Document X7vNK8JZBeY1O05NYQ2Nekb84

Occupational Risk Factors for Subgroups of Non-Hodgkin's Lymphoma Lilith Tatham, Paige Tolbert, and Carl Kjeldsberg3 The non-Hocigkin's lymphomas (NHL) are a diverse group of identified risk factors for NHL, we observed nvo interesting . neoplasms oi the Ivrnpharic system whose incidence has heen associations, one between solvent exposure and small cell increasing in recent ye:Irs. The Centers tor Diseasc Cmtrol .diffuse lymphomas [odds r ~ i (oOR) = 1.6C 95%co@e,ce Selected Cancers Study, a population-based case-control study of several cancers, included a large number of cases of NHL .'interval (CI)= 1.10-2.23], and ' packaging/processing and follicular the other between me+. lymphomas (OR = 1.60; anJ a pathology review. providing a rare opportunity to study 95% CI = 9.99-2.60). The results ot this exploratory analysis risk factors for groups of NHL subtypes. We examined the relation between occupational exposures and three subgroups are primarily negative. Our lack oi positive findings may indicate that the subgroups of NHL used may not be etiologically I of NHL: small cell diffuse lymphomas (N = 185), follicular distinct and that further work needs to be done to develop lymphomas ( N = 2681, and large cell diffuse lymphomas (N= an NHL classification system that is etiologically informative 526). There were 1,659 controls available tor snmparison. and useful for epidemiolnoic studies. (Epidemiologv 199733: .\iter controlling for demographic variable, d i d previously 551-558) Kevwords: non-Hodgkin's Ivmphorna, case-control study, occupational risk factors, solvents, meat processing. The non-Hodgkin's lymphomas (NHL) are a diverse group of malignant neoplasms of the lymphatic system. Since rhe early 1 9 7 h the incidence rate of NHL in the United States has been increasing ahout 3-4"h a year, and the mortality rate of NHL has been increasing about 2% a year.' Changes in diagnostic criteria and known risk factors such as familial factors, human immunode- ficiency virus, immunosuppressive conditions or drugs, and certain occupations only partially explain these rate increases.',J Etiologic studies aimed at explaining the increasing incidence rypically have studied all NHL combined, because the subtypes of NHL ,ire thought to be related and to arise trom a common hematopoietic stem cell.' The various NHL subcype~,as defined by the Working Formulation classification system,' ha\.e also been studied s e p a ~ ~ l t e loywing to likely eriologic hetcrcqeneiry among the different types of NL't-IL.'."-' Studying groups of the Working Formulation subtypes may also be useful,' because it results in larger nuinbers of cases than sub- type-specific analysis, and possibly more etiologically homogeneous groups than studying all NHL together. The tollowing groups ot NHL Working Formulation subtypes h,ive been prqwse'cl' *: small cell diffuse Iym- phomas (,<mall Ivmphocytic and diffuse m a l l cleaved cell sut.cype>), tolliciilar lymphomas \tullicular small cleaved, tollicular mixed small and large cell, anl] tollicu h r large cell subtypes), large cell diffuse Ivmphomas (dittuse mixed small and large cell, large cell d i h s e , and large cell immunohlasric subrypes), folliciilx center cell Ivmphomns (follicular Ivmphomas, diffuse mixed small and large cell, large cell diffuse. and small noncleaved c 1 small lymphocytic during the Vietnam War. This study has been described in detail elsewhere.'".'' In brief, investigators drew cases from eight popula- tion-based cancer registries in the following locations: 1: II 1: Ij I: I I diffuse mixed small and large cell follicular large cell Atlanta, Connecticut, Iowa, Kansas, Miami, San Fran- cisco, Detrpit, and Seattle. All males born between 1929 and 1953, and hence ot military service age during the V i e t n h War, and diagnosed with lymphoma, sarcoma, or nasal. nasopharyngeal. or liver cancer between December l , 1984, and November 317, 1988, were eligible for inclusion in the Selected Cancers Scudy. If a case subject died betore participating, next-of-kin or another proxy respondent was identified. Investigators frequency matched controls on registry and date of birth in 5-year categories to the expected distribution of all lymphoma cases, identifying controls through random digit dialing. Deceased cases and controls were not used in the present analysis. 