A significant number of patients develop chronic pain after surgery, but prediction of those who are at risk is currently not possible. Thus, prognostic prediction models that include bio-psycho-social and physiological factors in line with the complex nature of chronic pain would be urgently required. Here, we performed a translational study in male volunteers before and after an experimental incision injury. We determined multi-modal features ranging from pain characteristics and psychological questionnaires to blood plasma proteomics. Outcome measures included pain intensity ratings and the extent of the area of hyperalgesia to mechanical stimuli surrounding the incision, as a proxy of central sensitization. A multi-step logistic regression analysis was performed to predict outcome measures based on feature combinations using data-driven cross-validation and prognostic model development. Phenotype-based stratification resulted in the identification of low and high responders for both outcome measures. Regression analysis revealed prognostic proteomic, specific psychophysical, and psychological features. A combinatorial set of distinct features enabled us to predict outcome measures with increased accuracy compared to using single features. Remarkably, in high responders, protein network analysis suggested a protein signature characteristic of low-grade inflammation. Alongside, in silico drug repurposing highlighted potential treatment options employing antidiabetic and anti-inflammatory drugs. Taken together, we present here an integrated pipeline that harnesses bio-psycho-physiological data for prognostic prediction in a translational approach. This pipeline opens new avenues for clinical application with the goal of stratifying patients and identifying potential new targets, as well as mechanistic correlates, for postsurgical pain.
Read full abstract