Lymphocytopenia and occupational exposures among pattern and model makers

Lymphocytopenia and occupational exposures among pattern and model makers. Scand J 1994;20:107-12. OBJECTIVES - The study was performed to examine the relationship between a high prevalence of lymphocytopenia observed during a cancer screening program for pattern and model makers, who pro duce industrial prototypes and patterns, and 15 occupational exposures. METHODS - The cases (N = 83) were workers with lymphocyte counts of less than 1000, while the referents (N = 529) had counts of 1500 or higher. Exposures at the current workplace, and at up to 19 previous pattern and model shops, were assessed with the use of a questionnaire. RESULTS- Elevated risks were associated with exposure to epoxy resins [odds ratio (OR) 1.94,95% confidence interval (95% CI) 1.02-3.70) and plastic dusts (OR 2.60, 95% CI 1.19-5.68) after ad justment for age and smoking status. No clear associations were found with duration of exposure or percentage of time potentially exposed. Although the results were based upon small numbers, epoxy resins and wood dust displayed the most consistent relationships in an analysis of changes in lym phocyte count and exposure status over time. CONCLUSIOJliS- While firm conclusions cannot be drawn, the strongest associations observed in this study were for exposure to epoxy resins.

< 1500· mrrr-' . In an analysis of a subgro up of particip ant s in the 1988 scree ning, a relationship wa s found between low counts ofT ce lls and natural killer ce lls and employmen t as a pattern or model maker (6) . In addition, a rel ationship be tween low total lymphoc yte, T4, and natural kill er cell counts and the presence of co lon and rectal polyps was ob served. In order to explore whether these low lymphocyte counts were ass oc iated with occupational expo sures, a case-refere nt study among pattern and model makers who particip ated in the screen ing pro gr am was performed.

Subjects and methods
The subj ects were selected from among participants in the 1985 ca ncer sc reening program in Detroit, Mi ch igan, and the surrounding region. All members of the PML ass oc iations representing worke rs in sout heaste rn Michigan were e ligi ble to attend the scree ni ng pr ogram. The participants (all men ) were employed at "job shops," which primarily produce patterns and models fo r automobile manufacturers and other ma ss-production industri es.
Blood counts were performed wi th a sta nda rd instrument ge nerating a thr ee-part differential count, the Coulter "S plus IV ." Ca ses we re defined as persons with lymphocyte counts of < 1000 · mm'. The referents were screening participants with co unts of 1500· mrrr", Persons with intermediate counts were excluded from the case-re ferent an alysis . Becau se of Scund J Work Environ Hell/lh 1994. vo l 20. no 2 the relatively short half-life of lymp hocytes and their ability to regene rate quickly, onl y screening parti cipant s who were curre ntly employe d were includ ed in the analysis (7).
Prior to the screening, all of the particip ants were asked to complete a work history questionnaire designed to ca l1ect a deta iled history of expos ure to 15 different substances comm only en countered in the trade, incl uding two substances identi fied by tradename, Impreg'" and Kirkcite". Impreg" is cativo wood lami nated and impreg nated with a phenol-formaldehyde resin , such that the produ ct is 30% resin, while Kirkcite" is a metal alloy composed primarily of zinc (90%) with smal1er amoun ts of nickel, titanium , copper, and iron . Pattern and model making is frequently a life-long profession, and it is common for workers at "job shops" to change employers as freq uently as every one or two year s, either becau se of downturns in the industry resulting in layoffs or because of an advance in pay sca le. Thus the workers were queried as to the percentage of time they worked with eac h of the substances at their current shop and up to the 19 pattern and model making shops at which they had been previously employed. Screening participants who did not fil1 out a work history questionnaire were excl uded from the analysis.
