Frontiers in Oncology最新文献

Background: Accurately predicting axillary lymph node metastasis (ALNM) preoperatively is crucial for optimizing management in patients with clinically node-negative (cN0) hormone receptor-positive (HR+) breast cancer (BC).

Methods: We retrospectively analyzed 816 cN0 HR+ BC patients (2016-2024). Data from 2016-2023 (n=726) were randomly assigned to a training set (n=503) or an internal test set (n=223) in a 7:3 ratio. Patients treated in the most recent year, 2024 (n=90), were reserved as a held-out temporal validation set. Following feature selection via Recursive Feature Elimination (RFE), five machine learning models-XGBoost, Random Forest, Logistic Regression, Support Vector Machine, and K-Nearest Neighbors (KNN)-were developed. Performance was assessed by the area under the receiver operating characteristic curve (AUC) and decision curve analysis (DCA). The optimal model was interpreted using SHapley Additive exPlanations (SHAP).

Results: The incidence of ALNM was 30.9%. The KNN model demonstrated optimal performance, achieving an AUC of 0.898 (95% CI: 0.857-0.939) in the test set and 0.774 (95% CI: 0.655-0.892) in the external validation set. DCA indicated that the KNN model provided the highest net clinical benefit within the 30%-65% threshold probability range. SHAP analysis identified parity as the most critical predictor, followed by age, tumor location, menopausal status, tumor diameter, lymphocyte count, platelet count, alpha-fetoprotein (AFP), neutrophil count, and carcinoembryonic antigen (CEA).

Conclusion: The KNN model, integrated with the SHAP interpretability framework, shows favorable performance, interpretability, and clinical applicability for predicting ALNM in cN0 HR+ BC, offering a valuable tool for preoperative risk assessment and individualized decision-making.

Objective: This study aimed to develop interpretable machine learning (ML) models using apparent diffusion coefficient (ADC) radiomics to differentiate hepatocellular carcinoma (HCC) from intrahepatic cholangiocarcinoma (ICC).

Methods: Radiomic features were extracted from ADC maps of 83 pathologically confirmed HCC and 46 pathologically confirmed ICC patients who underwent MRI examinations. The least absolute shrinkage and selection operator (LASSO) method selected essential features for five ML models: logistic regression (LR), random forest (RF), gaussian naive bayes (GNB), support vector machine (SVM), and k-nearest neighbors (kNN). external validation was performed using 20 HCC and 20 ICC cases from the cancer imaging archive (TCIA) public database. Model performance was assessed using the area under the receiver operating characteristic curve (AUROC), sensitivity, specificity, accuracy, F1 score, calibration plots, and decision curve analysis (DCA). The best-performing model was interpreted using shapley additive explanations (SHAP).

Results: LASSO selected eight features. The models achieved training AUROCs of 0.84-0.95 and internal validation AUROCs of 0.78-0.91. The LR model demonstrated superior performance (training AUROC: 0.95; internal validation AUROC: 0.91; external validation AUROC: 0.85). Moreover, calibration plots and DCA confirmed that this model exhibited the best calibration and clinical utility. SHAP identified wavelet-LLL-firstorder-RootMeanSquared as the most impactful feature.

Conclusions: The ADC-based LR model robustly differentiates HCC from ICC, with validated generalizability using public data, offering a promising non-invasive clinical tool.

Background: Borderline resectable pancreatic cancer (BRPC) poses significant surgical challenges due to tumor-vessel involvement and high risk of positive margins and early recurrence. While neoadjuvant chemoradiotherapy has demonstrated potential benefits in this setting, the role of intensity-modulated radiation therapy (IMRT) combined with gemcitabine and nab-paclitaxel has not been specifically evaluated.

Methods: In this single-center retrospective cohort study, we analyzed patients with histologically confirmed borderline resectable pancreatic ductal adenocarcinoma treated between 2019 and 2022 who ultimately underwent curative-intent resection. Patients either underwent upfront surgery or received neoadjuvant chemoradiotherapy consisting of gemcitabine (1000 mg/m²) and nab-paclitaxel (125 mg/m²) combined with IMRT (36 Gy in 20 fractions), followed by surgery when feasible. Overall survival (OS) and recurrence-free survival (RFS) were calculated from the date of surgery. To address baseline imbalances, propensity score overlap weighting was performed to estimate the average treatment effect in the overlap population (ATO).

Results: A total of 152 patients were included, with 109 in the upfront surgery group and 43 in the chemoradiotherapy group. In unweighted analyses, median RFS was 27 months (95% CI 20.4-33.6) in the chemoradiotherapy group versus 13 months (95% CI 8.2-17.8) in the upfront surgery group (HR 0.61, 95% CI 0.39-0.94; p=0.026), and median OS was 33 months (95% CI 19.5-46.5) versus 21 months (95% CI 14.1-28.0) (HR 0.58, 95% CI 0.36-0.94; p=0.027). In ATO-weighted analyses, median RFS was 25 months (95% CI 14-not reached) versus 11 months (95% CI 8-17) (HR 0.56, 95% CI 0.35-0.88; p=0.013), and median OS was 33 months (95% CI 19-not reached) versus 17 months (95% CI 14-24) (HR 0.56, 95% CI 0.34-0.94; p=0.027).

