BackgroundHypoxia within the tumor microenvironment contributes to the progression of non-small cell lung cancer (NSCLC), yet the extent to which tumor growth alters regional pulmonary ventilation and its prognostic significance remains unclear.MethodsThis retrospective study included 365 patients with pathologically confirmed NSCLC diagnosed between 2009 and 2020. Tumor location and bilateral pulmonary ventilation were assessed using VPS. Ipsilateral ventilation impairment was defined by reduced relative ventilation compared with the contralateral lung. Subgroup analyses were conducted by histological subtype, tumor size, and anatomical location. ResultsQuantitative VPS demonstrated significantly lower ventilation in the tumor-bearing lung compared with the contralateral side (χ² = 17.74, p < 0.001). This pattern was observed in both lung squamous cell carcinoma (LSCC; χ² = 16.30, p < 0.001) and lung adenocarcinoma (LUAD; χ² = 4.81, p = 0.028). Ventilation was reduced by 4.08% in right-sided tumors (52.74% vs. 56.82%, p = 0.003) and by 4.07% in left-sided tumors (43.19% vs. 47.26%, p = 0.003). Significant impairment was evident in LSCC, while LUAD showed a similar but non-significant trend. Ventilation decline persisted across tumor size strata, and correlation analysis indicated that impairment was not solely attributable to tumor volume. Patients with tumors in the lower-ventilated lung had significantly worse survival (HR = 2.40, 95% CI: 1.28–4.51, p = 0.017).ConclusionPulmonary ventilation is significantly reduced in the tumor-bearing lung in NSCLC, independent of tumor size, and is associated with poorer prognosis. Ventilation metrics may serve as functional biomarkers for risk stratification.

Background Accurate detection of MET amplification is essential for guiding targeted therapy in lung adenocarcinoma (LUAD). Conventional methods such as FISH are limited by tissue availability, and ctDNA–based liquid biopsy shows limited sensitivity for amplification detection. DNA methylation profiling offers a promising alternative. Objectives To identify MET amplification-associated DNA methylation features and develop a diagnostic model for MET amplification detection in LUAD. Design A retrospective cohort study. Methods Targeted DNA methylation sequencing (Illumina TruSeq Methyl Capture EPIC) was performed on 38 LUAD tissue samples at our institution. MET amplification was defined based on NGS-derived copy number (CN) using two thresholds: CN3 and CN5. DNA methylation data from 443 LUAD cases in TCGA were integrated with the institutional cohort to identify high-confidence differentially methylated positions (DMPs). Random Forest models were constructed and validated internally and in the TCGA cohort. Subgroup analyses were performed by EGFR mutation status and sample type. GO and KEGG analyses were conducted for functional relevance. Results A total of 20,661 (CN3) and 13,303 (CN5) DMPs were identified, of which 241 and 107, respectively, overlapped with TCGA-derived DMPs. The CN3-based model achieved AUCs of 0.92 (95% CI: 0.84–1.00) in the test cohort and 0.69 (95% CI: 0.64–0.74) in the TCGA cohort, whereas the CN5-based model yielded AUCs of 0.93 (95% CI: 0.81–1.00) and 0.77 (95% CI: 0.70–0.84), respectively. A simplified CN5-based model using the top five DMPs maintained comparable performance (AUC 0.89, 95% CI: 0.72–1.00). Diagnostic performance remained stable in EGFR-mutant patients and in biopsy specimens. Functional enrichment analyses indicated that DMPs were significantly involved in kinase regulation and MET downstream signaling pathways, including the PI3K–Akt signaling pathway. A hypermethylated MET amplification-associated DMP at Chr7:116367600 was consistently observed under both CN thresholds in the institutional cohort. Conclusion This study identifies MET amplification–associated DNA methylation features in LUAD. The CN5-based diagnostic model demonstrated promising accuracy. These findings suggest a novel approach for predicting NGS-defined MET amplification and warrant further validation in larger cohorts and prospective liquid biopsy studies.

