Selecting the best non-invasive matrix to measure mercury exposure in human biomonitoring surveys

Exposure to mercury, even at low doses, can affect human health, well-being and life quality at a broad scale. Human biomonitoring is the most straightforward approach to measure and quantify mercury exposure in humans. The objective of the present study is to compare and discuss the relationships between Hg levels in the most used matrices, hair, urine and blood, with the aim to ascertain to what extent mercury exposure and internal mercury levels could be predicted by monitoring non-invasive matrices. The study population (n = 527) is a subsample from Spanish BIOAMBIENT. ES study (18-65 y, both sexes), with data of Hg levels in blood, hair, and urine from the same individuals. We found strong inter-matrix Spearman correlations between blood and hair mercury (r2 = 0.84), while the correlations for urine and blood mercury (r2=0.64) and urine and hair mercury (r2=0.65) were weaker. The geometric mean of the ratios between matrices were (GM, 95%CI): Hair/Blood 280 (271-290), Urine/Blood 0.174 (0.163-0.186) and Hair/ Urine 2070 (1953-2194) and Urine/Blood 0.135 (0.128-0.144) for urine corrected by creatinine. High individual variation was observed particularly in those ratios involving urine. Considering the wide range of values observed in the ratios, we do not recommend applying them at individual level. The predictive models indicate that hair Hg was a more accurate predictor than urine. The inclusion of urine values did not increase the predictive accuracy, so, we recommend a cautious interpretation of urine mercury levels. Our study presents clear evidence that in a population highly exposed to food-borne mercury, a large portion of urinary mercury primarily emanates from methylmercury demethylation. We conclude that urine, as a noninvasive matrix, can be used as a reliable qualitative biomarker for Hg exposure when hair measurements not are available. For quantitative individual assessments, still blood measurements are to be preferred.