Work

Dual-method immune deconvolution identifies subtype-specific correlates of PD-L1 expression and a γδ T Cell–M2 macrophage association in breast cancer

Year: 2026

Type: article

Abstract: Most computational studies of the breast cancer immune microenvironment rely on a single deconvolution algorithm. Since different tools analyze different parameters, integrated information across a dataset can be missed. Here, we employed two complementary approaches, xCell and a CIBERSORT-style ssGSEA using LM22 signatures, on 1,099 PAM50-classified primary tumors from the TCGA-BRCA cohort. PD-L1 differed significantly across subtypes (Kruskal–Wallis p = 4.22 × 10⁻⁹), with the highest median expression in HER2-enriched tumors (log2CPM = 2.81) followed by basal-like (2.60). Within basal-like tumors PD-L1 was correlated with CD8+ T cells (ρ = 0.65) and M1 macrophages (ρ = 0.67), a pattern consistent with, though not demonstrative of, IFN-γ-driven adaptive upregulation. In HER2-enriched cancers, PD-L1 tracked with both effector and regulatory populations simultaneously, while luminal tumors were largely immune-quiet. All correlations were corrected for multiple comparisons using the Benjamini–Hochberg procedure; associations involving NK cells and γδ T cells were not significant after correction in most subtypes. The most notable methodological difference involved γδ T cells and M2 macrophages: xCell showed a non-significant correlation (ρ = 0.048, p = 0.11), whereas CIBERSORT-ssGSEA produced a significant correlation (ρ = 0.565, p < 2.2 × 10⁻¹⁶) that was observed across all five subtypes. A multivariate model explained 49% of PD-L1 variance (adjusted R² = 0.49). In a multivariable Cox model stratified by stage and PAM50 subtype and adjusted for age, no immune feature remained independently associated with overall survival after false discovery rate correction (all FDR > 0.20), indicating that the modest prognostic signal observed in unadjusted models is attributable to clinical covariates. These analyses are correlational and derived from bulk transcriptomes; they cannot establish directionality or causation, and functional validation would be required before any of these associations could be considered therapeutically actionable.

Source: In Silico Research in Biomedicine

DOI: 10.1016/j.insi.2026.100577

Authors: Diya Jain, Hari S. Misra

Institution: Narsee Monjee Institute of Management Studies

Language: English

Topic: Cancer Immunotherapy and Biomarkers

Subfield: Oncology

Field: Medicine

Domain: Health Sciences

SDG: Good health and well-being

Open Access status: gold

Estimated APC paid: $2,000