Use of inflammation-based scores to optimize the selection of patients with advanced cancer considered for early-phase clinical trials.

Abstract

e13033 Background: The presence of adequate organ function, favorable performance status (PS) and exhaustion of standard treatments are the main factors guiding accrual onto Phase I trials. As inflammation is inherent to the pathogenesis and prognosis of cancer we investigated the prognostic and predictive performance of inflammation based indices including the neutrophil (NLR) and platelet to lymphocyte ratio (PLR). Methods: Unselected referrals to the Phase I unit (2007-2011) were considered. Demographic, treatment, staging data and routine blood tests were collected. Patients were defined as high risk if NLR>5 or PLR>300 respectively. Changes in the NLR (ΔNLR) were recalculated at disease reassessment. Kaplan Meier and log rank test were used to assess Progression Free (PFS) and Overall Survival (OS) with each variable. Significant factors were further tested in a multivariate Cox model. C-index was used to calculate the discriminative ability of each score and ROC curves to predict 90 days survival (90DS). Results: We included 126 patients with median age 63 years (range: 22-80); median OS 4.4 months (0.2-39), 36% male; 92% with metastases, 23% PS>1. On univariate analysis LDH>450, PS, NLR≥5, hypoalbuminaemia, NLR normalization (ΔNLR) (p<0.001), >2 previous treatment lines (p=0.01) predicted for OS. After multivariate analysis low albumin, high LDH and ΔNLR remained independent predictors (p<0.05). Shorter PFS was predicted by elevated LDH and ΔNLR (p<0.05). PS and NLR ranked as most accurate predictors of both 90DS with area under the ROC curve values of 0.66 (0.55-0.77) and 0.64 (0.54-0.75) and OS with c-index scores of 0.69 (0.56-0.82) and 0.60 (0.50-0.70). When combined to PS, the NLR improved its accuracy to 0.72 (0.59-0.83). NLR≥5 at screening was associated with advanced PS, low albumin (p<0.002), LDH>450 and >2 metastatic sites (p<0.05). Conclusions: The NLR can improve the selection of patients considered for experimental therapies, with its normalization following treatment predicting for a survival benefit of 7 months in our series.