Conclusion: Neoadjuvant chemoradiotherapy with IMRT plus gemcitabine and nab-paclitaxel was associated with improved surgical and survival outcomes in patients with BRPC compared to upfront surgery. These findings support the integration of modern chemoradiotherapy into the neoadjuvant treatment paradigm for BRPC and warrant prospective validation.

Background: The prognosis for locally advanced gastric cancer (LAGC) remains suboptimal with standard perioperative chemotherapy. Integrating PD-1 inhibitors into this regimen is a promising strategy requiring further validation.

Objective: To evaluate the efficacy and safety of total gastrectomy plus perioperative PD-1 inhibitor (Sintilimab) in the management of locally advanced gastric cancer (LAGC).

Methods: In this retrospective cohort study, 200 patients with LAGC undergoing total gastrectomy (January 2021 - November 2022) were categorized based on treatment received into an experimental group (perioperative sintilimab + XELOX, n=100) and a control group (perioperative XELOX alone, n=100).

Results: The experimental group demonstrated the markedly higher rates of R0 resection rate (97%: 90%, P < 0.05), pCR (25%: 10%, P < 0.05), and MPR (35%: 20%, P < 0.05) as opposed to the control group. Survival analysis revealed significantly better outcomes in the experimental group: 3-year OS (48%: 30%, P = 0.009), median OS (34.7: 23.6 months, P = 0.004), 3-year DFS (40%: 25%, P = 0.001), and median DFS (30.4: 21.3 months, P = 0.007) Two groups showed no clinically relevant difference in the frequency of grade ≥3 therapy-related adverse events (56.0%: 51.0%, P > 0.05). Immune-related adverse outcomes in the experimental group were mainly grade 1-2 hypothyroidism and rash, which were relieved after symptomatic treatment.

Conclusion: Total gastrectomy combined with perioperative sintilimab +XELOX regimen for LAGC significantly improves radical surgery, pathological response rate and long-term survival, with a manageable safety profile, offering a potentially effective treatment strategy.

ERCC6, also known as CSB (Cockayne Syndrome B), is a key protein involved in transcription-coupled nucleotide excision repair (TC-NER), a DNA repair process that removes lesions blocking RNA polymerase. ERCC6's multifaceted roles include chromatin remodeling, transcription regulation, oxidative stress response, and coordination with other DNA repair proteins. Mutations in ERCC6 lead to Cockayne Syndrome and other neurodegenerative disorders, but some variants, such as M1097V, have been associated with cancer risk, particularly prostate cancer (PCa) in African Americans (AAs) in Louisiana. Recent studies have explored the functional impact of ERCC6 variants in PCa, especially among AAs, who face higher incidence and more aggressive forms of the disease. A notable finding is that the M1097V variant increases cellular tolerance to UV damage, suggesting not only a possible evolutionary benefit but also a potential risk for mutagenesis when exposed to complex environmental carcinogens. Other ERCC6 mutations, such as S636N (also only found in Louisiana PCa), located near regulatory regions, may alter repair activity, though their effects remain unclear. Given the high mutation burden in mismatch repair (MMR) and NER genes observed in AA patients with PCa, a synthetic lethality strategy targeting both TC-NER and homologous recombination repair (HRR) pathways could be effective. This includes combining agents like CDDP (cisplatin) with inhibitors of RAD54, such as J54. These approaches may offer alternatives to androgen deprivation therapy (ADT), which is often ineffective in advanced or treatment-resistant PCa common among AA men. This work underscores the importance of integrating genetic, environmental, and therapeutic insights to address PCa disparities.

Background: Malignant melanotic nerve sheath tumor (MMNST) is a rare and aggressive subtype of malignant peripheral nerve sheath tumor (MPNST) characterized by melanin-producing cells. Due to its rarity, there is currently no standardized therapeutic strategy, posing significant challenges in selecting effective treatment modalities.

Case summary: We report the first documented case of a 54-year-old female with recurrent MMNST and multiple pulmonary metastases. Following local radiotherapy combined with systemic anlotinib treatment, the patient achieved prolonged progression-free survival (PFS) exceeding 32 months.

Conclusion: This case suggests that anlotinib combined with radiotherapy may be an effective therapeutic option for advanced MMNST, providing valuable clinical evidence for managing this malignancy.

Introduction: Liver cancer is among the deadliest malignancies worldwide, and both its incidence and mortality continue to rise. Precise tumor segmentation often remains difficult due to heterogeneous enhancement patterns, infiltrative margins, and frequently obscured underlying parenchymal disease. While deep learning has advanced the field, existing heavy 3D architectures (e.g., nnU-Net) often require substantial computational resources, which limits their clinical deployment. Standard architectures also still struggle to reconcile fine-grained tissue cues with whole-organ context.

