Predicting Human Infection Risk: Do Rodent Host Resistance Models Add Value?

Abstract

Use of genetically engineered rodents is often considered a valuable exercise to assess potential safety concerns associated with the inhibition of a target pathway. When there are potential immunomodulatory risks associated with the target, these genetically modified animals are often challenged with various pathogens in an acute setting to determine the risk to humans. However, the applicability of the results from infection models is seldom assessed when significant retrospective human data become available. Thus, the purpose of the current review is to compare the outcomes of infectious pathogen challenge in mice with genetic deficiencies in TNF-α, IL17, IL23, or Janus kinase pathways with infectious outcomes caused by inhibitors of these pathways in humans. In general, mouse infection challenge models had modest utility for hazard identification and were generally only able to predict overall trends in infection risk. These models did not demonstrate significant value in evaluating specific types of pathogens that are either prevalent (ie rhinoviruses) or of significant concern (ie herpes zoster). Similarly, outcomes in mouse models tended to overestimate the severity of infection risk in human patients. Thus, there is an emerging need for more human-relevant models that have better predictive value. Large meta-analyses of multiple clinical trials or post-marketing evaluations remains the gold-standard for characterizing the true infection risk to patients.