Effect of organic solvent exposure on chronic kidney disease progression: the GN-PROGRESS cohort study

Abstract

It has been suggested that solvent exposure may have a role in the progression of glomerulonephritis (GN) to end-stage renal failure (ESRD), but this has never been tested with an appropriate cohort study design. We included 338 nonESRD patients with a first biopsy for primary GN between 1994 and 2001: 194 IgA nephropathies (IgAN), 75 membranous nephropathies (MN) and 69 focal and segmental glomerulosclerosis (FSGS). ESRD, defined as an estimated glomerular filtration rate < 15 mL/min/1.73m² or dialysis, was registered over a mean follow-up period of 5 years. Patients’ lifelong solvent exposures before and after diagnosis were recorded by interview and assessed by industrial hygienist experts. We used Cox models to estimate adjusted hazard ratios (HR) of ESRD related to exposures. Overall, 15% of the patients had been exposed at a low level before diagnosis and 14% at a high level. Forty-two with IgA N reached ESRD, 12 with MN, and 22 with FSGS. A graded relationship was observed for MN: age- and gender-adjusted HR [95% confidence interval] for low exposure vs none 3.1 [0.5–18.2], and for high exposure vs none 8.2 [1.9–34.7], as well as for IgA N: 1.6 [0.7–3.9] and 2.2 [1.0–4.8], respectively, but not for FSGS. Solvent risk was only partly mediated by baseline proteinuria: adjusted HR for high exposure vs none = 5.5 [1.3 – 23.9] for MN and 1.8 [0.8 – 3.9] for IgA N. In patients with IgA N, there was a trend in increasing HR with exposure duration before and its persistence after diagnosis. These findings support the hypothesized association of solvent exposure with the progression of GN to ESRD. They should prompt clinicians to give greater attention to patients’ occupational exposures and possibly to consider professional reclassification.

INTRODUCTION

The possible role of solvent exposure in the development and/or the progression of chronic glomerulonephritis (GN) is still a controversial scientific issue more than 30 years after the earliest studies by Beirne and Zimmermann (1, 2). First reported cases were rapidly progressive GN with or without anti-glomerular basement membrane antibodies, a severe but rare type of GN (3). A few cases of membranous nephropathy (MN) (4), IgA nephropathy (IgAN) (5) and focal and segmental glomerulosclerosis (FSGS) (6) were later described in solvent exposed workers, but concern definitely rose from a number of case-control studies including chronic GN as well as other types of chronic kidney disease (CKD) (2, 7–21). Although some of these studies were criticized for their methodological weaknesses with respect to sample size and inaccuracy in case definition or exposure assessment, most of them showed significant associations of solvent exposure with various types of primary or secondary chronic GN (2,7,8,10–12,14–19), except three which were negative (9, 13, 21). However, based on our previous findings of an excess risk restricted to GN cases with chronic renal failure, we suggested that solvents may rather have a role in the progression to end-stage renal failure (ESRD) than in the initiation of GN (19). This was further indirectly supported by a meta-analysis showing that associations with solvents tended to be stronger in studies including ESRD patients than in those based on non-ESRD patients (22). This hypothesis, however, was never tested with an appropriate study design.

We thus investigated the effect of solvent exposures before and after diagnosis on the progression of primary chronic GN to ESRD, in a cohort of 338 new patients with a mean follow-up of 5 years after renal biopsy.