BACKGROUND:Childhood maltreatment (CM) has been associated with a higher risk of developing cancer later in life. However, the association between CM and lung cancer has not been thoroughly validated, and a comprehensive assessment of the mechanistic pathways underlying this relationship remains unaddressed. OBJECTIVE:This study aims to prospectively investigate the impact of CM and its specific subtypes on the risk of incident lung cancer in adulthood, and to explore the mediating effects of unfavorable lifestyles, psychological adversity, and biological alterations. METHODS:We included 117,281 participants enrolled in the UK Biobank between 2006 and 2010, who were followed up until June 1, 2022. CM was retrospectively assessed in 2016 using the online Childhood Trauma Screener. Cox proportional hazard models were employed to examine the association between CM and lung cancer. Mediation analysis was performed to estimate the mediating effects of potential mediators. RESULTS:Over a median follow-up period of 13.46 years, 414 participants were diagnosed with lung cancer. CM was associated with a greater risk of lung cancer. Each additional type of CM corresponding to an 11% higher risk (hazard ratio [HR], 1.11 [95% CI, 1.02-1.20]). The presence of physical abuse (HR, 1.42 [95% CI, 1.12-1.78]) and emotional abuse (HR, 1.33 [95% CI, 1.03-1.71]) was associated with an increased risk of lung cancer, respectively. Mediation analyses indicated that the observed association was largely mediated by smoking status (41.39%), self-rated mental problem (23.26%), and CRP-to-albumin ratio (6.64%). CONCLUSIONS:CM experiences are associated with higher risk of developing lung cancer in later life. Furthermore, smoking and self-rated mental problem constitute the main pathways in this relationship. These findings not only suggest the potential clinical value of incorporating childhood maltreatment assessments into future precision screening models, but also underscore the necessity of deeply integrating mental health support with traditional smoking cessation interventions for this vulnerable demographic. Fundamentally, preventing the occurrence of maltreatment during childhood can effectively mitigate the subsequent risk of lung cancer incidence.

Background:The integration of immune checkpoint inhibitors (ICIs) with chemotherapy has revolutionized lung cancer treatment, yet it amplifies the risk of severe hematologic toxicities. Case Description:We describe two patients with advanced lung cancer who developed life-threatening refractory hemocytopenia following ICIs-chemotherapy, unresponsive to glucocorticoids and hematopoietic growth factors. A striking feature of these cases was the exclusive presence of anti-Sjögren's-syndrome-related antigen A (Ro) (SSA) antibodies in refractory patients, but absent in non-refractory controls. Targeted administration of intravenous immunoglobulin (IVIG) and rituximab was temporally associated with clearance of anti-SSA antibodies and restoration of normal hematopoiesis, raising the hypothesis that aberrant B cell activity may contribute to refractory hemocytopenia in this setting. Conclusions:These observations suggest that anti-SSA antibodies could represent a candidate biomarker for identifying refractory hemocytopenia in early time. These findings support a proposal for routine anti-SSA screening in ICI-related hematologic toxicity to enable timely B cell-targeted management.

BACKGROUND:Social isolation, an objective lack of social connections, and loneliness, the subjective distress from perceived social deficits, are established risk factors for poor cancer prognosis. However, their associations with cancer incidence remain unclear. We investigated these associations using UK Biobank data. METHODS:We analyzed data from 354,537 UK Biobank participants aged 38-73. Participants linked to national health registries, without cancer within one year post-baseline, and with complete exposure and covariate data were included. The primary outcome was cancer incidence. Covariates were classified into demographic, physiological, socioeconomic, lifestyle, and health-related indicators. Cox proportional hazards models were used, with subgroup interaction analysis and mediation analyses performed. RESULTS:Here we show that 20,767(5.8%) of participants are isolated and 15,942(4.5%) of participants are lonely. During a median 11.60 years (IQR8.40-12.72) of follow-up, 38,103 participants are diagnosed with cancer. After adjusting for covariates, social isolation is associated with an 8% higher cancer risk(CSHR1.087 95% CI 1.043-1.133; sHR1.073 95% CI 1.029-1.120), while loneliness is not. Social isolation shows a strong interaction by sex (P-interaction<0.01), with isolated females at higher risk than males. Social isolation increases the risk of breast, lung, uterine, ovarian, bladder, and stomach cancers in females, and bladder cancer in males. Socioeconomic factors, health behaviours, and inflammation status largely explain these associations. CONCLUSIONS:Social isolation is a risk factor for cancer with significant sex and organ-specific effects. Addressing socioeconomic challenges, unhealthy lifestyles, and poor mental well-being through health policies could help reduce cancer risk in isolated populations.

