Immunologic and renal markers among photogravure printers exposed to toluene

and renal markers among photogravure Objectives This study assessed immunologic and early renal effects of chronic toluene exposure. Methods In a longitudinal study of 92 printers and 74 referents, 145 subjects had pre- and poststudy samples of blood and urine taken for the following measurements: immunoglobulin E (IgE), antiglomerular basement membrane (anti-GBM) and antilaminin (anti-LAM) antibodies in blood; creatinine and R2-microglobulin in blood and urine; and micsoalbumin, N-acetyl-b-D-glucosaminidase (NAG) and alanine-aminopeptidase in urine. Cseati-nine clearance was calculated according to-the Cockroft-Gault formula. Eight-hour personal air samples were collected twice to assess present exposure to toluene. A job-exposure matrix was developed to estimate past cumulative exposure. 1nfoAation about potential confounders was recorded by questionnaire. Multiple regression analysis was performed to study dose-effect relations adjusted for age and smoking. Results No subject was positive for anti-GBM antibodies, and only 12 were positive for anti-LAM. No relation was observed between the markers studied and present exposure to toluene except that creatinine clearance was higher among the exposed subjects than among the referents. A dose-response relation was observed between cumulative toluene exposure and both IgE and NAG excretion. No interaction was observed between hypertension and exposure, but the relationship with NAG did not persist when subjects with hypertension were excluded. Past or present exposure did not alter the 2-year trend of any marker studied. C O ~ C ~ U S ~ O ~ S According to the results of this study, toluene at 50 ppm is not related to detectable renal dysfunction. The increased IgE levels associated with present and past exposure require further investigation.

a1 (8) was negative for printers, but their results were not conclusive because the study lacked power. Furthermore, mortality is a poor indicator of the incidence of renal disease now that renal replacement therapy exists. Crosssectional studies of various early markers of renal damage have been performed among populations exposed to toluene. Most of these studies concerned painters exposed to several solvents, but others involved printers whose solvent exposure was limited almost exclusively to toluene (table 1). The studies of painters have yielded contradictory results, glomemlar and tubular renal alterations being found in 3 studies (9-1 I), tubular alterations only in 2 others (12,13), and no effect in the final 2 (14,15). Different exposure levels did not explain the variations. The study by Stevenson et a1 (16) also showed an increase in anti-glomerular basement membrane (anti-GBM) and anti-laminin (anti-LAM) antibodies among exposed painters. The 2 studies in the printing industry were negative, but both concerned small samples (14,17).
The objective of this study was to evaluate the renal effects of chronic toluene exposure among printers whose solvent exposure was limited to toluene and who were highly exposed to it. Several immunologic and renal markers were studied twice over a 2-year period to assess dose-effect relations with cumulative or present exposure, as well as trends.

Subjects and methods
Population A longitudinal survey was carried out in a photogravure plant from January 1991 through May 1993. It included 166 male workers, 92 exposed to toluene and 74 referents. Those working with rotary printers (operators, handlers, winders), as well as retouchers and mechanics made up the exposed group. The reference group, stratified for age, comprised workers in the following 2 subgroups: (i) an internal reference group of subjects from the same plant (N=43), recruited from workers who were not directly exposed to toluene but were subjected to background exposure (engravers, fork-lift operators, compositors, and other unexposed personnel) and (ii) an external reference group (N=31) of unexposed workers from a binding plant that employed the same occupational physician as the printing plant.
