No diploma Primary school Juniorschool High school or university graduate 2 1% 32% 31% 16% No. of jobs held Mean Standard error of the mean 4.7 0.2 *Including farmers. 01995 Blackwell Science Ltd. British JournalofHaernatology 91: 154-161 Controls (425) 56.0 05 78% 22% 279% 26.0% 46.1% 0.0% 18% 33% 32% 16% 4.8 0.1 Women 55.8 1.5 76% 24% 24.2% 46.8% 16.1% 12.9% 20% 3 5% 30% 15% 2.3 0.2 56.0 1.1 72% 28% 20.0% 478% 16.5% 15.7% 25% 34% 26% 15% 2.2 0.2 156 J. CZaveZ et d had died (100men and 1 2 women) - most of them (68%) having died before 1984 - and 71 living cases (52 men and 1 9 women) did not send back the questionnaire. A total of 1020 subjects were eligible as controls and were sent the questionnaire: 591 (58%) sent back the questionnaire. 50 responders were excluded because their matched case was dead or a non-responder and they could not be matched with another case. Thus, 541 controls (425 men and 116 women) were eventually included. The reasons for hospital admission are given in Table I. Data collection. The questionnaire included questions on socio-demographic characteristics, tobacco smoking, and long-time occupations and leisures. Questions on smoking involved the number of cigarettes,cigars or pipes smoked per day, and the dates of beginning and stopped smoking. With regard to occupation, subjects were asked to describe, in a semi-structured form, their job history, the job titles and economic area of their various employments, the tasks they accomplishedand the products they handled. They were also given a checklist of tasks involving exposure to solvents (use of paints, varnishes, adhesives, paint removers, machinery, metal degreasing, dry cleaning, use of solvents as chemical reactants), pesticides and ionizing radiation. Finally, the questionnaire included questions about contacts Nth animals (pets, breeding cattle). Subjects identified from the questionnaire as possibly exposed to solvents were asked to undergo a supplemen% telephone interview using structured task-oriented questionnaires, administered under case/control status blinding, in order to detail the exposures (exact nature of exposus, intensity). Subjects having been employed in farming were interviewed at home by occupationalphysicians s p e c i e g in agricultural work. The supplementary questionnaire were designed to enable case-by-case exposure assessment and are to be analysed in a second stage of this study. The present paper reports only results from the self-administered questionnaire. Statistical analysis. Occupations were summarized by &e job titles, coded using the 1968 International Standud Classification of Occupations (EO, 1968). The employem' industrial sector was coded using the International Standud Industrial Classification (United Nations, 1975). Only employments with a duration of 6 months or more were considered. 1 I I Table III. Association between HCL risk and occupations a priori involving exposures to solvents or pesticides (odds ratios estimated by conditional logistic regression). Jobtitles (ILOcodes) Cases Controls OR CI95 (%) Two-tailed Pvalue 1 I I , lMBN Jobsa priori involving exposure to organic solvents Painters (8.19.55-8.73.70-9.3) Artists and designers (1.61-1.62) Printers (9.2) Mechanics (8.43-8.44-8.49) Electrical fitters (8.5) Machine tool operators (8.32-8.33-8.34-8.41) Cabinet makers, joiners (8.1-9.54) Rubber and plastic product makers (9.01-9.02) Shoe and leather goods makers (8.0) Chemical industry* Housewives (5.5) Agricultural jobs Agricultural and forestry workers (6) Farmers (6.1) Agricultural and animal husbandry workers (6.2) Nursery workers and gardeners (6.27) Forestry workers (6.3) WOMEN Jobsa prior1 involvlng exposure to organic solvents Housewives ( 5 . 