874P - Artificial intelligence (AI)-powered spatial tumor-infiltrating lymphocyte (TIL) analysis in recurrent/metastatic (R/M) head and neck squamous cell carcinoma (HNSCC) patients treated with immune checkpoint inhibitor (ICI) treatment

Background

ICI has demonstrated survival benefit and became a part of standard therapy for the systemic treatment of R/M HNSCC. However, the identification of practical and reliable biomarkers to predict therapeutic response remains an unmet need. AI-powered spatial TIL analysis has demonstrated significant predictive value of immunotherapies in multiple cancer types including squamous cell carcinoma. The objective of this study is to investigate the influence of AI-powered TIL analysis in predicting the response to ICIs in R/M HNSCC.

Methods

H&E-stained whole-slide images (WSIs) of pre-treatment tumor samples and clinical data were collected from r/m HNSCC patients treated with ICI at Severance Hospital in Korea. Lunit SCOPE IO, an AI-powered WSI analyzer, was applied to assess the density of intratumoral TIL (iTIL), stromal TIL (sTIL), and TIL in tumor microenvironment (tTIL). For clinical outcome analysis, the median densities of TILs were applied as cut-off (high vs low).

Results

Of 108 pre-treatment samples included, primary tumor sites were oral cavity (42.6%), oropharynx (27.8%), hypopharynx (15.7%), and other (13.9%). Overall, the median tTIL density was 392/mm² (interquartile range [IQR] 150-1095), and the median values of oropharynx and of non-oropharynx were 446/mm² (IQR 221-1836) and 326/mm² (IQR 109-1006), respectively (p=0.163). Objective response rate of ICI per RECIST v1.1 was higher in high-tTIL compared to low-tTIL (21.6% vs 5.7%; p=0.0220). Median progression-free survival (mPFS) of ICIs was favorable in high-tTIL than low-tTIL (3.2 vs 1.6 months; hazard ratio [HR] 0.59, 95% confidence interval [CI] 0.39-0.90, p=0.0151). Interestingly, mPFS was more significantly different by iTIL density (HR 0.64, 95% CI 0.42-0.97, p=0.0367) compared to sTIL density (HR 0.70, 0.46-1.07, p=0.0999). Overall survival was not significantly different by TIL densities.

Conclusions

AI-powered assessment of TIL density in tumor microenvironment, especially intratumoral TIL density, is associated with favorable treatment outcomes to ICI in R/M HNSCC.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

M.H. Hong: Financial Interests, Personal, Advisory Board: AstraZeneca, BMS, Roche; Financial Interests, Personal, Invited Speaker: Merck, Yuhan, Jassen, Takeda, Amgen. S. Kim: Financial Interests, Institutional, Full or part-time Employment: Lunit; Financial Interests, Institutional, Stocks or ownership: Lunit. C. Oum, Y. Lim, S. Song: Financial Interests, Institutional, Full or part-time Employment: Lunit. C. Ock: Financial Interests, Institutional, Stocks or ownership: Lunit, Lunit, Medpacto, Y-Biologics; Financial Interests, Institutional, Leadership Role: Lunit; Financial Interests, Institutional, Advisory Role: Lunit, Medpacto, Y-Biologics. All other authors have declared no conflicts of interest.