Pathology

AI-derived virtual spatial proteomics from H&E improves NSCLC prognosis and immunotherapy prediction

The HEX model computationally infers spatial proteomic profiles from routine H&E slides, predicting 40 immune, structural, and functional biomarkers in NSCLC. Multimodal integration of H&E and HEX-derived spatial proteomics improved prognostic accuracy by 22% across six independent cohorts totaling 2,298 patients. Immunotherapy response prediction improved by 24–39% compared with conventional clinicopathological and molecular biomarkers. Responder tumors showed co-localized helper and cytotoxic T cells, whereas non-responders showed immunosuppressive macrophage and neutrophil aggregates. 1

Optical genome mapping clarifies complex karyotypes and cryptic lesions in MDS and AML

In 150 MDS and AML patients, optical genome mapping was added to standard cytogenetic and molecular testing. It refined or altered cytogenetic interpretations in 80% of cases, including breakpoint definition and translocation partner identification. Cryptic structural variants and copy-number changes involving MECOM, KMT2A, and NUP98 were uncovered, particularly in complex karyotypes. Previously undefined marker chromosomes and chromoanagenesis events were resolved, revealing aberrations with potential impact on risk stratification and therapeutic planning. 2

Single-cell immune atlas of myeloma marrow improves survival risk stratification

Researchers profiled 1,397,272 single cells from bone marrow of 337 newly diagnosed multiple myeloma patients to construct an immune atlas. High-risk cytogenetic lesions, including 17p13 deletion, showed distinct T-cell associations, such as enrichment of a type I interferon signature. Rapidly progressing patients exhibited a proinflammatory immune senescence-associated secretory phenotype in the marrow microenvironment. Integrating immune signatures with tumor cytogenetics and clinical variables significantly improved survival prediction compared with cytogenetics alone. 3

Tumor–stroma ratio adds to ctDNA and stage for risk stratification in stage III colon cancer

This study analyzed 206 stage III colon cancers resected and treated with adjuvant chemotherapy from the PLCRC-PROVENC3 cohort. Postoperative ctDNA positivity was the strongest recurrence predictor, with a 3-year recurrence risk of 65.4% versus 16.8% when ctDNA was negative. High-risk pT4/pN2 status and stroma-high tumors (tumor–stroma ratio >50% stroma) each carried a recurrence hazard ratio of 3.0. Among ctDNA-negative patients, combining pT stage, nodal status, and tumor–stroma ratio defined low-, intermediate-, and high-risk groups with 3-year recurrence risks from 2.9% to 40.3%. Roughly one-third of patients lacked all three high-risk features and could be considered for treatment de-escalation, whereas ctDNA-negative pT4/N2 stroma-high cases may merit escalation. 4