916P - Deep spatial profiling of head and neck squamous cell carcinoma offers insights into the tumor microenvironment of hpv-stratified patients

Background

Head and Neck squamous cell carcinoma (HNSCC) is an aggressive, biologically heterogeneous disease. Human papillomavirus (HPV) is a major risk factor for HNSCC development and HPV status correlates with prognosis and aids in the selection of optimal therapy. While HPV-positive (HPV+) HNSCC patients have better prognosis, the mechanism underlying this improvement is unknown. We employed a deep learning multiplex immunofluorescence (mIF) pipeline to assess cellular and spatial differences between HPV+ and HPV-negative (HPV-) HNSCC tumors, to better understand how HPV status affects HNSCC biology.

Methods

We profiled 16 HNSCC biopsies (7 HPV+ and 9 HPV- samples) with a 37-plex mIF panel and a corresponding H&E slide. The panel consisted of core immune cell markers, immune activation status markers, as well as structural markers. Employing a deep learning mIF analysis pipeline, we identified 14 distinct cell types based on known lineage marker expression. For tissue segmentation, we employed a deep learning-based area model transferred from the corresponding H&E image. Each slide was segmented into tumor and stromal areas, which were further classified into tumor core, outer tumor rim, adjacent stroma, and remote stroma. Spatial features were extracted based on regions, cell types and phenotypic markers positivity, and were compared using Mann-Whitney U test between HPV+ and HPV- samples to identify the top differentially expressed features.

Results

Table: 916P

Conclusions

Deep spatial analysis unveiled significant differences between HPV+ and HPV- HNSCC. Enrichment of lymphocytic infiltrates with elevated expression of inhibitory immune checkpoints in the tumor and TME of HPV+ patients, suggests potential benefits from immune checkpoints combinations. Conversely, HPV- patients exhibited a higher proportion of stromal and suppressive myeloid cells in their TME, indicating potential benefits from targeting these cell types.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Gilead Sciences.

Disclosure

A. Aggarwal: Financial Interests, Institutional, Full or part-time Employment: Gilead Sciences. H. Zong: Financial Interests, Personal, Stocks/Shares: Gilead Sciences. P. Bimbaum, B. Arbiv, Y. Shachaf, S. Bookstein, R. Elran, A. Laniado, G. Golan, K. Bloom: Financial Interests, Personal and Institutional, Stocks/Shares: Nucleai L.J. Diehl: Financial Interests, Institutional, Stocks/Shares: Gilead Sciences. E. Markovits: Financial Interests, Personal and Institutional, Stocks/Shares: Nucleai.