NPJ Precision Oncology最新文献

Colorectal cancer (CRC) is the third most common malignancy and the second leading cause of cancer-related death worldwide, yet current prognostic stratification is hindered by tumor heterogeneity. Here, we developed a deep learning radiomics model (DLRM), optimized through systematic evaluation of ten machine learning algorithms across 117 combinations, using venous-phase computed tomography (CT) images of 1183 patients from four centers. The resulting risk stratification stratified patients into high- and low-risk groups with distinct survival outcomes, and integration with clinical factors further improved prediction. Integrative transcriptomic and metabolomic analyses revealed that high-risk tumors were enriched for extracellular matrix (ECM)-related pathways associated with tumor progression, whereas low-risk tumors exhibited immune-related signatures, including higher CD8⁺ T-cell infiltration. Both omics consistently identified butanoate metabolism and nitrogen metabolism as protective pathways, validated in an independent public cohort (n = 417). This integrative analytic framework provides robust risk stratification and uncovers biological processes with potential therapeutic relevance.

Head and neck squamous cell carcinoma (HNSCC) represents a leading global malignancy among head and neck cancers. While chemotherapy serves as a standard adjuvant treatment, cisplatin resistance frequently compromises therapeutic outcomes. PANoptosis is an integrated inflammatory cell death pathway governed by PANoptosome complexes. It critically influences chemotherapy response, though its regulatory mechanisms remain incompletely characterized. NADH dehydrogenase (ubiquinone) 1 alpha subcomplex subunit 4-like 2 (NDUFA4L2), a subunit of respiratory chain complex I, has been identified as a critical regulator of cell survival. Our multi-platform investigation employed HNSCC cell lines, patient-derived organoids, tongue orthotopic xenograft models in C57BL/6 mice and Tgfbr1/Pten 2cKO mice to elucidate the role of NDUFA4L2 in cisplatin resistance. Bioinformatic analysis and clinical samples indicate that elevated NDUFA4L2 is associated with poor survival rates and low sensitivity to chemotherapy in HNSCC patients. Through in vitro and in vivo studies, we found that NDUFA4L2-KO in combination with cisplatin suppresses glycolysis levels, thereby inhibiting AIM2 inflammasome activation. Consequently, it triggers tumor cell PANoptosis, remodels the immunosuppressive tumor microenvironment, and enhances antitumor efficacy. These findings establish NDUFA4L2 as both a prognostic biomarker and therapeutic target for overcoming cisplatin resistance in HNSCC through PANoptosis modulation.

Circular RNAs (circRNAs) have emerged as important regulators in cancer biology, but their roles in acute myeloid leukemia (AML) remain poorly characterized due to limited sample sizes and technical challenges in RNA sequencing. Here, we analyze RNA-sequencing data from 315 Swedish AML patients to create the most comprehensive circRNA profile in AML to date. We identify 5,711 high-confidence circRNAs across 315 AML samples, including 402 differentially expressed between AML and healthy controls, with host genes enriched in hematopoietic pathways. We further discover two circRNAs including hsa_circ_0024048 (p = 2.16×10⁻⁶, FDR = 0.012) and hsa_circ_0084678 (p = 1.33×10⁻⁵, FDR = 0.075) whose high expression is associated with significantly improved overall survival, a relationship not observed in their respective host genes. Furthermore, these circRNAs are associated with sensitivities of several drugs, as validated in external datasets (p < 0.05). We identify 451 circRNAs with ELN2022 risk group-specific expression patterns, highlighting circRNA heterogeneity. Subtype analysis further reveals that hsa_circ_0080850 is specifically associated with worse survival (p = 2.13×10⁻⁵ and lower remission rates (38.9% vs 74.7%) within the ELN2022 Favorable subgroup. To conclude, this study establishes the most comprehensive circRNA landscape in AML to date and demonstrates their potential as biomarkers and therapeutic targets, suggesting further investigation into circRNA-driven precision medicine in AML.

