Related Experiment Video
Updated: Oct 10, 2026
Dual-modality Molecular Cartography: Integrating Multiplex mRNA Detection with Protein Imaging Mass Cytometry
Published on: November 14, 2025
Multimodal characterization of HLA-G expression across solid tumors reveals transcript-protein discordance and
Alexandra Epp1, Martha L Serrano-Serrano2, Florian Heil1
1Roche Pharma Research and Early Development, Early Development Oncology, Roche Innovation Center Munich, Penzberg, Germany.
Introduction:
Human leukocyte antigen-G (HLA-G) is a non-classical major histocompatibility complex class I molecule that functions as an immune checkpoint via inhibitory receptors such as immunoglobulin-like transcript 2 (ILT2), immunoglobulin-like transcript 4 (ILT4) and killer cell immunoglobulin-like receptor 2DL4 (KIR2DL4), suppressing anti-tumor immunity. Its prevalence and regulation across solid tumors remain incompletely defined due to low expression, spatial heterogeneity, and detection challenges.
Methods:
We performed a multimodal target prevalence analysis integrating data from a validated immunohistochemistry (IHC) assay for protein detection and prospective screening data from a phase 1 study of an HLA-G × CD3 T cell bispecific antibody (Clintrials.gov identifier: NCT05769959) with bulk RNA sequencing from early-phase oncology clinical trials.
Results:
Analysis of 252 patient tumor samples using an investigational HLA-G IHC assay showed that 59.9% were HLA-G positive, with highest prevalence in ovarian cancer (89%), non-small cell lung cancer (NSCLC) (76%), renal cell carcinoma (RCC) (71%), and pancreatic cancer (70%). Expression was generally low to moderate and spatially heterogeneous. Notably, discordance between transcript and protein expression was observed in melanoma, where high mRNA levels were not matched by detectable protein, suggesting post-transcriptional regulation. Prospective screening in a phase 1 cohort confirmed a high target positivity rate (77.5%). At the transcript level, HLA-G expression was enriched in inflamed tumors and correlated with interferon gamma (IFN-γ) signaling, antigen presentation pathways, and effector T-cell markers including cluster of differentiation 8 alpha (CD8A), interferon gamma gene (IFNG) and granzyme B (GZMB).
Conclusions:
HLA-G appears to be an inducible immune checkpoint associated with IFN-γ-driven immune activation and adaptive resistance. Transcript-protein discordance and spatial heterogeneity highlight the need for protein-level assessment in biomarker development. These findings inform companion diagnostic strategies for HLA-G-directed immunotherapies.