0 Small cell diffuse lymphomas* 0 Follicularlymphomas' -r t Blastlc bmphomar Penpheral1 ccll lymphomas Investigators conducted telephone interviews for all cases and controls. The questionnaire covered background characteristics, medical history, work history (in- Large cell diffuse 'ymphomas' 31 Follicular centercell lymohomas cluding detailed questions about specific jobs and expo- ?- FIGURE 1. Subgroups of non-Hodgkin's lymphoma. * = subgroups used in analyses, selected on the basis of similarity of cell type, likely cell of origin, and behavior as well as their sures to specific dusts and chemicals such as pesticides, herbicides, solvents, and wood preservatives), military history, and life-style. exclusivity. In an attempt to confirm the pathologic diagnosis of all cases, an expert pathology panel reviewed slides of tissue or tissue hlocks. The three pathologists on the subtypes), blastic lymphomas (large cell diffuse, large lymphoma panel independently reviewed the material cell immunoblastic, and small noncleaved subtypes), Jnd made a diagnosis, and then the group reached a and peripheral T-cell lymphomas (diffuse mixed small consensus diagnosis. and large cell, large cell diffuse, large cell immunoblastic, Investigators identified 2,354 men as lymphoma cases and diffuse small cleaved cell subtvpes) (Figure 1). (NHL. H d g k i n ' s disease. or lymphoma not otherwise h h n y risk iactclrs for .dl N H L combined have been ~pec~t'ielal)d :,233 men JS potentidl contrds and in- identitied previously. and the evidence has been re- teniewed 2,073 lymphoma cases and 1,911! potential rie\ved."-12 Risks inherent to the individual include a controls. Microscoye *lilies or tissue blocks were ob- t i d y history ot NHL o r ocher cancer and inherited tained for 7,l?l?4 of rhe lymphoma cases interviewed, and immuncxleficiencies. Identified exogenous risk factors a Ipmphmna diagnosis was confirmed for 1,j11 cases as include radiation at very high doses, viral infections and NHL. NHL cases were classified according to the Work- related diseases such as AIDS, and phenytoin usage. inq Formulation. To achieve comparability with the Occupations that have been rlssociated with excess risk tviginal analyses, we excluded control and case subjects include farmers. pesticide applicators, grain millers, i f at least one of the following applied: unknown military :.wmd and forestry workers, chemists, cojmecologists. ma- ' service in Vietnam ( 5 NHL cases. 7 controls); in Viet- 'A-rrntsFs printers. and workers in the petroleum. rubber, nnm c>roff rhe coa.t ,jf \.'ietnam bur not stationed there Flascics and synthetics industries. Cigarette smoking has (13 N H L cases, 27 s,lntrols); prewnce oi AlDS or an been found to he a risk factor for NHL in but AIDS-related illness (239 NHL cases, 1 control); o r not nl)t all,i7studies of this association. a U.S. resident betore 1969 (47 NHL cases, 99 controls). In this study, we used the Centers for Disease Control We excluded those with missing information regarding Selected Cancers Study database to identify occupa- education ( 1 NHL case, 1 control). We also excluded tional risk factors for three subgroups ot' NHL: small cell men n tch the following medical conditions or medical diffuse Iymphmias, f d i c u \ a r lymphomas. and large cell historiss: systemic lupus erythematosus, non-AIDS-re- diffuse lvmphoinns. :is detined :ibnve. l:iteJ i i i i m u i ~ c ~ d e f i s i ~r~hne~u\m~ .n d arthritis, imnitinc3- siipprcssivt. drug use, chcmotheriip. o r medical r d i a - tiun \ 125 S H L crizc>,116 cc)ntrtiI-!. The i i i u l number of cases available for analysis was 1,848, and the final nliinhcr of contrt>li ;i\.nilnble \vas 1.659. study.4~W~ e chose these three subgroups for several (N = 87) cell, and large cell diffuse lymphomas (N= reasons. First, they are mutually exclusive (Figure 1 ) . 