Prediction of the response to repetitive transcranial magnetic stimulation of the motor cortex in peripheral neuropathic pain and validation of a new algorithm

Motor cortex repetitive transcranial magnetic stimulation (M1-rTMS) induces analgesic effects in neuropathic pain, but not all patients are good responders, and no clinical predictors of the response have been identified. The present study aimed to develop and validate a simple and easy-to-use predictive algorithm for the individual response to M1-rTMS in peripheral neuropathic pain that may be potentially applicable to any chronic pain condition. This was based on a secondary analysis from a recent double-blind, placebo-controlled trial demonstrating the efficacy of high-frequency M1-rTMS against placebo-rTMS and rTMS of the dorsolateral prefrontal cortex in 149 patients with peripheral neuropathic pain. Baseline variables were entered in the model without preconception, and categorized into sociodemographic, pain, and psychological variables. Good responders to rTMS were defined based on 50% pain relief on average pain intensity (rated on a 0-10 numerical rating scale) at 25 weeks. Ridge regression, feature selection, and Monte Carlo cross-validation were used to build and validate a predictive model specific for the response to M1-rTMS at 25 weeks. The algorithm included 3 variables: 2 were psychological variables (depressive symptoms, magnification dimension of the Pain Catastrophizing Scale) and 1 was related to pain distribution (distal lower extremity pain). It demonstrated 85% sensitivity (P = 0.005) and 84% specificity (P < 0.0001) to predict a good response to M1-rTMS at 25 weeks. It was not predictive of the response to placebo or dorsolateral prefrontal cortex-rTMS. This simple and user-friendly algorithm may contribute to individualize treatment with M1-rTMS in patients with peripheral neuropathic pain in routine and in further clinical trials.