Friedrich Feuerhake

Background: Intrahepatic cholangiocarcinoma (iCCA) is characterized by marked molecular and microenvironmental heterogeneity and moderate response to immunotherapy. We aimed to define the genomic and spatial determinants of immune checkpoint activity in iCCA and their association with clinical outcome. Methods: We performed integrated transcriptomic, DNA panel sequencing, and multiplex immunohistochemistry analyses in a cohort of 144 patients with iCCA. Tumors were classified according to STIM-defined microenvironmental subtypes. Spatially resolved immune profiling was conducted across the tumor core, stroma, invasive margin, and adjacent liver. Immune checkpoint expression, cellular proximity, and clinical outcome were analyzed using Cox regression and Kaplan–Meier methods. TIGIT ^high tumors were further characterized by differential gene expression and pathway enrichment analyses. Results STIM classification stratified tumors into inflamed (47%) and non-inflamed (53%) subtypes with distinct survival outcomes and genomic profiles. Recurrent alterations in CDKN2A/B , TP53 , MTAP , and BAP1 showed subtype-specific enrichment and prognostic relevance. Spatial immune profiling revealed compartment-specific distributions of T cells and immune checkpoints, with predominance of TIGIT and PD-1 expression in stromal and marginal regions. Most tumor-infiltrating T cells lacked inhibitory receptor expression, while co-expression of multiple checkpoints was uncommon. TIGIT ^high tumors displayed coordinated activation of immune effector, antigen presentation, and interferon signaling pathways, alongside enrichment of distinct genomic alterations, defining an immune-activated molecular subtype. Conclusion: Integrated genomic and spatial immune profiling reveals pronounced context-dependent heterogeneity of immune checkpoint regulation in iCCA. TIGIT delineates an immune-activated subtype with distinct molecular features, providing a potential biomarker framework for patient stratification and rational immunotherapeutic targeting.

Supplementary Figure 4. Factor developed by pooling baseline biomarkers, including the expression of single genes by BC360, mIHC analysis in the tumor compartment, and sTILs measurement. Contribution (A) and loading (B) of each marker to the factor. Markers with contributions to the factor higher than specified by the dotted line in (A) were important in accounting for the variability in the factor. Loadings in (B) express the relationship of each important marker with the factor. Logistic regression with elastic net regularization analysis in (C) was performed with non-standardized values for the factor and clinical markers, and in (D) with standardized values from BC360 single genes, mIHC in tumor, and sTILs combined with clinical markers.

Purpose:We performed multiomic factor analysis of biomarkers potentially associated with pathologic complete response (pCR) after a chemotherapy-free combination of immunotherapy and dual ERBB2 (HER2) blockade.Experimental Design:Forty-eight patients with HER2 2+ (ISH+) or 3+ early breast cancer, stage I to III, and HER2-enriched subtype received pembrolizumab, trastuzumab biosimilar, and pertuzumab. The primary outcome was pCR (ypT0/is ypN0). Multiomic factors representing combined markers from pooling different translational research marker data were identified based on the eigenvalue >1 criteria. Additionally, logistic regressions with elastic net regularization were performed using all single markers, multiomic factors, and clinical characteristics to evaluate their association with pCR.Results:The multiomic factor involving genes related to immune response and tumorigenesis, multiplexed IHC markers related to immune response and HER2 in the stroma, and stromal tumor-infiltrating lymphocytes were associated with pCR. pCR rates were 66.7% and 28.6% in the high (>median) and low (≤median) factor score groups, respectively (unadjusted OR, 5.00; 95% confidence interval, 1.35–18.56; nominal P = 0.016). In the analysis using standardized values for all individual markers from gene expression analysis, multiplexed IHC analysis in the stroma, and stromal tumor-infiltrating lymphocytes, markers with the greatest impact on pCR were the genes FNBP1, CD36, MYCN, and SIX1 (OR between 1.43 and 1.62), and progesterone receptor status (OR, 0.65), which surpassed grade (OR, 1.22) and nodal status (OR, 0.90).Conclusions:Our multiomic analysis uncovered mechanisms of tumor response to immunotherapy combined with dual HER2 blockade. These results could inform the design of larger trials investigating chemotherapy-free regimens in selected patients with HER2+ early breast cancer.

Supplementary Table 2. Associations between single genes and pCR.

Supplementary Figure 3. Oncoplot for genes with somatic variants in baseline samples from patients with and without pCR.

Supplementary Table 1. Treatments administered in the adjuvant phase.

Supplementary Figure 6. Factor developed by pooling baseline biomarkers, including the gene signatures by BC360, mIHC analysis in the tumor compartment, and sTILs measurement. Contribution (A) and loading (B) of each marker to the factor. Markers with contributions to the factor higher than specified by the dotted line in (A) were important in accounting for the variability in the factor. Loadings in (B) express the relationship of each important marker with the factor. Logistic regression with elastic net regularization analysis in (C) was performed with non-standardized values for the factor and clinical markers, and in (D) with standardized values from gene signatures by BC360, mIHC in tumor, and sTILs combined with clinical markers.

Supplementary Figure 1. Mean sTIL levels at baseline (A) and week 3 (B) by pCR status. At week 3, sTILs could not be measured due to a low cellularity in 2 patients with non-pCR and in 10 patients with pCR; measurements were not performed in further two patients (one with and one without pCR).

Supplementary Figure 5. Factor 1 (A, B) and Factor 2 (C, D) developed by pooling baseline biomarkers, including the gene signatures by BC360, mIHC analysis in the stromal compartment, and sTILs measurement. Contribution (A, C) and loading (B, D) of each marker to each factor. Markers with contributions to the factor higher than specified by the dotted line in (A, C) were important in accounting for the variability in each factor. Loadings in (B, D) express the relationship of each important marker with each factor. Logistic regression with elastic net regularization analysis in (E) was performed with non-standardized values for each factor and clinical markers, and in (F) with standardized values from gene signatures, mIHC in the stromal compartment, and sTILs combined with clinical markers.

Supplementary Figure 2. Correlation within RNA expression signatures and baseline levels of sTILs. Square represents the Pearson’s correlation coefficients between the pairs of variables with the color intensity corresponding to the value of correlation as indicated on the scale. Hierarchical clusters are represented by a cluster tree. Magenta and green boxes indicate revealed clusters of gene signatures and sTILs.