All the subjects worked (8-hour) rotating shifts, spending successive weeks on day, evening, and night shifts. Those with diabetes or any renal disease (unrelated to exposure) were excluded. The study protocol included 2 examinations, the 1st at study entry and the 2nd 2 years later. Both examinations took place during a week that the subjects were working the day shift; each time, occupational nurses took blood and urine samples and measured the subjects' blood pressure. The subjects also Table 1. Cross-sectional studies of renal markers among toluene-exposed workers. (U-p2M and S-P2M = urinary and serum 02 microglobulin, respectively; pglu = pglucuronidase; NAG = N-acetyl-b-glucuronidase; RBP = retinol binding protein; yGT = gammaglutamyl-transfkrase; LAP = leucine-amino-peptidase; anti-GBM = anti-glomerular basement membrane antibodies; MEK = methyl ethyl ketone; TLV = threshold limit value; TWA = time-weighted average; NS = indicates no significant difference between the two groups)

Study
Population Exposure Resultsa Askergren et al, 1981 (14) Photogravure printers (42 exposed, 48 referents) Toluene (80 to 100 ppm) Albuminuria (NS), U-PZM (NS) Krusell et al, 1985 (17) Photogravure printers (39 exposed, 36 referents) Toluene (50% TLV) Albuminuria (NS), U-P2M (NS) Askergren et al, 1981 (14) Paint factory employees (40 exposed, 48 referents) Toluene, xylene (< 40 ppm) Albuminuria (NS), U-P2M (NS) Franchini et al, 1983 (12) Painters (1 18 exposed, 81 referents) Toluene, xylene (25 ppm) Albuminuria (NS), lysozymuria t, Lauwerys et al, 1985 (15) Car painters (43 exposed, 43 referents) completed a self-administered questionnaire designed to provide information about their personal characteristics, tobacco and alcohol use, and overall health, specifically including ltnown hypertension. Information about medication taken "at least once a day for at least a week during the past three months" was recorded with the aid of a list of potentially nephrotoxic drugs such as analgesics, nonsteroidal anti-inflammatory drugs, some antibiotics, and some antihypertensive and antiepileptic medications. The subjects were 21 to 55 years of age. The exposed and reference groups were similar in mean age, body mass index, blood pressure, percentage with hypertension, percentage with asthma, alcohol use, and total length of employment (tables 2 and 3). The reference group contained proportionally more ex-smokers or heavy smokers and fewer subjects who stated that they had allergic rhinitis than did the exposed group. Fortyfive percent of the subjects had taken at least 1 drug regularly within the previous 2 months, and 27% had taken  at least 1 analgesic (glafenine, paracetamol, aspirin, or an nonsteroidal anti-inflammatory drug); these percentages did not differ between the 2 groups. During the protocol development, 1 year before the study started, the company hired 18 subjects who had never been occupationally exposed to solvents. Each provided a urine sample as part of their preemployment medical examination, and 2 further urine samples as part of the study, and all 3 samples were considered in this analysis.
immunologic and renal markers Samples.Ten milliliters of blood was sampled for the assays of creatinine, 02-microglobulin, immunoglobulin E (IgE), and anti-GBM and anti-LAM antibodies. A spot urine sample (first of the morning) was collected to assay creatinine, microalbumin, 02-microglobulin, Nacetyl-0-D-glucosaminidase (NAG), and alanine-aminopeptidase (AAP). To limit physiological sources of variation, standardized methods were used for collecting the urine and pelforming the assays. For both sets of tests, sampling took place weekly over a 2-month period and involved, on each occasion, the same proportion of exposed and reference subjects. All biochemical measurements, including those for the 18 subjects examined when hired, were performed in the same laboratory, within 8 hours of sample collection. To alkalinize the urine and thus prevent degradation of the l32-microglobulin, the subjects drank a 33-cl bottle of alkaline water (Vichy Saint-Yorre) the evening before the testing.
