Multiple sclerosis and exposure to solvents, ionizing radiation and animals.

Earlier studies have indicated an association between multiple sclerosis and environmental factors, especially occupational exposure to solvents. The present study examined such relationships further. From medical files of hospitals in Kalmar and Jönköping, 91 cases of multiple sclerosis, diagnosed in 1983-1988, were identified from population registers corresponding to the catchment areas of the hospitals, and 348 referents were randomly drawn. The cases and referents answered a questionnaire concerning occupational exposure and animal contacts. The men had significantly elevated risks, determined from logistic odds ratios, for solvent exposure, occupational contact with dogs or cats, and leisure-time contact with caged birds. X-ray treatment and previous diseases were risk indicators among the women. For the men and women together, solvent exposure, radiological work, and previous diseases were associated with clearly elevated risks. Although the study concerned rather few subjects, the findings indicate that several exogenous factors might contribute to the development of multiple sclerosis.

It is also well known that the incidence of multiple sclerosis varies with geographic latitude and that individual risk can be altered by changing residence before adolescence, the result being adoption of the inciden ce of the new location (26). Many studies have shown that animal contacts could mean a higher risk (27)(28)(29)(30), but in a later review of the etiologic importance of such contacts the results have been ambiguous, with some of the nonp ositive studies bearin g more weight (3 1). Hence the pathogenesis is likely to be complex -both genetic and environmental factors may be invol ved and interact (32) .
Regard ing exogenou s factor s, exposure to orga nic solvents among Italian shoe and leather workers was associated with an increased risk of multiple sclerosis (33). A previous study from our group (34) showed that occ upational exposure to solvents, especiall y in co mbinatio n with welding, was associated with an increased risk of multiple sclerosis among men. For both genders X-ray examinat ion until five years before diagnosis occurred more frequently than in a reference group. There are also report s of an increased risk among health care worker s (35) and persons workin g in paper manufacturing (36).
General determ inants affecting the occurrence of the disease have been discussed with respect to the great variation in risk with regard to geographi c and climatologic distribution (37). Examples of such determinants are water chemistry (3 8 ) and ultraviolet radiation (39), but also dietary (40,41) and socioeconomic variables have been sugges ted as important (3 1).
Popul ation surveys repeatedly show that mult iple scle rosis appears more often among women than amon g men. According to recent reports the difference can be two-to threefold (42).
The aim of the present study was to determ ine wheth er the results of the earlier Swedi sh study (3 4) could be reprodu ced in another adjacent populati on. Thu s potenti al risk factors like exposure to organic solvent s, welding, radiation , and anim al conta cts were of a priori interest in this study . 39 9 Scan d J Work Environ Health 1993. vol 19. no 6

Cases
In 1989, cases of multiple scleros is were collected from the patient files of the neurolo gical departments of the hospital s in Jonkoping and Kalmar, representing two administrative provinces with a base population of 540000 individu als. There are no other neurological departments in the area , and in the Swedish medical system it is not likely that this type of patient would be diagno sed by private practitioners. Therefore there should have been minimal selection bias in the study . All of the cases identified had been diagno sed in 1983 through 1988. Furthermore, the cases were labeled as definite mult iple sclerosis, fulfilling the criteria by Schum acher et al (43), or as probable or possible multiple scleros is according to the criteria of Rose et al (44).
Altogether 97 persons aged 20-61 years at diagnosis were given case status. Of these, 91 responded to the questionnaire (94 %). The rest refused to participate. Of the responding subjects the diagnosis had been definite for 48, probabl e for 21, and possible for 22.

Referents
The referents were randomly drawn from the population register s of the admin istr ative provinces of Jonkoping and Kalmar (ie, corresponding to the catchment areas of the hospitals). The y were 20 to 65 years of age in 1989 when the selection was performed. Out of 400 selected referents, 348 agreed to be included in the study (ie, the repl y frequenc y was 87%). The deficit was due to refu sal to participate.