Background and Objective:In 2025, lung cancer research advanced rapidly across the disease continuum, from population-level risk assessment and screening to mechanistic studies of early carcinogenesis and therapeutic innovation in perioperative and metastatic settings. A key shift moved beyond a smoking-centred paradigm toward a multidimensional risk framework reflecting the growing burden among never-smokers and the roles of air pollution, occupational exposures, and systemic metabolic-inflammatory states. This narrative review aims to synthesize influential 2025 evidence across prevention, diagnosis, treatment, and survivorship, and to identify convergent themes and translational gaps relevant to clinical practice and policy. Methods:We performed a narrative synthesis of influential lung cancer studies published in major international journals in 2025. Evidence was organized along a clinically oriented pathway spanning carcinogenesis and screening, precision diagnosis, treatment optimization in resectable and advanced disease, and survivorship, emphasizing practice-informing trials, high-impact translational research, and implementation-relevant technologies. Key Content and Findings:Lineage tracing, single-cell and spatial omics, and evolutionary inference refined concepts of field cancerization, clonal selection, and copy-number-driven fitness. In small-cell lung cancer, evidence further supported neuronal coupling and synapse-like programs as potentially tractable vulnerabilities. Clinically, low-dose computed tomography (CT) strategies and data-informed nodule thresholds aimed to balance under-detection against over-surveillance harms. In diagnostics, artificial intelligence (AI) models increasingly inferred molecular features from routine histopathology ("virtual molecular testing") and should be regarded as decision support requiring prospective validation, population calibration, and explicit failure-mode reporting. Multimodal approaches integrating imaging with circulating tumor DNA (ctDNA) improved feasibility in tissue-limited settings, but clinical utility remains contingent on assay standardization and pathway-level implementation. In resectable disease, longer follow-up consolidated neoadjuvant chemo-immunotherapy for selected patients, while ctDNA kinetics emerged as a candidate biomarker for response-adaptive escalation and de-escalation. In advanced non-small cell lung cancer (NSCLC), phase III evidence for antibody-drug conjugates and bispecific antibodies began reshaping sequencing, while highlighting challenges in toxicity, access, affordability, and immature overall survival in several programs. Conclusions:The 2025 landscape reflects coordinated progress in risk conceptualization, biology, diagnostics, and therapeutics, yet gaps in validation, standardization, and real-world deliverability persist. Priorities include prospective evaluation of AI- and ctDNA-enabled pathways, toxicity-informed sequencing, and equitable implementation aligned with health-system capacity.

Lung cancer remains a leading cause of cancer-related deaths globally, with non-small cell lung cancer (NSCLC) accounting for approximately 85% of cases. The tumor microenvironment (TME) plays a crucial role in lung cancer progression and treatment response. Multiplex immunofluorescence (MIF) technology provides a unique perspective for analyzing spatial relationships within the complex TME. However, existing methods for processing MIF pathological images often process each image in isolation, overlooking both intra-patient multi-image complementarity and inter-patient pathological similarities. To address these limitations, we introduce the Hypergraph Aggregation Contrastive Learning Network (HACLN), which constructs a hypergraph to jointly model intra-patient multi-image features and inter-patient pathological relationships. HACLN aggregates features from multiple MIF images per patient, decomposes them into specialized subgraphs, and integrates them to enhance feature discrimination. We validate HACLN using an immunofluorescence image dataset from the First Affiliated Hospital of Guangzhou Medical University, demonstrating its effectiveness in capturing microenvironmental features and modeling patient-to-patient similarities. Here, we show that HACLN achieves a C-index of 0.7023, outperforming existing methods, providing a new direction for future research in lung cancer prognostic prediction based on the tumor microenvironment. Code is available at: https://github.com/ sujuKyukyu/HACLN_code