MATERIAL AND METHODS

Description of the cohort

The GN-PROGRESS study is a retrospective cohort considering the three most frequent types of primary GN: IgA Nephropathy (IgAN), membranous nephropathy (MN) and focal and segmental glomerolosclerosis (FSGS). It was conducted between January 2002 and March 2004 in 11 nephrology departments of the Paris region (see Appendix for a list of participating centers). Patients were identified from the renal biopsy files of the affiliated pathology departments. All GNs were histologically proven. Caucasian adult patients (> 18 years old) first diagnosed between January 1994 and June 2001, were eligible. Medical records of 853 patients fulfilling these criteria were reviewed by a group of nine experts, two nephrologists and one pathologist for each histological type, who confirmed the diagnosis and the primary nature of GN in 562 of them. We excluded from the progression study 24 patients with a baseline estimated glomerular filtration rate (eGFR) below 15 mL/min/1.73m², i.e. CKD stage 5 of the K/DOQI (23), leaving 538 eligible patients. Eighteen patients had died before the study started, 91 could not be traced despite intensive search, 7 refused to participate, 82 were unable to come to a nephrology out-patient clinic to complete the study interview and blood testing, mainly because they had moved far away from the Paris region and two had poor exposure data, leaving 338 patients for this analysis.

Exposure assessment

Patients were asked about their lifetime occupational activities by trained interviewers, using the same questionnaire as previously described (19, 24). For each job period – defined as any period of continuous employment of more than 6 months in a given job category – we recorded job title, economic activity, period and duration, as well as any contact with products or processes involving solvents or petroleum products, such as paint, printing inks and adhesives, degreasing, cleaning, and paint stripping, aerosols and pesticides, chemical and pharmaceutical processes, and gasoline. Working conditions and frequency of use (every day, at least once a week, occasionally) were also recorded. Occupations were coded according to the International Standard Classification of Occupations (ISCO) (25). In order to control for information bias, we used two independent methods to determine patients’ exposure status: an expert assessment as described by Siemiaticky et al. (26) and a job-exposure matrix, a tool used to convert job categories into information on occupational risk factors (27) which can be seen as a cross-tabulation of job titles and toxic agents, each cell including an exposure score.

First, job periods were reviewed – blind with respect to patient renal function stage - by two industrial hygienists (one of whom, JCP, was also an expert in our previous study) who determined solvent exposure level (low, medium or high) and frequency (occasional, i.e., less than once a week, or regular, i.e., once a week, subdivided as < 2, 2–20, and > 20 hours per week). For each job period, a 3-level classification combining exposure level and frequency was used: 0) no exposure; 1) low exposure – defined as under 2 hours/week whatever the level, or low level whatever the frequency; 2) high exposure – defined as above 2 hours/week at a medium or high level.

Second, the job-exposure matrix (27) assigned an exposure score to all job periods in the same category, without taking into account patients’ specific working conditions. We used the job-exposure matrix from Berrino et al. (28), which we previously tested for use in a French population (24), and where jobs are classified according to a 7-level scale as follows:

• 10: non exposed

• 20: possibly exposed at a low level

• 31: probably exposed (less than 33% of job categories exposed)

• 32: probably exposed (between 33% and 66% of job categories exposed)

• 33: probably exposed (more than 66% of job categories exposed)

• 40: certainly exposed

• 50: certainly exposed at a high level.

These categories were also grouped into three levels : 0) no exposure– defined as a score equal to 10; 1) low exposure– defined as a score from 20 to 32; 2) high exposure– defined as a score 33 or over. This job-exposure matrix was able to assign an exposure category to 92% of all patients’ job periods.

Other information

Other information from patients’ interview included social class, smoking habits, alcohol drinking as well as past or current analgesic intake (including aspirin, paracetamol, codeine, propoxyphene, nonaspirin non steroidal anti-inflammatory drug, and pyrazolones) defined as a regular use of more than 3 times per week for at least three months.

Nephrology department records provided the clinical data. Baseline information at diagnosis included proteinuria (g/24h), hypertension (defined as a blood pressure > 140/90 or anti-hypertensive treatment), renin-angiotensin system inhibitors (RASi) treatments, diabetes status as well as glomerular filtration rate (eGFR in mL/min/1.73m²). eGFR was estimated using the abbreviated Modification of Diet in Renal Disease Study equation (29), recently validated for use in French patients with CKD (30). This information, as well as age and sex, was available for all eligible patients whether participants or non participants in the study. Social class was available from medical records for 49% of the later.

