Abstract Background No tools exist to individualize treatment selection in Inflammatory bowel disease (IBD) to optimize outcomes. We prospectively recruited patients prior to initiation of molecularly targeted (advanced) therapy (MTT) to identify predictive biomarkers of response and define pathways impacted by successful treatment. Methods Serum was collected before initiation of treatment in all patients, and at visit 1 (V1~16 wks) for a subset of patients. Serum samples were analyzed for >140 protein biomarkers (OLINK) and >1200 lipid metabolites. Whole exome sequencing and GWAS data was generated for all participants. Initial data outline proteomic data. Results Of the first 166 IBD patients analyzed, therapies initiated included TNF antagonists (n=85), ustekinumab (n=43), vedolizumab (n=28) and janus kinase inhibitors (n=10). Bio-naïve (n=96) and patients using baseline corticosteroids (n=77) were included. Sixty-three patients achieved remission. In TNF-antagonist initiations, serum TNF concentrations increased between baseline and V1 (p=4E-3), while it decreased for other MTTs (p=0.07). In bio-naïve patients, serum TNF concentration had a strong correlation with CXCL10 (r=0.58, P=3E-10). From the Principal Component Analysis of analytes correlated to TNF in bio-naïve patients, we built a proxy for the level of active TNF. At baseline, compared to TNF antagonist naïve patients, significantly higher serum TNF concentrations existed in patients with prior TNF-antagonist exposure >1-month post-cessation and a markedly high TNF concentration with cessation <1 month prior to baseline (p< 2E-16). In patients with prior TNF-antagonist exposure, serum TNF concentrations did not correlate with other cytokines (incl. CXCL10). In these patients, elevated serum TNF were found at V1, suggesting accumulation of inactive (complexed) TNF with treatment. A similar pattern was observed for IL-12 after ustekinumab treatment (elevated serum IL-12 with recent ustekinumab (p= 8E-13), large increase in IL-12 after ustekinumab (p=9E-5)). Reduced IL-12 concentrations existed in current corticosteroid users, compared to previous users and those who never used (p=4E-8). In baseline samples in bio-naive patients prior to therapy initiation, serum concentrations of six analytes provided a discriminant model of corticosteroid use, whith an area under (AUC) the receiver operation characteristic (ROC) curve showed excellent discrimination (AUC=0.91). Conclusion These data uniquely characterize proteomic data in patients with various advanced therapies and will enable development of predictive panels. Unique availability of multiple therapies linked to longitudinal proteomic data may allow personalized panels for treatment selection.