Assays. The Cobas Bio centrifugal autoanalyzer (Roche, Neuilly, France) was used for measuring creatinine, NAG, and AAP, and a Laser BN 100 nephelometer (Behring, Rueil Malmaison, France) measured microalbumin. The assay methods included the kinetic Jaff6 reaction for creatinine, immunoprecipitation for microalbumin, enzyme immunoassay (MEIA) for 02-microglobulin (IMx 02-microglobulin test, Abbott-France, Rungis), hydrolysis of 3-cresolsulfonphtaleinyl-N-acetyl-b-D-glucosaminide (Boehringer France, Meylan) for NAG, and hydrolysis of L-alanine-4-nitroanilide (Merck, Paris, France) for AAP. Individual differences in urine dilutions were taken into account by relating the data to creatinine concentration. Creatinine clearance was calculated with the Cockcroft-Gault formula (18). Relative 02-microglobulin clearance, defined as the ratio of 02-microglobulin clearance to creatinine clearance, was also calculated. According to Jarup et a1 (19), values of 0.1-2.5% indicate slight tubular dysfunction, and those above 2.5% reveal severe dysfunction. The anti-GBM and anti-LAM antibodies were assessed with an enzyme-linked immunosorbent assay adapted from an assay described previously in a rat model (20). Collagenase-digested human GBM and Englebreth-Holm-Swarm (EHS) mouse sarcoma laminin (Sigma, St Louis, MO, USA) were used as antigens. Sera from 300 healthy blood donors served as reference. Serum samples from the study subjects were tested at dilutions of 1/10, 11100 and 111000. Abnormal values for each antigen were defined as equal to or greater than the blood donor mean + 3SD. As positive referents, we used, for the anti-GBM assay, serum from a patient with Goodpasture's syndrome and, for the anti-LAM assay, serum from a rat with autoimmune glomerulonephritis induced by exposure to mercury chloride (20). IgE was assayed with a Biotrol Kit (Biotrol, Paris).
The blood and urine analyses were performed twice, at the beginning and end of the study, for 145 subjects (82 exposed, 63 referents). For technical reasons, 02-microglobulin measures were not available at the time of the 2nd examination. Anti-GBM and anti-LAM antibodies were assessed only once.
Toluene exposure Current exposure assessment. Personal air sampling in the breathing zone was performed in March 1991 and June 1992, for 1 workweek each time. It covered all jobs, on all shifts, for all subjects who worked at the printing plant, whether directly or indirectly exposed. Air samples were collected on charcoal tubes (SKC ref 226-01) with constant-flow (50 mllmn) portable pumps over an 8-hour shift. After desorption with carbon disulfide, the toluene content was analyzed by gas chromatography with flame ionization detection. Overall, 85 of the 92 exposed subjects had at least 1 air sample taken, and 47 had 2 samples taken. Of the 43 internal referents at the printing plant, 26 had at least 1 measurement, and 19 had 2 measurements so that the background level of toluene exposure could be assessed. Since no change in work conditions occurred between the 2 periods, these measurements were considered to reflect the mean exposure levels during the study. The current toluene exposure level was defined as the mean of the measurements for subjects with 2 samples and as the value measured for those who had only 1 sample. The 7 exposed subjects and 17 referents without air sampling were attributed the mean value of all subjects with measurements in the same workplace. The exposure level of the external referents was assumed to be zero.
The pressmen were, on the average, exposed to a toluene level half that of the French time-weighted average occupational exposure limit [ie, 375 mglm"100 ppm)], but there were substantial variations from this mean depending on the machine to which they were assigned (table 4). Exposure levels for the other exposed jobs were approximately a 3rd of the occupational exposure limit. For the referents within the company, background exposure was highest for the forklift operators, about onetenth of the occupational exposure limit, and lower for the other referents.
Assessment of past exposure to toluene and to total hydrocarbons. The company's records allowed us to establish the history of their printing techniques. Letterpress printing ended in 1970, typesetting in 1979, and off-set printing in 1980. Photogravure techniques, especially the composition of the ink-diluting solvents, also changed. Benzene disappeared from the formulation in 1952. Until 1968, the diluting solvent was a mixture of spirits composed of aromatic solvents (naphtha petroleum distillates) containing toluene and xylene. Starting in 1968, the toluene content of ink solvents increased, reaching 98% in 1982. The company had also used several generations of presses. The following 2 dates mark important changes in the work conditions of the plant: (i) 1974: systems for solvent recovery and air exchange began to be installed on some machines and (ii) 1982: giant printing presses went into operation; some were enclosed and located in a small building, others were not enclosed and were located in large workshops.