51-5.52) Nurses (0.71-0.72) Chemical industry* Other solvent-exposed workerst Agricultural jobs Farmers or agncultural workers (6.1-6.2) (229) 6 1 3 14 12 2 1 2 86 37 54 3 9 (62) 6 3 1 2 14 (425) 15 0 4 44 30 3 3 10 122 38 90 8 9 (116) 7 3 1 6 13 0 8 0.3-2.0 1.4 02-8.7 0.6 03-1'1 0.7 03-1.4 1.3 0.2-8.0 0 7 0.1-6.4 0.3 0'1-1'6 1.7 1.1-2'4 2.3 1'3-3'9 1.3 0.8-2.0 0.6 0.2-2.5 2.4 0'8-6.8 1.7 0.5-5.4 1.7 0.3-8.7 1.4 0.1-24.0 0.6 0.1-3.2 2.7 1.1-6.7 ( Ins ns 0.07 ns ns ns ns I I <0.01 <0.01 ns ns ns 1 II ns ns ns ns <0.05 *ISICcode = 3510 t Pamters. pnnters. mechanics. machine tool operators.woodworkers, rubber and plastic product makers, shoe and leather goods makers, ' (01995 Blackwell Science Ltd. British Journal of Haemutology 91: 154-161 Hairy Cell Leukaemia, Occupation, and Smoking 157 Odds ratios were estimated by conditional logistic regression for matched sets with a variable number of controls by pair, using SAS@ procedure PHREG. Separate analyses were conducted for men and women because the distribution of the occupational factors of interest and of smoking was expected to vary with gender and because a sex-related factor seems to be involved in the aetiology of the disease (resulting in a sex ratio of 4) and might interact with fie factors under study. RESULTS The mean age was about 56 years for both men and women. Age, sex and residence matched exactly (Table II), enabling satisfactory comparability of cases and controls for socioeconomic status, as defined by the category of the last job held. level of education, and number of different jobs held. Job titles Table Ill reports the odds ratios associated with farming and with occupations a priori considered to involve exposure to solvents. Among the men, none of the jobs selected for their probable exposure to solvents was associated with the risk of HCL. A negative association was observed even with employment as a mechanic or as a machine tool operator. In contrast, overall, employment of men in agriculture was associated with an increased risk of HCL, and the odds ratio was higher for farmers. An elevated odds ratio was also found for forestry workers, but without reaching statistical significance. No other job was found to be associated with HCL risk and. in particular, there was no association with woodworking as defined by employment as a wood preparation worker, papermaker, cabinet-maker or joiner (OR = 0.8 - CI95% [0.4-1*4]). Only two cases and four controls were classified under the ILO title 'Other salesmen' which includes the job of petrol-pump attendant. Only a few women worked in solvent-exposed jobs, mainly those normally perceived to involve oxygenated solvents (alcohol for nurses, alcohol and glycol ethers for cleaners and housewives). None of these jobs was associated with HCL risk. As was the case for men. farming was associated significantly with HCL risk. Occupational exposures The odds ratios associated with selfdeclared occupational tasks and exposures are reported in Table IV. Solvent exposure, as defined by exposure to paints, varnishes, paint strippers, adhesives, diluents, metal degreasers or use of Table IV.Associationsbetween HCL risk and self-declaredoccupational exposhes (oddsratios estimated by conditional logistic regression). Cases & Use of solvents Solvent exposure* Petrol/diesel Glass cleaning Dry cleaning Exposure to engine exhausts Farm practices Use of pesticides Breeding Bovine cattle Ovine cattle Swine Ionizing radiation WOMEN Use of solvents Solvent exposure* Petrol/diesel Glass cleaning Dry cleaning Farm practices Use of pesticides Breeding Bovine cattle Ovine cattle Swine Ionizing radiation (220) 107 79 17 2 43 53 50 17 16 8 (62) 11 3 7 1 8 8 3 3 4 Controls (425) 176 109 23 4 53 58 58 18 20 17 (116) 21 3 8 1 7 8 2 9 5 OR CI95 (X) 1.2 0.8-1.7 1.5 1.0-2.1 1.3 0'7-2.5 19 0 0.2-5.6 1.6 1'0-2.6 2.0 1.3-3.1 1.9 1.1-3.1 1.8 0.9-3.6 1.4 0.6-2.8 0.9 0.4-2'2 1.1 0.5-2'4 2.1 0'4-10.6 1.8 0.5-6.1 2.0 0.1-32.0 2.7 0'8-8.7 2.0 0'6-6'9 3.3 0'5-19.6 0.5 0.1-2.2 1.3 0'3-5.0 Two-tailed P value ns <0.05 ns ns 0.05 <0.01 <0.02 0.08 ns ns ns ns ns ns 0.10 ns ns ns ns *Paints,varnishes. paint strippers. adhesives,diluants,metal degreasers,solvents as chemical reagents. 