2172). SeconJ, small cell diffuse I\mphomas are i>f A-cell origin We computed crude odds ratios to coinpare each KHL as are follicular \ymphomasl'c~22~a'3nd both subgroup with the control subjects with respect to po- groups involve subtypes that behave similarly. Large cell tential confounders, including life-style and demo- diffuse lymphomas, although a mixture of 8- (80%)and graphic variables. We used the SAS PHREG procedure:' T-cell lymphomas, are related by their diffuse, large cell to run conditional logistic regression models to assess the nature, aggressive behavior, and their possible origin in relation between each NHL subgroup and the following activated peripheral lymphoid cell^.:^*^^ T h e remaining exposures: pesticides, herbicides (farming and nonfarm- subgroups-follicular center cell lymphomas, blastic ing exposure treated separately), wood/saw dust, sol- lymphomas, and peripheral T-cell lymphomas-are vqncs, shw/leather dust, meat packaging or proce?mpc, more heterogeneous groupings phenotypically and be- niecal plating, cutting oils, formaldehyde, chlorophenok;'-- haviorally, and they overlap wich the other selected heterocyclic nitrogens, carbamates, organophosphates,. subgroups and each orhkr.'.8 phenoxy herbicides, chlorinated hydrocarbons, and py- rethroids (not considered for small cell diffuse lympho- OCCUPATIONAL ENPOSLREk X S S M E N T mas owing to absence of exposure among cases). W e considered each exposure separately in models chat were The job history portion of the interview first asked stratified on the matching variables registry and date of participants the followinu questions regarding all full- or birth (4c7 strata), and we controlled for the fklllowing part-time jobs held for at least 1 year: job title, main potential confounders: age at diagnosislcase selection duties, type of business or industry, year started the job, (5-year age groups), education, ethnicity, year entered and ye3r ended rhc j&. P m i c i p n c s \ \ w e then a s k d the yrudy ( \ c a r d'c ~ s ediagnosis o r control notiticxion about specific jobs (such as meat packaging/processing), of enrollment), being Jewish, having never married, and if this job had not been previously mentioned, the AIDS risk behaviors including intravenous drug use and questions listed above were asked. The third part of the sex with a male, use of seixre drugs, service in Vietnam, job history included questions about jobs that may in- and cigarette smoking. If an occupational exposure vari- volve the use of herbicides o r pesticides and the names able appeared to be a potential risk factor for an NHL of specific pesticides and herbicides, but questions con- subgroup, we evaluated the impact of duration and time cerning year of contact referred to all of the pesticides or since first esposure. herbicides together rather than specific pesticides or herbicides. The final part of the job history involved questions about specific materials participants may have worked wich or around on a job (such as formalJehyde). If the associated job had not been previously mentioned, Results DLSCRIPTIV.~NE.ALYI;IS We present crude associations between NHL subgroups the questions listed , i b u \ ~were asked. and various potenrial risk factors in Table 1. High levels We crenteci a variable describing time since first ex- of cigarette moking increased the observed risk tor all posure fur d l jubs ur exposures for which year srarted the job hall been asked. This variable could not be createcf for specific pesticides o r herbicides. We divided time since flrst exposure into exposure within 10 years and subgroup$.Hi?p;inic ethnicity m d higher eJuc3tion were asuciated wid1 decreased r i s k tor all NHL, and ;i stariis of never ha\.ing been married and reporred ,41DS ri.k behavior increased the estimated risk for all NHL. For more than IO years before case diagnosis or control follicular lymphomas, we