Since 1960, regular area samples of both toluene and total hydrocarbon concentrations have been collected at a fixed point near each press twice a year. Sampling lasts an average of 3 hours. The samples are collected with charcoal tubes and analyzed with a gas chromatograph, a method that is very similar to that used for personal sampling in this study. These measurements allowed us to calculate the annual mean exposure to toluene and total hydrocarbons among the printers who worked on different generations of machines and thus to develop a jobexposure matrix. The detailed work history of each subject within the plant was recorded on a self-administered questionnaire. Combining the work histories and the matrix allowed us to assign each printer the mean of the measurements for his machine for a given year. No past area measurements were available for jobs that were not associated with a specific machine. Thus there were no measurements for mechanics and engravers, for example, whether or not they were exposed. To estimate past exposure for these jobs, we used the data from the 2 sets of personal measurements taken in 1991 and 1992 and calculated the relation between these exposure levels and that of the printers for each job category. On the assumption that this ratio remained stable over time, we then assessed past exposure for each category, using the ratio with the mean of the measurements available for each year for the printers overall. A cumulative exposure index (mg/m3 x years) was calculated for all the printing plant subjects by summing these estimated annual exposures. Exposure to toluene and total hydrocarbons decreased considerably between 1960-1973 and 1974-1982 (table 5). Since 1983, toluene has been the main solvent in inks, and levels of exposure have decreased more slowly, but regularly, through 1991. In 1991, the toluene exposure level assessed by regular area sampling was lower than that observed with personal air sampling in this study.

Statistical methods
The serum creatinine and 132-microglobulin and creatinine clearance were normally distributed. The other variables had skewed distributions and were log- Table 5. Past toluene and total hydrocarbon exposure level transformed. An analysis of variance and multiple regression were used to compare the mean values of the markers at the beginning of the study and their trend over 2 years for the exposed and reference subjects. For the normally distributed variables, the trend between the 2 measurements was evaluated by the difference between the 2 values; for the other variables, the ratio of the 2 corresponding median values was used. A matched analysis of variance was used to test the overall trend of the markers between the 2 periods and to compare the values of the markers before and after employment began for the 18 subjects who provided samples when hired.
Multiple regression was used to study the dose-effect relations between the markers and current toluene exposure, length of employment in a given exposed job, and cumulative exposure to toluene and total hydrocarbons. The exposure units used were (i) 5-year periods for length of employment in an exposed job, (ii) 37.5 mg/m3 (ie, 10 ppm) for cussent exposure to toluene, (iii) 375 mg/m3 x years (that is, 100 ppm x years) for cumulative exposure to toluene, and (iv) 375 mg/m3 x years for cumulative exposure to total hydrocarbons.
The relations between the markers and the exposure indicators were adjusted for age (quantitative) and for smoking (in the 4 categories of nonsmokers, ex-smokers, <20 cigarettestday, 220 cigarettestday). They were also adjusted for body mass index (quantitative) when serum creatinine was analyzed.
Four referents had previously held jobs considered to be exposed in the plant. Variance analysis was performed both with and without these subjects. In the multiple regression analysis, all the referents were included. Duration and cumulative exposure were set at 0 for all but the 4 subjects with past exposure, which was thus taken into account in the study of dose-effect relations.
To test whether an interaction existed between hypertension and exposure, the analysis was repeated with an interaction term between hypertension and the   (17). "nly for subjects with a urinary pH of 26 (ie, 62 exposed subjects and 55 referents). * P = 0.02, comparison between the exposed subjects and the referents. various exposure indicators. Furthermore, a 3rd analysis excluded subjects with hypertension, defined as those who stated that they were taking antihypertensive medication at the beginning or end of the study and those whose systolic or diastolic blood pressure, as taken by the nurses, exceeded normal limits [systolic >I60 mm Hg (>21.28 kPa) or diastolic >90 mm Hg (>11.97 kPa)] in both examinations.

Comparison of the marker values between the exposed and reference groups
The mean marker values were similar in both reference groups, with the exception of serum 132-microglobulin, which was higher for the external than internal referents [I264 (SD 307) versus 1130 (SD 175)l. No subject was positive for anti-GBM antibodies, and only 12 for anti-LAM (6 exposed and 6 referents). The crude mean values did not differ significantly between the exposed and reference subjects for any of the renal or immunologic markers (table 6). Creatinine clearance was significantly higher among the exposed subjects than among the referents. No subject had a relative 132-microglobulin clearance above 2.5%, and 10 had a value between 0.1 and 2.5% (4 exposed and 6 referents). Neither adjustment for age and smoking nor exclusion of the 4 previously exposed referents modified any of these relations. Limiting the analysis to urine samples with a pH above 6 increased the mean value of urinary 132-microglobulin, but it did not alter its relation to exposure.