1995 Blackwell Science Ltd, British Journal ofHaematoZogy 91: 154-161 158 J. Cluvel et ul Table V. Association between risk of HCL and tobacco smoking in men and women (odds ratios estimated by conditional logistic regression). Cases Mk3 Smoking status Non-smokers Former smokers Current smokers Unknown Total 83 79 57 10 229 Cumulative cigarette consumption Non-smokers <lo pack-years 83 46 10-23 pack-years 40 2 24 pack-years 27 unknown 33 Total 229 worn Smoking status Non-smokers Pormer smokers Current smokers Unknown Total 39 3 15 5 62 Cumulative cigarette consumption Non-smokers <6 pack-years 39 8 2 6 pack-years 8 Unknown 7 Total 62 Conkok 100 145 163 17 42 5 100 71 85 92 77 42 5 84 8 15 9 116 84 8 8 16 116 OR a 9 5 (%) 1.0 0 6 0.4-0.9 0.4 0.3-0.6 1.0 0 6 04-1.0 0.5 0.3-08 0 2 02- 05 1.0 - 1.o 02-4.2 2.4 0.9-62 1.0 2.9 09- 9.7 2.8 0.8-9.6 Two-tailed P value <0.001 <0.001 ns 0.10 solvents as chemical reagents, was not associated with the risk of HCL in either gender. A significantly increased risk was observed among men declaring exposure to petrol or diesel. An elevated odds ratio was also found with exposure ' ' to petrol or diesel among women, but the number of cases was very small. i Table VI. Odds ratios for some occupations estimated by conditional logistic regression adjusting for smoking status. Job title (LOcodes) Cases Controls OR C195 (%) Two-tailed P value MEW Painters (8.19.55-8.73.70-9.3) Printers (9.2) Mechanics (8.43-8.44-8.49) Electrical fitters (8.5) Machine tool operators (8.32-8.33-8.34-8.41) Cabinet makers, joiners (8.1-9.54) Agricultural and forestry workers (6) Farmers (6.1) Agricultural and animal husbandry workers (6.2) WOMEN Housewives ( 5.51-5.52) Farmers and agricultural workers (6.1-6.2) (229) 6 3 19 17 14 12 86 37 54 (62) 6 14 I:425) 15 4 57 36 44 30 122 38 90 (116) 7 13 0.9 0'3-2.7 1.4 0.2-8.4 0.7 0.4-1.2 0.9 0'5-1'7 0.7 0.4-1.3 0.6 0'3-1'4 1.5 1.0-2.2 2.0 1.1-3.5 1.2 0'8-1'9 2.4 0'7-8'9 2.2 0.8-5.7 ns ns ns ns ns ns <0,05 <0.05 ns ns 0.10 I I01995 Blackwell Science Ltd. British Iournal of Huematology 91: 154-1b1 A Very few subjects were exposed to ionizing radiation, and it was not possible to categorize them according to the type of radiation or duration of exposure. No association with radiation was found for either gender. An increase in risk was observed in both genders for subjects who used pesticides and for those who declared involvement with bovine or ovine cattle breeding, although the association was not statistically significant for women. Tobacco smoking Table V reports the odds ratios associated with tobacco smoking. There were contrasting results for men and women. For men, smoking was significantly negatively related to the risk of HCL, with lower odds ratios for current smokers than for former smokers compared to non-smokers. and with the lowest odds ratios for the heaviest smokers. In contrast, odds ratios >1, on the borderline of statistical significance,were observed in women. Control of confounding and selection bias Because of the high proportion of prevalent cases in the study sample, the analysis was also restricted to pairs in which the cases were diagnosed after 1984, when mortality was 3 times lower (about 15%).This restriction did not modify the relationship with either tobacco smoking or occupation. The association between smoking and HCL was not affected by adjustment for blue collar/white collar status. Adjustment for having been employed as a farmer or as an agricultural worker did not modify the odds ratios associated with solvent-exposedjobs. In contrast, smoking appeared to confound. at least partly, the negative relationships observed with mechanics and machine-tool operators (Table VI). DISCUSSION Hairy Cell Leukaemia, Occupation, and Smoking 159 survival were related to the factors under study, i.e. if survival were better in farmers and poorer in mechanics or smokers. An attempt was made to obtain data from surrogates of cases who had died and of matched controls who had died, but the response rate was weak (30%),the information