A Metabolomic Signature For Predicting Chemosensitivity In Gastric Cancer.

e15504 Background: Perioperative chemotherapy (QT) with platinum and fluoropyrimidines with or without anthracyclines is recommended option in patients with resectable gastric cancer (GC) at least cT2 or nodal involvement. Another option is surgery followed by QT with radiotherapy (QT/RT) or QT without RT in patients with D2 lymphadenectomy. Unfortunately, a considerable percentage of patients progress during neoadjuvant-QT (neo-QT) and some cases become inoperable cancer. These patients could benefit from curative surgery after diagnosis without neo-QT. Currently, histological/molecular markers have not been established to predict which patients can benefit from neo-QT. As potent analysis method, study of blood metabolites of resectable GC patients to establish a profile to differentiate responder patients (R-P) or not-responder (NR-P) to neoadjuvant-QT is promising. To establish a metabolomic profile or metabolomic signature and correlate with chemosensitivity, defined as pathological and clinical response is our endpoint. Methods: To this end we performend an untargeted metabolomic analysis by LC-HRMS of serum samples from resectable GC patients before neo-QT (n = 20 vs n = 10 healthy controls). Chemosensitive tumors were defined as those with good pathological response (Mandard 1 or 2) and partial response by TAC and chemoresistance tumors, defined as those with poor pathological response (Mandard 5) or/and progression by TAC. Reverse phase and HILIC chromatographic modes were applied to deal with highly polar as well as hydrophobic as required for untargeted metabolomics. For identification of potential biomarkers, we used in combination 2 independent variable selection techniques: principal component analysis and Student t test. Results: 11 patients were R-P and 9 patients were NR-P. We observed differences in metabolic profile between patients with GC & healthy controls and R-P & NR-P to neo-QT. Seven identified metabolites contributed most to the differentiating between R-P and NR-P. Conclusions: There are different metabolomic phenotypes among patients R-P and NR-P to neo-QT. It is necessary to validate a metabolomic signature to allow effective chemosensitivity prediction in patients with resectable GC.