[Early markers of nephrotoxicity: variation factors and reproducibility].

Epidemiological validity of early markers of nephrotoxicity currently used in occupational epidemiology has been poorly investigated. The aim of this study was to identify variation factors of these markers, to quantify their intra- and inter-individual variability and to evaluate the consequences of these results on study size and power. A cross-sectional survey was carried out in 1991 in a rotogravure plant including 168 male subjects (92 exposed to toluene and 76 controls). Blood and urine samples were taken twice: at study onset and one to five months later for 40% of the subjects. Creatinine and beta-2-microglobulin (beta 2M) were measured in both blood and urine; microalbumin (microALB), N-acetyl-beta-D-glucosaminidase (NAG), and alanine-aminopeptidase (AAP), in urine. Sources of physiological variation were reduced by standardization of collection and assay methods. Subjects completed a questionnaire to record information about their personal characteristics, alcohol, tobacco and drug consumption and their health. Statistical analysis of all subjects was adjusted for exposure status. Several factors were significantly related to the markers: age with beta 2M, NAG and AAP; smoking, alcohol drinking, and blood pressure with both microALB and NAG; urinary pH with beta 2M. These factors explained from 13 to 21% of the total variance of these markers. Short-term reproducibility, i.e. the correlation between the two measurements, was high for microALB (r = 0.75), moderate for NAG (r = 0.51), and low for beta 2M (r = 0.33) and AAP (r = 0.17). These results showed that confusion bias in the evaluation of exposure-marker association can be reduced by adjusting for several factors and that accuracy and study power can be improved by repeating measurements, especially for beta 2M and AAP.