Mature tertiary lymphoid structures (mTLS) usually flag the lung cancers most likely to respond to checkpoint inhibitors — yet a large share of mTLS-positive patients never benefit, and why stayed unclear. Across an ICI-treated NSCLC cohort of 509 patients, spatial transcriptomics and multiplex immunofluorescence traced primary resistance to the tumor stroma: two cancer-associated fibroblast subsets, FAP+αSMA+ and MYH11+αSMA+, that drive CD8 T-cell exhaustion and regulatory T-cell infiltration. Both point to stromal biomarkers and targets to rescue immunotherapy.
Published in Cell Reports Medicine, this translational study was led by first author Florent Peyraud, MD — work carried out during his PhD at Explicyte — under the supervision of Prof. Antoine Italiano (Institut Bergonié and Gustave Roussy) and Alban Bessede, PhD (Explicyte). It draws on pre-treatment FFPE samples from the Bergonié Institute Profiling (BIP) precision-medicine study (NCT02534649) and was made possible through the imCORE Network collaboration supported by Roche/Genentech. Explicyte ran the spatial transcriptomic profiling (NanoString GeoMx Whole Transcriptome Atlas), the multiplex immunofluorescence and digital-pathology work — including TLS maturity scoring and the CAF, T-cell-exhaustion, and regulatory-T-cell panels — the regional transcriptomic profiling, and the supporting computational analyses that anchored the findings.
Explicyte performed the pathology-based TLS scoring and digital-pathology analysis across 509 ICI-treated NSCLC patients from the BIP study, using CD3/CD20/CD23 multiplex IHC on the Ventana Discovery platform with PhenoImager HT imaging to separate mature from immature TLS. mTLS were present in 28.7% of cases and predicted markedly better outcomes — 56.8% response vs 33.6%, median OS 28.4 vs 14.5 months, median PFS 10.8 vs 4.5 months (all p < 0.001) — independent of PD-L1 expression and genomic features.
Using the NanoString GeoMx Whole Transcriptome Atlas on six mTLS-positive tumors from patients with extreme responses (three responders, three with progressive disease), Explicyte profiled TLS, tumor, and stromal compartments separately via CD45/PanCK segmentation. Differential expression localized the responder/non-responder split to the stroma: non-responders showed fibroblast enrichment (p < 0.001) and elevated TGF-β and epithelial-mesenchymal transition signaling, while responders were marked by IFN-α and IFN-γ inflammatory pathways.
A 5-plex mIF panel (PanCK/CD8/FAP/MYH11/αSMA) across 77 patient samples quantified FAP+αSMA+ and MYH11+αSMA+ cancer-associated fibroblasts. Both subsets were denser in the stroma of non-responders: high FAP+αSMA+ CAF density corresponded to 57.6% non-responders vs 27.8% (p = 0.027) and shorter PFS (HR 0.35; 95% CI 0.17–0.72; p = 0.003), with a parallel pattern for MYH11+αSMA+ CAF (56.5% vs 36.4%, p = 0.038; PFS HR 0.36; 95% CI 0.16–0.79; p = 0.008).
Regional transcriptomic profiling of 40 mTLS-positive tumors plus mIF exhaustion and regulatory-T-cell panels on 64 patients separated the two mechanisms. FAP+αSMA+ CAF-high tumors carried inflammatory (IFN-α/IFN-γ) signatures and CD8 T-cell exhaustion, with denser intratumoral CD8+PD1+ cells co-expressing CD39, LAG3, TIGIT, and TIM3. MYH11+αSMA+ CAF-high tumors instead showed regulatory T-cell enrichment, with more stromal CD4+FoxP3+ cells expressing TIGIT and ICOS and a higher Treg-to-CD8 ratio.
The work reframes primary resistance in mTLS-positive NSCLC as a stromal problem — and names two specific fibroblast populations behind it, each with its own immunosuppressive signature.
For drug developers, the two CAF subsets are candidate stratification biomarkers and stromal targets: flagging mTLS-positive patients whose tumors are nonetheless fibroblast-rich could redirect them away from ICI monotherapy toward combination strategies. Because FAP+αSMA+ CAFs map to CD8 exhaustion and MYH11+αSMA+ CAFs to Treg infiltration, the data support pairing checkpoint blockade with exhaustion- or Treg-directed agents — the rationale behind the upcoming INDIGO trial of atezolizumab plus the anti-TIGIT tiragolumab in PD-L1-low mTLS-positive NSCLC.
