A CTLA4–TIGIT Prognostic Signature and Checkpoint Blockade Efficacy in Oral Squamous Cell Carcinoma

Technical Summary: A CTLA4–TIGIT Prognostic Signature and Checkpoint Blockade Efficacy in Oral Squamous Cell Carcinoma

Problem Statement
While immune checkpoint inhibitors (ICIs) targeting PD-1 have improved outcomes for recurrent/metastatic head and neck squamous cell carcinoma (HNSCC), objective response rates remain limited to 15–20%. Co-inhibitory receptors CTLA4, TIGIT, and LAG3 are frequently co-expressed on exhausted CD8+ tumor-infiltrating lymphocytes (TILs) in HNSCC, yet their integrated role as a therapeutically tractable prognostic axis in oral squamous cell carcinoma (OSCC) has not been systematically evaluated. Existing prognostic models often rely on large gene panels (7–24 genes) lacking direct pharmacological tractability, and few integrate transcriptomic characterization with functional in vivo validation of specific checkpoint blockade.

Methodology
The study employed an integrated multi-omics and preclinical approach:

1. Transcriptomic Analysis: Bulk RNA-seq data from TCGA-HNSC (n=467) and microarray data from GEO (OSCC: GSE23558, n=32; OSF: GSE64216, n=8) were analyzed. Differential expression of CTLA4, TIGIT, and LAG3 was assessed using Wilcoxon rank-sum tests with Benjamini-Hochberg correction.
2. Immune Infiltration Estimation: The immunedeconv R package (integrating six algorithms: TIMER, xCell, MCP-counter, CIBERSORT, EPIC, quanTIseq) was used to estimate immune cell composition. Spearman correlations were calculated between checkpoint expression and CD8+ T-cell abundance.
3. Prognostic Modeling: LASSO Cox regression was applied to the three checkpoint genes to construct a risk score for overall survival (OS). Patients were stratified into high- and low-risk groups based on the median score.
4. Single-Cell Localization: Summary-level scRNA-seq data from the TISCH database (datasets HNSC_GSE103322, HNSC_GSE180268, OSCC_GSE172577) were used to map CTLA4 and TIGIT expression across nine annotated cell lineages.
5. In Vivo Validation: A syngeneic mouse model (SCC7 cells in C57BL/6 mice, n=6/group) was used to evaluate the efficacy of single-agent blockade. Mice received intraperitoneal injections of anti-CTLA4, anti-TIGIT, anti-LAG3, or anti-IgG isotype control (200 µg every 3 days for 4 doses). Tumor volume, weight, and body weight were monitored. Immunohistochemistry (IHC) for CD8 and FoxP3 was performed on day 21.

Key Results

• Expression and Correlation: In the TCGA-HNSC cohort, CTLA4 and LAG3 were significantly differentially expressed between tumor and normal samples (FDR < 0.05), while TIGIT was not. However, all three checkpoints showed a strong positive correlation with CD8+ T-cell abundance (CTLA4: r=0.465; TIGIT: r=0.611; LAG3: r=0.652).
• Prognostic Signature: LASSO regression yielded a two-gene risk score: Risk score = −0.0945 × TIGIT − 0.0738 × CTLA4. LAG3 was excluded from the final score (coefficient = 0). This score significantly stratified patients into high- and low-risk groups (Log-rank P = 0.00048; HR = 1.64, 95% CI 1.24–2.18), with the high-risk group exhibiting poorer overall survival.
• Single-Cell Localization: scRNA-seq data confirmed that CTLA4 and TIGIT expression is predominantly localized to CD8+ T cells, followed by Treg and NK cells, supporting the immune-cell origin of the bulk transcriptomic signal.
• In Vivo Efficacy: Single-agent blockade of CTLA4, TIGIT, or LAG3 significantly reduced tumor weight and Day-21 tumor volume compared to the isotype control (all Holm-adjusted P < 0.05). Anti-TIGIT treatment showed a smaller magnitude of effect compared to anti-CTLA4 and anti-LAG3. No significant differences in body weight were observed, suggesting tolerability in this model. IHC scores for CD8 and FoxP3 were numerically lower in treatment groups, though the study notes these findings are uninterpretable regarding immune cell density due to undefined scoring scales.

Significance and Claims
The authors claim to have identified a T-cell-associated CTLA4–TIGIT prognostic signal in HNSCC transcriptomic data that independently stratifies overall survival. They demonstrate that individual blockade of CTLA4, TIGIT, or LAG3 suppresses established OSCC tumor growth in a syngeneic mouse model.

Crucially, the paper maintains a modest stance regarding clinical utility:

• The CTLA4–TIGIT score is presented as a prognostic biomarker derived from a standard-therapy cohort (TCGA), not a validated predictive biomarker for immunotherapy benefit.
• The authors explicitly state that the score has not been validated in ICI-treated cohorts and that its predictive utility remains unassessed.
• The in vivo findings are characterized as preliminary evidence that each receptor contributes to immune suppression in the SCC7 model, motivating but not validating combination strategies.
• The study highlights that while the two-gene score is parsimonious and derived from druggable targets, it requires independent validation in OSCC-specific cohorts and prospective evaluation in ICI-treated patients before clinical translation.

The paper concludes that while the integrated analysis supports the biological relevance of the CTLA4–TIGIT–LAG3 axis, further preclinical evaluation of combination blockade and rigorous predictive validation are necessary prerequisites for clinical application.