Abstract
The authors would like to thank Rose et al for the comments and the Editor for the opportunity to respond. Rose et al support the use of new approach methodologies, but question the robustness of our interpretation of US EPA’s ToxCast high-throughput screening (HTS) data in our review article, “Cancer Hazard Identification Integrating Human Variability: The Case of Coumarin.” 1
As described in our review on page 502, we integrated evidence from multiple data streams in considering the potential carcinogenicity of coumarin, including information from animal cancer bioassays, data from toxicogenomic, genotoxicity, and metabolism studies, and data from studies of human variability in CYP2A6 activity. A small section of this comprehensive review (pages 539-540) reported HTS data. Coumarin and its metabolite, 3,4-dihydrocoumarin (3,4-DHC), were active in only a small percentage of the ToxCast iCSS Dashboard (version 2) assays in which they were tested: 1.5% and 0.3%, respectively. Further, cell-based ToxCast assays currently offer limited or no metabolic capability (which coumarin requires) and do not adequately cover end points related to the key characteristics of carcinogens. Due to these limitations, HTS data were not able to contribute much, if any, weight in the overall review of the potential carcinogenicity of coumarin.
Rose et al contend that curve fitting flags were associated with the few HTS assays in which coumarin and 3,4-DHC were active and thus may have been false positives. First, we would like to emphasize that such flags are subject to change as the ToxCast data analysis pipeline evolves. 2 Our results were extracted from the iCSS Dashboard (version 2) in 2017. Since then, all data have been re-pipelined and new data have been incorporated. This resulted in changes to the parameters (including baseline median absolute deviation (bmad)) that contribute to defining the hit call, which most often affect borderline hits and flags associated with them (ie, “borderline active” and “borderline inactive” flags; personal communication with Dr. Keith Houck, US EPA, 2020). Thus, it stands to reason that borderline flags may change in different versions of ToxCast. Second, completely filtering out all active assay calls with flags is not recommended because potential biological signals could be omitted. 3 Judson et al 3 recommends that, rather than using flags as a cutoff for the relevancy of assay responses, users can benefit from applying a comprehensive understanding of the flags for each assay to identify potential interference, consider assay accuracy, and interpret biological significance.
The second concern of Rose et al is that “one would have greater confidence in the results if there are several robust assays to validate a specific target.” They concluded that 3 assays lacked biological concordance, and only 2 assays testing for the enzymatic activity for monoamine oxidase (MAO) showed concordance but were accompanied by four inactive MAO assays. We agree that concordant observations across multiple active assays for one specific target can provide greater confidence in a particular effect. Given the small number of active assays for coumarin and 3,4-DHC, however, it would be surprising to observe such consistency. In fact, lack of biological concordance among related assays is commonly seen in high-throughput assays, even for chemicals with more active assays than were observed with coumarin. 4 ToxCast enzyme inhibition assays are often paired with enzyme activator assays, and this must be taken into account in interpreting the data. 5 Regarding the 6 MAO assays mentioned by Rose et al, these represent 3 sets of paired assays: 2 for MAO-B and 1 for MAO-A. One assay in each pair is analyzed in the positive fitting direction reflecting decreased enzymatic activity, while the activator assay is analyzed in the negative fitting direction reflecting increased enzymatic activity. Thus, the 2 assays in an MAO assay “pair” are not expected to both be active.
Finally, Rose et al highlighted the fact that it is difficult to draw “conclusions from the bioactivity of a compound in ToxCast to an apical, histological endpoint in toxicity tests.” We concur that bioactivity in ToxCast assays assessing non-apical end points is not always echoed by observations of corresponding apical effects in in vivo studies. This is the case for coumarin with regard to observations from the ToxCast HTS assays.
Footnotes
Author Contribution
ChingYi Jennifer Hsieh and Meng Sun contributed equally to this work. Hsieh, C.Y.J substantially contributed to conception or design, drafted the manuscript, and critically revised the manuscript for important intellectual content; Sun, M. substantially contributed to conception or design, drafted the manuscript, and critically revised the manuscript for important intellectual content; Osborne, G. substantially contributed to conception or design, drafted the manuscript, and critically revised the manuscript for important intellectual content; Ricker K substantially contributed to conception or design and critically revised the manuscript for important intellectual content; Tsai, F. C. substantially contributed to conception or design and critically revised the manuscript for important intellectual content; Li, K. substantially contributed to conception or design and critically revised the manuscript for important intellectual content; Tomar, R. substantially contributed to conception or design and critically revised the manuscript for important intellectual content; Phuong J. substantially contributed to conception or design and critically revised the manuscript for important intellectual content; Schmitz, R. substantially contributed to conception or design and critically revised the manuscript for important intellectual content; Sandy, M.S. substantially contributed to conception or design, drafted the manuscript, and critically revised the manuscript for important intellectual content. All authors gave final approval and agree to be accountable for all aspects of the work in ensuring that questions relating to the accuracy or integrity of any part of the work are appropriately investigated and resolved. The views expressed are those of the authors and do not necessarily represent those of the Office of Environmental Health Hazard Assessment (OEHHA), the California Environmental Protection Agency, or the State of California.