Outcome

The studied outcome was ESRD, defined as either an eGFR < 15 mL/min/1.73 m² or dialysis (CKD stage 5). Serum creatinine was measured at the time of interview in all patients (except those on dialysis or leaving with a functioning graft).

Statistical analysis

Participants’ baseline characteristics were described and compared with those of non-participant eligible patients. Relationships between these characteristics and solvent exposure before diagnosis were then studied using a logistic regression model adjusted for age and gender. The hazard ratios (HR) and 95% confidence intervals (95% CI) of ESRD associated with solvent exposure level before diagnosis were estimated by Cox proportional-hazards regression for overall GN and by histological type. We also studied the HR of ESRD associated with exposure duration at a high level before diagnosis as well as with the persistence of high level exposure after the diagnosis in patients with IgA N. In all these analyses, both exposure assessment methods were studied, and HRs were estimated from three models as follows: Model 1 - crude analysis; Model 2 - adjustment for age at diagnosis and gender; Model 3 - further adjustment for baseline hypertension and quantitative proteinuria. We also tested the effect of adjusting for RASi treatments, diabetes, analgesic intake (ever vs. never), smoking habits and alcohol.

RESULTS

Patients’ characteristics at baseline

IgA N was the most frequent histological type. Men were affected three times more often than women (Table 1). Mean age was 37 ± 14 years for IgA N, 50 ± 15 for MN and 46 ± 16 for FSGS. No significant differences were observed between participants and non-participants in the study, particularly with respect to baseline eGFR and social class. Among participants, 7% had diabetes, 24% had past or current analgesic intake and 18% for RSAi treatment, 59% were past or current smokers, 62% had already regularly consumed alcohol (at least once a week).

Table 1.

Patients’ characteristics at baseline

	Participants N = 338	Non participants N = 200	p-value
Age ≤ 40 years	51%	46%	ns
Men	74%	73%	ns
Histological type
IgA N	58%	45%	ns
MN	20%	28%
FSGS	22%	27%
Hypertension	58%	61%	ns
Proteinuria ≥ 3 g/24h	41%	47%	ns
eGFR (ml/mn/1,73 m2)
≥ 90	27%	33%	ns
[60 – 90 [	33%	32%
[30 – 60 [	25%	23%
[15 – 30 [	15%	12%
Social class
- Professional and managerial workers	34%	28%	ns
- Sales and clerical workers	23%	23%
- Service, agricultural, and production workers	43%	49%

IgA N, IgA Nephropathy; MN, Membranous Nephropathy; FSGS, Focal and Segmental Glomerulosclerosis

ns: p>0.05

Relationships between patients’ characteristics and solvent exposure before diagnosis

Participants reported a mean of three job periods (range: 1 – 9). The experts classified 9% of these jobs as exposed to solvents at a low level and 8% at a high level. These percentages were respectively 18% and 14%, using the job exposure matrix. Table 2 displays solvent exposure by main job categories according to expert assessment. As expected, most exposed jobs were those of production. Surprisingly, there was a high proportion of exposed jobs in the life scientists and related technicians category, but this was due to the presence of several laboratory technicians manipulating solvents. On the other hand, there were few exposed individuals among in sculptors, painters and creative artists group because many of them were drawers and photographers, classified as unexposed. Moreover, the painters in this group were considered as poorly exposed in sight of the patients’ jobs description and experts’ evaluation.

Table 2.