Abstract Background: Determination of the tissue of origin (TOO) of cancer is essential for appropriate clinical management and treatment selection. Liquid biopsy using circulating cell-free DNA (cfDNA) offers a non-invasive approach for cancer detection and TOO prediction. Circulating tumor DNA (ctDNA), a tumor-derived fraction of cfDNA, carries genomic and epigenomic signatures reflective of its origin. Recent advances in machine learning have enable the development of models to predict TOO from cfDNA profiles. However, current methods show variable performance, particularly in samples with low ctDNA fractions (ctDNA < 3%), and accuracy remains inconsistent across different cancer types. Methods: We developed a TOO classifier using whole genome cfDNA profiles from 1814 patients across 17 cancer types. Multiple distinct cfDNA features were extracted to reveal diverse cancer-associated alterations, including copy number variations, repeat elements, fragment end motifs associated with DNA methylation, fragment size distribution and coverage, microsatellite instability, mutational signatures, nucleosome occupancy, tissue-specific fragmentation patterns, and the presence of cancer-associated viral DNA. Model performance was evaluated in an independent external cohort of 1221 patients. Additional tests were conducted in cohorts of patients with cancers of unknown primary (CUP) and multiple primary cancers (MPC). Results: Our cancer classifier achieved an overall top 1 accuracy of 78% and top 2 accuracy of 89% in the training cohort, with consistently high accuracy across all cancer types. In the independent validation cohort, the model maintained robust performance, with top 1 and top 2 accuracies of 80% and 90%, respectively. Sensitivity increased with the advancing cancer stage, improving from 66.8% in stage I to 86.2% in stage IV. Among 612 low-ctDNA samples, 435 cases (71.1%) were correctly classified. The classifier also showed strong potential in CUP, with 11 of 15 cases (73.3%) aligning with the clinically suspected primary site. Furthermore, among 20 MPC cases with two primary sites, both were correctly identified within the top three predictions in 9 cases. In 3 MPC patients with three primary sites, two of the three sites were accurately captured among the top three predictions. Conclusion: Our cfDNA-based machine learning classifier provides a robust, non-invasive approach for accurate cancer tissue-of-origin identification. Integrating 11 distinct cfDNA-derived fragmentomic, genomic, epigenomic, and microbiomic features, the model achieved high accuracy across multiple cancer types and maintained strong performance in low-ctDNA samples. Its promising results in CUP and MPC further highlight its potential clinical utility in resolving diagnostically challenging cases and guiding precision oncology applications. Citation Format: Yunjian Zhang, Liang Liu, Hua Bao, Haimeng Tang, Ke Xu, Hao Zhang, Song Wang, Shuang Chang, Dongqin Zhu, Zongyao Huang, Zheng Wang, Liu Yang, Bingzhong Zhang, Ji Tao, Wenhua Liang, Jierong Chen, Shanshan Yang, Xue Wu, Yang Shao, Wenquan Wang, Dongyuan Zhu. Machine learning classifier for cancer type identification via multi-feature genome-wide cfDNA profiling [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2026; Part 1 (Regular Abstracts); 2026 Apr 17-22; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2026;86(7 Suppl):Abstract nr 1123.

