Prediction of postoperative opioid analgesia using clinical-experimental parameters and electroencephalography.

BACKGROUND: Opioids are often used for pain treatment, but the response is often insufficient and dependent on e.g. the pain condition, genetic factors and drug class. Thus, there is an urgent need to identify biomarkers to enable selection of the appropriate drug for the individual patient, a concept known as personalized medicine. Quantitative sensory testing (QST) and clinical parameters can provide some guidance for response, but better and more objective biomarkers are urgently warranted. Electroencephalography (EEG) may be suitable since it assesses the central nervous system where opioids mediate their effects.

METHODS: Clinical parameters, QST and EEG (during rest and tonic pain) was recorded from patients the day prior to total hip replacement surgery. Postoperative pain treatment was performed using oxycodone and piritramide as patient-controlled analgesia. Patients were stratified into responders and non-responders based on pain ratings 24 h post-surgery. Parameters were analysed using conventional group-wise statistical methods. Furthermore, EEG was analysed by machine learning to predict individual response.

RESULTS: Eighty-one patients were included, of which 51 responded to postoperative opioid treatment (30 non-responders). Conventional statistics showed that more severe pre-existing chronic pain was prevalent among non-responders to opioid treatment (p = 0.04). Preoperative EEG analysis was able to predict responders with an accuracy of 65% (p = 0.009), but only during tonic pain.

CONCLUSIONS: Chronic pain grade before surgery is associated with the outcome of postoperative pain treatment. Furthermore, EEG shows potential as an objective biomarker and might be used to predict postoperative opioid analgesia.

SIGNIFICANCE: The current clinical study demonstrates the viability of EEG as a biomarker and with results consistent with previous experimental results. The combined method of machine learning and electroencephalography offers promising results for future developments of personalized pain treatment.