Analyses were cond ucted using logistic regression to compare the history of exposure amo ng the cases and referents. The mean age of the cases was 45 years and 20% were current smokers, while the mean age of the referents was 39 years and 43 % were current smokers. Smokin g was negat ively associa ted with lymph ocytopenia [odds ratio (OR) 0.34, 95% confidence interval (95% CI) 0.19-0.60], while age was somew hat positively assoc ia ted (O R 2.29 for 30-39 years, OR 1.90 for 40-49 years, and OR 3.83 for~50 years relative to those < 30 years of age). Thu s a dec isio n was made to adj ust for age «30, 30-39, 40-49,~50) and curre nt smoking statu s in all of the analyses. 108 An alternate analysis was also performe d to examine whether changes in expos ure status over time led to a drop in lymph ocyte counts. For this analysis , the subjec ts co nsis ted of al1 PML screening participants who atten ded both the 1985 and 1981 -1982 scre ening programs and wer e currentl y employed at the time of the 1985 scre ening. Our a priori hypoth esis was that tho se workers not exposed to the potenti al ca usal age nt or age nts at either screening or during the time between should have little change in their lymph ocyte co unts, while those exposed at the first screening, but who cea sed to be exposed prior to the second, should have an increase in their lymphocyte co unts. On the other hand , decreases in lymphocyte cou nt might be expecte d among those workers who began exposure between the two screenings or who were exposed at both.

Results
A tota l of 864 men attended the 1985 scree ning and 858 had total lymph ocyte cou nts performed. Of the 858, 86 (10%) had counts of :5 1000 . mm'. 550 (64% ) had counts of~1500 · rnm', and 222 (26%) had intermediate co unts. Thu s there were 86 pote ntial "cases" and 550 potential " refe rents." Two cases and 10 referents were not currently employed at the time of the scre ening, and one case and I I refere nts did not provide an adequate work history for the analysis. After exclusions, 83 cases and 529 refere nts were avai lable for the analysis.
No association was found with number of year s emp loyed as a pattern or model maker (O R 1.00, 95% Cl 0.96-1.03) after adj ustment for age and smoki ng status. As part of the occupational history, screening participant s were queried as to the main material with which they worked. No assoc iation was found for primarily work ing with woods (O R 0.97, 95% CI 0.59-1.59), metals (OR 0.97, 95% CI 0.54-1.73), or plastics (OR 1.07, 95% CI 0.6 1-1.86). Table I lists the 15 specific exposures queried in the questionnaire with the percentage of cases and referen ts curre ntly exposed and ever exposed. Table 2 lists the odds ratios and 95% confidence inte rvals associated with the 15 expos ures. The first co lumn presents the odd s ratios for exposure at the time of the screen ing with the reference category consis ting of those not currently expose d. The second column of table 2 presen ts the res ults for currently exposed versus never exposed. Th e risks associate d with expos ure to epoxies and plastic dusts were the most high ly elevated. The risks associated with exposure to soft woods , metals other than Kirkcite"), and fiber glass appeared to be somew hat elevated; however, the co nfidence limits were wide and incl uded one. The second co lumn of table 2 yie lds results which are almost identical to those obtained using an eve r versus never catego riza tio n of exposure. For example, the odds ratio associated with ever havin g worked with epoxy resins was 1.95 (95% CI 1.03-3.71) , while the corresponding odds ratio for plastic dusts was 2.8 1 (95% CI 1.29-6.12).
In table 3 the data from column 2 of table 2 are stratified by the percentage of time spent working with the material. No exposure appeared to have a clear association with this surroga te measure of dose that co uld not be ruled out as due to chance . Asbestos was excluded from the table because no cases or referents reported having used it more than 50% of the time. The odds ratio s associated with exposure to both epoxy resins and plastic dusts were elevated for both those who had reported using the substa nces for less than 50% of the time and for those who reported using the substances more than 50% of the time . Table 4 lists the odd s ratios by category of duration of exposure relative to no history of exposure. As in table 3, no expos ure appeared to have a clear assoc iation with this additional surrogate measu re of dose.
