Human exposure to toxic metals: factors influencing interpretation of biomonitoring results

An important approach to assessment of risk from environmental and occupational exposures is biomonitoring which provides an estimate of the total dose absorbed and gives indirect access to determination of target site concentrations. It is also a tool for assessing residual exposure, when respiratory protection is used. The interest in biological effects from toxic metals has increased during the last decades, as large amounts of metals have been released into industrial areas, and it is estimated that approximately 150000 Danish workers are exposed to various metals. Since biomonitoring results play an important role in decision-making regarding great health and economic impact, understanding the factors influencing validity of such data is essential. In the present survey, the toxic elements arsenic, cadmium, chromium, cobalt, lead and nickel are used as examples to illustrate the disturbing factors in the interpretation of biomonitoring results. The aim of collecting samples is to obtain a small and representative sample of subjects or of a subpopulation being investigated for specific purposes, e.g. the pollution at Mundelstrup, where arsenic exposure of inhabitants and workers engaged in removing contaminated soil was monitored. As exposures vary over time and between subjects, it should be recognized that sampling as well as analytical variations contribute both to bias and random errors. Also biomonitoring data are a function of demographic, lifestyle and geographic factors. Therefore, stratified sampling designs are of the utmost importance. Half-lives play an important role. For short half-lives below 10 h, no decision can be made on long-term exposure if only one urine sample has been taken. For arsenic, cobalt, chromium and nickel in urine measured at the end of a workweek (t12; 20–100 h), 2–3 samples should be taken to monitor a single worker, and the results should be interpreted from the average of the results. For groups of workers, it is recommended that results from at least 5–10 workers be used to obtain a useful group mean value. In general, pharmacokinetics modelling contributes to information on sampling time and sampling size. Intake of cobalt in mineral tablets containing soluble cobalt compounds was a factor with a large influence on blood and urinary levels. Age and gender influence the blood and urine concentrations of arsenic, cadmium, cobalt and lead, e.g. BPb in females is 20–30% lower than BPb in males. Urinary cobalt was three times higher in females than in males under controlled experimental conditions of oral up-take, although the number of studied subjects was low. Quality assurance and method evaluation are needed to reduce false decisions due to measurement errors. Reference values are essential to assess exposure, and the IFCC recommendation is useful for a number of reference individuals above 120. For observations below 80, a 95% expectation tolerance interval is recommended, as the interval contains a prediction regardless of the number of reference values. At low level exposure to a mixture of toxic metals, valuable information is obtained from principal component regression modelling, which forms an important tool for interpretation of individual results. Future research should include physiologically-based pharmacokinetics modelling, quality assurance for harmonization of biomonitoring results, identification of reference limits, and multivariate modelling to maintain the preventive approach of biological biomonitoring.