Document 91vJb7M2aB4ZY7QNoo90x5gkV

Original Contributions ~ -; .- 1 Agricultural Herbicide Use and Risk of Lymphoma and Soft-Tissue Sarcoma Shefia K. Hoar. ScD; Aaron Blair, F'hD; Frederick F. Holmes,MD Cathy D. Boysen; Robert J. Robel. PhD: Robert Hoover. MD; Joseph F. Fraumeni. Jr. MD A population-based case-control study of soft-tissue sarcoma (STS), Hodgkin's disease (HD). and non-Hodgkin's tynphoma (NHL) in Kansas found farm herbicide use to be associated with NHL (odds ratio [OR], 1.6; 95% confidence interval [Cl]. 0.9. 2.6). Relative risk of NHL increased significantly with number of days of herbicide exposure per year and lalency. Men exposed to herbicides more than 20 days per year had a sixfold increased risk of NHL (OR. 6.0; 95% CI. 1.9. 19.5) relative to nonlarmers. Frequent users who mixed 01 applied the harbicides themsehtes had an OR of 8.0 (95% CI. 2.3,27.9)for NHL.Excessem were associated with use of phenoxyacetic acid herbicides. spedcicaly 2,4-dichiorophenoxyacetk acid. Neither STS nor HD was a s s o d a l e d with pesticide exposure. This study confirms the reports from Sweden and several US states that NHL is associated with farm herbicide useo.especially phenoxyacetic acids. W does not cwlfinn the casecontrot studiis or the cohort studies of pesticide manufacturer8 and Wehem veterans linking herbicides to STS or HD. (JAMA 1986Sk1141-1147) EPIDEMIOLOGIC studies from Swe benzofurans There have also been swden have r u ~ t s dthat worken era1 reports of increased SI'S and NHL exposed ta phenoxyacetic acid herbi- among work- producing phenoxycides and chlorophenols are at 0 1 0 ~ s acetic acid herbicidesc' and among risk of soft-tissue sueoma (STS). fnrmemc'o Hodgkin's diseae (HD). and non- Hodgkin's lymphoma (NtEL).'" For dl three cancers, rirks w e e i n a d fivefold to sixfold, regardless of wheth- Conccrn over possible carcinogenic cr exposures were contaminated by risks from phenoxyacetic acids and polychlorinated dibenodiodns or di- chlomphenols is heightend by the potential for widespread exposure. In addition ta herbicide fonnulrtionsusd in agriculture and in the Vietnam WU, these chemicals occur in blue stain retrrdants used in sawmills, slime OWtrol preparations in paper and pulp manufacturing, cutting oils and fluids, wood preservatives, watmpmof1ng agents for leather and textilea, and medications' For these reasons, a pop ulation-based cnsacontrol study was conducted to clarify whether agricult u r d use of herbicides and insecticides affects risk of ST$ HD, and NHL in the United States METHODS Kansu. a major wheat-producing state, was chosen aa tbe location for the study, since herbicides have bean UICd mora frequently than other paticidcr on wheat. ZCDithtorophenoxyacetic acid (&4-D) hm been the most commonly u d herbicide in Kaa=% subatantid amounts of &4$-triddonk phenoxy.atic acid (24,S-T)h v a a b bean usedh along with myriad other chemicals. Caoee All newly diagnosed ofcpsc~l SIS. HD.and NHL amonp white male Kan- popuiation-based registry covering the state of Knasaa Cancer reportink mandated by Kansas law, in eonsidered campletc, as evidenced b a higher annual incidence rate for STS (U/ 100 OOO) than reported by the n-by Iowa National Cancer Institute-Bponsored SurveiUana, Epidemiology, and End Results registry (3.UlOOOOO). There w e n 2M) men diagnosed dtb JAMA. Sepr 5.19BS-Vd 256. No. 9 UIC-HOU CJI 1141 -- - 1-5 6-15 2 16 unknown __~ x lor trend . .-P (0n~al.M) ~ fiaquwryof- uw. d l y 0 -~ ~ 1-5 6- 10 11-20 L21 Unknor*n x for 1reI.d P (au-hiM) 53 5 0.7 0.2, 2.0) 9 0.9 (0.4, 2.21 9 1.3 (0.6.3.1) 58 5 0.7 (0.2. 1.9) 8 0.8 0 3 . 1.9) 12 1.6 (0.7, 3.4) fit$ y57 11 1.4 (0.6. 3.1) 10 1.2 (0.5. 