obtained was poor, and the control selection very complex. The cost-effectiveness of the process was, in consequence, discouraging. Mortality from HCL was high before the introduction of interferon-alpha in 1983. In our sample, the proportion of cases who had died was 59%. and 15% for cases diagnosed prior to and after 1984, respectively. The main results of the study were not affected when the analysis was restricted to pairs in which caseswere diagnosed after 1984. Moreover, selection of farmers by better survival is unlikely, because farmers accounted for 2 7%of the cases diagnosed before 1984 and for 35%of those diagnosed after 1984, when mortality was lower. Similarly, current smokers accounted for 37.5%of the cases diagnosed before 1984 and for 21.5% of the cases diagnosed afterwards. This is not consistent with selection by survival of those cases who smoked less. The selection of controls listed in the telephone directory, although all cases were included, could also bias the results. 49 cases (39 men and 10 women) were included in the study although they were not listed in the directory at the time of recruitment. Among them, 27% were farmers, whereas farmers were 32% of the remaining sample. It is unlikely, therefore. that retaining unlisted cases may have introduced an excess of farmers. Similarly. 22.6%of the cases listed in the telephone directory were current smokers, compared to 35.9% among those unlisted. Retaining the latter is therefore unlikely to lower the proportion of smokers among cases. The proportion of non-smokers was the same for the cases listed in the directory and those unlisted. This study was designed specificallyto investigate the role of occupational exposures to solvents, pesticides and ionizing radiation in the risk for hairy cell leukaemia. The present analysis, based on self-reported data, produced mainly the following results: (i) no positive association was observed with occupations supposed to involve an exposure to organic solvents or with self-declared exposures to solvents, but a sigiiikant association was found with self-reported exposure to petrol or diesel. Significantly decreased odds ratios were found with employment as a mechanic or as a machine-tool operator, but the relationships seems at least partly confounded by smoking. (ii) Employment as a farmer, selfdeclared exposure to pesticides, and bovine or ovine cattle breeding were related to the risk of HCL. Among men. the association seems to involve farmers rather than agricultural workers. (iii) No association was found with exposure to ionizing radiation, without any distinction as regards the nature or extent of exposure. (iv) Finally, a significant negative association in men was observed with smoking, with an inverse exposure-risk relationship contrasting with the finding of odds ratio clearly >1in women. Due to the rarity of the disease, recruitment was Performed restrospectively and prevalent cases were enrolled. This design could give rise a selection bias if Smoking The negative association between smoking and HCL risk does not seem to be explained by an excess of smokers in the hospitalized control group. Hospital controls may tend to be of lower social class and may be more likely than population controls to be cigarette smokers, as has been observed in the U.K. (Staines & Cartwright, 1993). However, in the present study, smoking habits of the control group were very similar to those of the general population: the proportion of current smokers for the ages 35-44.45-54, 55-64,65+ was 49%. 43%,41%and 27%.respectively,for the study controls, and SO%, 45%. 