Axitinib plus avelumab for recurrent/metastatic adenoid cystic carcinoma (R/M ACC): Biomarker analysis of the phase II trial.

6104 Background: In the phase II axitinib (VEGFR inhibitor) plus avelumab (PD-L1 inhibitor) trial in R/M ACC, response rate was 18% with a median progression-free survival (PFS) of 7.3 months, with subsequent inclusion of the combination in NCCN guidelines as a possible treatment for ACC (category 2B). We sought to identify potential biomarkers predictive of axitinib plus avelumab benefit in R/M ACC. Methods: The phase II trial included 28 R/M ACC patients (pts). Pre-treatment tumors were analyzed by whole exome sequencing (n=24), using Twist Human Core Exome V2 kit, and transcriptome profiling (n=26), using HTG Transcriptome Panel (19,616 probes). For 17 pts, imaging mass cytometry (IMC) was performed with 35 metal-tagged markers and analyzed with Visiopharm software. Microbiome composition was assessed in tumor (n=20), oral rinse (n=20) and fecal (n=19) samples via 16s rRNA gene sequencing. All results were assessed for association with PFS on therapy with a Cox proportional hazards model, maintaining ACC subtype (ACC-I vs. ACC-II) as a covariate. Results: Only 1 (6%) of 17 assessed tumors were PD-L1 positive. By WES, tumor mutational burden (TMB) was overall low (median 1.4 mut/Mb, range 0.7 – 18.7), but higher TMB was associated with worse PFS. Four mutational signatures significantly correlated with worse PFS, including SBS86, associated with previous chemotherapy exposure. Immune deconvolution of RNA-seq data revealed that a higher neutrophil-to-lymphocyte ratio and more T follicular helper cells associated with worse PFS, while presence of cytotoxic cells, T effector memory cells, and mast cells associated with longer PFS. Notably, a 167-gene predictive signature was developed and validated in the phase II ipilimumab plus nivolumab salivary gland cancer trial. This 167-gene signature was not prognostic in ACC pts who did not receive immunotherapy. IMC single-cell immune mapping revealed that pts with longer PFS had increased presence of CD8 T cells, CD73+ macrophages in tumor, a higher stromal CD8 T cell:Regulatory T cell ratio, increased stromal SIGLEC15 positive cells, and a higher tumor-to-stroma ratio of fibroblasts. Conversely, Ki67-positive T cells, cancer stem cells, and M2 macrophages were associated with worse PFS. Through microbiome analysis, 13 bacterial genera in tumor samples, 25 in stool, and 8 in oral rinse were significantly associated with therapy outcomes. Of note, in stool microbiome, presence of Akkermansia spp. and Bifidobacterium spp. were significantly correlated with improved PFS, with the latter association also seen in oral rinse. Conclusions: Correlative analysis of the phase II axitinib plus avelumab trial revealed potential biomarkers predictive of combination clinical benefit in ACC, including a gene-expression signature. These findings may guide patient stratification for combinatorial therapy.