observed decreased risk for selection. We calculated total years exposed if year w r t e d and year ended the job werc known, and we used black and Hispanic ethnicity. Never having been mar- ried and AIDS risk behavior were associated with in- the median of the non-iero values c>t'tor;il years c ~ p ~ s e d creased risk tor Inrge cell ditfuse lymphomas, and hipher tor the case subjects a s a cutpoint in creating Juration of cspvure v.iri,ible. . education ,ind low levels of smoking were associated with a recfticcJ r i k . T,ible 2 presents the frequencies cd occupaticm,il expsiires for study suhjects. ~ ~ L X Y ~ L IRSEGSCTS Table 3 presents rhe results of models assessing the association between occupational exposures and KHL subgroups. For ;ill NHL combined and for large cell diffuse lymphomas, the risk among the exposed and unexposed ,qyxarecf similar after concrolhng for the var- ious covxixes. For this reason, we JiJ not perform Jiirarion o r Litcncy iinalyses. hlen C S F \ ~rIo~\~olJ\,enrs :ippe;ireJ r o he ,it iiir rc.ise< r i s k $ 4w i . i I l <-'-I1 J i t t ' i i w lviiiphtiiiiLis[ t i A ~r ; i t i o ( ( X )= 1.62; L)S''~, cTlmt'i\tc*iiicnrcrv,iI ((:I) = l . I ~ ' - 2 . 2 ~ ' ] . The ic.>ii:c.x .h;,.,i\..c> ,*I .Iur,icih~ii.iii,I riiiiz h i i i i c :ii..c ~ ~ \ p ~ - t 7 Epidemiology September 1997. Volume P Number 5 TABLE 1. Distribution of Cases and Controls According to Selected Charactzriatics, with Odds Ratios (ORs) and 95% Cu..?dence Intervals (CIS)for All Types of Non-Hodgkin's Lymphoma Combined and for Subgroups of Non-Hodgkin's Lymphoma .,: , 'f Age (years) at study e n r d 31- 33 36-4 4115 4650 51- 55 56-60 134 317 Hi 318 333 210 Ed<ucaHtiiognh ichcxdf High school. rech . 1- 3 vears collree 2 4 years college 1YC 40s 341 642 Year oi stirdv entry 1934,1985+ 1956. 1987 1YS8,199Y 2L.5 I.\x'6 44:J Cigarette moktnq (pickwars) Nonsmuker: 562 >2 to < I 5 367 I 5 to <j$ 33: 30 to <I5 227 z45 189 L'nknown pack-year.; 4 60 1.00 5 148 3.91 (039-1.40) 14 187 0.79(0.45-1.40) 31 214 0.73(0.37-1.40) 34 281 0.66(0.31-1.40) 56 158 0.50(0.22-1.10) 45 I .m 1.m(0.31-3.50 1.90 (0.43480) 1.60 i0.2Y4.60) 1.50 ('2.25-9.30) 1.50(0.22-Q.50) 15 1.LV 32 1.&' 50 1.10(0.52-2.30) 73 0.89(0.51-1.60) 53 0.59(0.3-1.50) 98 0.82(0.39-1.70) 59 0.50 (0.16-1.50) I12 0.79(0.33-1.90) 61 0.42 (0.12-1.5@) 144 0.72 (0.!!3-1.90) 3 3 0.34 (0.09-1.30) 67 2 4 i (0.li-1.30) I 5 5 1.W 32 1.00 152 0.88(C.68-1.10) 59 d.86 (0.53-1.40) 113 0.85(0.64-1.10) 38 2.97(0.57-1.60) 3 3 0.71(0.554.93) 56 G.77 (0.47-1.30) 36 l.W 76 120 97 0.90(0.58-1.40) 178 0.87 (0.62-1.20) 53 @.30(0.j&1.40) 111 0.82 (&ji-l.2C) 90 0.X((2.45-1.10) 161 0.65(0.47-d.91) 227 1.M 35 1.00 66 1233 109 1.00 281 P.50(C.42-0.63) IC3 $151 (0.39-2.94) 152 2.47 (0.344.66) 3 9 3.50 (0.382.66) -40 C.46(0.35,7.59! 47 P.55 (C.344.90) 50 0.35 (0.23-0.54) 128 0.5C (0.364.68) 309 I.iV 49 1.iv 69 1 . N I70 1.N I75 0.87 (0.69-1.10) 3 1 0.96(0.63-1.50) 5 2 1.10(0.78-1.70) i 8 0.69(0.51-@.94) 202 ljl 1.10(c'.84-I.?0) 34 1.13(C.ss-i.jC) jI I.LW(C.66-1.70) 57 1.20(?.76-:.N) 41 1.40 (0.96-2.10) 98 0.92 (0.69-1.20) l.5C (0.97-2.30) 70 I.@@ (0.72-1.40) 199 1.M(1.3~w.10) 36 1.40 (0.89-2.30) 47 1.92 il.20-2.90) 103 1.60(1.202.20) I! 5.N(1.611-16.N) 4 9.JO(!.IN!.@@I 1 3.X(O.31-28.W) i 5.40(1.5~19.00) ; i d first exposure within 10 years betore rnrolliuent was more strongly associared with risk oi follicular lymphcmia than t'inr ttspwire more than 1I: yrt,irs b & w cmrr 1I1iiieiit ( T.11.k 4 1. Epidemiology Septctrnber 1997. Volume 8 Number 5 OCCUPATION ANL3 NON-HODGKIN'S LYMPHOMA 555 TABLE 2. Frequencies of Occupational Exposures for C o n t r o l s a n d Cases, According t o Subgroups of Non-Hodgkin's Lymphoma Esposure Controls (N = 1.659) .%I1 Non- Hodgkin's Lymphoma (N = 1,048) Numher ( 9 6 ) Small Cell Diffuse Lymphomas (N = 185) Follicular Lymphomas (N = 268) Large Cell Diffuse Lymphomas (N = 526) Farming exposure to herhicides Yes 291 (16.9) 90 1,376 (93.1) 26 (14.0) a 159 (d6.C) 57 (21.3) 211 (i8.i) Farming exposure co pesticides Yes - 384(23.1) N O 1.275 (i6.9) / 34 (15.4) 51 (81.6) 83 (31.0) 185 (69.5) Sonfarming exposure to herhicides Yes 191 ( 11.5) NO 1.468 (88.5) 15 (18.1) 70 (91.9) 37 (13.8) 231 (86.2) 5 5 (10.5) 471 (89.5) Chlorinnred hydrocartans Yes 32 (4.9) N O 1.577 (95.1) 9 (4.9) 76 (95.1) 15 (5.6) 253 (94.41 !!