Scand J Work Environ Health 1994 . vo l 20 . no 2 A total of 405 particip ants in the 1985 screenin g also attended the 1981-1 982 cancer screening. This gro up consisted of 48 cases , 253 referents, and 104 men with lymph ocyte counts between 100 and 1499 . mm-' in 1985. The mean lymph ocyte count at the earlier screenin g for persons defin ed as cases in the curre nt study was 1677 . mm", while their mean for the 1985 screeni ng was 798 . mm'. The mean value for the intermediate group also dropp ed between the two screenings, from 1660 to 1286 . mm', while the mean value for the referents rose slightly during the same time period (from 2098 to 224 1 . mrn'). Only exposure to soft wood and epoxy resins (ta ble 5) displayed relationships with lymphocyte coun ts in the directi ons hypoth esized a priori. A small num ber of participant s, four in the case of both the so ft wood and epoxy resin analyses, who had multiple changes in exposure status were excluded from the expos ure-s pecific analyses . The analyses of all other exposures displayed either Table 2. Odds rat ios (OR) and 95 % confiden ce in terval s (95 % Table 3. Odds ratios (OR) and 95 % confide nce int ervals (95 % CI) for lym phoc yt o penia amo ng the pattern and model makers by ex posure at th e ti me of screening.
CI) f o r lymphocytopenia among the pattern and model ma kers by exposure at the t ime of screen ing and the percentage Currently exposed Currently exposed of time ex posed .
versus not versus never Exposure currently exposed exposed Less than 50%   a dec rease in lymph ocyte cou nt after the cessa tion of expos ure or an increase after the beginni ng of exposure . Plastic dusts are included in table 5 as an example. Those who ceased expo sure to plastic dusts had a mea n decrease of 442 . mrrr' in thei r lymphocyte cou nt, whereas an increase in lymphocyte co unt would have been predicted . The differences between those who had been exposed at the time of both scree nings and those who had not been exposed at either reached statistica l significance (P<0.05) for both epoxy resins and plastic dusts. The difference between those who had been exposed to plastic dusts at the time of the first screening but had cease d exposure by the time of the second screening and those who had not been exposed at either was also statistically significant. None of the other differe nces were statistically significant; however, many of the compariso ns were based upon small numbers.

Discussion
The most consistent association wit h lymphocytopenia observed in this study was with expos ure to epoxy resins. Lymphocytopenia was assoc iated with both a current and past use of epoxy resins , and decreases and increases of lymphocyte counts corre lated in the predicted direction with sequential exposure versus nonexposure to these substances . The fact that expos ure at the time of both screenings was associated with a larger decre ase in lymphocyte count than expos ure only at the time of the latter scree ning would suggest that, if a true relationship exists, chronic exposu re may be important. However, no consistent associations were found with duration of exposu re or percentage of time spent working with the material in the current job, and the analyses of changes over time were sometimes based upon small numbers.

110
The resu lts for other exposures co nsidered in this analysi s were less consistent. Although curre nt exposure to plastic dusts was associated with lymphocytopenia, the results of the sequential analysis were not in the predicte d direction. Likewi se, the sequentia l association with softwood was not substantiated by a clear association with curren t expo sure. In addition, as was the case with epoxy resins , neither expo sure showed evidence of a dose-response relatio nship with lymph ocytopenia.
Many substances encou ntered in the work environment have been found to be immunotoxic (8) . However, relative ly few studies have been conducted to exami ne the relat ions hip between wor k-related exposure s and lymphocytop enia. Stud ies have found a relationship between expo sure to asbestos and decreased T helper:suppressor ratios and lower numbers of natura l killer cells (9)(10)(11). Decreased numbers of lymphocytes have also been associated with occu pational exposure to solvents, particularly benzene ( 12,13). In this analysis neither exposure to asbestos nor expos ure to solve nts was related to lymphocytopenia. The epoxy compounds typically used in pattern and model making are epich lorohydrin and diglycidyl resorcinol ether, along with hardeners such as diethy lenetriamine and triethylenetetramine ( I). Exposure to epoxy resins has not been previous ly associated with lymphocytopenia, althoug h other lymphatic and hematopoietic effect s have been observed. An increa se in chromosome aberrat ions in the white blood cells of epoxy-exposed workers has been noted (14) , and anima l studies have shown glycidy l ether to cause a loss of lymphoid tissue in rats ( 15).
