Discontinuity in the cancer slope factor as it passes from high to low exposure levels - arsenic in the BFD-endemic area

Background: The ingestion of inorganic arsenic causes bladder and lung cancers demonstrably at >400-500. ug/L but questionably below 100-200. ug/L. Using the standard 42-village cancer mortality dataset from the Blackfoot-disease (BFD) endemic area of southwest Taiwan ( Wu et al., 1989), we examined the risk from low exposures by excluding the high exposures. Method: Poisson regression analyses with the sequential removal of the highest exposure village have been performed using the median, mean, or maximum village well water arsenic level and demonstrated graphically. Results: Risk estimates are positive when villages with exposures of 200-400. ug/L are included and significantly so when villages with >400. ug/L are included. Risk estimates for exposures below 100. ug/L are negative but rarely significantly so. The inflection point where the slope is no longer positive occurs in the range of 100-200. ug/L, depending upon whether the exposure metric used is the median, the mean or the maximum. Conclusion: There is a discontinuity in the cancer slope factor or risk from arsenic exposure that occurs in the range of 100-200. ug/L. Above these levels, there are significantly positive risks, while below these levels there are not. The analysis reveals within this dataset an intrinsic non-linearity in the cancer risk. The literature speaks to this discontinuity, but this is the first demonstration within a single dataset that shows the discontinuity across the full exposure range and where the low-dose data are not compromised with high-dose data.