P3.02c-051 A Pre-Treatment Serum Test Based on Complement and IL-10 Pathways Identifies Patients Benefiting from the Addition of Bavituximab to Docetaxel: Topic: IT

Abstract

SUNRISE, a global, double-bind, Phase III trial of docetaxel (D) plus bavituximab (B) or D plus placebo (P) in previously treated non-squamous non-small cell lung cancer, demonstrated similar overall survival (OS) in both treatment arms. Mass spectrometry and correlative analysis were used to create a test able to identify a subgroup of patients benefitting from the addition of B to D. Pre-treatment serum samples were available for 197 of the first 200 subjects enrolled in the trial. Mass spectra could be generated for 193 samples using the Deep MALDI method (Duncan et al, ASMS 2013), processed and features (peaks) identified. Mass spectral (MS) features associated with various biological functions were identified using a gene set enrichment analysis approach. Analysis of scores based on these MS feature, subsets indicated that in patients with high complement activation outcome depended on IL-10 activation in D+B but not in D+P. A test using the MS features associated with these functions was created to reliably identify a patient subgroup associated with clinical benefit using modern machine learning methods. Complement activation, as assessed by a classifier trained using related MS features, was a prognostic factor in both treatment arms, with high activation associated with poorer clinical outcome (OS HR = 0.54, log-rank p = 0.013 for D+B; OS HR = 0.60, log-rank p = 0.040 for D+P). Within the subgroup with high complement activation [N=50 (D+B); N=54 (D+P)], a second classifier using features related to IL-10 activation was able to isolate a subgroup of patients showing numerical benefit from the addition of B [median OS 5.9 months (D+P), 12.5 months (D+B)]. The remaining subgroup showed no benefit from addition of B [median OS 10.4 months (D+P), 5.6 months (D+B)]. Blinded validation of the test in the remainder 397 patients randomized in SUNRISE is will be presented. Proteomic and correlative approaches identified complement activation and low IL-10 levels as important pathways for predicting improved outcomes of patient treatment with D+B, in line with preclinical work on B’s mechanism of action. The test resulting from this work will undergo blinded independent validation.