2.6) 2.0 (1.0. 4.0) 24 1 ... 1 -.. ~ ~- ~ 0.13 0.11 3 ... 2.06 .a ~~ .UT ~~ .02 447 ~~ 102 1 4 1.1 (0.7. 1.7) S 0.6 (0.3. 1.5) 45 0.7 (0.4. 1.1) ~~ i e 1.0 (0.5. 1.0) 94 1.3 (0.8.2.0) 19 1 4 (0.7. 2.6) 29 2 0.5 (0.1, 2.3) 2 0.4 (0.1. 1.7) 6 1.6 (0.5. 4.3) 13 5 2.9 (o.a.9.5) 2 1.3 (0.2. 6.8) j, 2.6 (0.8, 8.81 d12 1 0.8 (0.04.6.5) 3 1.0 (0.2. 4.1) (7) 6.0 (1.9. 19.5) 94 ~ ~~ ... 3 ... - 2 om2 -0.56 333 K) .29 CmM STS and 173 men with HD.A random sample of 200 men was drawn from the 297 men diagnosed with NHL from 1979 through 1981. Pathology specimens for 87% of the cases were reviewed by a panel of rhm pathologists t6 confirm the diagnoses and to standardize the subgroup terminology.w' Specimana for the remaining CLD~S either codti not be obtained 111%)or were of noor or insufficient Controls The controls were white men from the general population of K a n m . T h m controls (N-1005)were matched to each patient on age ( + 2 y e a d and vital status. For living patients, a n trols aged 65 years or older were selected from the Health Care Financing Administration file (Mediare), whereas controls aged 64 year or younger were selected by telephone, using a two-staged random digitdialing technique." Control selection was not hied by differences in telephone coverage for urban (91%) and rural (90% ) white Kansan houuholdr" For 1142 JAMA. S d 5. 1986-Vd 256. No 9 923% oi the working residential tele farm locations and sizes, herbicides phone numbem called, a person an- and insecticides used, years and acres swered who was willing t~ provide the treated. names and locations of compa- names and ages of household members nies where pestiddm were purchased, aged P to 64 years and if someone in method of application, days per year the household was aelected as a control, expomid, and use of protective e q u i p the household address. For deceased ment. Information on other crops was patients, the controls were selected not obtained; however, in 1978, these from Kansas stah mortality fila, with four craps mnotitutai 94% of Kansas I \the additional matching factor of year of death. Personswith a cpue of death of STS, HD.N'HLa malignancy of an illaefiaed site (Illtrrnotiond Chs+ cotion of Disapta cad0 195). homicide, or suicide were excluded. One half of the patients with SI'S and NHL died farm acreage and 87% of acres treated with hubicida'" All but four subjab who lived or worked on farmland z u d u l b grew at least one of the four -led crops. ' Interviews were obtainud from 133 ptkb With SI'S, 121 with HD, 170 Ibefore the initiation of the study. while with N&. and 948 controls, which only one third of the patients with HD represented 95% of the eligible s u b had died. The next of kin were infer- jsctr (patients, 96%; cantmls, %%I. viewed for deceased subjeck The same The overall respom rate, a weighted controls were used for mmparisons average accounting for the initial with the three different muxr series n f d s in thc random digitdialing when of comparable age. control selection proccrs, was 93%. lntenriew The patients and controls, or their Pesticide Suppliers' Survey next of kin, were intorviewed by tale To evaluate potential observation phone between December 1982 and bias, corroborative evidence of the selfJanuory 1984 The questions on farm- reported pesticide exposures was ing practices covered the uiendar sought for a sample of 130 subjects years working or living on farmland with farming experiencs. Suppliers for during which any of four specific crops 110 subjects wen locatcd and provided (wheat, corn, sorghum, or pasture) information on the subject's crops and herbicidt and insecticide purchases &lrpx\'cs4 Tahle P.