Only serum creatinine and 132-microglobulin had increased significantly after 2 years of follow-up, but these increases were similar in both groups. Although 33 printers had once worked in off-set and letterpress printing, their markers did not differ from those of the other subjects.

Relation between the quantitative exposure measures and the markers
Multiple regression of the immunologic and renal markers with each of the 4 quantitative exposure measures only showed a slight increase in NAG and IgE with the cumulative toluene exposure estimate (+2% per 100 ppm x years for the former and +3% for the latter). No significant interaction was observed between hypertension and the exposure indicators, but the weak relation for NAG vanished when the 17 subjects with hypertension were excluded. The trend of the markers over the 2-year period did not vary by length of employment or cul-sent or past toluene exposure.

Trend of the renal markers before and after hiring
The 18 printers whose urine was tested before they began working at the plant were, on the average, 25 years of age. No significant change in their urinary markers was observed after 1 or 3 years of exposure (table 7).

Discussion
This study did not find any renal dysfunction related to chronic toluene exposure at the levels encountered in this printing plant. No dose-effect relation was observed with the markers of either tubular or glomerular damage or with those of renal function, nor did the trend of these markers differ over 2 years according to the subjects' exposure. Only the concentration of IgE increased with toluene exposure. In a subgroup of newly hired printers, no significant change in the renal markers was noted after 3 years of exposure.
Questions can be raised about whether selection bias and confounding, as well as study power, might have affected the internal validity of the study. The possibility of selection bias must always be considered in a cross-sectional study that yields negative results.  Nonetheless, although the risk of bias is indeed high in the study of a clinical disorder, it is unlikely in a study of subclinical events that have not been monitored at all in the company, as was the case for the renal markers studied here. The lack of any discernible change among the newly hired printers, even though there were few, is another argument against the existence of such bias.
Recruiting half the referents within the company might also have resulted in an underestimation of the difference between the exposed and reference groups, because of the background exposure of the latter. However, the 2 reference groups did not differ except for their concentration of serum 132-microglobulin, and it was higher among the external referents than among the internal group. Thus a wholly external reference group would not have altered the conclusions about this marker. In addition, the exposure of the internal referents contrasted markedly with that of the printers and was only one-fifth as high. This choice of referents did not affect the investigation of the dose-effect relation either, since, overall, the study subjects present a wide range of exposure. The 1991 regular area measurements showed lower toluene exposure levels than did the personal air samples measured simultaneously as part of this study. This finding suggests that the area measurements may have been underestimating subjects' exposure, which in turn may have affected cumulative exposure estimates and thus resulted in an underestimation of its effect.
Particular attention was paid to the comparability of the groups in this study, by choosing the referents and sampling methods with special care, as well as by reducing the physiological factors of marker variation. Risk of bias related to socioeconomic status cannot be suspected because the referents were recruited among workers in the same occupational activity. All the subjects were male, and the groups were comparable for age. The referents, like the exposed subjects, worked rotating shifts and were all tested during a week on the day shift; thus any differences due to circadian effects on the urinary markers were avoided (21,22). The subjects were examined under identical conditions, and the samples were not stored. Both of these factors limited any preanalytic variations between the subjects. Subjects with hypertension were not systematically excluded, so that we could study the possibility of an interaction between hypertension and exposure, as suggested by Hotz et a1 (23). No significant interaction was observed, but, when the subjects with hypertension were excluded, the relation between cumulative toluene exposure and urinary NAG, which was at the borderline of statistical significance, disappeared. Finally, among the external factors of marker variation, as reported elsewhere (24), only smoking differed between the 2 groups, and it was systematically taken into account in the study of exposure-marker relations.
