Pretreatment spatial immune architecture and 1-year clinical benefit to first-line ICB in advanced melanoma.

Abstract

9539 Background: Predictive biomarkers for immune checkpoint blockade (ICB) in advanced cutaneous melanoma (CM) remain limited. Imaging mass cytometry (IMC) enables spatially resolved profiling of the tumor microenvironment (TME) and can capture immune–tumor architecture beyond cell density. Methods: Patients with advanced CM treated with first-line ICB and available pretreatment FFPE tissue were analyzed using IMC (40-marker panel). The primary endpoint was durable 1-year clinical benefit (1y-CB: metabolic CR/PR or stable metabolic disease ≥12 months by PERCIST on FDG-PET/CT); the secondary endpoint was 1-year progression-free survival (1y-PFS). Cell densities, tumor marker expression, and spatial neighborhood metrics were evaluated. Reproducibility was assessed by re-analyzing an independent published IMC melanoma cohort (PMC8026677). Results: Twenty-nine patients (anti–PD-1 n=27; ipilimumab+nivolumab n=2); 15 achieved 1y-CB. 1y-CB was associated with higher leukocyte infiltration and increased densities of B cells, conventional CD4⁺ T cells (Tconv), CD8⁺ T cells, and plasma cells, whereas higher tumor density was associated with lack of benefit and inferior 1y-PFS (HR 10.57; p<0.001). Higher CD8⁺ and B-cell densities were associated with improved 1y-PFS (HR 0.26, p=0.013; and HR 0.26, p=0.047). Tumor PD-L1 and HLA-DR expression correlated with 1y-CB (p=0.022 and p=0.040). Spatial metrics further stratified outcomes: increased tumor–tumor proximity was associated with poor 1y-CB (p=0.0016), while increased B–Tconv proximity was associated with 1y-CB and longer 1y-PFS (HR 0.20; p=0.037). The strongest spatial association was macrophage proximity to HLA-DR⁺ tumor cells (p≤0.0001). In the external cohort (n=60), responder enrichment for Tconv/CD8⁺ T cells and macrophage proximity to HLA-DR⁺ tumor cells were reproduced. Conclusions: Pretreatment IMC identifies both compositional and spatial TME features associated with durable benefit from first-line ICB in advanced CM. T-cell enrichment and macrophage proximity to HLA-DR⁺ tumor cells emerged as reproducible candidate spatial biomarkers warranting prospective validation. Main predictors of 1y-CB and 1y-PFS. Parameters 1y-CB events/N 1y-CB p-value 1y-PFS HR (95% CI) 1y-PFS p-value B cells ≥9.5 5/15 vs 9/14 0.026 0.26 (0.00–0.99) 0.047 CD8⁺ T cells ≥36 7/20 vs 7/9 0.023 0.26 (0.00–0.76) 0.013 Plasma cells ≥1 3/12 vs 11/17 0.017 0.28 (0.08–1.00) 0.050 Tumor density ≥4194 5/5 vs 9/24 0.006 10.57 (3.08–36.23) <0.001 B–Tconv proximity ≥13.3% 2/10 vs 10/16 0.018 0.20 (0.00–0.91) 0.037 HLA-DR ≥58.4% 1/7 vs 13/22 0.040 0.17 (0.02–1.28) 0.085 PD-L1 ≥3.3% 8/20 vs 6/9 0.022 0.48 (0.16–1.38) 0.171