Clinical pharmacology of analgesics assessed with human experimental pain models: bridging basic and clinical research

Abstract

The medical impact of pain is such that much effort is being applied to develop novel analgesic drugs directed towards new targets and to investigate the analgesic efficacy of known drugs. Ongoing research requires cost-saving tools to translate basic science knowledge into clinically effective analgesic compounds. In this review we have re-examined the prediction of clinical analgesia by human experimental pain models as a basis for model selection in phase I studies. The overall prediction of analgesic efficacy or failure of a drug correlated well between experimental and clinical settings. However, correct model selection requires more detailed information about which model predicts a particular clinical pain condition. We hypothesized that if an analgesic drug was effective in an experimental pain model and also a specific clinical pain condition, then that model might be predictive for that particular condition and should be selected for development as an analgesic for that condition. The validity of the prediction increases with an increase in the numbers of analgesic drug classes for which this agreement was shown. From available evidence, only five clinical pain conditions were correctly predicted by seven different pain models for at least three different drugs. Most of these models combine a sensitization method. The analysis also identified several models with low impact with respect to their clinical translation. Thus, the presently identified agreements and non-agreements between analgesic effects on experimental and on clinical pain may serve as a solid basis to identify complex sets of human pain models that bridge basic science with clinical pain research.

Figure 1

Drug class and sum of successful and unsuccessful applications in clinical pain conditions (max. number possible: 35) and experimental pain models (max. number possible: 33). The graphs suggest a correlation between positive and negative outcomes in clinical and experimental pain conditions. This correlation was indeed present as indicated by a Spearman's σ of 0.71 (P < 0.001).

Figure 2

Clinical pain conditions for which analgesia or no analgesia was correctly predicted by an experimental pain condition. The table presents all combinations of experimental pain models and clinical pain conditions in which analgesics were effective or ineffective in both, the clinical and the experimental pain condition (correct prediction of clinical effectiveness by an experiment). The size of the pie charts correlates with the number of drugs that were correctly predicted. The colours of the segments indicate the drug or drug class that was correctly predicted.

Figure 3

Clinical pain conditions for which analgesia or no analgesia was incorrectly predicted by an experimental pain condition. The table presents all combinations of experimental pain models and clinical pain conditions in which analgesics were effective in only one condition, e.g. the clinical pain condition, while being ineffective in the other, e.g. experimental pain condition (incorrect prediction of clinical effectiveness by an experiment). The size of the pie charts correlates with the number of drugs that were incorrectly predicted. The colour of the segments indicates the drug or drug class that was correctly predicted.