The recent successes of HER2-targeting agents, even in tumors characterized by FDA-approved molecular testing as HER2-negative or non-amplified, have underscored the limitations of current diagnostic approaches for accurately identifying patients with actionable HER2/EGFR activation/phosphorylation. We therefore performed a multi-omic investigation integrating clinical next-generation sequencing with a CLIA-certified reverse-phase protein array (RPPA) assay and laser microdissection-enriched tumor samples to characterize ERBB2/HER2 at the DNA, RNA, protein, and phosphoprotein level in patients with advanced pan-cancer solid tumor malignancies. Functional pathway activation mapping by RPPA revealed several patients with ERBB2 genomic or transcriptomic alterations and/or HER2^Total-positivity by immunohistochemistry who exhibited no significant HER2^Y1248 activation/phosphorylation. In contrast, other patients lacking ERBB2 genomic/transcriptomic alterations demonstrated significant HER2^Y1248 activation/phosphorylation with co-activation of EGFR^Y1173, a marker associated with prognostic significance. Our results highlight the weak concordance between ERBB2 genomic/transcriptomic alterations and downstream activation of HER family signaling and support the inclusion of functional proteomic/phosphoproteomic analysis as an essential component of precision oncology pipelines to more accurately guide selection of HER2- and EGFR-targeted therapies.

Peritoneal dissemination is the major cause of mortality in epithelial ovarian cancer (EOC) and requires tumor cells to survive in a detached state by evading anoikis. However, the molecular mechanisms supporting anchorage-independent survival remain poorly defined. Here, we identify glutamate dehydrogenase 1 (GLUD1) as a key regulator of anoikis resistance and metastatic progression in EOC. GLUD1 expression was elevated in metastatic EOC tissues and associated with unfavorable clinical outcomes. Loss of GLUD1 impaired anoikis resistance and reduced metastatic capacity of ovarian cancer cells in vitro, while markedly suppressing peritoneal dissemination and prolonging survival in vivo. Mechanistically, GLUD1 was found to interact with a key protein ARAF, the A-Raf proto-oncogene. By limiting ubiquitin-proteasome-mediated degradation of ARAF, GLUD1 exerted a non-enzymatic function that stabilized ARAF protein levels and sustained MEK/ERK signaling.Together, these findings reveal a non-canonical role of GLUD1 in regulating protein stability and identify the GLUD1-ARAF axis as a critical mechanism supporting anchorage-independent survival during peritoneal dissemination of EOC.

Lung adenocarcinoma (LUAD) remains a leading cause of cancer-related mortality worldwide, highlighting the urgent need for non-invasive strategies for early detection. Here, we present a machine learning-assisted metabolomics approach for the early detection of LUAD. Untargeted metabolomic profiling was performed on 199 serum samples from healthy individuals, patients with lung precancerous lesions, and those with stage I LUAD. An ensemble machine learning workflow was developed to identify metabolite panels capable of discriminating clinical status with high accuracy. We observed progressive metabolic alterations in bile acid, lipid, amino acid, and purine metabolism during LUAD initiation and stepwise progression. Notably, ensemble learning identified a six-metabolite panel, including 12-hydroxydodecanoic acid, hypoxanthine, xanthosine, cholic acid, agmatine, and paraxanthine, for accurate detection of early-stage LUAD, and a distinct four-metabolite panel, comprising 7-α,27-dihydroxycholesterol, 11-undecanedicarboxylic acid, biliverdin, and Prolyl-Valine, for precise differentiation between pre-invasive and invasive lesions. Both panels demonstrated promising diagnostic potential, with performance metrices comparing favorably to established methodologies within the current study cohort. This study delineates the evolutionary trajectory of the serum metabolome associated with early LUAD pathogenesis and provides promising biomarkers for non-invasive early detection.

The treatment principle for locally advanced cervical cancer is concurrent chemoradiotherapy (CCRT). However, local recurrence or distant metastasis may still occur after the completion of CCRT. Currently, there is no effective method to monitor the efficacy of CCRT and predict prognosis. This study aims to predict the therapeutic efficacy and prognosis by dynamically monitoring human papillomavirus (HPV) circulating tumor DNA (HPV ctDNA) levels. We enrolled 31 patients with locally advanced cervical cancer and used the HPV-Seq (14 subtypes) to quantify HPV genotype-specific DNA in plasma and tissues before and during CCRT, and in the post-treatment plasma. Regular imaging was performed to evaluate tumor regression. HPV reads in plasma and tissues decreased significantly during treatment (p < 0.05). Higher HPV ctDNA levels were associated with poorer treatment outcomes (p < 0.05). In univariate and multivariate analyzes, dynamic changes in plasma HPV ctDNA were an independent factor for predicting early treatment outcome. These findings indicate that longitudinal HPV ctDNA monitoring reflects the efficacy of CCRT in locally advanced cervical cancer and provides a basis for earlier identification of candidates for post-CCRT treatment intensive and prognostic stratification.

