Hormesis Outperforms Threshold Model in National Cancer Institute Antitumor Drug Screening Database

SUPPLEMENTARY DATA

Summary of Findings
The supplementary data (Tables S1-S4, Figures S1-S3) represent the same analyses
using a BMD(2.5) that were presented with a BMD(10) in the body of the paper. The
supplemental data using the BMD(2.5) are incompatible with threshold model predictions, in
agreement with the findings using the BMD(10), but even more so. The findings with the
BMD(2.5) are also highly consistent with the predictions of the hormetic model for all
methodologies employed to estimate below threshold responses. Using the BMD(2.5)
instead of BMD(10) had two principal effects. First, use of BMD(2.5) reduced the estimated
BMDs compared to the BMD(10). As a result, the number of chemical and strain
combinations with estimated BMDs < 3.7 uM for which we could not assess the behavior of the responses below the BMD increased from 7,558 out of 28,457 [for BMD(10)] to 12,167 out of 28,457. Second, for the chemical and strain combinations for which we could assess the behavior of the responses below the BMD, the evidence supporting the hormetic hypothesis is notably stronger using the BMD(2.5) method than the BMD(10) method, regardless of the analytical approach employed. ToxSciHormesisYeast