Chronic pain is the hall-mark symptom of osteoarthritis (OA) and although several therapies are available, a sizeable number of patients do not gain adequate pain relief from these therapies. Predicting those patients who will not respond to current therapies remains a challenge. Although psychosocial, sensitivity, inflammation and genetic factors have been identified as pain mechanisms that predict response to pain therapy, none of these is a sufficiently strong predictor alone. An emerging approach to this challenge is the use of machine-learning algorithms that integrate several pain mechanisms, which are superior to previous prediction models. Importantly, these machine-learning algorithms identify networks of pain mechanisms that could be targeted therapeutically. From these models, a new mechanistic framework is proposed, in which pain in OA can be viewed either as a simpler joint disease with inflammation or as a complex pain problem that involves multiple factors. The latter is associated with a higher risk of poor response to standard OA pain therapy. Understanding the complexity of pain in OA and predicting those who are at risk of not responding to standard OA pain therapy are essential for improving the management of pain in OA.
Petersen et al. (Thu,) studied this question.