Crit Rev Oncol Hematol. 2026 Mar 27;222:105305. doi: 10.1016/j.critrevonc.2026.105305. Online ahead of print.

ABSTRACT

The management of hepatocellular carcinoma (HCC) faces substantial and evolving challenges, driven by its aggressive biology, drug resistance, and the clinical urgency to detect recurrence. The treatment paradigm has undergone a profound transformation, evolving from surgical interventions and molecular targeted agents to the current era dominated by immunotherapy. Immune checkpoint inhibitors, particularly when used in combination with anti-angiogenic drugs or as part of dual-checkpoint blockade regimens, have established a new first-line standard of treatment for advanced HCC, delivering unprecedented survival improvements. Despite this progress, significant obstacles remain, including primary and acquired resistance, variable patient responses, and notably reduced efficacy in specific etiological subgroups. This comprehensive review synthesizes the emerging modalities such as bispecific antibodies, adoptive cell therapies, and innovative rational combinations that integrate systemic immunotherapy with locoregional treatments or novel targeted agents. Furthermore, we delve into the critical search for predictive biomarkers, encompassing liquid biopsy and multi-omics approaches, and dissect the complex cellular and molecular mechanisms underlying therapeutic resistance within the immunosuppressive tumor microenvironment. Finally, we outline future translational directions, emphasizing the expansion of immunotherapy, the development of tailored strategies for therapy-resistant disease, and the imperative move towards a personalized, biomarker-driven treatment framework. This review provides a cohesive overview of the field and charts a roadmap for future research to overcome the current challenges in HCC immunotherapy.

PMID:41905572 | DOI:10.1016/j.critrevonc.2026.105305

Cancers (Basel). 2026 Mar 9;18(5):878. doi: 10.3390/cancers18050878.

ABSTRACT

BACKGROUND: Lung adenocarcinoma (LUAD) and squamous cell carcinoma (LUSC), the major subtypes of non-small cell lung cancer (NSCLC), exhibit distinct molecular landscapes that demand precision in prognosis and therapy. While deep learning models can achieve high predictive accuracy, their black-box nature limits clinical translation.

METHODS: We introduce AttentioFuse, an interpretable deep learning framework employing a Reactome-guided mid-fusion strategy for multi-omics integration. AttentioFuse builds on three pillars: (i) dual-phase learning with omics-specific encoders to preserve modality-unique patterns, (ii) hierarchical attention mechanisms (cross-omics, feature-level, and fusion-layer) to quantify layer contributions dynamically, and (iii) integrated explainability combining DeepSHAP and global attention weights for gene-to-pathway interpretation. Two depth variants are instantiated under identical priors: a three-layer configuration (3F) for main discrimination and a five-layer configuration (AttentioFuse-5X) for deeper hierarchical interpretation; the 5X variant is trained end-to-end and yields comparable accuracy while enhancing pathway-level resolution.

RESULTS: Evaluated on The Cancer Genome Atlas (TCGA) LUAD/LUSC cohorts, AttentioFuse matches state-of-the-art performance in TNM staging while uncovering actionable biological insights, including pan-NSCLC AKT/mTOR metabolic control, histology-divergent Notch signaling roles, and additional pathways related to developmental reactivation, microbiota-associated metastasis, and extracellular matrix remodeling.

CONCLUSIONS: By design, AttentioFuse-5X bridges predictive performance with hierarchical, pathway-resolved explanations, advancing oncology by transforming black-box predictions into biologically grounded decision support.

PMID:41827812 | PMC:PMC12985206 | DOI:10.3390/cancers18050878

Cancers (Basel). 2026 Mar 9;18(5):878. doi: 10.3390/cancers18050878.

ABSTRACT

BACKGROUND: Lung adenocarcinoma (LUAD) and squamous cell carcinoma (LUSC), the major subtypes of non-small cell lung cancer (NSCLC), exhibit distinct molecular landscapes that demand precision in prognosis and therapy. While deep learning models can achieve high predictive accuracy, their black-box nature limits clinical translation.

METHODS: We introduce AttentioFuse, an interpretable deep learning framework employing a Reactome-guided mid-fusion strategy for multi-omics integration. AttentioFuse builds on three pillars: (i) dual-phase learning with omics-specific encoders to preserve modality-unique patterns, (ii) hierarchical attention mechanisms (cross-omics, feature-level, and fusion-layer) to quantify layer contributions dynamically, and (iii) integrated explainability combining DeepSHAP and global attention weights for gene-to-pathway interpretation. Two depth variants are instantiated under identical priors: a three-layer configuration (3F) for main discrimination and a five-layer configuration (AttentioFuse-5X) for deeper hierarchical interpretation; the 5X variant is trained end-to-end and yields comparable accuracy while enhancing pathway-level resolution.

RESULTS: Evaluated on The Cancer Genome Atlas (TCGA) LUAD/LUSC cohorts, AttentioFuse matches state-of-the-art performance in TNM staging while uncovering actionable biological insights, including pan-NSCLC AKT/mTOR metabolic control, histology-divergent Notch signaling roles, and additional pathways related to developmental reactivation, microbiota-associated metastasis, and extracellular matrix remodeling.

CONCLUSIONS: By design, AttentioFuse-5X bridges predictive performance with hierarchical, pathway-resolved explanations, advancing oncology by transforming black-box predictions into biologically grounded decision support.

PMID:41827812 | PMC:PMC12985206 | DOI:10.3390/cancers18050878