Prognostic utility of inflammation biomarkers in a PDAC trial involving the human anti-CTGF antibody pamrevlumab.

Abstract

752 Background: Pancreatic ductal adenocarcinomas (PDAC) often exhibit desmoplasia, elevated CTGF (connective tissue growth factor) expression and inflammation. The influence of inflammation on patient outcomes was examined in a dose-ranging trial of the anti-CTGF antibody pamrevlumab in combination with a fixed regimen of gemcitabine and erlotinib in locally advanced or metastatic PDAC patients (NCT01181245; Picozzi et al. J Cancer Clin Trials 2017 2:123; n = 75). Methods: The prognostic utility of pre-treatment plasma levels of C-reactive protein (CRP), transforming growth factor β1 (TGFβ1), albumin, CTGF and CA 19-9 were assessed by univariate and multivariate Cox analysis. Demographic parameters, treatment cohort, and pamrevlumab exposure determined on treatment day 15 were also evaluated, as were changes in biomarker levels over the first four weeks of treatment. Results: Elevated baseline CTGF and CRP were prognostic for shorter overall survival (OS) by univariate analysis (HR = 3.2 for CRP > 10 mg/L, p= 0.00002 and HR = 1.6 for CTGF > 10 ng/mL, p= 0.045). In a five-factor multivariate Cox model that included CRP and TGFβ1 as continuous Ln-transformed variables, performance status, age, and pamrevlumab treatment cohort, cohort assignments associated with increasing pamrevlumab exposure predicted improved OS (HR = 0.87, p= 0.03). Removing CRP from this model reduced the prognostic utility of pamrevlumab cohort assignment and exposure, indicating an important contribution of inflammation to interpretation of treatment outcome. Changes in inflammation biomarkers over the course of treatment were also evaluated, but were not prognostic in this study. Conclusions: In multivariate Cox models, assessment of pre-treatment CRP levels improved ability to detect significant differences in PDAC patient survival outcomes associated with pamrevlumab treatment. Our results emphasize the utility of accounting for pre-treatment CRP levels as an independent prognostic factor in PDAC treatment effect models.