Serum and tumor immune cell distribution as predictor of response to immune checkpoint blockade in solid tumors.

Abstract

e16029 Background: Immune Checkpoint Blockade (ICB) is established in many solid cancers often lacking robust biomarkers for response stratification. We explored the predictive role of neutrophil and lymphocyte counts as well as the immune-cell composition of the tumor neighborhood in patients with solid tumors. Methods: Clinical data, including neutrophil-to-lymphocyte ratio (NLR), from seven prospective trials investigating ICB in solid tumors were pooled (NCT03416244, NCT03409848, NCT03620123, NCT03193931, NCT03044626, NCT02917772 and NCT03219775). Tumor biopsies and peripheral blood samples were collected from 21 patients with esophageal squamous cell carcinoma (ESCC) before ICB (Nivolumab +/- Ipilimumab; NCT03416244). Tumor sections were stained by a 40-plex immunofluorescence panel (Akoya PhenoCycler) to examine spatial tumor-immune interactions associated with clinical outcomes as overall (OS) or progression-free survival (PFS) and best overall response (RECIST version 1.1). Peripheral blood mononuclear cells (PBMCs) were isolated and profiled by 10x single-cell RNA-sequencing to compare cellular (sub)types between ICB responders and non-responders. Results: Higher pretreatment NLR, defined as NLR > 3, predicted poorer OS (HR 0.4; 95% CI 0.30-0.53; p < 0.0001) in multivariate analysis of 693 patients, as well as worse PFS (HR 0.55; 95% CI 0.43-0.71; p < 0.0001), overall response rate (OR 2.01; 95% CI 1.32-3.05; p = 0.0012) and disease control rate (OR 2.31; 95% CI 1.34-3.99; p = 0.0026). Spatial analysis of ESCC patients´ tissue discovered the enrichment of a neutrophil-CD4 + regulatory T cell neighborhood in the tumor-adjacent tissue of ICB non-responders (p < 0.05, n = 6), while intratumoral macrophages, CD4 + /CD8 + T cells and dendritic cells were more abundant in responders (p = 0.0245, n = 9). scRNA sequencing of PBMCs validated correlation of T cell abundance with better survival (p < 0.05, responders vs non-responders), while myeloid cell abundance was reversely correlated (p < 0.05, responders vs non-responders). The myeloid cells showed significant differential expression of G0S2, BASP1, CXCL8, LYN , confirming prominent neutrophil presence in non-responders. Conclusions: Pro-tumoral inflammation and a lack of anti-tumoral immune response, reflected by a higher NLR in peripheral blood, and a neutrophil-CD4 + regulatory T cell neighborhood in the tumor-adjacent tissue can be used to predict ICB response in solid cancers, such as ESCC. A deeper understanding of underlying mechanisms of neutrophil activation in ICB might help to stratify patients in the future.