-NmHod~hm's Lymphan8 in AalaIMn to Herbicide Grwp risk of STS or HD were scen with Horbicide Group N e w tumW Ever use* PhanCrXyAUbC d d S No. 01 Cam. 37 24 No. 0 ) Contmh 286 78 Odds Ralio (95% cmn#ii.ncm Intowall J1.a 2.2 (1.2. 4.1) either duration or frequenq of herbi- cide use. More detailed analyses showed no association between agricul- tural factors and the oCcurrenEe of STS or HD, 50 the remainder of this report will describe only the h3a results. TrOZlMS Amides &fUOlCS Carbamates 14 8. 1 2 43 22 2 3 2.5 (1.2.54) 2.9 (1.1. 7.6) d 4.0 (0.1. 62.6) 5.6 (0.6. 45.9) iSubjects who reported usually mir- ing or applying the herbicides them- selves (OR,19; 95% CI. 1.1, 33) had higher risks for NHL than those who reported that someone else performed these functions (OR, 1.1). Indeed. the Trilkrrrh 3 2 12.5 (1.6, l16.l)J trends in the OR with increasing fre- Utacas Nonrpecilled 99 114 8 11 1.3 (0.7, 2.5) S.8 (1.9, 17.2) v quency and duration of use derived mainly from the workers who mixed or applied the herbicides themselver For Hierarchid u w t example, men who mixed or applied Phenoxyacelr adds the herbicides and who were exposed ~~~~ ~~ uracils;M phmnoxyrCRtiCa d Triazines; no phenorylrutic acids ar urmcils ~ Amdes; no phonoxyacslk rear. uracils. or Vmzinea 13 ~ 3 0 101 11 1 1.0 (0.5, 2.1) 2.2 (0.4, 9.1) ... for one to five, six to ten, 11to 20. and more than 20 days per year had ORs of 1.4 (95% CI,0.7, 3.0), 1.5 (95% CI,0.5, 4.6), 1.8 (95% CI,0.4, 7.4), and 6.0 (95% CI, 2.3, 27.3seven patients, nine con- trols). Fanners who did not use protec- tive equipment. such as rubber gloves or masks, had a higher OR associated with herbicide use (OR, 21; 95% CI,1.0. 4 8 ) than those who protected them- selves (OR), 1.5; 95% CI, 0.7, 3.1). Risk Measurements ghum. or pasture) was reported by 22 Higher risks for herbicide use were The measure of association between patients with STS, 28 with HD, and 40 also seen among farmers who used pesticide exposures and risk of eancer was the odds ratio (OR). All estimates with NHI,compared with 192 controls. yielding ORs oi 09 (95% CI,0.5, 1.6). backpack or hand sprayers (OR, 23; 95% CI. 1.0.52). an application method were adjusted for the effects of age by 0.9 (95% CI, 05. 1.5). and 1.6 (95% CIL, with greater potential for personal stratification. Maximum likelihood es- 0.9. 26), respectively. -was a exposure than other methodsm After timates of the overall risk and the 95% signficant trend (P=.O2) in risk of excluding persons who used backpack confidence interval (CI)were computed MIL with increasing years of herbicide or hand sprayers, the ORs for tr.a_c_ tor- I i by G a d s method." Matched analyses use and with number of days of herbi- mounted or mist-blower spraying and yielded results similar to those p m cidc urposure per year (P=.oooQ) (To- aerial application of herbicides were vided by the unmatched approach. The ble 1). Persons exposed to herbicides both 1.5.Too few subjects remained for unmatched analyses are presented isince they allowed control of additional more than 20 days per year had an OR of 6.0 (95% CI.19.1951. There was no evaluation of the group that applied herbicides only in the soil faactors that were not matching factors d a t i o n with years of `herbicideuse A logistic kgression analysis was and increased power by pooling con- after adjustment for annual days. On performed to examine risk accounting trols for ea& case series. For duration- the other hand,adjustment for years of for all pesticide exposure variables response relationships. significance herbicide use changed the OR for the simultaneously. The results showed was assessed by means of Mantel's five patients and six controls exposed that age and annual days of herbicide I onetailed linear trend test" Logistic more than 20 days per year from 6.0 to exposure were significantly related to rep?ssion was also performed to eval- 7.4 (9S% CI. 1.6. 