Tumor Microenvironment-Guided Therapeutic Selection in Peripheral T-Cell Lymphoma: Non-Overlapping At-Risk Populations Targetable by Anti-TIM-3 and Anti-CCR4 Agents

Anti-TIM-3 antibodies (sabatolimab, cobolimab) and the anti-CCR4 antibody mogamulizumab represent emerging targeted treatment strategies for peripheral T-cell lymphoma (PTCL), yet biomarker frameworks to guide treatment selection are scarce. We investigated whether ligand-receptor RNA co-expression among immune cells present in the PTCL tumor microenvironment (TME) alongside the malignant T-cell clone identifies non-overlapping patient populations with prognostically significant, distinct targetable biology.

To determine the immune cell compositions within individual bulk tissue samples, the previously validated CIBERSORTx algorithm was applied to 193 PTCL tumors (GSE58445). Based on immune cell profiles, samples were grouped into three distinct TME subtypes (unsupervised hierarchical clustering; annotated based on dominant cell populations): TME-1 (Immune Quiet; enriched for CD4-naïve,T-cells and regulatory T-cells), TME-2 (Inflammatory; enriched for gamma-delta T cells, plasma cells, and M0 macrophages), and TME-3 (Cytotoxic/Innate; enriched for CD8+ T cells, activated CD4+, memory T-cells, and M2 macrophages). To identify biologically active and prognostically significant ligand-receptor signaling enriched within each TME subtype, gene expression of 85 ligand-receptor pairs (curated from established immune communication literature) was scored and tested within each TME, and for association with overall survival (Cox regression, n=162). Non-overlap between the two independently adverse pathways from multivariable modeling, CEACAM1/TIM-3 and CCL17/CCR4, was assessed by median-split co-stratification, TME distribution analysis, and pathway-specific survival.

TIM-3/CEACAM1 and CCL17/CCR4 communication scores were uncorrelated (ρ=0.016, p=0.84) and enriched in different microenvironments: TIM-3 in TME-3/Cytotoxic (FDR=3.93×10⁻⁷) and CCR4 in TME-2/Inflammatory (FDR=8.15×10⁻⁴). Among at-risk patients (either pathway elevated), 68% were driven by one pathway: TIM-3-high-only patients concentrated in TME-3 (55%) while CCR4-high-only patients concentrated in TME-2 (69%). Patients with both pathways elevated had the worst prognosis (median OS 0.77 years; HR=2.18, 95% CI 1.25-3.82, p=0.006 vs neither elevated). Four-group stratification was highly significant (p=0.0003). In multivariable analysis, both pathways were independently prognostic (TIM-3: HR=1.58, p=0.0005; CCR4: HR=1.33, p=0.012).

PTCL harbors two uncorrelated adverse ligand-receptor pathways (ρ = 0.016) that localize to distinct microenvironment subtypes: TIM-3/CEACAM1 in cytotoxic TME-3 and CCL17/CCR4 in inflammatory TME-2. Two-thirds of at-risk patients are driven by one dominant pathway, supporting TME-guided monotherapy selection: anti-TIM-3 agents (sabatolimab, cobolimab) for TME-3 patients and anti-CCR4 therapy (mogamulizumab) for TME-2 patients. The 32% of at-risk patients with both pathways elevated had the worst prognosis (median OS 0.77 years; HR = 2.18, p = 0.006) and represent candidates for combination strategies. Prospective validation of this biomarker-directed framework is warranted.