BACKGROUND:Evidence suggests that the hypoxic tumor microenvironment (TME) contributes to the initiation and progression of non-small cell lung cancer (NSCLC). However, the degree to which tumor growth affects ventilation in the ipsilateral lung, and the clinical significance of this functional change, is not well understood. This study aimed to measure ventilation impairment on the tumor-bearing side using ventilation-perfusion scintigraphy (VPS) and to evaluate its association with patient survival. METHODS:This retrospective study included 365 patients with pathologically confirmed NSCLC diagnosed between 2009 and 2020. Tumor location and bilateral pulmonary ventilation were evaluated using VPS. Ipsilateral ventilation impairment was determined by comparing the relative ventilation contributions of each lung. Subgroup analyses were performed by histological subtype, tumor size, and anatomical location. Survival outcomes were assessed using Kaplan-Meier analysis and univariate Cox regression. RESULTS:Quantitative VPS analysis showed that pulmonary ventilation on the tumor-bearing side was significantly lower than on the contralateral side (χ2 = 17.74, p < 0.001), a pattern consistently observed in both LSCC (χ2 = 16.30, p < 0.001) and LUAD (χ2 = 4.81, p = 0.028) subgroups. Further analysis confirmed an association between tumor presence and reduced ipsilateral ventilation: in patients with right-sided tumors, ventilation was 4.08% ± 1.34% lower than in unaffected right lungs (52.74% vs. 56.82%, p = 0.003); a similar difference was observed in left-sided tumors (43.19% vs. 47.26%, p = 0.003). Subgroup analysis indicated significant impairment in LSCC (left lung: 42.51% vs. 49.02%, p = 0.002; right lung: 50.97% vs. 57.50%, p = 0.002), while LUAD showed a similar but non-significant trend (p > 0.05). Stratified analysis by tumor size demonstrated a consistent decline in ventilation across different tumor volumes. Spearman correlation analysis suggested that ipsilateral ventilation impairment was not solely explained by space-occupying effects (left lung: ρ = -0.225, p = 0.011; right lung: ρ = -0.322, p < 0.001). Survival analysis showed that patients with tumors in the lower-ventilated lung had a significantly higher risk of mortality (HR = 2.40, 95% CI: 1.28-4.51, p = 0.017). CONCLUSION:This study is the first to systematically demonstrate that pulmonary ventilation is significantly reduced in the tumor-bearing lung of patients with NSCLC. This reduction is not solely explained by the space-occupying effect of the tumor, but may also reflect a more complex interaction between tumor biology and regional pulmonary function. In addition, the presence of a tumor in the lower-ventilated lung was associated with worse prognosis. These findings support the potential value of functional imaging in prognostic assessment and suggest that ventilation metrics may serve as functional biomarkers for risk stratification in NSCLC.

BACKGROUND:Accumulating evidence links dietary inflammatory potential to cancer, yet its association with lung cancer incidence and mortality remains inconsistent. This study aimed to examine the relationship between the Dietary Inflammatory Index (DII) and lung cancer risk, and to explore the underlying molecular mechanisms. METHODS:Associations between DII scores and lung cancer incidence were evaluated using Cox proportional hazards models within the UK Biobank cohort, with external validation performed in the NHANES dataset. Mediation analyses were conducted to investigate plasma proteins as potential mediators. Subsequent analyses included functional enrichment assessment, protein-protein interaction network mapping, and Mendelian randomization to characterize candidate proteins and ascertain the causal effects of specific dietary components. RESULTS:During a median follow-up of 11.8 years among 198,735 participants, 1,024 lung cancer cases were identified. Elevated DII scores demonstrated a J-shaped nonlinear positive association with lung cancer risk (Q4 vs Q1 HR 1.30, 95% CI 1.08-1.56). Proteomic screening revealed 133 candidate proteins enriched in immune regulation and inflammatory signaling pathways, with IL-6 identified as a central hub. Thirteen proteins partially mediated the association between DII and lung cancer risk, with HGF and PRSS8 each accounting for approximately 25% or more of the total effect. HGF, TNFSF13B, and LGMN were characterized as potential drug-targetable mediators. CONCLUSIONS:Pro-inflammatory dietary patterns were associated with increased lung cancer risk alongside alterations in circulating protein profiles. These findings provide potential mechanistic insights into the role of dietary inflammation in lung carcinogenesis and underscore the preventive potential of anti-inflammatory dietary interventions.

Introduction Respiratory diseases are significant risk factors for lung cancer; however, the association between acute respiratory infections and lung cancer incidence requires further exploration.Methods We performed a secondary analysis of the prospective UK Biobank cohort, including participants aged 37–73 years recruited from 22 assessment centres across the UK between 2006 and 2010. Cox proportional hazards models estimated HRs for incident lung cancer according to respiratory disease status, which was defined based on linked hospital inpatient records. Mediation analysis explored potential biomarkers, and Mendelian randomisation assessed causal relationships.Results During a mean follow-up of 10.44 years (4 790 738 person-years), 2189 participants developed lung cancer. Among 107 007 individuals with respiratory diseases, 1322 cases occurred (incidence rate 27.6 per 10 000 person-years), compared with 867 cases among 351 876 participants without respiratory diseases (7.9 per 10 000 person-years). Overall, respiratory diseases were associated with increased lung cancer risk (HR 2.97, 95% CI 2.75 to 3.21). Acute respiratory infections, including acute nasopharyngitis (HR 3.41; 95% CI 1.83 to 6.34), influenza (HR 3.90; 95% CI 2.53 to 6.01), viral pneumonia (HR 9.86; 95% CI 6.55 to 14.85) and bacterial pneumonia (HR 6.28; 95% CI 4.40 to 8.96), showed strong associations with lung cancer incidence. In subtype analyses, squamous cell carcinoma exhibited the highest risk elevation (HR 3.65; 95% CI 3.06 to 4.36). Mediation analysis indicated that neutrophil counts partially mediated these associations (proportion mediated up to 8%).Conclusion Acute respiratory infections were associated with higher lung cancer incidence, providing hypothesis-generating evidence that may inform future risk stratification research.