Prostate cancer (PCa) is a heterogeneous disease, impeding early detection and risk stratification. Liquid biopsies (LBx) enable minimally invasive tumor profiling, but circulating tumor-derived DNA (ctDNA) detection remains difficult, particularly in early-stage PCa. We developed a multimodal LBx approach combining genomic and epigenomic cell-free DNA (cfDNA) features in plasma and urine from newly diagnosed PCa patients to improve early characterization of PCa and risk stratification of aggressive disease. Plasma and urine samples from 55 localized PCa (lPCa) patients, 18 advanced PCa (aPCa) patients, and 36 cancer-free controls were subjected to low-coverage whole-genome sequencing and methylated DNA immunoprecipitation sequencing to assess fragmentation, chromosomal instability, and methylation in cfDNA. This complementary approach yielded a 45% ctDNA detection rate in newly diagnosed PCa. Major differences were observed between aPCa and controls, reflecting increasing signals with tumor progression. Epigenomic cfDNA features differentiated lPCa from aPCa, and ctDNA was detected in 46% of PCa patients with prostate-specific antigen <10 ng/mL, suggesting potential for risk stratification. This study highlights the value of multimodal LBx approaches for early characterization of primary PCa and identification of aggressive disease at initial diagnosis. Integration into clinical workflows could complement diagnostics and support personalized decision-making tailored to patients' PCa risk profiles.

While homologous recombination deficiency (HRD) presents therapeutic opportunities in endometrial cancer (EC), its molecular determinants and clinical implications remain poorly characterized. Through genomic analysis of 688 cancer-related genes combined with genomic scar assessment, we investigated HRD molecular features and clinical relevance of HRD across three cohorts: an EC cohort from Sun Yat-sen University Cancer Center (SYSUCC, n = 114), the Cancer Genome Atlas EC cohort (n = 500), and a high-grade serous ovarian cancer (HGSOC) cohort (n = 118). HRD was identified in 23.7% of SYSUCC EC cases, and HRD tumors paradoxically had fewer short-nucleotide variations in HRR genes than proficient (HRP) tumors (18.52% vs. 48.28%, P = 0.007). Mechanistic analysis revealed large-scale transition (LST) losses as the potential predominant HRD driver in EC, occurring significantly more frequently in HRD versus HRP tumors (74.1% vs 5.7%; P < 0.001). Comparative genomics demonstrated enrichment of HRR gene LST losses was EC-specific, contrasting with HGSOC where LST distribution was HRD-independent. Clinically, elevated HRD scores predicted reduced progression-free survival (HR 1.74, 95% CI 1.03-2.94; P = 0.04) yet enhanced platinum sensitivity (HR 0.41, 95% CI 0.18-0.94; P = 0.034). Our findings indicate that the HRD phenotype in EC, driven primarily by LST losses rather than short-nucleotide variations, serves as both a prognostic and predictive biomarker.

BBP-398 is a selective allosteric SHP2 inhibitor designed to inhibit mitogen-activated protein kinase (MAPK) pathway-driven tumors. We performed the first-in-human phase 1 trial described herein to assess the safety, tolerability, pharmacokinetics, and preliminary efficacy of BBP-398 in patients with advanced solid tumors harboring MAPK pathway mutations. Once-daily BBP-398 was administered at 350-550 mg in the dose-escalation phase (1a; n = 35) followed by a dose-expansion phase (1b; n = 37). The study endpoints were dose-limiting toxicities, treatment-emergent adverse events, pharmacokinetics, target engagement, disease control rate, progression-free survival, and overall survival. In phase 1a, 26% of the 23 evaluable patients had stable disease, with a median progression-free survival duration of 1.8 months (range, 1.7-4.1 months). In phase 1b, 30% of the 27 evaluable patients had stable disease (31% at 350 mg, 27% at 450 mg), with median progression-free survival of 2.2 months and 1.9 months at 350 mg and 450 mg, respectively. We halted dose escalation at 550 mg owing to an increased rate of thrombocytopenia and edema. At daily doses of up to 450 mg, BBP-398 exhibited an acceptable safety profile and produced disease stabilization in nearly 30% of heavily pretreated patients.