Solvent exposure by main job categories according to expert assessment

		% exposed
Jobs	N	Any level	High level
Professional technical workers	285	18%	6%
Technicians: Architects, engineers	45	31%	9%
Life scientists and related technicians	13	77%	61%
Sculptors, painters, creative artists	13	38%	0%
Administrative and managerial workers	50	6%	4%
Clerical workers	171	2%	0%
Sales workers	94	0%	0%
Service workers	142	3%	0%
Agricultural workers	20	0%	0%
Production and transport workers	355	32%	22%
Production supervisors and general foremen	16	31%	25%
Blacksmiths, toolmakers and machine tool operators	22	27%	27%
Machinery fitters, machine assemblers	33	48%	36%
Electrical and electronics workers	37	30%	8%
Plumbers, welders	37	46%	35%
Painters	7	100%	86%
Construction workers	58	41%	17%
Material-handling and transport workers	53	8%	3%
Manual workers	24	26%	17%
Total	1117	17%	9%

Based on the expert assessment, 29% of the patients had been exposed at least once to solvents before diagnosis, but only 14% at a high level (Table 3). These percentages were higher when using the job exposure matrix. Solvent exposure was significantly more frequent in men and older patients with both assessment methods. A high level of solvent exposure was also more frequent in patients with baseline proteinuria in the nephrotic range as well as in those with low eGFR, but the age- and gender-adjusted associations were only statistically significant for expert assessed exposures. In contrast, solvent exposure was neither associated with hypertension nor with diabetes, analgesic intake, RSAi treatment, smoking habits or alcohol consumption.

Table 3.

Relationships between patients’ baseline characteristics and solvent exposure before diagnosis

	N	Solvent exposure before diagnosis
		Expert assessment				JEM assessment
		Any level	p	High level	p	Any level	p	High level	p
Total	338	29%		14%		51%		25%
Age
≤ 40 years	173	20%	*	8%	*	39%	*	15%	*
> 40 years	165	38%		22%		58%		34%
Gender
Women	88	15%	*	7%	*	40%	*	11%	*
Men	250	34%		18%		51%		29%
Histological type
IgA N	194	26%		13%		45%		21%
MN	75	35%		20%		51%		30%
FSGS	69	32%		14%		57%		26%
Hypertension
No	141	22%		10%		40%		21%
Yes	197	34%		18%		55%		27%
Proteinuria
< 3 g/24h	198	24%		10%	*	42%		19%
≥ 3 g/24h	134	35%		21%		57%		31%
eGFR (ml/mn/1,73 m2)
≥90	88	22%		8%	*	37%		16%
[60 – 90 [	113	27%		12%		45%		23%
[30 – 60 [	85	35%		21%		63%		30%
[15 – 30 [	52	35%		23%		50%		31%

JEM : Job-exposure matrix; eEGFR : estimated Glomerular Filtration Rate using the MDRD equation; IgA N, IgA Nephropathy; MN, Membranous Nephropathy; FSGS, Focal and Segmental Glomerulosclerosis

^*age- and gender-adjusted p-values < 0.05

Relationships between solvent exposure before diagnosis and end-stage renal disease risk

Mean patient follow-up was 56 months. Fourty-two patients with IgA N reached ESRD, 12 with MN, and 22 with FSGS. Crude annual incidence rates of ESRD were 4.4% [3.1 – 5.7] for IgA N, 3.4% [1.5 – 5.3] for MN, and 8.0% [4.7 – 11.5] for FSGS.

The crude analysis showed dose-response relationships for GN as a whole (p for trend < 0.05) with both methods of exposure assessment (Table 4). These trends were observed in patients with IgA N and MN, but not in those with FSGS. Overall, solvent exposures assessed by the job-exposure matrix yielded lower HR than those assessed by experts, particularly in patients with MN. Adjustment for age and sex did not significantly alter any of the above associations. Further adjustment for hypertension and proteinuria tended to weaken the HR for both IgA N and MN, but, for this latter GN, estimates based on the expert assessment remained statistically significant. In contrast, additional adjustment for diabetes, analgesic intake, treatment with RASi, smoking and alcohol consumption did not modify risk estimates. Finally, it is worth noting that the lack of association in patients with FSGS was equally observed in both age groups, below and above the age of 40 years.

Table 4.

