Tumor microenvironment exclusion¶

Confidently known¶

human Pre-existing CD8+ T-cell infiltrate at the invasive margin predicts pembrolizumab response in melanoma (Tumeh 2014)¹. The inflamed / excluded / desert taxonomy has been validated pan-tumor and anchors the cancer-immunity set-point framework (Chen & Mellman 2017)².
human mouse TGF-β signaling in peritumoral fibroblasts excludes T cells. Mariathasan 2018 in urothelial carcinoma (IMvigor210) defined the F-TBRS (fibroblast TGF-β response signature) in atezolizumab non-responders, whose T cells were trapped in collagen-rich stroma³. Tauriello 2018 established the mouse mechanism in a quadruple-mutant MSS CRC GEMM where TGF-β blockade unleashes Th1/CTL responses and renders liver metastases susceptible to anti-PD-L1⁴.

Study	N	Feature	Effect	95% CI / p	Method
Tumeh 2014	n=46 (metastatic melanoma on pembrolizumab; discovery n=46 + validation n=15)	pre-existing CD8⁺ T cells at the invasive tumor margin	response prediction (AUC) AUC ~0.9 in discovery cohort	—	IHC + multiplex IF + TCR-seq
Mariathasan 2018	n=298 (IMvigor210 atezolizumab-treated metastatic urothelial carcinoma)	fibroblast TGF-β response signature (F-TBRS)	response direction non-responders enriched for F-TBRS and T-cell-excluded phenotype	—	bulk RNA-seq

human TGF-β is biologically robust but has clinically disappointed. Bintrafusp alfa (TGFβRII trap / anti-PD-L1 bifunctional) failed pivotal trials — INTR@PID lung 037 discontinued 2021, biliary tract halted. Galunisertib + durvalumab, NIS793 combinations in MSS CRC — modest or negative. Likely reflects redundancy (TGF-β has pleiotropic roles), dose-limiting cardiotoxicity, or context-specific effects. A textbook translational gap between compelling biology and clinical benefit.
human IPRES underperformed as an independent predictor. Hugo 2016⁵ nominated an EMT/wound-healing/angiogenesis transcriptomic signature in innately resistant melanoma. Subsequent validation across independent cohorts has been mixed. IPRES overlaps heavily with TGF-β and stromal signatures and adds little beyond CD8, PD-L1, TMB, and pan-fibroblast measures.
mouse WNT/β-catenin activation as exclusion driver — stronger in mice than in patients. Spranger 2015 defined the CCL4 / Batf3+ DC recruitment mechanism⁶. Clinical correlations weaker; WNT-inhibitor + ICI combinations limited by toxicity.

Study	N	Feature	Effect	95% CI / p	Method
Hugo 2016	n=38 (pretreatment melanoma biopsies on anti-PD-1)	IPRES transcriptional signature (EMT / ECM / angiogenesis / wound-healing)	response direction IPRES-high enriched among non-responders	—	bulk RNA-seq + WES

mouse TROP2 associates with claudin-7 to regulate tight junctions and exclude T cells from TNBC⁷. TROP2 loss or hRS7 (sacituzumab govitecan antibody component) targeting restores infiltration and enhances anti-PD-1 efficacy. Reframes TROP2 from pure ADC target to a barrier-mechanism target. Single humanized TROP2 syngeneic model; needs broader preclinical and patient validation.
human hMENA-high TGF-β-driven CAFs define an ICT-resistance transcriptional program validated against OAK phase III in NSCLC (this period)⁸. Signature-level clinical anchoring but mechanistic causality is inferential from in vitro CAF experiments.

Study	N	Feature	Effect	95% CI / p	Method
Melchionna 2026	TCGA NSCLC + SU2C + OAK (NCT02008227)	hMENA⁺ TGF-β-driven CAF 9-gene signature	prognosis / ICT resistance direction signature-high → worse prognosis, ICT resistance	—	9-gene RNA signature

human Single-cell spatial transcriptomics defines six tissue niches that stratify neoadjuvant PD-1/PD-L1 response in cSCC better than PD-L1 IHC⁹. High antigen-presentation / B-plasma / inflammatory-keratinocyte niches enrich responders; proliferative-keratinocyte / low-APC-myeloid / fibroblast-rich-EMT niches dominate non-responders. Niche-based biomarkers may eventually outperform single-marker IHC.
human LOAd703 (CD40L/4-1BBL oncolytic adenovirus) + atezolizumab in anti-PD-1-refractory melanoma reprograms the myeloid compartment and restores ICI-responsive immune signatures in 24 patients¹⁰. Biomarker-level evidence; small single-arm; but few interventions reliably rescue refractory melanoma.

Study	N	Feature	Effect	95% CI / p	Method
Lee 2026	n=27 (3 cSCC cohorts incl. 2 phase II trials)	six spatial tissue niches (high-APC / B-plasma / inflammatory-keratinocyte vs. proliferative / low-APC-myeloid / fibroblast-EMT)	pathologic response prediction niche profiling outperformed PD-L1 IHC	—	single-cell spatial transcriptomics
Grauers 2026	n=24 (single-arm phase I/II, anti-PD-1-refractory stage IV melanoma)	intratumoral LOAd703 (CD40L/4-1BBL oncolytic adenovirus) + atezolizumab	immune-signature biomarker increased DC markers, T-cell infiltration, EM CD8⁺; decreased circulating Tregs	—	multi-parameter flow + TME transcriptomics

Inflamed vs. excluded vs. desert remains a useful pre-treatment mental model. A patient with high CD8 at the invasive margin but non-response may have a downstream exhaustion or metabolic problem; a patient with a desert tumor faces a different biology (likely WNT/β-catenin, STK11, low TMB, or low DC recruitment).
Do not count on TGF-β-targeting agents to rescue excluded tumors yet. The clinical data on bintrafusp alfa, galunisertib, NIS793 are disappointing despite strong preclinical rationale.