(4.2) 524 (95.8) Heterocwlic nirrocen, Yes 67 (4.2) N O 1,592 (96.0) 4(2) 181 (97.8) 11 0 . 1 ) 257 (95.9) 'I (4.2) 505 (96.0) Chlorophenoxy herhicides Yes 98 (5.9) NO 1.561 (94.1) 6 (3.2) 179 (96.8) 19 (7.1) 249 (92.9) 26 (4.9) 5 0 3 (95.1) Carbamates Yes NO 47 (2.8) 1.61: ( Y i . 2 ) 3 (1.6) IS2 (98.4) 11 (4,1) 2 5 i (9>.9) I ? (2.3) 514 (97.;) Organophosphates Yes 83 ( 5 . G ) No l j i 6 (95.2) 4 (2.2) Id1 i37.Y) 10 ( 3 . i ) 258 (96.3) 22 (4.2) 504 (35.8) Pyrethroids Yes NO IC (0.6) 1.649 (99.4) !(0.8) 266 (99.2) 1 (0.2) 525 (99.3) Chlorophenuls Yes NU 13: (11.0) I.4ii (39.3) 44 (16.4) 224 (83.6) Mcrol plating I Yes 54 (3.2) No 1.605 (96.3) 7 (2.6) 261 (97.4) 16 ( S . i ) 169 (91.3) 29 (117 3) 239 (89.2) Formaldehyde Ye, No 13Q(7.5) I , Z Y (92.2) 21 (11.3) 164 (SB.7) 114 (42.5) ljt (j7.j) 17 (6.31 251 (93.0 i ? (26.9) 196 (73.1) 139 176.4) 33; (73.6) i(1.6) 261 (974) 5 (1.2) 521 (99.3) i 4') ( I P . 1) !lU(Sl 7) TABLE 3. Associations between Occupational Exposures a n d Subgroups of Non-Hodgkin's Lymphoma* Exposure All Ntm-Hdekin'* Lymphoma Odds Ratio (95% Confidence Interval) Small Cell D1tiu.e Lymphomas Follicular Lvmphomas I_ Large Cell Diffuse Lymphonub Farming exposure to herbicide3 0.87 ('2.69-1.10) 0.73 (0.46-1.2Q) 1.10 (3.i6-1.50) 2.84 (C.62-l.lC) Farmtng exposure to FesciciJe, $.91 (C.75-1.1:) 0.67(C.44-l.c'c7) 1.3C (2.96-1.80) 0.81 (0.62-1.30) Nonfarming exposure to herbicides Chlorinated hvdrocarh)ns 1 3 (0.80-1.3C) 0.91 (0.62-1.30) 3.73 (0.41-1.317) 0.78(0.J 7-1?.'IO) 1.32(0.84-1.90) 1.10 (0.6Q-Z.LW) P.92 (0.66-1.30) 0.80 (0.49-1.317) Heterocvclic nitrogens Chlorophenoxy herbiciJe, 3.51 ($53-1.33) 2.ib (C'.52-1.1<~ 0.49(C. I i-1.40) 2.4; (P.I9-1.1?) 0.61(0.30-1.20) 1.cT(0.58-1.80) 17.9; (C.56-1.i0) 17.76 (0.47-1.217) Carbamates Organophosphdtes Pyrethroids 0.93 (0.56-1.5C) C 66 (0.43-i.cC') 0 3 (@.14-1.Y0) Q.64(0.19-2.2~1) 0.44(0.15-1.33) -8 L 1.12 (C.55-2.30) 00..5948((00.2.7?-014.1.0i) @) 2.8; (2.+4-l.i~?) 30..4816 ((0P.5.20-51-.43C.)30) Chlorophenoh . 1.10 (0.86-I.4C) 1.317 (0.85-2.12) 1.42 (0.94-Z.CcI) 0.84(0.6c?-l.!C) Metal plating 0.83(0.52-1.30) 0.69 (0.!4-Z.&') 2.80(0.35-1 30) 0.79(C.43-1.50) Meat packaging/prccerring 1.013 (0.i2-1.40) 1.20 (0.66-2.133 1.M{0.99-2.6@) 0.59(0.36496) Chemical solvents 1.117 (0.9Ly1.30) 1.60 (l.lL&2.X) l.iV (0.79-I.*') 3.97 (@.79-I.?0) Formaldehyde 1.X(0.86-1.50) 1.40 (0.87-L.40) 0.71 (@.41-l.?0) 1.10 (0.79-l.iO) W d l s a w dU>t I .iyI (0.84-1.2c') 1.30 (0.93-1.9C) 0.88 (0.64-1.20) 1.00 (0.80-1.30) Shoe/learher dust 1.10 (C.53-2.10) 0.93(@.2l-4.!3) 1.62(0.644.20) 3.71 (0.26-1.9C) 1 Cuttine oils 3.9: (0.75-1.13) 1.29(0.84-1.80) 0.i: (C.5l-l.C@) 3.91 (0.70-1.!0! '.., 'I 0 . tant opportunity to perform robust analyses oi occupa- insecticides and organophosphates on livestock were I . tional risk factors for NHL suhgroups considered possibly associated with small lymphocytic lymphomas.'@ Expo, less etiologically heterogeneous than all NHL combined. sure to Fesricides and insecticides and to dust, sawdust, T h e few previous studies reporting subgroup-specific re- and fiber particles was associated with large cell diffuse sults have been small, and risk estimates have generally lymphomas In one st~cly."Based on these scant findings, been unstahle. In these studies, risk factors associated there is little prior evidence to form the basis of testable with follicular lymphomas have included the ~ c u p a - hypothoscs, and rather, the present analyses should be tions emt.almer/funeraI carpenterlplumber viewed as hypothesis generating. Although primarily a and construction worker,'; residing near a stone, clay, or negative study, the two associations observed in this glass industrial t a ~ i l i t ya, n~d~ exposure to benxne, oils, exploratory analysis provide leads to be investigated in and greases.'" Fxmine ancj the use of natural product future studies. TABLE 4. Results ofDuration and T i m e since First Exposure Analyses for Occupational Exposures Noted t o Be Associated with Small Cell Diffuse and Follicular Lymphomas Subgiiup oi YonI HodRkin's Lvmphoma Small cell rliifuse lymphomas (N= 1853 .