Evidence For a Benzene-Leukemia Supralinear Dose Response?
What is the risk of leukemia in workers exposed to relatively low cumulative levels (<100 ppm yrs) of benzene? To get at the answer researchers examined nine published studies from which exposure-response data could be extracted. The exposure response curve that best fit the data shows a supralinear response, by which it is meant that the rate of response falls less rapidly for an interval before returning to its former rate of descent. The primary explanation offered for the deviation from the linear is that at some level the mechanism by which benzene is ordinarily metabolized is overwhelmed, or saturated, and metabolism by some other mechanism, one responsible for metabolites that cause leukemia, takes over until it too is saturated beyond which point risk rises at a lower rate.
The obvious problem with the best fitting curve graphed in the study is that it manages to predict a positive risk for benzene-induced leukemia even when the exposure to benzene is zero - an apparently absurd result. This best fitting model shows greater risk of benzene induced leukemias down to 10 ppm years than would be predicted by a typical linear no threshold model and a doubling of the risk level at just above 40ppm yrs. However, the second best fitting curve shows a sharp decline, especially below 40 ppm yrs and one which might be consistent with a threshold for risk at some non-zero level.
The paper is free and you can find it here: "Flexible Meta-Regression to Assess the Shape of the Benzene-Leukemia Exposure-Response Curve"