This study is, as far as we know, the first to use renal markers to examine past and present exposure to toluene and other hydrocarbons in such detail for this occupational activity. It was able to show a difference of 5 mmolll in semm creatinine between the exposed and reference subjects, with a risk of error of 5% and a power of 80%, and a difference, respectively, of 107 mgll and 7 mllmin for serum B2-microglobulin and creatinine clearance. For the log-transformed parameters, it could have shown a 39% difference between the exposed and reference subjects in the mean level of microalbumin, 24% for the 132-microglobulin and NAG levels and 20% for AAP.
Other questions may be raised about the renal markers studied and their sensitivity. We chose to investigate 1 early marker of glomerular damage (microalbumin) and 3 early markers of damage to the proximal tubule (B2microglobulin and 2 enzymes, NAG and AAP). The sensitivity of these 4 markers is unknown from an epidemiologic point of view. Nonetheless, studies over the past 20 years in various populations exposed to solvents or metals have shown that microalbumin, NAG, and AAP are capable of detecting groups at risk (25). For a long time, 02-microglobulin was the only low-molecularweight protein used as a marker of tubular damage, but it is now often replaced by retinol-binding protein or a-1-microglobulin, more stable in acid pH (26). However, the efficacy of B2-microglobulin as a marker was recently proved again in the demonstration of a dose-effect relation with cadmium exposure (27). In our study, despite absorption of sodium bicarbonates, approximately a 3rd of the subjects in both groups had a pH of <6. Their exclusion from the analysis increased the mean of this marker but did not modify its relation to exposure. We also assessed the relative 132-microglobulin clearance, which is considered a sensitive marker of tubular dysfunction by some authors (19). It was not related to exposure in this study.
Our results for the biochemical markers are comparable only with those in the studies by Askergren et a1 (14) and Krusell et a1 (17), who measured urinary excretion of albumin and 82-microglobulin among printers whose characteristics were similar to those of the population we studied. The power of these 2 negative studies is nonetheless questionable. In other studies (table I), the subjects were painters exposed to various solvents and sometimes also to lead and cadmium (9). In the studies that observed a significant increase in the urinary excretion of some of the renal markers, the effect of toluene could not be isolated.
In view of the lesions that have been described for acute toluene toxicity, which differs from those due to other hydrocarbons, a study of a marker of damage to the distal tubule would have been interesting and might have improved our understanding of the mechanism of renal damage. There are, however, very few such markers, and they are not yet well understood in an epidemiologic context. Only Normand et a1 (9) have so far studied the urinary excretion of Tamm-Horsfall glycoprotein. They did not observe any change in this variable in relation to the painters' exposure to toluene.
The roles of organic solvents in glomerular nephropathies have been the subject of many studies over the past 20 years. Forty cases of rapidly progressive glome~ulonephritis (Goodpasture's syndrome) and of chronic nephropathies have been reported and associated with exposure to various solvents (28); 2 of these cases involved toluene (6, 7). Fifteen case-referent studies have been performed during that time, 12 of which showed an association between chronic glomerular nephropathies and solvent exposure (1). An explanation of the mechanism by which solvents act has yet to be furnished. The most recent proposals suggest that solvents may play a role in the progression of a nephropathy towards chronic renal failure (29,30).
This study also analyzed 3 immunologic markers, anti-GBM antibodies, which are specific to the glomerular damage of Goodpasture's syndrome, anti-LAM antibodies, which have been found in autoimmune glomerulonephritis induced by exposure to mercury chloride among rats, and total IgE, which indicates type I hypersensitivity. Unlike Stevenson et a1 (16), we did not observe any increase in these antibodies with toluene exposure. This difference may have been due to exposure differences between the 2 populations or to the threshold chosen to define positive subjects. IgE did increase significantly with cumulative toluene exposure, a modification that is not explained by a higher frequency of allergic events. As far as we know, no hypersensitivity to toluene has been described.
In conclusion, this study found no indication that toluene at the current mean exposure level of 50 ppm (50% of the French occupational exposure limit) is related to renal dysfunction in this plant. The observation that the IgE concentration increased with cumulative exposure to toluene requires confirmation.