Identifying a Novel Opioid-Specific EEG Biomarker That Can Be Targeted to Improve Patient Pain- and Opioid-Related Outcomes - PROJECT SUMMARY Over one million lives have been lost to the ongoing Opioid Overdose Crisis in the U.S. Recently it has been acknowledged that surgery and postoperative pain are major contributors to persistent opioid use and dependance. A study from our lab found that administering incrementally higher doses of opioids in the operating room could lead to clinically significant decreases in postoperative pain, opioid consumption up to 180 days after surgery, and chronic pain diagnoses. This work highlights that postoperative pain and opioid consumption could be minimized by optimizing intraoperative opioid administration, but this would require objective markers we could use to monitor opioid effects in real-time. Current opioid dosing guidelines rely on drug pharmacokinetics, but these are idealized models that do not account for an individual’s unique physiology. Pain medicine physicians suggest using respiratory depression as a correlate for opioid overdosing. However, most patients taking prescription opioids receive doses far below what is required to exhibit respiratory depression and remain at risk for addiction. EEG is an intriguing candidate to identify an opioid-specific marker because of the impact of opioids on brain signaling and recent discovery of a specific EEG signature that tracks intraoperative fentanyl administration in a dose-dependent fashion. Opioids are regularly used in the operating room where EEGs are collected as part of standard anesthesia monitoring. The goals of this project are to identify a unique intraoperative EEG marker that is specific for opioid administration and to determine whether interventions that influence such a marker can improve pain- and opioid-related outcomes. In Aim 1, a prospective observational study will be conducted to characterize and localize EEG markers for intravenous opioids used in surgery, including remifentanil, hydromorphone, and morphine. Time-frequency decompositions of intraoperative EEG recordings will identify if these opioids have a similar dose-dependent increase in theta-band power as seen for fentanyl. Aim 2 will characterize the relationship between opioid-related intraoperative EEG biomarkers and post-operative clinical outcomes. A statistical mediation analysis will be done to quantify how much the effect of intraoperative opioids on postoperative pain- and opioid-related outcomes is mediated by changes in patients’ theta oscillations. Aim 3 will use machine learning methods to identify baseline and modifiable risk factors that may influence postoperative pain and opioid requirements using a feature-rich data set of more than 100,000 patients. Transformer-based deep learning approaches will be used to isolate demographic factors, medical comorbidities, and medications given during surgery that significantly modify postoperative pain and opioid outcomes. Successful completion of this proposal would be highly significant as it will provide the basis for individualized intraoperative administration of opioids to optimize postoperative pain while minimizing excessive opioid use and potential for addiction.
