ABSTRACT: Abdominal pain is the primary driver of morbidity in chronic pancreatitis (CP) and affects approximately 90% of patients over the course of their disease with devastating effects on quality of life. Etiology of pain in CP is multi-factorial. Patients with evidence of pancreatic duct obstruction due to stones and/or strictures are offered invasive treatments such as endotherapy or surgical drainage to relieve pain. However, response to invasive treatments is unpredictable, and currently no clinical tool is available to identify patients who will respond to technically successful treatment. The lack of pain response is at least partially due to supraspinal central sensitization (SCS), a phenomenon of neuropathic and neuroplastic remodeling resulting from persistent pain stimuli. Quantitative Sensory Testing (QST), an investigative technique of standardized stimulations to test nociception (the neural signaling that encodes noxious stimuli and the downstream experience of pain), is used in other pain conditions to differentiate between patient subgroups to guide treatment. QST has the potential to change the management algorithm of patients with painful CP. Our preliminary data show that pancreatic QST (P-QST) can phenotype patients with CP into three groups: normal pain processing, segmental (T10 dermatome at the pancreas) sensitization, and widespread hyperalgesia (consistent with SCS). In this proposal, we will evaluate the ability of P-QST to predict response to invasive treatment for painful CP, and to develop a predictive model for individualized prediction of treatment response. Our specific aims are: Aim 1. Test the predictive capability of pre-treatment P-QST phenotype for pain improvement following invasive treatment for painful CP. Using pre-procedure P-QST, we will phenotype 150 patients undergoing clinically-indicated invasive treatment for painful CP at UPMC and Johns Hopkins University. Our primary outcome will be average pain score measured by Numeric Rating Scale at 6 months post-intervention. Aim 2. Incorporate P-QST with known and suspected patient, disease, and treatment- related factors to create a model for individualized prediction of response to invasive treatment. Using machine learning tools, we will develop a model that optimizes the prediction of probability of response to invasive treatment in individual patients. This will also determine the relative strength of P-QST as an overall predictor of treatment response. Aim 3. Augment the predictive model (Aim 2) with biochemical inflammatory markers to assess the potential to increase predictive capability for pain improvement following invasive treatment for painful CP. The predictive model developed in aim 2 will be further strengthened by incorporating serum neuroinflammatory markers at baseline. Our findings will be a major step toward development of individualized prediction of treatment response following invasive treatment for painful CP. They will lay the f...