38.9), relative to the NHL r i s k Restricting the analysis to uate simultaneously several compo- least exposed users. The risk of NXL persons exposed to herbicides and nents of pesticide exposure." associated with herbicide use did not using catcgoriul variables yielded ORs change significantly when restricted to of 1.1(95% CI,03.40), 13 (95% CI. 0.3. RESULTS exposures incurred before 1976, the 6.0). and 103 (95% CI, 21, 49.5) for Ninety-five patients with STS, 71 earliest possible date of death for the persons exposed to herbicides six to with HD,and 133 with NHL reported pooled control series. The Mn. risk ten,11to 20,and more than 20 days per having worked or lived on farmland. also rose with increasing time since year, respectively, relative to persons compared with 662 controls, yielding fimt exposure. Farmers who s t u t e d exposed one to five days per year. ORs of 1.0 (95% CI. 0.7. 1.6). OB 195% CI,05, 1.2); and 1.4 (95%di, 0.9,` 21). respectively (Table 1). No trend with years cpent working or living on a farm was observed Risks did not vary by crop or farm acreage. Herbicides using herbicides after 1965,from 1956 -thmigh 1965, 1946 through 1955, and before 1946 had ORs of 13,1.7.1.7, and 33.respectively. This trend was dimin- ished by amtrolling for frequency of herbicide use. but fanners who began u s e befokr-6e491 (OR22). abon was seen with Increased risk was also associated with use of a backpack or hand sprayer (OR, 13;95% CI. OS. 3.7)and failure b use protective equipment (OR, 2% 95% CI, OS, 51). Total period of herbicide use was not ;rssociated with NHL,when the other pesticide exposure variables were coatrolled The effect of personaIly Farm herbicide use on any of the nu'm'aerof a y e s treated with hcrbi- mixing or applying herbicides could not four specific crops (wheat, am.sor- cides No consntent patterns of excess be evaluated; the ovvwhclming major- I JAW. S.pC 5.1986-Vd 256, NO 9 - &ntiald UIL.' y 1-5 6.IS 18-25 a28 xlorlrand P (Onern9d) Frmquauy d um.t dly ...1-2 ...3.5 Bo- ele 11-20 221 xfortmd P lawrubd) Finly.plo(uI.$ t966 a ).tu 19~m85 196-1955 amtom 1946 Xb- P (OneUaed) 3 7. 0 8 3.560 .ooo2 -*.Y ...*.: yr,.. 64 4 4 5 3 733 .OM1 5 0 8 2 3-561 OOM 16 22 15 , 17 ... < i 3:. -' -,.#srpe 17 16 16 9 e ... .-. 1.3 (0.3.5.1) *--2.5 (0.0.6.8) ....2.3 (0.z. sm ,<-n-.. i IICC'-- .-. 2.7 (0.9.8.1) 1 4 (0.4. 57) 1.9 (0.5.6.7) 3.0 (0.7. 11.s) ... ... ity of subjects involved in thc logistic analysis either m i d or applica the herbicides themnelvu Significant excesses were rssaiatcd with tver use d pheaoxyacetk acids, triazines (eg,atrazine, ey.nazine m e tribuxin, prometone propdne, terbu- trun). amid- (co.d*, p r o d - lor), trifluralin, and nompecSd herbicidar, such 01 T i q u a " "rpn~&a"nd "dusts" Pabk 2). Mort farmers re pta$Jne of chemicals in d of ;&e herbidde wbgmups SiaOa the a priori hypoth- d d t With phmoxy- acetic acids#we d rirk msoci- atcd with herbicide ranked in a hierarchical manner ("able 2). In the absence of phenmy.ectic acid exposure, the MIL risk associated with triazine e q ~ ~ ~ uwarsereduced to l2 (95% CI. 0.4. EO) and the risk with uracil herbicides (e& b r a m d . tcrba- a,eil) was reduced to 1.0 (95% 0 5 rtudv. -acetic acid 21 23 24 2 ._. ... 1.9 (0.6. 6.0) 2.9 (1.1. 7.2) 2 1 (0.1. 8.6) 6.2 (0.6.85.3) ... ... _ _ -.. sistent. i n c r e k in NHL risGin &a- tion to the dumtion, frequency, and Iateney of s4-D uae (Table 3). The 24-DU#TS e x p o d to berbidea more - ........ -AWIdDmdUr N- U n d lnooclclk. WaOl W d odd8 m&h' mslbcorr, -0 1-5 26 w.adrwted lor-uu 90 474 1.0 11 57 1.1 (0.5. 2.3) 6 30 2.1 (0.7. 5.9) Naror U d Weld.. Maof crr Yaol brrw. odd. Ilruo' (psIm(c dwrt.ktrrJI L.r Uwd Irwy#n N0.d w a d whm (ps%bn* AdhmmdkbwcuddmUu M0.d w0.d 0d-m /-a* 4 21 1.1 (0.3. 3.s 8 43 1.2 (0.5,2-9) 12 as 2-3(1.0, 4.9) 1.1 (0.6.2.2) v u u n d ))w#cld.. W a d Waof h a a s Can- odd.m- (OIItonb dMcmWUv@ 94 496 1.0 19 100 1.1 (0.6, 2.1) i a 59 2.1 (0.0.4.9) A*W*((rbldd.Uu -- odd. N8.d Ysd -6n* t w m Conbwb Om=**M hvraodr 0 1-2 23 nwbddsw..drardb -w #) 2 2 474 1.o 17 12 0.9 (0.1. 4.2) B Q 1.5 (0.2. ai) 12 7s 1.3 (0.7. 2.4) IO7 549 1-0 30 1.7 (0.7. 4.1) 10 42 1.2 (0-5.2.1) 48 1.7 (0.8. 14 57 1.4 (0.6. X l ) 1.4 (0.8.2.4) than 21) days per ycnr had nn OR of 7.6 (95% Ci, I S. 323). 13owerer. t h e e vjriabics c n n o t he delermined with complete kccuracy because the ques.tionnaire elicited dates and frequency of use of any herbicide on each farm instead of dates and frequency for each specific herbicide. Therefore, the aetuaJ years or days of 2.4-Duse may be less than that reported for all herbicides on a farm. Risk associated with herbicide exp+ sure was also examined by subgroups of NHL There were no differences in the ORs associated with herbicide use when cases were categorized by histologic types (follicular, diffuse, and not specified) and by grade (prognostic groupings of low, intermediate. and high), as classified by the Natimw! Cancer Institute Working fEarmulation." In addition, no significant differences in risk for any tumor were associated with either age at diagnosis or-vital status at interview. Deceased subjects had just slightly lower risks of NHL associated with overall herbicide exposure (OR, 1.5; 95% CI,0.7, 33) or Inngterm exposure (OR. 5.5; 95% CI. 0.7, 41.3)than did living subjects (OR, 1.7; 95% CI,0.8, a?;and OR 5.6; 95% CI, 1.1, 28.5 respectively). Insecticides Insecticide use on crops or animals was reported by 54 patients with NHL and 275 controls. yielding an OR of 15 (95% CI, 0.9,24). There was no association with increasing yeam of insecticide use. Risk increased significantly but inconsistently with days of esure per year. Men exposed to inacticides more than six days pes year had a S f o l d increased N I U risk (95% CI, 1 2 65).Farmers who started using insecticides after 1965. from 1956 through 1965. 1946 through 1955, and before 1946 had ORs of 15.0.7.15, and 1.7. respectively. No association was seen with number of acres treated with rnsKticides Other exposure variables, such as mixing and applying insecticides, application method. and insedieide type, showed little or no associa- tion with NHL risk. Herbicides vs Insecticides Table 4 shows that NHL risk increased with annual days of herbicide exposure. but was not further i n c d by simultaneous use of inseeticides. Adjustment for insecticide exp u r e did not alter the ORs for herbicide use. On the other hand, risk increased only slightly with annual days of insecticide exposure within strata of herbicide use- Adjustment for herbicide use d e e d the OR for insecticide use fram 1.5 to 1.1 (95'k CI. 0.6, 2.21. Similar analyses hrrscd on years of exposure also suaes(cd that risk was more strongly associated with exposure to herbicides than to insecti- cides. The number of subjects was too small for a more detailed stratified analysis of herbicide or insecticide use. In a logistic analysis of all pesticide variables. insecticide exposure was not found to be significantly related to risk The important variables remained age and annual days of herbicide expo- sure. The elevation in NHL risk (OK.13) seen among farmers who did m-' yeport