------ Expisure Sol\m t s Expcwre Hisron Duration 1)f expcwre (bc.irsI Nor t.xpwaif 59 , >9 Number o i Number Controls' OJJs ,it CJSKS* (N = 1,659) R.itiot 95O.6 ConiiJence Internal Y9 9S2 I .A! 47 194 1.52 0.3'4-2.20 47 Z 79 1.717 l.l~?-2.63 IN = :6d) S9 980 I OP 16 38 3 00 I .60-5.,33 iY 575 1.52 1.N-2.10 239 1,563 1 20 14 e;? 1.20 G.63-2.23 I4 33 2.10 l.lL7-4.?2 Epidemiology Seprzmber 1997, Volume 8 Number 5 0CCL:P.ATION AND SON-HODGKIN'S LYXt PHOXtA 557 In this study, only two associations were noted in the lymphomas tvould be 2.22 (estimated crude OR = 1.98). espected direction: exposure to solvents appeared to These calculations indicate little bias to the null. increase the risk of small cell diffuse lymphomas, and A general concern with these analyses IS that known exposure to meat work appeared to increase the risk of etiologic subgroups of the non-Hodgkin's lymphomas do follicular lymphomas. not currently exist.+.s'The subgroups used in these anal- We found little evidence of a duration-response rela- yses have been proposed by researchers+,$and have been tion for the association between small cell diffuse lym- used recently to a limited extent,20-32but our lack of phomas and solvent exposure. The observed association findings suggests that they may not De etiologically in- ,.may not De meaningful, or the lumping together of many formative. Further classification of NHL for epidemio- .- different chemicals into the category "solvents" may logic studies may be improved by the inclusion of infor- have obscured a dose-response relation in this analysis..- mation such as tumor location, phenotype, a ? & S p e d I c .Our finding is supported by most studies of the relatioh genetic changes. between all NHL and solvent exposure (usually defined as a general exposure to organic solvents or work in an A final concern of this study involves the large number of comparisons. We found two associations in thy industry where exposure to organic solvents is expected, expected direction and three associations in an unex- such as the rubber industry), which did find an associa- pected direction (cutting oils and follicular lymphomas, 329JI-4C Fewer studies reported no a~sociation.:'.'~~'~ and farming exposure to pesticides and meat packaging/ There are also suggesrions that solvent expsure may be processing and large cell diI?use lymphomas) out of a involved in lymphomagenesis by decreasing T-lympho- total of 67 assessed associations. As this is a n exploratory cyte counts4' or causinq clonal aberrations.+' analysis, these findings will need to be replicated. Exposure to meat packaging/processing may increase the risk of follicular lymphomas, as is suggested by a n elevated odds ratio and positive duration-response relation. Meat work involves many potential exposures, including nitrite, incomplete combustion products, fumes from meat packaging materials (benxne includ- Acknowledgments We thank Brooke Fielding for programming assistance. U' Dana Flanders for inpur on andlyrical merhds. Edward Brann for comments on rhe manuscript. ,inJ the Sclecred Cancers Sru& Coopenrive Research Group for rhe Jdta used in rhir scucly. ed), preservatives, ',4,6-trichlorophenoI, and viruses that cause cancer in cattle (bovine leukemia virus) and chickens (avian leukosis virus and Marek's disease virus) and are widely present in raw meat."