Hazard ratios of end stage renal disease associated with exposure before diagnosis for each histological type

			Model 1	Model 2	Model 3
	Total number	Number of events	Hazard ratios (95% Confidence Interval)
Expert assessment
IgA N
No a	145	26	Ref.	Ref.	Ref.
Low b	24	6	1.6 (0.7 – 4.0)	1.6 (0.7 – 3.9)	2.0 (0.8 – 4.8)
High c	25	10	2.6 (1.3 – 5.5)	2.2 (1.0 – 4.8)	1.8 (0.8 – 3.9)
MN
No	49	3	Ref.	Ref.	Ref.
Low	11	3	4.0 (0.8 – 21.4)	3.1 (0.5 – 18.2)	3.1 (0.5 – 20.0)
High	15	6	8.7 (2.1 – 35.1)	8.2 (1.9 – 34.7)	5.5 (1.3 – 23.9)
FSGS
No	47	19	Ref.	Ref.	Ref.
Low	12	0	–	–	–
High	10	3	0.5 (0.2– 1.8)	0.7 (0.2 – 2.4)	0.5 (0.1 – 2.1)
All
No	241	48	Ref.	Ref.	Ref.
Low	47	9	1.0 (0.5– 2.1)	1.0 (0.5 – 2.2)	1.0 (0.5 – 2.1)
High	50	19	2.2 (1.3– 3.7)	2.2 (1.2 – 3.8)	1.6 (0.9 – 2.9)
JEM* assessment
IgA N
No d	107	16	Ref.	Ref.	Ref.
Low e	43	10	1.5 (0.7 – 3.3)	1.4 (0.6 – 3.1)	1.0 (0.4 – 2.2)
High f	41	16	2.9 (1.5 – 5.8)	2.6 (1.2 – 5.3)	1.6 (0.7 – 3.4)
MN
No	36	4	Ref.	Ref.	Ref.
Low	15	2	1.4 (0.3 – 7.8)	1.4 (0.3 – 7.7)	1.0 (0.2 – 5.6)
High	22	6	2.5 (0.7 – 9.0)	2.0 (0.5 – 7.5)	1.4 (0.4 – 5.4)
FSGS
No	30	12	Ref.	Ref.	Ref.
Low	21	5	0.6 (0.2– 1.8)	0.7 (0.2 – 2.0)	0.6 (0.2 – 1.8)
High	18	5	0.7 (0.2– 1.9)	0.9 (0.3 – 2.7)	0.7 (0.2 – 2.3)
All
No	173	33	Ref.	Ref.	Ref.
Low	79	17	1.2 (0.7– 2.2)	1.2 (0.6 – 2.2)	1.2 (0.7 – 2.3)
High	81	27	2.0 (1.2– 3.3)	2.0 (1.1 – 3.4)	1.4 (0.8 – 2.4)

^*Job Exposure Matrix

^ano exposure;

^blow intensity or frequency < 2hours/week;

^chigh intensity and frequency ≥ 2hours/week;

^dJEM score = 10;

^eJEM score = 20, 31 or 32;

^fJEM score ≥ 33

Model 1: crude hazard ratios;

Model 2: hazard ratios adjusted for age, gender;

Model 3: hazard ratios adjusted for age, gender, proteinuria and hypertension

Relationships between duration and discontinuation of solvent exposure, and risk of end-stage renal disease in patients with IgA nephropathy

The higher the duration of exposure at a high level before diagnosis, the higher the ESRD risk in IgAN. This trend was observed with both assessment methods, and remained statistically significant in the full-adjustment model for the job-exposure matrix (Table 5).

Table 5.