herbicide use m a r b+ to other e-surar cni-*:r,iered in farming ac- ti*cS;a or to biased misclassification of exposure status. However, it is not likely that more patients than controls underreported herbicide exposure. Slightly fewer "nonexposed" farming patients with NHL (52%) than mntrole (55%) were deceased, with interviews supplied by next of kin who might underreport exposures Approximately the same percentage of "nonexposed" farming patients (4%) and controls ( 5 % ) were usually employed in agricul- ture. Fungicides Thirty-two patients with NHL and 105 controls treated seeds with fungi- cides (OR,21; 95% CI, 12, X7). Risk waa elevated among both herbicide users (22 patients, 68 mntrols; OR,23, 95% CI. 1.2, 4.3) and farmers who had never used herbicides (ten patients, 37 controls; OR, 1 3 95% CI,0.8, 4.4). No other information on fungicide use was collested to allow further evaluation of this association. Pesticide Suppliers' Data Suppliers usuaUy reported less pesticide usa than subjects Agreement on speczc years of exposure was better for insecticide use than herbicide use. There were no consistent differences between agreement rates for patients and controls. Multiplying the number of patients and controls who reported any herbicide use by the percentage of verified herbicide use yielded a recalm- lated OR of 1.8 for NHL,which was slightly higher than the OR based on intemew data only (1.6). Agreement between suppliers and subjects improved when pesticide use during tbe last ten years was considered. Nontatming Exposures Nanfarming exposures did not confound the association between NHL and agricultural use of herbicides. Nonfarming pesticide use in home gardens and yards was not associated with N1.L Thc OR associated with cver smoking at last I00 cigarcttcs was slightly below*1 (OR,0.7; 95% CI,0.5, l.O), as it was for lifetime consumption of at least 100 c u p of coffec (OR, 0.8; 95% CI.05, 1.4). Consumption of raw. unpasteurized milk products had no effect on NHL risk (OR,11; 95% CI. 0.8, 1.6). Eight patients with NHL had diabetcs, half the expected number a,(OR.0.5; 95% u a 12).Nc Wbw-S hJd q3b&c ? i i , r w ~q??.slnatosus, CeIiac dice=, or immunodeficiency a ~ - f r ~ m e sor had received immune suppressive drugs. Seven patients with NHL reported previous radiation treat- ment (OR,0.q 95% CI. 0.4 22). S u b jects reporting a family history of cancer had a significant risk of NHL (OR. 23; 95% CI, 1.6. 32). Three patients and four controls r e p o d a relative with I p p h o m a (OR, 4.Q 95% CI, 0.7, 222). COMMENT This investigation cadurns the re- sults of the ase-csontrol study by I h r dell et al' that initidly suggested pn association between herbicide useand NHL.Our finding of a sixfold increase of NHL among farmers exposed to herbicides more than 20 days per year is consistent with the sixfold excess risk associated with exposure to either phenoxyacetic acids or chlorophenols in the Swedish study. In both Sweden and Kansas, risk was elewatd among persons e x p o d to phenoxy-tic acids, eg,2&D, not likely to be contaminated by dioxinsaP These findings are mnsistent also with a number of descriptive studies wggcdng increased lymphoma n& among agricultural workers, partieubrly in where herbicide use is aommonL.' Cohort studies of pesticide manufacturing workem and applicators have in general been too mail to examine adequately the risk of NHL Danish workem manufacturing 2 methyl4 chlomphenoxyaetic acid had Seven NHL deaths with 5.4 expected. a nonsignificant 30% ex-a No NHL cases have been observed in five other cohort studies of exposed workerq but the total number of workers involved was only 2'705.