Assessment of exposure
Information about various types of expos ure among the cases and referents was obtained by an identical procedure and without revealing the purpose of the study with respect to multipl e sc lerosis. A ten-page questionnaire preceded by an introductory letter was sent to the subjects. The que stionn aire contained 16 main questions, 10 of which focused on occupational exposures. Some of the questions were further subspecified with regard to certain detail s. Five questions were devoted to medical car e, particularly to the use of drug s, X-ray treatment, and X-ray examinations.
A minimum criteri on of one year was required for expo sure time , alon g with a five-year latenc y period (ie, the last fi ve years of exposure were disre garded both for the cases and the referent s). The mean year of diagnosi s was 1986. In the consideration of latency, this year was taken as an anchor point in time for the referents with regard to the various exposures.
Information about solvent expos ure in qualitative terms was directl y obtained from the questionnaires. To avoid overreporting of the exp osure, especially among the cases, all exposure dat a of a prior i inter-400 est were critica lly checked for credibility in telephone interviews with both the cases and the referents. The quantitative classification had five intensity categories formed on the basis of an appraisal applied in some of our other studies, and as originally suggested by Ra vnsko v et al (45). Hence five categories of increasing inten sity were created for solvent exposure, from category 0 = not exposed to category 4 = highe st inten sity (eg, floo r layers). An exa mple of an occupation in category I would be staff at day nurseries, one for category 2 would be cleaner, and for category 3 repairm an. Categorie s 0 and I (ie, not or only very slightly exposed subjects) were considered practicall y unexposed. Categories 2, 3, and 4 were merged into an exposed group since the number of subjects was too small for separate analyses of single exposure categories.

Statisti cal methods
The statistical analyses of the data were based on stratifi cation by age and application of Mantel-Haenszel procedures (46). The approximate confidence intervals of the Mantel -Haensz el rate rat ios (odds ratios) were calculated according to Miettinen (47). Multipl e logistic regre ssion , based on the Epilog Plus statistical pack age (48), was used as an additional method in the data analyses for the simultaneous control of several potenti ally confounding factors. The data for the men and wo men were analyzed separately since female gender appeared to be a determinant per se (34).

Results
The Mantel-Haenszel rate ratios were e levated among the men for occupational solvent exposure in ge neral, and specifically for exposure to kerosene, with its lower 95% confidence limit exceeding unity. Occupational exposure to domestic anim als, like dogs , cat s, horses, swine, and poultry, were significant risk indicators similar to caged birds at home (table I), whereas contact with pet dogs or cats were not (table 2). Among the women occupati onal solvent and radiological exposures appeared as less clear risk factors (table 3) since the confidence interval includ ed unity and the results were similar for animal contacts. X-ra y treatment occu rred for fi ve cases and no referents, giving a formally infinite risk. Various previous di seases taken togeth er were more common among the cases than the referent s. Similar results were obtained by multipl e logistic regression, which resulted in risk estimates with lower 95% confidence limit s exceeding unit y for so lvent exposure, occupational contact with dogs or cats and leisure-time contact with caged birds among the men. Since exposure to farm animal s was compl ex. it was impossible to analyze the separate ef fects from single species ; instead, we had to merge all types of farm animals in the multiple logistic regress ion (ta-  Dividing the material into the subgroups of multiple sclerosis (ie, possible, probable and definite) resulted in small numbers without any distinguishable pattern in the rate ratios when calculable. The three subgroups were therefore merged in all of the analyses.

Methodological aspects
This study had a primary study base since the subjects were recruited from the population of two adjacent administrative provinces (49). Hence the cases of multiple sclerosis were searched for in the patient files of the two, and only existing neurologic departments of these provinces and the referents were randomly drawn from the population registers of the same areas. Therefore it seems unlikely that any selection of cases conditionally on exposure should have occurred, nor is any selection bias likely with regard to the referents.
A five-year time criterion was applied in order to avoid the possibility that the disease itself would cause the cases to have particular exposures (eg, exposure to X-rays).
The proportion of men with multiple sclerosis in our material was low, 26%, which is in agreement with recently reported values (42).
Since female gender is associated with multiple sclerosis, most of the analyses were performed separately for the men and women to avoid confound-a TheSUbjects werestratifiedinto agegroups20-34, 35-49, and50-65 years. ing from gender when occupational exposures associated with gender were considered. In the correlation analyses of various exposures among the referents it turned out that there was an association between some exposures of interest, most strongly between solvents and farming, and a possibility for confounding existed. Therefore multiple logistic regression was performed as well, but it resulted in rela- tively little new information, except for almost eliminating the effect of animal contacts.