40% and 30% in a population-based survey conducted in 1980 in France on a representative 16 000person sample (Hirschet al, 1988).The distribution of heavy smokers by age among the controls also parallels the figures from this survey. The association between smoking and HCL risk is also unlikely to be explained by socio-economic status or level of education, which were similar in the cases and controls. In addition, adjustment for blue collar/white collar status produced no change in the estimates. The negative association with smoking in men is consistent with the results published by Oleske et a1 (1985) and by Staines & Cartwright (1993).Although a positive relationship seems to exist between smoking and acute non-lymphocytic 01995 Blackwell Science Ltd. British Journal oJHaernatoZogg 91: 154-161 160 J. Clavel et aI leukaemia, none has been demonstrated for chronic lymphocytic leukaemia (CLL) (Austin & Cole, 1986; Garfinkel 81Boffetta, 1990; Severson et ul, 1990; Siegel, 1993; Brownson et al, 1993) or multiple myeloma (Linet et ul, 1992: Heineman et ul, 1992). The contrast in the results between men and women for smoking was unexpected. However, this result correlates in an interesting manner with some other facts: (i)the sex ratio of the disease suggests the existence of a sex-related factor that could interact with smoking; (ii) some sex hormone related malignancies such as breast and uterine neoplasms are also negatively associated with smoking (IARC monograph no. 38. 1986); (iii) sex hormones and smoking share common metabolic pathways, providing an opportunity for interactions. In addition, the activity of the oxidative enzyme systems differs between males and females and is influenced by hormone levels (Sipes & Gandoffi. 1991). It may therefore be suggested, for instance, that male hormones could have adverse effects counterbalanced by smoking, or that a carcinogen is cleared through a metabolic pathway activated both by smoking and female hormones. Occupations The comparability of cases and controls in terms of socioeconomic status, educational level and number of jobs was good. Selectionof controls is therefore unlikely to explain the findings of the study with respect to occupation. The absence of a positive association between self-declared solvent exposure and the risk of HCL does not de6nitely rule out the existence of an association with more specific chemicals, such as benzene (present mostly in petrol), other aromatic solvents, chlorinated solvents and glycol ethers (present mostly in glass-cleaningproducts). The finding of an elevated odds ratio associated with petrol or diesel use suggests an association with benzene exposure. Unfortunately, diesel and petrol exposure were sought by the same question and so cannot be separated. Such a finding would be consistent with the 2-fold HCL risk found by Staines & Cartwright (1993) among subjects exposedt o benzene and, more generally, with the well-known association between benzene and leukaemia (Austin et al. 1988). although this association could be restricted to myelocytic leukaemias (Lamm et al, 1989: Enterline. 1993). However, this increase in odds ratio contrasts with the low odds ratio associated with employment as a mechanic, which can be regarded as a more objective anamnestic marker of petrol exposure (only two cases and four controls were classed under the L O title `Other salesmen' which includes petrol-pump attendants). Recall bias might explain the discrepancy: benzene is a known risk factor for leukaemias and, in France, compensation may be awarded for benzene-related leukaemias. Cases might be better informed by their doctors and those whose jobs involved, even only slightly or occasionally. exposure to petrol might then be more likely to declare it in the selfadministered questionnaire than controls with similar exposure. The supplementary questionnaires have been structured and standardized in order to provide more objective, specific and semi-quantifiedinformation. and will be explored further in the subsequent analyses. The absence of an association between exposure to ionizing radiation and HCL risk is difficult to interpret categorically in the present study. because the number of exposed subjects was small and the exposures not well- defined. An association was suggested in the study by Stewart & Keating (1980).However, the risk of