== Also. the Vietnam veteran experience offers little evidence to date for an association between NHL and herbicide exposure. The mortality study of 1247 men who applied herbicides in Project Ranch Hand reported no deaths due to lymphomq as of December 19eabut less than one such death was expected in this small cohort An unusual presentation of ?WLof W,5cpl5. 1986- Vd 256. NO. 9 the scalp was reported in b o men of 158 who developed chloracne while employed in a British plant manufacturing pentachlorophenol." The expected number of cases of cutaneous NEIL was far less. A similar associa- ers of the contaminants was not avail- able in the early time p e r i d n The Swedish case-control studies and several of the industrial cohort studies have been criticized on a variety of grounds, including possibIe inaccurate nities for mom than one pesticide supplier per subject The evidence sug- gests that the errors introduced by these factors are likely to be similar for cases and controls. This random mjs-. classification of exposure would tend to tion of cutaneous hWL with occupa- diagnoses, observation andlor d l dilute risk estimates, rather than pro- tions involving spraying phenorsacetic bias, and lack of control for confound- duce spurious associations. While this acid herbicides was found in Sweden.= In our study. only one patient had cutaneous MIL, but he reported no r herbicide exposure. * An association between phenoxyacetic acid herbicides and NHL may have biologic plausibility through the ing variables?' The current study was .designed to address P number of these concerns. The cases were drawn from all incident cases in a defined popula- tion. &duded for analssis wem-only cases histologically confirmed a~ NHL Fy an expert ret?n\' 1. The may have led us to miss an association between pesticide use and STS or HD, it is unlikely to have created the association with MIL If inaccurate exposure assessment occurred, we would have underestimated the magni- tude of the risks involved. If the risks relationships between dioxin mntami- reswnse rate for both cases and mn- reported are accurate. and if they nants and the immune system, as trois was high. The retationship be- reflect a true causal relationship, then postulattd by Hardell et a1.J Diosin tween h312, and herbicide use was the amount of NHL in the current suppresses cell-mediated immunit? specific there was no relationship study attributable t6 herbicide expo- and causes thymic involution in labora- between such use and the two other sure would be 119b.= tory animals," while IWL occurs cancer sites studied, and the associa- Despite the limitations of the e x p e excessively in various immunodeficien- tion was with herbicide and not insecti- sure information. the sixfold excess cy states of hurnans.yDioxin is also a cide use. Indeed, the association was risk of NHL among high-intensity potent animal carcinogen, wifh the risk limited to two subcategories of herbi- users of herbicides is cause for concern, seen for squamous cell carcmomas of cides, phenoxyacetic acids and tria- particularly since the association was the lung, hard palate/nasal turbinates/ zines This degree of specificity argues mainly for phenoxyacetic acids and tongue, hepatocellular carcinomas, fol- against a significant role for either was apparent using several different licular cell thyroid adenomas, and skin observation or recall bias in the as- measures of exposure. Since.over 42 fibrosarcomas.