Cancer type classification is challenging due to tumor heterogeneity and undefined tissue of origin (TOO), particularly in cancers of unknown primary (CUP) and multiple primary cancers (MPC). Accurate TOO identification is critical for guiding treatment and prognosis. We developed a stacked ensemble machine learning classifier that integrates 11 multidimensional cfDNA features spanning genomic, fragmentomic, methylation/repeat, and microbial signals. Base models were constructed using five algorithms, including Deep Learning, Distributed Random Forest, Gradient Boosting Machine, Generalized Linear Model, and XGBoost, within a five-fold cross-validation framework, and their predictions were aggregated into a final ensemble optimized for top-1 accuracy. The classifier achieved robust performance across 17 cancer types, with top-1 and top-2 accuracies of 78

Immune checkpoint blockade (ICB) has transformed cancer treatment, yet its efficacy is often limited by the progressive exhaustion of tumor-reactive CD8 T cells. By analyzing transcriptomes of CD8 T cells from patients treated with ICB across cancer types, we found that prothymosin alpha (PTMA) is highly expressed in progenitor exhausted T (TPEX) cells and is associated with treatment response. PTMA expression was directly controlled by T cell factor 1 (TCF1), a central regulator of TPEX cell maintenance in the tumor microenvironment. In mice, genetic deletion of Ptma from T cells compromised CD8 T cell persistence in tumors and abolished the therapeutic effect of programmed cell death protein 1 (PD-1) blockade. PTMA preserved mitochondrial DNA integrity through interaction with mitochondrial transcription factor A (TFAM), sustaining T cell oxidative phosphorylation under metabolic stress. Our findings identify the TCF1-PTMA axis as a molecular link between mitochondrial fitness and durable T cell-mediated antitumor immunity, offering insights and potential directions for future therapeutic strategies to boost immunotherapy efficacy.

STUDY OBJECTIVES:The biological link between sleep patterns and lung cancer remains an enigma. This study aims to explore how healthy sleep patterns modulate lung cancer risk through proteomic mediation and develop a sleep-related biomarker panel. METHODS:This study utilized data from the UK Biobank, comprising 34 881 participants. We used a composite sleep score that considered the following five sleep behaviors: sleep duration, chronotype, insomnia, snoring, and daytime sleepiness (0-5 scores). Proteomic profiling measured 2862 proteins, which were then used to build proteomic signatures through logistic regression to identify significant proteins, followed by LASSO regression to select representative biomarkers for forming weighted signatures. Cox proportional hazards models and mediation analysis were employed to evaluate associations between sleep patterns, proteins, and lung cancer risk. RESULTS:Higher healthy sleep scores were significantly associated with reduced lung cancer risk, with hazard ratios (HR = 0.589, p = .046) for a score of 3 and (HR = 0.516, p = .013) for a score of 4. Similar trends were observed in non-small cell lung cancer and lung adenocarcinoma. Proteomic signature analysis revealed that protein signatures associated with healthy sleep scores had a protective effect against lung cancer (HR = 0.865, p < .001) and non-small cell lung cancer (HR = 0.863, p < .001). Additionally, mediation analysis identified 18 proteins with mediation proportions exceeding 10 per cent in the relationship between healthy sleep scores and lung cancer, with CXCL17 showing the highest mediation proportion. CONCLUSIONS:This study reveals a potential association between healthy sleep patterns and reduced lung cancer risk.

