Human models of pain for the prediction of clinical analgesia

Abstract

Human experimental pain models are widely used to study drug effects under controlled conditions. However, efforts to improve both animal and human experimental model selection, on the basis of increased understanding of the underlying pathophysiological pain mechanisms, have been disappointing, with poor translation of results to clinical analgesia. We have developed an alternative approach to the selection of suitable pain models that can correctly predict drug efficacy in particular clinical settings. This is based on the analysis of successful or unsuccessful empirical prediction of clinical analgesia using experimental pain models. We analyzed statistically the distribution of published mutual agreements or disagreements between drug efficacy in experimental and clinical pain settings. Significance limits were derived by random permutations of agreements. We found that a limited subset of pain models predicts a large number of clinically relevant pain settings, including efficacy against neuropathic pain for which novel analgesics are particularly needed. Thus, based on empirical evidence of agreement between drugs for their efficacy in experimental and clinical pain settings, it is possible to identify pain models that reliably predict clinical analgesic drug efficacy in cost-effective experimental settings.

Keywords: Drug discovery; Evidence-based analysis; Informatics; Model; Pain.