Immune phenotype profiling based on anatomic origin of melanoma and impact on clinical outcomes of immune checkpoint inhibitor treatment

• Published in: Journal of clinical oncology Vol. 42; no. 16_suppl; p. 9569
• Main Authors: Hwang, Soohyun, Ryu, Hyang Joo, Kim, Youhyun, Cho, Soo Ick, Oum, Chiyoon, Valero Puche, Aaron, Lee, Jinhee, Kim, Kyoo Hyun, Chung, Kee Yang, Shin, Sang Joon, Ock, Chan-Young, Jung, Minkyu, Shen, Jeanne
• Format: Journal Article
• Language: English
• Published: 01.06.2024
• ISSN: 0732-183X; 1527-7755
• DOI: 10.1200/JCO.2024.42.16_suppl.9569

9569 Background: The inflamed immune phenotype is associated with a favorable response to immunotherapy in metastatic malignant melanoma. Although variations in molecular features across melanoma subtypes and differing ethnic distributions have been well reported, studies investigating immune phenotypes across multiple institutions within the context of different melanoma subtypes are scarce. We examined the association of immune phenotypes with outcomes after immunotherapy, based on subtypes of malignant melanoma. Methods: Two institutional advanced melanoma cohortsrepresenting Asian and Western populations (YUHS, n = 123; Stanford University, n = 102) were included. H&E-stained tumor slides were assessed using Lunit SCOPE IO, an AI-powered whole slide image analyzer, to define the inflamed score (IS, % of inflamed area) and the inflamed phenotype (>33.3% IS). Clinical outcomes after immune checkpoint inhibitor (anti-PD-1/PD-L1) treatment, including progression-free survival (PFS) and overall-survival (OS), were assessed. Results: In the overall cohort (n = 225), the inflamed phenotype was associated with prolonged PFS (HR = 0.634, 95% CI: 0.453 – 0.887, p = 0.007) and OS (HR = 0.655, 95% CI: 0.432 – 0.992, p = 0.044). On analysis of melanoma subtypes, we identified non-acral cutaneous melanomas (CM, n = 94), acral melanomas (AM, n = 52), mucosal melanomas (n = 43), uveal melanomas (n = 16), and melanomas of unknown primary site (n = 20), with a disproportionate enrichment of AM (37.4%) and mucosal (29.3%) melanoma cases in the YUHS cohort and CM cases (68.6%) in the Stanford cohort. The IS varied across different melanoma subtypes, with the highest IS observed in CM (median, [IQR]; 34.0, [5.8–52.2]), and significantly lower scores in AM (22.3, [7.7–36.9]), mucosal (13.6, [0–31.4]), and uveal melanoma (5.5, [0–13.6]). Among CM patients, the inflamed phenotype was significantly associated with better post-immunotherapy PFS (HR = 0.476, 95% CI: 0.274 – 0.826, p = 0.007). In contrast, the inflamed phenotype did not correlate with survival in AM patients (p = 0.58), who exhibited universally poorer PFS compared to non-AM patients (HR = 2.124, 95% CI: 1.486 - 3.005). Conclusions: AI-powered immune phenotype assessment highlights a heterogeneity of immune phenotypes across specific melanoma subtypes, as well as differential association with outcomes after immune checkpoint inhibitor treatment. While CM patients were well-stratified by immune phenotype, the predictive value of immune phenotyping was less pronounced in acral melanomas, which demonstrated poorer outcomes following immunotherapy.