A big data approach to the concordance of the toxicity of pharmaceuticals in animals and humans

Many prior concordance studies have focused predominantly on approved drugs, introducing survivor's bias because these drugs' adverse events are less severe or clinically manageable.

“many prior studies have focused predominantly on approved drugs, introducing “survivor’s bias” as these drugs’ adverse events (AEs) are less severe or clinically manageable (Clark and Steger-Hartmann, 2018).”

The higher frequency of gastrointestinal disorders in dogs and monkeys is consistent with their known susceptibility compared to rodents.

“consistent with their known susceptibility comparing to rodents (Clark and Steger-Hartmann, 2018).”

The finding that most SOC-level LR+ values are low to moderate is consistent with previous work showing that SOC-level relationships are often highly statistically significant but likelihood ratios may not indicate strong predictivity.

“This finding is consistent with previous work showing that, although SOC-level relationships are often highly statistically significant, likelihood ratios may not indicate strong predictivity (Clark and Steger-Hartmann, 2018).”

In a previous study, the highest predictivity was observed for infections and infestations (LR+ = 9.6), while general disorders, investigations, and pregnancy-related conditions were among the lowest (LR+ < 3).

“In that study, the highest predictivity was observed for infections and infestations (LR+ = 9.6), while general disorders, investigations, and pregnancy-related conditions were among the lowest (LR+ < 3)”

The result that preclinical models have limited ability to predict the absence of human adverse events is consistent with previous reports.

“This result is consistent with previous reports (Bailey et al., 2014; Clark, 2015; Clark and Steger-Hartmann, 2018).”

The dog findings reflect their known sensitivity to hematologic and cardiovascular toxicities and underscore their utility in evaluating related safety risks in preclinical studies.

“reflecting their known sensitivity to hematologic and cardiovascular toxicities and underscoring their utility in evaluating related safety risks in preclinical studies (Clark and Steger-Hartmann, 2018; Koshman et al., 2024).”

The elevated percentages in rabbits reflect their role in evaluating local administration-related effects and associated systemic responses.

“reflecting their role in evaluating local administration-related effects and associated systemic responses (Clark and Steger-Hartmann, 2018).”

The pattern of LR+ results in this study broadly aligns with the pattern seen in the previous study.

“broadly aligning with the pattern seen in our results.”

The majority of adverse events showing low iLR-values under 3 is as repeatedly reported from previous concordance analyses.

“as repeatedly reported from previous concordance analyses (Clark and Steger-Hartmann, 2018; Giblin et al., 2021).”

Assignment of MedDRA terms, particularly for preclinical findings such as animal behaviors or nuanced histopathological changes, may involve subjective interpretation by original authors or database curators.

“the assignment of MedDRA terms, particularly for preclinical findings such as animal behaviors or nuanced histopathological changes, may involve subjective interpretation by original authors or database curators (Clark and Steger-Hartmann, 2018).”