- However, it should ciation. In addition, independent as- million pounds of phenoxyacetic acid be noted that 2&D, the herbicide most sessments of herbicide exposure (as herbicides were applied to US farm- frequently used by subjects in this reported by self or next of kin, and by lands in 1976,' the carcinogenic effects study, has not been shown to be carcin- pesticide supplier) yielded similar esti- suggested by this study and others ogenic in animals"or immunosuppres- mates of risk. Finally, while the origins have important public health implica- sive=,&(S.Wong, PhD, written commu- of N H L in the general population are tions nication. Dec 12, 1985). 24-Dichloro- largely unknown. some known risk phenoxyacetic acid does not contain factors (eg, immune-altering conditions ~3.7,&tetrachlorodibnz~~-p-dioxitnh,e and drugs, family h i ~ t o r y ) ~ ~ ~ 'a's" The study waa supporlcd in p u t by wntracl most carcinogenic dioxin isomer, although other less toxic isomers may The increased risk for well as several speculative factors (eg, cigarette smoking, coffee consumption, ionizing radiation)-' were assessed NOICP-ESll0!#7 from N8tiond h o w Inrbhlte We thank Ariel E Raker and Nlomi E Wuh- burn fa assisting the study rnanrpr; Rita Kingma, Loi.1. Murphy. and Junim W. Pekmun for farmers first using z4-D before 1946 and found not tn confound the herbi- inta-uiwiw V s f a GI .ad Claudia J. OM& for may indicate the presence of carein& genie impurities in the early formula- tions, with subsequent improvements cide d a t i o n . Several arcurnstances offered oppor- tunities for inaccuracies in the assess-' d i n g Shdley Hobta and M e h i e Wall for aaretsrkl ud.t.na; J a a L V.a Niol. Prm Bun#& md Dobrr wkm8nr1fa identifying .ad louting a t d y wbj.ctr; Ritr Lis HD, and H in the manufacturing process Alterna- merit of pesticide exposures. These Clarke Andmn. MD. for -wing patholw tively, the high risk of NHL among persons exposed 40 years ago may reflect a long latency period. The tech- include the memories of subjects, the knowledge of relevant practices by next of kin. the vagueness of some questions Ympla; Edward Bmvn. na D.VCand KO JMI Liu for computer pmglunmine and Richard H A&mwn, PhD. Suun Y. Si&. PhD, uad Eliubeth K Wd8burpcr. PhD, for rwiewing the nology necessary to identify the isom- concerning exposure, and the opportu- manuscript Relerences 1. Hudell .t Sandruom A Case-eontrol study. Soft-tmmue -mu and erpaum ta phenoxy- u i t i c mds or chlomphenols Br J CII- I=, D.7ll-717 2 EnLson M. tluddl ,.I Bmg NO, et ak Soft- tissue s m u and urpmun lo c h c m i d aub -s A u u - r r l e n n t study. Er J J d M d 1981~39. 3. Hudell L Erihoa M.Lnner P. et 11 Malip- nant lymphoma and exposure to e h c r n i d crp- c r d y *MU s d v r n ~c,hlomphenolsand phenoxy .j&A cuEsn-1 swdy. Br J Co- 1W; 4916917L 4. Honehar PA. Halprio WAr ZCST,trichlom phenol and loft-tissue YrCDmL L o w 1981. 1.26B-269. 5. Cook R R Dionn, c h l o r a a ~ .and soft-tissue ssrtbma L a n d 1981;1:618619. 6. M- M. S l i b f f U: Soft-tinuc urcornu. phenoxy hrrbicidm md chlorinalcd phenols I a n . ccf 19Blll370. 7. Johnroo FE.Kugler MA, Emurn SM:Son-tissue u n m m u and chlorinated phenols. L o a d 1981; 240. 8. Cantar W. Parmini and monrlity from nom- Hod+in', lymphoms A -ntml sedy. M J C o r n 1982292SZ47. 9. Burrnairtr 1s.Evcrett CD,Van Lier SF.et & Selrrcd w e e r m o d i t y and fam praetik in I O W ~Am J pid don id iwimm. 10. Buaching DP.Wollrmit L.Gnecr mortality among famnen. J N U 1981;72503. . .11. W.ldronAC.PukELPsfidbOsonAlopr crop IR Lh. cenhd kkgwn~ . 1918. ruearch bulletin 11JL W m t e r . Ohio, Ohio Agri- mlhlral R B u r c h and Drvelopment Center. 1981. 1 2 I d Q r n a t d w i e d . o n s/ Direr fa O n c d q . Cenev- World Heal& Organiution. 1976. 13. Fon-Hodgkin'u Lpnphoma Rtholcgic C l u l i - 1146 JAMA, Sepl 5. 1986-Vd 256. No 9 Herbicide Use-Hoar el ai t &MA. scd 5.1966-Vd 256. No. 9