Although this study was undertaken to explore the relation ship between depressed human lymphocyte co unts and workplace expos ure among pattern and model makers, it was also motivated by concern about the risk of cancer among members of this trade. An exce ss of co lon and rectal cancer has been observed among pattern and model makers in studies of both mortality (2,3,16) and cancer (3,4,17,18) incidence , and an excess of colon and rectal polyps has also bee n observed (19). Although two rec ent studies have not observed an excess (20, 2 1), the majority of the evidence indicates that these workers are or were at excess risk of colon and rectal cancer. The cause for the excess risk of cancer observed among these workers has not been identified. Initial suspicion had falle n on wood dust becau se it is a commo n exposure and an exce ss of nasal eancer had been observed in the wood furniture industry (22) . However, the one study designed to exami ne this question did not find an exces s of co lon cancer associated with wood dust expos ure (2 1).
The observation of a high prevalence of lymphocytope nia in this populatio n has prompted the consideration that a toxic effect on the immu ne system, as reflected in depressed lymphocyte counts, could predispose the development of colon and rectal cancer. In a previous study of this population, a relatio nship bet ween low total lymphoc yte, T4 , and natu ral kille r cell counts and the presence of co lon and rectal polyp s was ob served (6). Another study, ba sed on patien t reco rd s fro m the Ma yo Clinic , found that, of 51 0 healthy pati en ts, 6 % had lymphoc yt e counts of < 1000 . mrrr ' , while , of 22 7 patients w ith untreated gas tro intes ti nal malignanci es, 22.4% had co unts of < 1000 . mm" (22 ). Thus it may be significa nt to the findings of co lon and rectal neoplasia among pattern and model ma kers that epic hlorohy drin has be en classified as a probable human carcinogen while dig-Iycidyl re sor cinol ethe r ha s been cl assified as a possible human ca rc inogen (2 3) .
There are some limitations that should be kept in mind when the re sults of th is study are considered. Firs t, the study population co ns is ted of vo luntary part icipants in a ca ncer sc ree ning program. Approxim atel y 35 % of the e lig ible workers participated in th e 1985 screening, while 4 6% pa rti cipate d in th e 1981-1982 screening (24 ). The participa nts in th e pro gram have been found to be more likel y to be older, to ha ve never smoked, and to have worked wit h wood . How th is hi stor y mi gh t influence the res ults o f the study is unknown . Howe ver , it is impo rta nt to note that this was a generally healthy working population and it seems unlikely that individual decisions to partic ipate had any rel ati on ship wi th lymphocyte co unts . Second , only tot al lymphocyt e co unt was avail able for all of the parti cip ants. Subset co unts, as we ll as other markers of immune impa irme nt, wo uld have been opti mal but, given the ex pense, were not collected as part of the screening .
Differen t types of ep oxy resin s, wo ods , and plastic du sts may have very different toxic effects. Another limitation of thi s study was th at mor e specif ic in form ati on regard ing exp osure was not availa ble . Th e drop in lymphocyte counts between the two screenings among the person s who subseque ntl y became cases was substantia lly greater th an that observed in rel ationsh ip to an y of th e ex pos ures examined . This res ult might indi cate th at the true agent resp on sibl e for the cha nge in lymphocyte co unts wa s eith er not measured or poorly measu red. Th e lack of a do se-response rel ationship in th is study may have be en due to the fact that both the percentage of tim e and the du ration were po or sur rogates for tru e do se.
In co nclus ion, epoxy resin s formed the ex pos ure most co ns iste nt wi th lymphocyt op enia amo ng the pattern and model makers, while weaker evidenc e was fo und for exp osure to plastic dusts and so ftwood dust. Because th is effec t has been observe d in a gro up of wor kers with an elevate d risk of cancer, future in vesti gations will focu s on assessing the relat ion ship be tween the se ex posures and th e ris k of col on and rectal cancer.