Week 2026-W26

This study comprehensively compared seven homologous recombination deficiency (HRD) prediction methods in 235 patients with early-stage triple-negative breast cancer. The methods included sequencing-based, copy number-based, functional, mRNA-based, and image-based approaches, with an exploratory comparison to the FDA-approved Myriad myChoice CDx assay. Results revealed substantial concordance among methods, alongside specific discordances attributed to data preprocessing and training strategies. Despite these variations, all methods demonstrated similar prognostic performance in patients treated with adjuvant chemotherapy. The study emphasizes the critical need for rigorous optimization of data processing workflows and threshold definitions to ensure consistency and reproducibility of HRD classifications.

Accurate identification of somatic variants is crucial in oncology, yet existing methods often struggle with high sensitivity in certain genomic regions. VariantMedium, a novel tool, combines a tree-based classifier with a 3D DenseNet convolutional neural network. This model was trained and validated on a large dataset of experimentally confirmed variants and further refined using an active learning strategy. It demonstrates superior sensitivity and comparable or better F1 scores than benchmark tools, especially in genomic regions prone to high sequencing error rates.

This study assesses the performance of in silico prediction tools for genetic variant curation within a panel of cancer predisposition genes. Researchers applied ClinGen SVI Working Group recommended thresholds and AlphaMissense predictions to variants in genes such as BRCA1, BRCA2, TP53, TERT, and ATM, which have established pathogenicity or benignity. The findings indicated insufficient sensitivity for pathogenic TERT variants and benign TP53 variants. The study emphasizes that the performance of these tools can be gene-specific and is influenced by their training datasets. Consequently, it is crucial to validate these tools for individual genes, especially for missense variants.

Accurate RNA splicing is vital for gene expression, yet its mechanisms and the impact of mutations, which can cause diseases like cancer, are not fully understood. Current prediction tools struggle with long-range dependencies and often lack interpretability. SpliceSelectNet (SSNet) is a novel hierarchical Transformer-based deep learning model designed to predict splice sites from DNA sequences up to 100 kb. It integrates local and global attention mechanisms to efficiently capture both proximal and distal regulatory signals. SSNet achieves state-of-the-art performance in splice site prediction and aberrant splicing detection, offering a biologically interpretable framework for modeling long-range splicing regulation.

This study identifies ERO1A as a novel and promising biomarker for early-stage lung adenocarcinoma (esLUAD). High ERO1A expression is associated with poor prognosis, yet paradoxically, it correlates with an immune-activated tumor microenvironment. Tumors with elevated ERO1A levels demonstrate a superior response to immune checkpoint inhibitors. This biomarker could therefore aid in patient stratification and guide peri-operative therapeutic decisions.

This single-center retrospective study investigated the prognostic impact of MGMT promoter methylation and TERT promoter mutations in 54 patients with WHO grade 4 glioblastoma. The findings indicate that MGMT methylation is independently associated with improved overall survival, whereas TERT mutations predict worse survival. Adjuvant chemoradiotherapy was also confirmed to improve outcomes. A combined stratification using both biomarkers allows for the identification of four distinct prognostic subgroups, with MGMT-methylated/TERT-wild-type patients exhibiting the longest survival.

This pooled analysis of two phase II clinical trials, MEDITREME and METIMMOX, investigated the efficacy of first-line chemoimmunotherapy in patients with microsatellite stable (MSS) metastatic colorectal cancer. The findings revealed a significant improvement in median overall survival for the chemoimmunotherapy group compared to chemotherapy alone. Additionally, higher complete response rates were observed with the combination therapy. The study also suggests that baseline CD8+ T-cell infiltration could serve as a predictive biomarker to identify patients most likely to benefit from this treatment approach.