Exposure-response data from animal studies were used to derive AEGL values for arsine. AEGL values derived with animal data that had complete exposure data were more scientifically valid than AEGLs estimated from limited anecdotal human data. The greater conservatism afforded by the animal data is justified by the incomplete and often equivocal data for human exposures, the documented extreme toxicity of arsine, and the known latency involved in arsine-induced lethality. The AEGL values for the various exposure periods of concern (10 min, 30 min, 1 h, 4 h, and 8 h) were scaled from the experimental exposure duration using exponential scaling (Cn × t = k, where C is exposure concentration, t is exposure duration, and k is a constant). Data were unavailable to empirically derive a scaling factor (n) for arsine. The concentration exposure-time relationship for many irritant and systemically acting vapors and gases may be described by Cn × t = k, where the exponent n ranges from 0.8 to 3.5 (ten Berge et al. 1986). In the absence of an empirically derived exponent, and to obtain conservative and protective AEGL values, temporal scaling was performed using n = 3 when extrapolating to shorter time points and n = 1 when extrapolating to longer time points using the Cn × t = k equation.
Based upon the available data, derivation of AEGL-1 values was considered inappropriate. The continuum of arsine-induced toxicity does not appear to include effects consistent with the AEGL-1 definition. The available human and animal data affirm that there is little margin between exposures that result in little or no signs of toxicity and those that result in lethality. The mechanism of arsine toxicity (hemolysis that results in renal failure and death) and the fact that toxicity in humans and animals has been reported at concentrations at or below odor detection levels (0.5 part per million [ppm]) also support such a conclusion. The use of analytical detection limits (0.01-0.05 ppm) was considered as a basis for AEGL-1 values but was thought to be inconsistent with the AEGL-1 definition.
The AEGL-2 values were based on exposure levels that did not result in significant alterations in hematologic parameters in mice exposed to arsine for 1 h (Peterson and Bhattacharyya 1985). Uncertainty factor application included a factor of 10-fold interspecies variability because of uncertainties regarding species-specific sensitivity to arsine-induced hemolysis. Uncertainty regarding intraspecies variability was limited to a factor of 3-fold, because the hemolytic response is likely to occur to a similar extent and with similar susceptibility in most individuals. This was based on the assumption that physiologic parameters (e.g., absorption, distribution, metabolism, structure of the erythrocyte and its response to arsine, renal responses) would not vary among individuals of the same species to such an extent that the response severity to arsine would be altered by an order of magnitude. Additionally, individual variability (i.e., variability in erythrocyte structure/function or response of the kidney to hemolysis) is not likely to have a significant impact on any of the proposed subcellular mechanisms of arsine toxicity. The steep exposure-response curves from animal data also affirm the limited variability in response. Furthermore, the AEGL-2