Hazard ratios of end stage renal disease associated with the duration of high exposure before diagnosis and the persistence of high exposure after diagnosis, for IgA Nephropathy

			Model 1	Model 2	Model 3
	Total number	Number of events	Hazard ratios (95% Confidence Interval)
Expert assessment
Duration
0	168	32	Ref.	Ref.	Ref.
≤ 10 years	14	4	1.8 (0.6 – 5.0)	1.5 (0.5 – 4.3)	1.2 (0.4 – 3.5)
> 10 years	11	6	3.3 (1.4 – 7.8)	2.8 (1.1 – 7.1)	2.1 (0.8 – 5.3)
Period
Never until diagnosis	167	32	Ref.	Ref.	Ref.
Before diagnosis only	15	6	2.8 (1.2 – 6.8)	2.3 (0.9 – 5.9)	1.7 (0.7 – 4.2)
Before and after diagnosis	9	4	2.4 (0.8 – 6.8)	2.3 (0.8 – 6.5)	2.1 (0.8 – 6.8)
JEM * assessment
Duration
0	149	25	Ref.	Ref.	Ref.
≤ 10 years	26	7	1.6 (0.7 – 3.8)	1.6 (0.7 – 3.8)	1.1 (0.4 – 2.8)
> 10 years	12	8	5.0 (2.3 – 10.7)	4.0 (1.7 – 9.6)	2.6 (1.1 – 6.2)
Period
Never until diagnosis	146	25	Ref.	Ref.	Ref.
Before diagnosis only	26	10	2.5 (1.2 – 5.3)	2.3 (1.1 – 4.8)	1.4 (0.6 – 3.2)
Before and after diagnosis	11	5	2.7 (1.0 – 7.1)	2.4 (0.9 – 6.4)	2.6 (1.0 – 6.9)

^*Job Exposure Matrix

Model 1: crude hazard ratios

Model 2: hazard ratios adjusted for age, gender

Model 3: hazard ratios adjusted for age, gender, proteinuria and hypertension

Based on the expert assessment, 15 (60%) out of the 25 highly exposed patients discontinued exposure after diagnosis; this was the case for 26 (63%) out of 41, according to the job exposure matrix. Hazard ratios tended to be weaker in highly exposed patients who discontinued exposure than in those who continued which was clear with the job-exposure matrix, but less so with expert-assessed exposures.

DISCUSSION

Our hypothesis that solvent exposure may hasten GN progression is strongly supported by the dose-effect relationships with ESRD risk observed for IgA and MN in this study. These relationships, independent of age and gender, were partly mediated by baseline proteinuria. These findings are original in that having generated this hypothesis on the basis of a case-control study (19), we now provide additional evidence with a new set of patients and an appropriate cohort study design taking into account known determinants for CKD progression. Their implications for patient care as well as for CKD monitoring and prevention in the occupational setting are important.

The most important obstacles in studying risk factors for progression of GN are the need for biopsy proven diagnosis, the rarity of each histological type, and their relatively slow progression to ESRD. Indeed, despite considering all incident adult Caucasian patients over a 6-year period from all major nephrology centers in the Paris area, we only identified 538 eligible cases. A major strength of this study, however, is that it was based on an unselected cohort of histologically well-defined primary glomerulonephritis as confirmed by experts. The choice of a retrospective rather than a prospective cohort design increased the number of ESRD events, but the drawback was the number of non participants, as moving from the Paris area is frequent. Although we were able to trace more than 80% of the initial cohort, only 63% of the patients did participate in the study. Nevertheless, participants did not significantly differ from eligible nonparticipants with respect to social class as well as baseline clinical and biological data, and the number of death was low (3.5%). Selection bias, if any, should thus be minimal and more likely to weaken than to explain the observed associations.