CLL has not abeen found to increase in atomic bomb survivors and is generally considered to be the only form of leukaemia unrelated to ionizing radiatiun (Cartwright, 1992). The role of farming in B-cell malignancies is receiving considerable attention. With respect to HCL, Oleske et (1985) reported a %fold increase in risk for farmers and forestry workers, not indicated in the study by Staines & Cartwright (1993). although contact with farm animals yielded an odds ratio of 1.7. Some studies on B-cell leukaemias, lymphomas and myelomas suggest an effect of pesticide use. The main suspects are chlorinated aromatic compounds such as phenoxyacetic acid herbicides, pentachlorophenols. DDT and other organochlorinated insecticides (Pearce & Reif. 1990; Hoar Zahm & Blair, 1992).The IARC classified insecticides, in general, in group 2A of `probably carcinogenic products', based on both epidemiological and experimental data (IARC monograph no. 53, 1991).In addition to pesticide use, farming usually involves exposures to cattle. This appliesparticularly to the past when mixed farming was usual. Several studies have also raised the possibility of animal virus transmission to man among farmers, abattoir workers or veterinarians. Bovine leukaemia v i r u s has most frequently been proposed for investigation, but there is no evidence yet for such transmission, as reviewed by Pearce & Reif (1990) and Blair et al(1992). The present study found an association with pesticide use and with bovine and ovine cattle breeding in both genders, but these variables are highly correlated in the present sample of farmers who began work around 1940 and in traditional mixed farming. A further stage of the analysis will examine ad hoc detailed questionnaires analysed by agricultural-work and chemical industry specialists. The results of this study suggest that it would be of interest to investigate further the relationship between farming and hairy cell leukaemia, study solvent exposure with respect to specific chemicals such as benzene (for which the strongest hypotheses have been formulated), and try to account for the confounding or modifying effect of smoking. ACKNOWLEDGMENTS This work was supported by the Ministere de la Recherche et de la Technologie. Departement Homme, Travail, Technologies, by the Caisse RCgionale d'Assurance Maladie de 1'Ik de France. and by a grant from the Ministere charge de I'Enseignement SupCrieur. We thank Professor F. Bauters, Professor J. Bayle. Dr Blanc. ProfessorM. Boasson. Professor A. Broustet. Professor B. Coiffier. Dr J. Gardais. Dr M. Gardembas, Professor D. Fiere. Dr C. Foussard, Professor 0. Guerci. Dr F. Guilhot. Professor J.-L. Harousseau. Professor M. Leporrier. Professor P.-Y. Le Prise. Dr G. Malinvaud. Professor F. Oberlingv Professor J. Pris. Dr M. Renoux. Dr X. Troussard, Professor 0 1995 Blackwell Science Ltd, British Journal ojHaernatology 91: 154-161 Hairy Cell Leukaemia, Occupation, and Smoking 161 j.-F. Schved. Professor F. Streiff. Professor A. Thyss, Dr H.Tilly a n d Dr Walters. haematologists. for their help in the identification a n d recruitment of cases. W e also t h a n k the heads of the departments where controls were collected: professor M. Audran. Professor G. Asencio, Professor P. Ballanger. Professor C. Bazin, Professor P. Besse, ProfessorP. Brissot, Professor J.-P. Clarac, Professor B. Devulder, professor J.-M. Dubernard, Professor A. Duquennoy, professor A. Durandeau, Professor M. Ferrand, Professor A. Gaucher, Professor J. Jeannaret, Professor J.-F. Kempf. Professor J.-L. Lerat, Professor J. Letennew, Professor M. Mansat. Professor C. Pecout, ProfessorJ.