P3.03-046 Prognostic Fibrinogen/Leucocyte Score at Diagnosis Predicts Survival and Benefit from Multimodality Treatment in MPM: Topic: Mesothelioma Clinical

The aim of this study was to identify and validate prognostic and predictive biomarkers in a large cohort of patients with malignant pleural mesothelioma (MPM). We performed a retrospective chart review, including all patients with histologically confirmed MPM, treated at two specialized centers between 1994 and 2014. The effect of different clinical and pathological characteristics and laboratory values on outcome was investigated by using uni- and multivariate logistic and cox regression models. Two-hundred ninety-one patients were enrolled (222 males and 69 females). Main histological subtype was epitheloid (n=199, 68%). Multimodality treatment, defined as macroscopic complete resection combined with chemotherapy and/or radiation therapy and/or intracavitary treatment, was performed in 134 (46%) patients. Median overall survival (OS) was 17.7 months from diagnosis. In the multivariate cox regression model, leucocyte count at diagnosis (continuous, hazard ratio (HR) 1.087, p=0.04), fibrinogen at diagnosis (continuous, HR 1.002, p=0.002), histological subtype (epitheloid vs. non-epitheloid, HR 0.064, p=0.006) and age (continuous, HR 1.035, p=0.001) remained as independently significant co-factors influencing OS. ROC curve analyses for predicting 1-year survival revealed an area under the curve (AUC) of 0.72 (p=0.001) for fibrinogen and 0.65 (p=0.001) for leucocytes. Dichotomizing fibrinogen and leucocytes at the median values (550 mg/dl and 8 G/l) revealed a sensitivity of 0.65 and 0.55 and a specificity of 0.69 and 0.61 for predicting 1-year survival, respectively. Combining dichotomized fibrinogen/leucocytes to an inflammation based prognostic score (none, one or both elevated) significantly influenced 1-year survival (p<0.001) and OS (score 0 vs. I, p=0.005; I vs. II, p=0.03). When introducing to the multivariate cox regression model, the fibrinogen/leucocytes score remained as independently prognostic for OS (I vs. O, HR 1.48, p=0.02; II vs. 0, HR 2.26, p<0.001). Strikingly, a significant predictive interaction between the fibrinogen/leucocytes score and treatment modality was observed (p<0.001). The inflammation based fibrinogen/leucocytes score predicts OS independently from sex, age, stage, subtype and treatment modality. Multimodality treatment including surgery increases survival selectively in patients with low fibrinogen/leucocytes score.