Methods: This study introduces the Liver Cancer Mamba Network (LCMambaNet), an efficient 2D segmentation framework built on selective state-space models. A tailored scan-patch mechanism extracts salient texture- and density-based features, sharpening the discrimination between normal parenchyma and malignant regions. The Liver Cancer Attention Module (LCAM) further decouples the confounding relationships between parenchymal descriptors and tumor characteristics. The selective state-space backbone captures long-range dependencies and continuous feature dynamics. We evaluated the model on both the LITS (CT) and CirrMR160+ (MRI) datasets.

Results: The proposed approach surpasses current state-of-the-art methods, achieving Dice scores of 92.94 ± 3.12% and 92.08 ± 2.85% on the LITS and CirrMR160+ datasets, respectively. Notably, stratified analysis shows superior performance on small lesions (< 2 cm), with statistical significance (p < 0.01) against strong baseline models. Comprehensive ablation studies verify the contribution of each component.

Discussion: The results demonstrate that LCMambaNet offers an efficient, clinically viable solution for 2D liver tumor segmentation. Its design addresses the key limitations of existing models, balancing computational efficiency with high segmentation accuracy. The strong performance on small lesions also highlights its potential to support early diagnosis and precise treatment planning, advancing the clinical utility of AI-based segmentation tools.

Objective: The purpose of this study was to evaluate the prognostic value of metabolic bulk volume (MBV), a baseline 18-fluorode-oxyglucose positron emission computed tomography (¹⁸F-FDG PET/CT) derived indicator characterizing bulky disease, in DLBCL patients treated with R-CHOP.

Methods: 311 consecutive newly diagnosed Diffuse Large B-Cell Lymphoma (DLBCL) patients were retrospectively evaluated. Estimating MBV, Distance between the centers of the two farthest lesions (Dmax) and Total metabolic tumour volume (TMTV) as semiquantitative metabolic parameters. Receiver Operating Characteristic (ROC) curve analysis was used to determine the optimal cut-off values. Progression-Free Survival (PFS) and Overall Survival (OS) were the endpoints for evaluating prognosis. PFS and OS were estimated using Kaplan-Meier curves, and comparisons were performed via log-rank test.

Results: Multivariate analysis showed that only two baseline ¹⁸F-FDG PET factors, MBV and Dmax, remained significant for OS (P = 0.004 and P < 0.0001). Combining high MBV and high Dmax generated three risk groups for PFS (P < 0.0001) and OS (P < 0.0001). This was equally effective as the three-risk-group stratification by high TMTV combined with high Dmax for both PFS (P < 0.0001) and OS (P < 0.0001). Further stratification of the three risk groups generated by the combination of high TMTV and high Dmax using MBV showed that MBV could further stratify PFS in both the low-risk group (P < 0.0001) and high-risk group (P = 0.015), as well as OS in high-risk group (P = 0.001).

Conclusion: MBV was an independent prognostic factor for DLBCL patients. The combination of MBV with parameters reflecting tumor dissemination distribution or total tumor burden further improved the risk stratification for staging in DLBCL patients.

Background: Detection of circulating tumor DNA (ctDNA) has attracted growing attention for predicting postoperative breast cancer recurrence; however, the differences between the landmark and surveillance strategies remain unclear.

Methods: We systematically searched the PubMed, Cochrane Library, Embase, and Ovid MEDLINE databases for studies published up to April 17, 2025. Effect models were selected based on heterogeneity tests to pool diagnostic indicators, including sensitivity and specificity. Subgroup analyses were conducted according to molecular subtype, detection method, analytical strategy, and disease stage.

Results: A total of 17 studies were included in the analysis. The sensitivity and specificity of the landmark strategy were 0.40 (95% CI: 0.22-0.62) and 0.95 (95% CI: 0.81-0.99), respectively. For the surveillance strategy, sensitivity was 0.79 (95% CI: 0.71-0.85) and specificity was 0.98 (95% CI: 0.92-0.99). The surveillance strategy significantly improved sensitivity without a substantial loss of specificity. Among molecular subtypes, triple-negative breast cancer(TNBC) exhibited the best performance under the surveillance strategy. Whole-genome sequencing (WGS), droplet digital PCR (ddPCR), and whole-exome sequencing (WES) all demonstrated high sensitivity within the surveillance framework.

Conclusion: ctDNA serves as a highly specific biomarker for predicting postoperative breast cancer recurrence. The surveillance strategy substantially improves its sensitivity; however, the current performance remains below the ideal threshold for clinical implementation. Future research should focus on refining detection strategies and technologies to achieve personalized recurrence risk stratification and guide therapeutic decision-making.

Systematic review registration: https://www.crd.york.ac.uk/prospero/, identifier CRD420251056270.