-M. Thomine,Professor J. Toubol, Professor E. Vignon and Professor G. Ziegler. We are grateful to A. G. Mullarky for his translation of the manuscript. REFERENCES Austin, H. & Cole, P. (1986) Cigarettes smoking and leukemia. Journal of Chronic Diseases, 39, 417-421. Austin, H.,Delzell, E. & Cole, P. (1988) Benzene and leukemia: a review of the literature and a risk assessment. American Journal of Epidemiology, 127, 419-439. Bernstein, L.,Newton, P. &Ross,R.K. (1990)Epidemiology of hairy cell leukemia in Loss Angeles county. Cancer Research, 50, 36053609. Blair, A., Hoar Zahm. S., Pearce. N., Heineman, E.F. & Fraumeni. J.F., Jr (1992) Clues to cancer etiology from studies of farmers. ScandinavianJournalof Workand Environmental Health, 18,209-215. Bouroncle, B.A.. Wiseman, B.K. & Doan. C.A. (1958) Leukemic reticuloendotheliosis. Blood, 13. 609-630. Brownson, R.C., Novotny, T.E. & Perry, M.C. (1993) Cigarette smoking and adult leukemia: a meta-analysis. Archives of Internal Medicine, 153, 469-475. Cartwright.R.A. (1992)Leukemia epidemiology and radiation risks. Blood Reviews, 6, 10-14. Enterline. P.E. (1993) Review of new evidence regarding the relationship of gasoline exposure to kidney cancer and leukemia. Environmental Health Perspectives. 101, (Suppl.). 101-103. Flandrin. G. & Collado,S. (1987)Is male predominance in hairy cell leukaemia related to occupational exposure to ionizing radiation, benzene and other solvents? British Journal of Haematology. 67, 119-120. Garfinkel.L. & Boffetta, P. (1990)Associationbetween smoking and leukemia in two American Cancer Society prospective studies. Cancer. 65. 2356-2360. Heineman. E.F., Hoar Zahm, S., McLaughlin. J.K., Vaught. J.B. & Hrubec. 2. (1992)A prospective study of tobacco use and multiple myeloma: evidence against an association. Cancer Causes and Control, 3. 31-36. Hirsch. A.. Hill. C., Frossart. M.. Tassin. J.-P. & Pechabrier. M. (1988) Lutter contre le tabagisme: propositions au ministre d6leguC charge de la sante et de la famille. La Documentation Franpise. Paris. Hoar Zahm. S. & Blair, A. (1992) Pesticides and non-Hodgkin's lymphoma. Cancer Research. 52, (Suppl.), 5485-5488. IARC (1986) Monographs on the evaluation of the carinogenic risk of chemicals to humans: 38. Tobacco Smoking. Lyon. IARC (1991) Monographs on the evaluation of the carinogenic risk of chemicals to humans: 53. Occupational Exposures in Insecticide Application and some Pesticides. Lyon. International Standard Classification of Occupations, revised edition. (1968)International Labour Office, Geneva. International Standard Industrial Classification of all Economic Activities (1975) United Nations, New York. Lamm,S.H., Walters, A.S., Wilson, R.. Byrd, D.M. & Grunwald, H. (1989) Consistencies and inconsistencies underlying the quantitative assessment risk from benzene exposure. Environmental Health Perspectives, 82, 289-297. Linet. M.S., McLaughlin,J.K., Hsing. A.W., Wacholder, S., Co Chien, H.T., Schuman, L.M.. Bjelke, E. & Blot, W.J. (1992) Is cigarette smoking a risk factor for non-Hodgkm lymphoma or multiple myeloma? Results from the Lutheran brotherhood cohort study. Leukemia Research, 16, 621-624. Oleske, D., Golomb, H.M.. Farber, M.D., Levy,P.S. (1985) A casecontrol inquiry into the etiology of hairy cell leukemia. American Journal of Epidemiology, 121, 675-683. Pearce. N. & Reif, J.S. (1990) Epidemiologic studies of cancer in agricultural workers. American Journal of Industrial Medicine, 18, 133-148. Severson. R.K.. Davis, S.. Heuser. L., Daling, J.R. & Thomas, D.B. (1990) Cigarette smoking and acute nonlymphocytic leukemia. American Journal of Epidemiology, 132, 418-422. Siegel. M. (1933) Smoking and leukemia: evaluation of a causal hypothesis. American Journal of Epidemiology. 138, 1-9. Sipes,I.G. & Gandoffi,A.J. (1991)Biotransformation of toxicants. In: Casarett and Doull's Toxicology: the Basic Science of Poison, 4th edn (Amdur, Doull and Klasseen, eds). Permagon Press, Oxford. Staines. A. & Cartwright. R.A. (1993) Hairy cell leukaemia: descriptive epidemiology and a case-control study. British Journal O/HaerllatolOgg, 85, 714-717. Stewart. D.J. & Keating, M.J. (1980)Radiation exposure as a possible etiologic factor in hairy cell leukemia (leukemic reticuloendotheliosis). Cancer. 46, 1577-1580. 01995 Blackwell Science Ltd. British Journal ofHaematoZogy 91: 154-161