Etiologic aspects
Solvents. Occupational solvent exposure in general, and specifically exposure to kerosene, turned out to be associated with increased risk. In our previous study (34) white spirit gave a significantly high risk for multiple sclerosis. These types of solvents are closely related and are both mixtures of mainly aliphatic hydrocarbons, but they also include a few percent each of many different aromatics. Kerosene has a higher flame point (81°C) than white spirit (57°C). It is not unreasonable to assume that there was no clear distinction between these two solvents in the answers to our questionnaire since these solvents are probably rather intermixed in Swedish colloquial language. They are also used to a great extent for the same purposes among mechanics and repair workers.
Cerebrospinal fluid analyses of men with heavy solvent exposure have in some cases shown bloodbrain barrier dysfunction indicated by leakage of protein into the cerebrospinal fluid (50), but the findings are inconsistent (51). Under the hypothesis that multiple sclerosis ultimately has a viral etiology, one could speculate of the possibility of solvents enhancing viral entrance to the nervous system. X rays. This study indicates that exposure to ionizing radiation might have an increased risk for multiple sclerosis, as observed both for patients treated 402 with X rays and for radiological personnel. There were five cases of multiple sclerosis being treated with X rays until five years before diagnosis, but no referents had had such an experience. A previous study (34) showed an increased risk in connection with X-ray examinations until five years before diagnosis. This finding was primarily interpreted as simply reflecting the fact that a number of investigations have to be made at an early stage of the disease when the diagnosis is still unclear. However, together with the earlier findings, the new data obtained in this study support the alternative that radiation per se might contribute to the development of multiple sclerosis, since we now have shown an increased risk for multiple sclerosis also in two other situations (ie, X-ray treatment and occupational exposure). Since both ultraviolet and ionizing radiation can induce viral synthesis in lysogenic bacteria (52,53), one could speculate that radiation might cause primary damage and therefore play an inducing role in a two-step pathogenetic development of multiple sclerosis (ie, if there is a viral or immunologic etiology as well).

Animal contacts.
Our results give no clear support to earlier findings (12,(27)(28)(29)(30)34) that animal contact may be a risk factor for developing multiple sclerosis. In the pathogenetic discussion of multiple sclerosis, several infectious agents have been suggested as directly or indirectly inducing or precipitating the development of multiple sclerosis (6)(7)(8)(9)(10)(11)(12)(13)(14)(15). By univariate analyses a raised risk of multiple sclerosis also appeared in this study among the men occupa-tionally ex posed to animals. The qu esti on arises of why occupati onal contac t wi th dogs and cats inc urs a ri sk, whe reas "pe t contac t" do es not. Howe ver , the formal risk fo r occupa tional con tact with so me animal s (cow, swi ne , hor se , po ultry) as ob served in the uni variate ana lyses appea rs to be confounded by so lve nt use am ong fa rmers (mos tly chain-saw petrol) since this ri sk wa s almost eliminated by th e logistic regression an aly sis. For occupat ional contact with dogs and cats there remained a sig nificant risk in the multiple logi st ic reg ressio n an al ysis whic h ca nnot be ex plai ned by confo unding by so lvents .
Previous diseases. In order to detect possible co nne c tions between multiple scleros is and othe r di seases (eg, infla mmatory or in fectious in c haracte r) the questionn aire co ntained o ne qu esti on co ncerning " any lon g-l asting or serious disease s." In th e statistic al analysi s we focused on the interval fi ve ye ars o r more before the diagnosis of multiple sclerosis. Su ch diseases (eg, diabetes mellitus, allergy , bronc hial asthma, e pile psy , thrombocyt openia) turned o ut to be a sig nificant ris k fact o r for th e wo me n. In a study of th ese di agnoses , ho wever, no di stingu ishable pattern was obs erve d. Th e odds rat io was ne ve rthe less rath er high whe n these var ious di so rders we re g ro uped together, a nd the interpre ta tio n co uld be th at va rio us diseases might pl a y a rol e in prec ipitat ing clini call y ov ert multiple sclerosis.

Concluding remarks
Exposure to so lve nts appears to be associated with a n ex cess risk of developing multipl e scleros is . Both X-ray treatmen t a nd radiological wo rk show a ris k, indi cating th at ex pos ure to ion izin g radiation might be a fa ctor o f inte res t in the pathogenesis of multiple scleros is. In our earlier study , X-ray examinati ons appeared as a ris k indicator, and thi s finding sup po rts the sa me hypothe sis. In analogy wi th othe r investiga tio ns, we hav e dem on st rated a risk from ani mal co ntacts, whic h was restricted to men with occupatio nal contacts wi th cats or dog s, or contact s with caged birds. For livestock, howe ver, the risk appea rs to be confounded by solvent use amo ng farme rs. Befo re firm con c lusions about th e risk s from the expo su res a nal yzed in thi s study can be mad e, furt he r s tudies ha ve to be performed .