Retrospective exposure assessment was a critical issue in many of the previous studies. Our method based on a highly detailed and structured questionnaire and blind evaluation of exposures by experienced industrial hygienists has proved to be very reliable (26). We used it in a previous case-control study (19), and it is worth noting that the experts classified a similar percentage of individuals, about 30%, as exposed to solvents at any level, and about 15% at a high level, in both studies. A possible weakness of this method, however, is that it strongly relies on individual recall of past exposures. Although most interviews were carried out blind with respect to the patients’ CKD stage, this was not possible to achieve for some patients seen at the dialysis center. Recall bias may have thus contributed to our findings, if patients without ESRD tended to underreport exposures in comparison with those with ESRD. Moreover, the expert assessment was sometimes based on poor information from patients’ questionnaires, resulting in possible nondifferential misclassification of exposure level or frequency or both. An important strength of this study is that we used a second method to assess exposure, the so-called job-exposure matrix. Its advantage over the expert rating method is that it does not rely on patients’ memory of his or her personal occupational exposures. Instead, exposures are assigned on the basis of the job categories alone, which are reported with accuracy by most individuals. We have previously shown that this method was more sensitive and less specific that the expert assessment resulting in both higher percentages of exposure and weaker hazard ratios due to nondifferential misclassification, just as was observed here (24). The fact that the job-exposure matrix yielded similar results as the expert assessment argues against major information bias.

It is difficult to compare our results with others, because of differences in study objectives and design. They are, nevertheless, consistent with previous case-control studies showing increased odds-ratios of GN or all-cause CKD with solvent exposure (7, 8, 10–12, 15, 16, 18–20), as well as with our earlier findings of a different pattern of risk according to CKD stage in 298 GN cases (19). This pattern was also observed in another case-control study (18), and further pointed out in a meta-analysis (22), in which studies including ESRD patients clearly tended to report stronger ORs (2, 10, 14, 16) than those including patients with chronic renal failure (18–20), whereas negative findings were observed in studies, or in subgroups of cases within studies, mainly including patients with GN at an early stage (9,12, 19,20). However, the negative findings from a recent population-based case-control study (21), including 926 patients with all-cause chronic renal failure (GN, 24%), did not seem to fit in that overall pattern. An explanation may be that it was based on prevalent cases with long disease duration (mean, 10 years), mixing up rapidly and slowly progressing patients, which may have diluted odds-ratios estimates. Another explanation may also come from the high prevalence of solvent exposure in controls (40%) as compared to other studies as reviewed in (22). Finally, our results are also consistent with those from Yaqoob et al (31), based on a series of 68 patients with primary GN (60 proliferative GN, 8 MN), showing that solvent exposure was higher in 29 patients with progressive renal failure (defined as a persistent rise in serum creatinine >50 micromol/l above the baseline over 5 years) than in their 39 counterparts without such a rise.

In hypothesizing that solvents would hasten CKD progression to ESRD, we expected an effect independent of the histological type of GN, also suggested by findings in patients with diabetic nephropathy (17). Yet, we observed strong associations in patients with MN and IgAN, but not in those with FSGS. This was all the more surprising as they were at variance with our previous findings of higher odds-ratios with exposure in patients with FSGS and chronic renal failure as compared to matched controls (19). Moreover, at least one case was reported in a man exposed to toluene for 40 years (6). In this study, we focused on patients with primary FSGS, defined by the experts as the presence of FSGS injury with or without nephrotic range proteinuria excluding cases with advanced chronic renal failure and unclear history of renal injury. The overall selection process was highly specific and patients were excluded when in the slightest doubt about secondary FSGS. This led to the exclusion of many old hypertensive patients, in whom the primary nature of FSGS was difficult to assert. As solvent exposure is related to age, selection bias may be a possible explanation for the negative findings in that group of patients. However, we cannot rule out that the effect of solvents is specific of the GN type.