-4' Few studies of the relation between meat work and NHL exist. Ahat- toir workers were found to be at increased risk of NHL.4' 3s were workers with potentially similar esposures in the food processing" and leathe? industries. Another study did not find meat workers to be at increased risk ot NHL." Veterinarians, a group with potentially similar exposures, were found not to be at increased risk ot N H L in one study.4n There are several limitations of the Selected Cancers Study that may have affected our results. Many occupa- tivnal exposures were rare, resulting in scanty data. '4s the study involves self-reported exposures, there is a porential for recall bias that could bias the estimates to\v;irJ o r away irom the iiull." L,i \'tichin ir d" however, have suggesred that chrre is no reason to assume a differential recall between cases and controls for broad occupational categories similac to c>urs. The validity of self-reported work history is also an issue. In general, self-reported work history has been shown to be v a l l ~ , 5 C -53 although the valicjicy tends to increase with length of time on the job5p,51a,n5d3 decrease with num- ber ot jobs held.",5i Overall, we would most suspect in thu study nondifferential misclassificatirn h e to ~ilack of FreCl.sic)n in wlf-reporteJ expocure chat \vauId bias o u r rewits tow'irJ the null? I t LPo k ) t the -ubjects were c l m i t i c d <isi i r i c % x p w Jw h ~ t ih e y \ \ w e .Ic.ru;ilIy v x p v J , the true cr\i\!I .!.I> r.itio L>r ~ t i l \ . e r i ri~ d ri.k m . i I t cell Jitttiw 1% i , ? t m mwouM b r I .o i t ~ - r i i i i ~ i r e Lci rit,l' ( X = I . % > ) . .+mI r h . i r tibr i i i c ~ . i t u,,rL .id t , d ! u i . l r , .* - Smoking d i d risk $>nion-Htd.:kin', Iyiiiphtini.i , i d niulrirle mw1oni.i. Cancer CauwI Gintriil IY92,j:4Y-55. 16. Francexhi S. Serraino D. Bidoh E. Talamini R. TirJ!i U. &:kine A, LJ Vccchid C. T h e epidemioloq-j oinon-Hodekin'i lymphoma in the north-east of Italy: a h o s p i t a l - h a d cdse-control study. Leuk Res 1989;13:i65-472. 17. Tavani A. N e p i E. Francexhi s. Serr;lino D. La Vecchid C.Smokine hahits and non-Hdgkin't Ivrnphonia: J cxse-sonrrd <tuJvin northern Iralv. Prev Mcd 1994;!3:ii7-45!. 16. E l e c r d Cancers Cwperarive S d v G m u r The .Arrociarion oi SelecreJ Cancers with Service in the U.S. btilitarv in Vietnam: Final R e p ~ t Atlanta: C.S. Depxrmenr ,ti Health and Human Services. IW. 19. Selected Cancers Cwperarive StuJv Group. T h e association oi x l e c r d cancers with service in the US. millran in Vietnam. Arch Intern MeJ -....18UO;I5?:21i3-2483. W<i,enlwrger [7D.H.1rrinetk.n E. .4rmirice !O B-cell nmylnqia: J cun- ceprual understanrling bdserl on the normdl humoral immune rcspme. In: R o x n PP. Fechner RE. ds.Patholow Annual. Nonvalk. CT:Appleton and Ldnp. I9vo;?j:YY-I 15. 21. Hcming SJ. Diffuse small cell lymphomas. In: h!ah!arh IT. ed. The Non- Hdgkin's Lymphomas. London: Edwzard Arnold. 1%';3~9-316. 22. &hen PJ.Jatfr ES. Hisroparholoqv and immunophenorvpinq. In MM.igrdth IT, d.T h e Non-Hdgkin'r Lymphomas. LonJon: &ad Arnold. 1W; 49-76. 23. Long0 DL. Wilson W Follicular lymphoma. In Magrarh IT. d. T h e Non-Hdgkin's Lymphomais. London: EJu-Jrd ArnolJ. 1990293- 108. 24. Ga~v,nor ER. Risher RI. Piifuse dmezsive Ivmrhoma, in dults. In: Magrath IT, ed. The Non.Hdqkin'r Linyphomas. 'LinJun: EJwirJ .Arnold, 1990: 317-329. 2 5 SAS Insrirure. SAS Technical R e p m P.229. Cin.. NC: SAS Institute. 108:. 26' Linor A. Blair A. CmrLir KP. Burmeisrer L. Van Lier 5. Gibson RW. %human L. Everett G.Leukemia ,inJ nnn-HoJ\lgkin's Ivmphuma Jmong cmhalmers anJ funeral Jirrcron (Letter\. j Ndrl C.inr.er lnst 1'?%;d!:66. 27. &hen PA. Hutchiron GB, Neiman RS. Yon-Hdckin's hmphoma and ocupational exp~surr.CJnier Res 1902.51(suprl):5 5C3s-551CQs. 28. Linos A. Blair A. Gibwn RW. Everett G. Van Lirr 5. Cmror KP,Schurnan L. Burmeister L. Leukemia .inJ non.Hdpkin', lvmrlwma 3nJ reidenttal proximity to indusrriai pIanr3. Arch Environ Hc.ilrh lQQI;46:7?-71. '9. Blair A. Linor A. Stewan fa\. Burmeister LF. Gibwn R. Everett G. Schw man L. Canror KP. Comments 'in xcuparional J n J environmental iacrors in rhe origin oi non-Hdgkin's lymphoma. C m c e r Rcr iW?;i:(suppl): 5501s-55CL. 30. Cantor KP. Blair .A. Everett G. Gihson R. Purmeister LF. Brown Lbl, Schuman LM. Pick FR. Pesticides mJ <orher .~qisi~lturaril,k tactors ior non-Hdgkin'r h p h o m d ;lmony men in Iou..