Antibiotic treatment prior to immune-checkpoint inhibitor therapy is a tumour-agnostic predictive correlate of response in routine clinical practice

Introduction: Broad spectrum antibiotic therapy (ATB) has been suggested to impair responsiveness to immune checkpoint inhibitors (ICPI) through modulation of intestinal microbiota. Evidence is however limited to clinical trial participants with non-small cell (NSCLC) and renal cell carcinoma (RCC). In this multi-centre study, we intended to validate the impact of ATB in patients (pts) treated with ICPI in routine practice, irrespective of tumour site. Methods: We analysed a large, multi-centre, prospectively maintained dataset of pts treated with ICPI in two academic centres. We documented timing and duration of ATB (+/-) administered within 1 month prior to ICPI treatment (pATB) or concurrently (cATB) until ICPI cessation. We evaluated Objective Response Rates (ORR) and Overall Survival (OS) across ATB+/- and studied the prognostic effect of ATB using Kaplan-Meier curves and multi-variable Cox regression. Results: We enrolled 196 pts with NSCLC (n = 119), Melanoma (n = 38), and other histotypes (n = 39) including head and neck, RCC/urothelial, and gastrointestinal cancers. The majority of pts were male (n = 137, 70%) with performance status 0-1 (n = 159, 84%) and a median number of 2 metastatic sites (range 0-7). Pts received mostly anti-PD-1/PD-L1 ICPI (n = 189, 96%) as first-line metastatic therapy (n = 120, 62%). Twenty-nine patients (15%) received pATB with penicillins (n = 22, 75%) for ≤7 days (n = 26, 89%). Sixty-eight pts (35%) received penicillin-based (n = 49, 72%) cATB for ≤7 days (n = 39, 88%). Respiratory tract infections were the commonest indication for both pATB (n = 16, 55%) and cATB (n = 38, 85%). pATB (p < 0.001) but not cATB (p = 0.76) was associated with worse OS (26 vs. 2 months, Hazard Ratio 7.4, 95% CI 4.2-12.9) and increased likelihood of primary refractoriness to ICPI (44% vs 81%, p < 0.001). pATB consistently worsened OS in NSCLC (26 vs. 2.5 months, p < 0.001), melanoma (14 vs 3.9 months, p < 0.001), and other tumours (11 vs 1.1 months, p < 0.001). Following multi-variable analyses, pATB (p < 0.001, HR 3.4, 95% CI 1.9-6.1) and response to ICPI (p < 0.001, HR 8.2, 95% CI 4.0-16.9) emerged as predictors of OS independent of histotype, tumour burden, and performance status. Conclusions: This study suggests that pATB exerts an independent detrimental effect on response and survival in unselected pts treated with ICPI in routine clinical practice. Mechanistic studies are urgently required to investigate ATB-mediated alterations of gut microbiota as a determinant of poorer outcome following ICPI treatment. (Figure Presented) .