The fact that solvent exposure was associated with increased baseline proteinuria and that the relationship with ESRD tended to weaken after adjustment, suggests that solvent nephrotoxicity may be partly mediated by enhancing proteinuria, a major risk factor for kidney disease progression (32). In support of this hypothesis, several occupational cross-sectional studies have shown that workers exposed to various types of solvents had higher levels of albuminuria as well as of low-molecular weight proteinuria, a marker of tubular injury, as compared to nonexposed workers (33). More interestingly, although there are very few relevant experimental studies in the field, it is worth noting that from Mutti et al (34). This unique study was specifically designed to assess the role of styrene, a widespread aromatic solvent, in the progression of adriamycin nephrosis in rats, a well known model of glomerulopathy. Female Sprague-Dawley rats were divided into four groups, exposed to either adriamycin or styrene or both, or neither of them. It showed that styrene induced proteinuria, including albuminuria and urinary fibronectin, in both adriamycin-treated and nontreated rats as compared to controls with neither toxics, with a significant adriamycin*styrene interaction. Moreover, the score for interstitial fibrosis was significantly higher and cellular infiltrates significantly greater in rats co-exposed to both toxics as compared to adriamycin-treated rats. Although limited to a single solvent, this experiment provides strong evidence of the role of solvent exposure in CKD progression through several potential pathogenic mechanisms including enhanced proteinuria and tubuluinterstitial damage.

Two limitations of this study should be noted. First, even though based on the largest series of GN ever studied in relationship with solvent exposure, this study lacked power in sub-group analysis, particularly in MN patients. The a posteriori power of the study was 80% (with α = 5%) to detect a HR of 2.6 for IgAN, 6.0 for MN and 2.9 for FSG (35). Therefore, negative findings by chance cannot be completely ruled out for FSGS. Second, serum creatinine were not measured in a single laboratory. Although inter-laboratory calibration errors only slightly affect estimates in high values of serum creatinine, outcome misclassification is possible in some patients. However, inter-laboratory calibration errors only slightly affect estimates in high values of serum creatinine. Moreover, misclassification, if any, would be nondifferential, i.e., independent of solvent exposure, and thus unlikely to have produced a spurious association.

Implications for clinical practice are important. Patients diagnosed with GN should systematically be asked about their occupations and potential solvent exposures. In the same way, individuals referred to nephrologists by occupational physicians for early detection of proteinuria should be given greater attention in case of solvent exposure, and professional reclassification may be considered in those with progressive GN.

In conclusion, this study provides new evidence of the potential role of solvents in the progression of CKD. As GN, particularly IgAN, are a major cause of ESRD, interventions to promote screening for proteinuria and possibly hematuria in solvent-exposed workers and discontinuation of exposure in those identified with GN at an early stage may prevent or slow the progression to ESRD. Further larger epidemiological studies and additional experimental studies are nevertheless needed to confirm our results and to better characterize the types of solvents at risk as well as to clarify whether or not they can affect the course of CKD other than GN.

APPENDIX: GN-PROGRESS Study Group

Experts

Nephrologists: P. Ronco; X. Belenfant; D. Chauveau; O. Kourilsky; F. Martinez; F. Vrtovsnik ; Pathologists: Y. Allory; D. Droz; D. Nochy

Centers participating in the study (heads of nephrology department, investigators)

Hôpital André Grégoire, Montreuil (X. Belenfant) ; AP-HP Hôpital Bicêtre, Kremlin-Bicêtre (B. Charpentier, A. Durrbach); AP-HP Hôpital Bichat, Paris (F. Mignon, F. Vrtovsnik); Hôpital Claude Galien, Quincy/Senart (G. Rostoker); AP-HP, Hôpital Européen Georges Pompidou, Paris (J. Bariety, C. Jacquot); AP-HP Hôpital Henri Mondor, Créteil (P. Lang, P. Remy); Hôpital Louise Michel, Evry (O. Kourilsky); AP-HP Hôpital Necker, Paris (J-P. Grünfeld), D. Chauveau); AP-HP Hôpital Pitié Salpétrière, Paris (G. Deray, H. Izzedine) ; AP-HP Hôpital Saint-Louis, Paris (C. Legendre, F. Martinez) ; AP-HP Hôpital Tenon, Paris (J-D. Sraer, C. Vigneau ; P. Ronco, J. Rossert)