~J n J IlinnewtJ. Cancer Res 1~~;52(suppI):~4i;~~i5js. 31 McMichdel Aj, Spina, R. Kupper LL. An epldemi~ri~)ciiCrudv of mortality uitliin ,I iSjh,wr.'I ribber wd.erG. ! d W - I Q i I 1 &cup 5teJ I'K4:16:459- 464 12 Mwvon RR. Fine LJ. Cancer monalirv ,inJ morbiJirv .immy r u h k r work. i'r? J Narl Cmcer lnzt l~i3,~l:l~47-l~5~. L'dkc)>kv TC. Chcck,ir;iv H. ilmh.ill EG. T\nder ki.4. .:.,&crmon.ilirv .inJ sdvenr cxwsures in rhe rubber i d u s r n - . ,Am ind ti<< ,A,wc J 138i: i5KW-311. 34 Wads JS, Pdiuar L. severwn RK. H e u w E.ihlJnJer 6G. Soh rissue ,nrcom,i .ind nun-Hdpkin', Ivmphuma in relaticm phen,nv hrrbiciJe 3nJ chlorin.ircd yhen$d c x p * ~ t i r rin uu-rern 'X'i.hingron. .I Sad Cancer In,[ lWi;iR:SYU -0lC. 35. Olswn H. R r d r L. R I A oinon-HcJgkin'c lvmphoma among men o c c i 1 ~ tionally e x p ~ e Jto org.inic jolvenrs. Scand J Work Environ Health IVM; 1424b-251. 36. Cartwriqhr R4.McKinnev PA. OBrien C. Richards IDG.Roberts B.La&r I.Darwin CM. Bernard S M . Bird CC. Non-Hdgkin's Ivmphoma: case- cuntrol eFiilcmiidogic.i~mJv in Y , d h i r e . Leuk Res 1988:1?:81-88. 37. Perssm B. Dahlander ,A. FrrJrikscm M.Nwrlind Brdee H, Ohlwn CG. .Axellon 0.Ltalignant Ivmphomas and cxcuptional expxures. Br InJ Med IY8Y:i6:516-3:0. , 38. Landrigan PJ. Critical i s ~ , s m e n toi epdemiologic rrurlies o n the human carcinwnciicy oi I.3-hutdiene. Environ Health Perspct 1 ~ : 8 6 : 1 4 3 - 143. 39. Pmanen T. Kauppinen T.Luukkonen R. Hakulincn T;Fukkala E. Malig- nant 1vmrhom.h a d ICU~CIIII,.n~J~.expwurcs in the u'otd i n h n ,111 indwtrv-baseJ case-rckrenr m d v . Inr Arch &cup Environ Health 19Y3; &:59 3-596. IC'. Hadell L. Erikswn M. Degerman A. Exposure to phenoxyaceric xi&. c h h e p h e n o b or orgmic * h w t r > in relation to hisropatholo~y.sage and macomical locali:ation ot non-HI*fpkin's Ivmphom. Cancer R a 1994;54 :386-:3Liv. 41. La Vecchia C. Negri E. D'.Avanro B. Francexhi S. &cupation J I l~ym- phoid neqhsms. Br J Cancer 1969;60:385-388. 42. jpirras R. S ~ C WPJA~. Lue JS. %larmo DE. Forbe. CD. Grauman DJ, Pertiqreu. HM. Blair A. Hwver RN. Cohen JL. Retrurpecrive cohon mor- ralirv truJv oi workers JI .in arCr.itt ni.untendnce facility. 1. Epidemiological rewirs. Rr J InJ MeJ 1991;i8:515-535. 43. Moszcrvnski P. Organic d v e n r r and T-l$mphocytes. Lancer lYUl;l:4h3. 44. BranJt L. Olsson H. Mitelman F. Relation hetween occupational expuure to ,irg.iiiic wlventc , i d chromosome akrrarions in non-H-ttdgkin's Ivm- phoma. Eur J Haemand 1989:iL:ZY8-3J?. 45. Kristen,en TS. Lynee E. Lune cancer .imong butchers and slaughterhouse worken. ScdnJ J Work Environ Health IYY3:19:137-Ii?. 46. Johnson ES. Fischmm HR. h ~ ~ t a n o r kGtM.Diamond E. Cincer mortaliry .imonr white male< in the medt indurrrv. J C ~ CMUe d~lW6;?8:?3-3?. 47. Pearce N.Smith AH. R d IS.Incrcaxrl risks of soit tiwe sarcoma. malipniint Ivmrhoma . i d cute invelcd leukemia in ~hatroitworkers. Am J lnd MeJ lYSS;l4:63-72. 48. Blair A. Haver Hb! Jr. hlortnlitv patterns among U.S. veterinarians: 1947- IY77: ,in expanded s t d v . In1 I Epiderniol lY8l;l1:391-397. 49 CheckowJy H. Pc.irce NE. CrauiorJ-Brown DJ.Research M e t h d s in Oc- cupational EpiJernioLlp) New t'ork: OxiorJ University Press. 1989. Si). Baum~anenM. Sirmiarvcki J. Gihbs C W . V ~ l i d i to~i work histories obr a i n d bv mrcrvieu. ior cpi&mic>lOalc ~urposes.Am J EpiJemiol 1983;118: 583-59 I 5 I . k d w n n a i s R. %!ever F. Theridult G.V ~ l i J i ot ~i self-reporreJ work history. Rr I InJ 5leJ lU88.4i.Y-l2. E!.5 : . R , n i ' . t ( d u h I '/ tiidin .drcpcit h i i t \ ,.i . d i - r q v r t d ~ ~ ~ ~ . i t i ~ . ~ . i I And inJumiaI hirrun from parienr5 in EED iwntries. Inr J EpiJemii)I IQS'J:I ?:0;4-6?4 i 3 S i e ~ ~Wr tF. TGm.i.ci.i 1.4. \ I I*. GSt. The vdidirv d questionnaire, rept)rreJ u.rrk h i m w in I i \ c .: iiJenr. J ~ ' c : c ~Jp!eJ lY8i;:Y:i95-SCC. 54. ]at& ES.R~iir.l\l&I. &ledcir<,*LJ.Eicrler-SrevcnWn If. .An overview <,ithe ~ l . i ~ ~ i i i c ~a ti ni ~mm- H d g k i n ' r ivniphumu: .in inteqrxion di rnolyht,loqical .ind phenorvyicd concepr5. Cancer Res 19V::i?(,uppt):jJ-);ti.(j2s. .---.--