Despite curative-intent treatment, recurrence risk in resectable non-small cell lung cancer (NSCLC) remains difficult to define, and postoperative management still depends on imperfect clinicopathologic assessment. Circulating tumor DNA (ctDNA)-based minimal residual disease (MRD) has emerged as a promising adjunct for postoperative risk stratification, early relapse detection, and perioperative response assessment. Its clinical translation, however, is still limited by low sensitivity in low-shedding diseases, imperfect specificity from biological background signals, interpretive uncertainty, and a lack of assay standardization. In this context, artificial intelligence (AI) is relevant not as a generic add-on, but as a potential means to improve weak-signal detection, refine variant-origin assignment, and translate serial ctDNA measurements into clinically interpretable risk estimates. This review considers the current clinical role of ctDNA-MRD in perioperative NSCLC, the barriers to routine implementation, and the extent to which AI may help make molecular monitoring more reliable and clinically useful.

Background:Neoadjuvant immunochemotherapy (neo-ICT) has demonstrated promising pathological response for resectable non-small cell lung cancer (NSCLC), while its role in epidermal growth factor receptor (EGFR)-mutated NSCLC remains unclear. Here, we aim to explore whether patients with EGFR-sensitive mutation and positive programmed cell death ligand 1 (PD-L1) expression may benefit from it. Methods:We retrospectively reviewed 15 patients with stage IIB-IIIC EGFR-mutant NSCLC and a PD-L1 tumor proportion score (TPS) of ≥10%, who received neo-ICT (with or without bevacizumab) followed by surgery at The First Affiliated Hospital of Guangzhou Medical University between 2020 and 2026. The primary endpoints were pathological complete response (pCR) and major pathological response (MPR), and the secondary endpoints included R0 resection rate, tumor and nodal downstaging rates, event-free survival (EFS), and safety. Results:All patients underwent minimally invasive thoracoscopic surgery with 100% R0 resection. The MPR rate (including pCR) was 33% (5/15), of which 13% (2/15) achieved pCR and 20% (3/15) achieved non-pCR MPR. Pathologic downstaging was achieved in 67% of patients, and nodal downstaging in 73%. With a median follow-up of 29.8 months, the median EFS was 30.1 months [95% confidence interval (CI): 22.5-not reached], the 12-month EFS rate was 93.3% (95% CI: 81.5-100.0%), and the 24-month EFS rate was 66.7%. Neoadjuvant therapy was well tolerated, with no grade 3 or higher treatment-related adverse events, no surgical delays, and no treatment-related deaths reported. Conclusions:Among patients with PD-L1-positive EGFR-mutant NSCLC, neo-ICT yielded pathological and surgical responses with acceptable safety in this small retrospective cohort. Observed pathological benefits remain exploratory, and prospective studies are required to validate its clinical utility before defining its role as a neoadjuvant regimen for this patient subgroup.

Background:Non-small cell lung cancer (NSCLC) shows substantial molecular and immune heterogeneity, and prognosis varies widely even among patients with similar clinicopathologic stage, highlighting the need for biologically grounded biomarkers. Disulfidptosis is a newly described form of regulated cell death triggered by disulfide stress and linked to tumor metabolic vulnerability, but its molecular patterns and prognostic relevance in NSCLC remain unclear. This study aimed to define disulfidptosis-related molecular subtypes in NSCLC, construct and externally validate a disulfidptosis-based prognostic gene signature, and investigate its associations with the tumor immune microenvironment (TIME) and potential immunotherapy response. Methods:We analyzed NSCLC gene expression and clinical follow-up data from the Gene Expression Omnibus (GEO) and The Cancer Genome Atlas (TCGA). Initial unsupervised clustering identified disulfidptosis-related molecular subtypes. Further analysis used weighted gene co-expression network analysis (WGCNA), univariate and least absolute shrinkage and selection operator (LASSO)-COX regression, and stepwise Cox regression to identify disulfidptosis-related prognostic genes, which were then validated by quantitative reverse transcription polymerase chain reaction (qRT-PCR) in surgically resected NSCLC tumor specimens in our center. A Disulfidptosis Score-based risk model was built and evaluated with Kaplan-Meier survival analysis and receiver operating characteristic (ROC) curves. Single-cell RNA sequencing (RNA-seq) data characterised gene-expression patterns in the TIME. Results:Two molecular subtypes (C1 and C2) based on 16 disulfidptosis genes were identified, among which subtype C2 had a lower immune infiltration score and poorer prognosis. Furthermore, eight disulfidptosis-related prognostic genes were identified and validated through our local cohort by RT-PCR. The Disulfidptosis Score-based on these eight genes effectively stratified patients into high- and low-risk categories. Kaplan-Meier analysis demonstrated that disulfidptosis-related genes were significantly associated with survival outcomes. A prognostic model based on the Disulfidptosis Score and Stage was validated using independent datasets, and the area under the curve (AUC) was 0.728, 0.763, and 0.73 at one, three and five years respectively. Single-cell analysis revealed the expression of disulfidptosis-related prognostic genes across various immune cell types, particularly in plasma cells. Conclusions:This study identified disulfidptosis-related gene characteristics and their correlation with tumor microenvironment and prognosis in NSCLC. These findings could provide new insights into NSCLC heterogeneity and offer potential biomarkers for prognosis and therapeutic strategies.

