Neoadjuvant nivolumab alone or in combination with relatlimab or ipilimumab in resectable head and neck squamous cell carcinoma (HNSCC).

Abstract

6018 Background: PD-1 inhibition using nivolumab (nivo) monotherapy is modestly effective in metastatic HNSCC. Neoadjuvant trials using nivo may permit development of more effective combinations in surgically resectable HNSCC. We evaluated combinations with nivo plus additional immune checkpoints, CTLA-4 (ipilimumab, ipi) and LAG-3 (relatlimab, rela). Methods: Phase II randomized trial of neoadjuvant nivo alone (240 mg q2 weeks), or with ipi (1 mg/kg q3 weeks) or rela (160 mg q4 weeks) for 4 weeks prior to surgery. Response was scored using RECIST and standard pathologic response criteria. Patients were stratified by p16, PD-L1, and LAG-3, with staining assessed by immunohistochemistry. Freshly digested tumors were subjected to single cell RNA sequencing (scRNAseq) for T cell receptors and gene pathways to identify biomarkers of response or progression. The Cochran-Armitage trend test was used to explore associations with increasing pathological response efficacy. Results: 41 patients (pts) have been enrolled, with 33 evaluable for this analysis. Of these 33, median age is 63 (32-81), primary site oral cavity (n=25), oropharynx (n=5, 3 HPV), larynx (n=3), clinical T2 (n = 5), T3 (n = 12), T4 (n = 13), and cN0/1 (n = 22), cN2 (n = 9), and PD-L1 CPS > 1 (n=25). There were no serious study drug-related AEs or unexpected surgical delays/complications. Pathologic response (Table) was more frequent with nivo/rela (11/13) vs. nivo/ipi (6/10) or nivo (4/10). Partial (>50%) or major (>90%) pathologic responses were more frequent and deeper in the combination arms. Minor (<50%) pathologic responses were more frequent with nivo/rela, and similar between nivo/ipi and nivo. There was no association between RECIST response, PD-L1, or LAG3 and pathologic response. Combined PD-L1 and LAG3 expression was not associated with pathologic response in the nivo/rela arm, however more patients with combined positivity had a >50% response (4 vs. 0). Expansion of CD8+ T cells, as well as expanded proportion of CD8+ CXCL13+ T cells in responder tumors were identified in post-treatment specimens using scRNAseq. Conclusions: Neoadjuvant nivo/ipi or nivo/rela combinations were safe and associated with promising pathologic responses compared to nivo monotherapy. Anti-tumor CD8+ T cell populations and targetable pathways are emerging in responder patients. The trial continues to enroll and further evaluation of this strategy is warranted. Clinical trial information: NCT04080804 . [Table: see text]