86 EVALUATION OF CLINICOGENETIC RISK MODELS FOR OUTCOME OF FOLLICULAR LYMPHOMA PATIENTS IN THE PRIMA TRIAL S. Huet 1 * | E. Szafer‐Glusman 2 | L. Xerri 3 | C. Bolen 2 | E. Punnoose 2 | L. Tonon 4 | H. Tilly 5 | P. Brice 6 | P. Feugier 7 | B. Tesson 8 | A. Viari 4 | J.M. Venstrom 2 | G. Salles 9 1 Laboratoire d'Hématologie Sud, Hospices Civils de Lyon, Pierre‐Bénite, France; 2 Oncology Biomarker Development, Genentech, South San Fransisco, San Fransisco, USA; 3 Department of Bio‐Pathology, Institut Paoli‐Calmettes, Aix‐Marseille University, Marseille, France; 4 Plateforme de Bioinformatique 'Gilles Thomas', Synergie Lyon Cancer, Lyon, France; 5 Inserm U1245, Henri Becquerel Comprehensive Cancer Center, Rouen, France; 6 Service d'hématologie Hopital Saint –Louis, Assistance Publique‐ Hopitaux de Paris, Paris, France; 7 Hematology, Nancy University Hospital, Vandoeuvre‐les‐nancy, France; 8 Département de Bio‐statistiques, Institut Carnot‐Calym, Pierre‐Bénite, France; 9 Service d'hématologie Marcel Bérard, CHLS, Hospices Civils de Lyon, Pierre‐Bénite, France Introduction: Composite scores integrating genes mutation status with the clinical predictor FLIPI have been recently proposed to improve risk stratification for follicular lymphoma (FL) patients. We evaluated the ability of m7‐FLIPI and POD24‐PI scores to predict progression free survival (PFS) in a large cohort of patients receiving first‐line immunochemotherapy, with or without rituximab maintenance. Methods: Tumour biopsies were obtained at FL diagnosis from 252 patients from the PRIMA study, either as FFPE tissues (n = 98) or fresh‐frozen tissues (n = 154). After DNA extraction, DNA‐targeted sequencing was performed using the Foundation One Heme™ panel. m7‐FLIPI and POD24‐PI models were applied as originally described. Results: The frequency of non‐silent mutations were similar to those previously reported: CREBBP = 75%, EZH2 = 28%, CARD1 = 19%, ARID1A = 19%, EP300 = 16%, FOXO1 = 16% and MEF2B = 12%. We first evaluated the prognostic value of each gene mutation status separately. While some mutations were associated with a longer (MEF2B, EZH2) or shorter (EP300, CREBBP) PFS as previously described, ARID1A or CARD11 mutations were not associated with patients outcome. Moreover, FOXO1 mutations were associated with a good outcome, in opposite to their prognostic weight in the m7‐FLIPI. The m7‐FLIPI and FLIPI scores classified 28% and 43% of patients as high‐risk, respectively. m7‐FLIPI correlated with PFS (p = 0·005; OR = 1.74, 95%CI: 1.00‐3.02), slightly outperforming the FLIPI 96 ABSTRACT