Background Lung cancer incidence and mortality continue to rise in China, highlighting the need for a deeper understanding of its epidemiology and optimal screening targets. Methods A total of 10,560 participants with 5–30 mm pulmonary nodules incidentally detected via low-dose computed tomography (LDCT)/CT were enrolled from 23 hospitals (October 2018–May 2021) and followed for 2–3 years for evaluation of circulating tumor DNA (ctDNA) biomarkers to distinguish malignant and benign nodules. Demographic, clinical, and imaging data were collected and analyzed. Lung cancer risk factors were assessed using stepwise single and multiple-factor binary logistic regression. This study reports only the baseline epidemiological characteristics of the population and the associated risk factors of lung cancer. Results A higher proportion of females (56.56%) were present across all age groups, with a median age of 53 years (45–62), younger than males (55 years [46–63], P < 0.001). While 97.1% of females and 37.8% of males were non-smokers, and 68.65% of participants had no occupational exposure, 71.10% had a history of second-hand smoke exposure. Most nodules were solitary (81.24%), sub-centimeter (78.22%), and non-solid (62.49%). Malignancy was identified in 17.33% of nodules, with 97.32% classified as early-stage lung adenocarcinoma (AJCC stage 0–I: 90.35%). The malignant rate increased with age and nodule size and was also associated with multiple nodules, ground-glass nodules (GGNs), upper lobe location, and female sex. Proportion of patients with malignant nodules was similar between non-high-risk (21.04%) and high-risk (17.92%) groups based on current Chinese screening guidelines, while proportions of benign nodules were notably high for solid nodules (37.21%) and small nodules (20.15%). Conclusion Our large-scale investigation provides updated data to guide future precise clinical management of pulmonary nodules, with a particular focus on younger females and reducing overtreatment.

OBJECTIVE:Utilizing lung cancer risk prediction models at the screening stage can enhance the accuracy of identifying high-risk individuals eligible for lung cancer screening. However, there is a relative lack of research on such prediction models in China, particularly regarding machine learning algorithms. METHODS:A stratified random sampling method was employed to randomly divide the dataset into a training set (70%) and a validation set (30%). Key variables were screened using LASSO regression. Then logistic regression and XGBoost algorithm were utilized to construct a lung cancer risk prediction model in the training set and validate it in the validation set, respectively. RESULTS:A lung cancer risk prediction model was constructed using 11,708 participants enrolled in a prospective cohort, the Guangzhou Lung-Care Project Program. In the constructed lung cancer risk prediction models, the AUC of the logistic regression model in the validation set was 0.647 (95% CI: 0.574-0.720); in contrast, the AUC of the XGBoost model based on the machine learning algorithm in the validation set was 0.658 (95% CI: 0.589-0.727), demonstrating slightly better discriminative ability compared to the logistic regression model. In addition, this study found the important effect of childhood exposure to cooking fuels on the risk of lung cancer, which has been rarely considered in previous research. CONCLUSION:The lung cancer risk prediction model constructed based on the XGBoost algorithm is better than the logistic regression algorithm in terms of prediction accuracy and robustness, aiding in the risk assessment of individuals undergoing screening.