Prioritizing artificial intelligence (AI)-detected imaging findings may reduce the time to diagnosis of lung cancer. This prospective, multicentre, randomized controlled trial tested whether immediate AI prioritization of primary care-requested chest X-rays (CXR) influenced time to computed tomography (CT) and lung cancer diagnosis, the primary outcomes. Secondary outcomes included the number of urgent suspected lung cancer referrals, incidence and stage of lung cancer, times to urgent referral and treatment, concordance between AI and radiology reports, and algorithm accuracy. AI was available in both study arms, with AI prioritization randomized by day. Of 97,731 participant CXRs, 4,405 were excluded due to data compliance issues or failure of randomization, resulting in 93,326 CXRs analyzed (45,987 and 47,339 in the prioritization 'on' or 'off' arms, respectively). A total of 13,347 CTs were identified, with 2,766 performed within 14 days of CXR. Median (interquartile range) times to CT were 53 days (17-145) and 53 days (19-141), with and without AI prioritization, corresponding to a ratio of geometric means of 0.97 (95% confidence interval (CI) = 0.93-1.02; P = 0.31). When restricted to CTs performed within 14 days of CXR, the median time to CT was 8 days (5-11) in both groups. Lung cancer was diagnosed in 558 people (0.6% of CXRs). Median times to diagnosis were 44 days (26-90) and 46 days (24-105) respectively, with a ratio of geometric means of 0.98 (95% CI = 0.83-1.16; P = 0.84). No significant differences were observed in time to lung cancer referral (14 versus 15 days; P = 0.13), time to treatment (76 versus 72.5 days; P = 0.99) or stage at diagnosis (P = 0.34). Discordance between AI and radiology reports occurred in 28,261 CXRs (30.3%) and expert radiology review identified actionable findings in 6,750 cases (23.9%). AI prioritization of CXR requested by UK primary care has no significant impact on the lung cancer pathway. Therefore, CXR AI deployments should not include worklist prioritization in this context. Future research should differentiate between primary pathway changes and the direct impact of AI. ISRCTN registration: 78987039 .

Background: Large language models (LLMs) show promise for extracting information from clinical free-text documents, but their outputs are often unstructured and lack traceability, complicating validation and adoption in clinical workflows. In this work we introduce SIFTING, an LLM-based framework designed to address these shortcomings. Methods: SIFTING combines the language comprehension capabilities of LLMs with segment-level processing and structured prompts with strict output control, linking findings to the source text to enable both accurate and transparent information extraction. To demonstrate its capabilities, we applied the framework to the task of extracting tumor T-stage information from 130 lung cancer radiology reports (SIFTING-T-stage). A compact 4-bit quantized version of the open-source LLM Llama-3.3-70B (35 GB) was used in a fully self-hosted setup, providing full control over data and model. Performance was evaluated against a reference standard created by four clinical experts and compared with a range of LLMs as used in a conventional single-prompt approach, using bootstrap resampling to estimate confidence intervals. Results: SIFTING-T-stage achieved an accuracy of 90 Conclusion: SIFTING enables accurate, structured, and traceable information extraction from clinical free-text documents. It ensures data control, reproducibility, and verifiable outputs that can support clinical validation and workflow integration.

BACKGROUND:Smokers are eligible for lung cancer screening using low dose CT (LDCT) if they have a sufficiently high risk of lung cancer. In England, for example, people are eligible if their estimated risk using either of two validated models (PLCOm2012 and Liverpool Lung Project) exceeds a certain threshold. Risk is often estimated using only cigarette consumption (duration and number per day), yet many smokers also use other cancer-causing tobacco products. METHODS:SUMMIT is a prospective cohort study of a LDCT screening service. We evaluated the impact of cigars, pipes, and cigarillos on lung cancer risk. RESULTS:Among 13,000 participants who were current/former cigarette smokers, 511 (1 in 25) were also regular users of cigars, pipes, or cigarillos. The 5-year cumulative lung cancer incidence in people who currently smoked both types of tobacco products was 8.85%, versus 5.74% for cigarette only current smokers. Among these 511 participants, the percentage whose risk was below the screening eligibility threshold ignoring other tobacco products which then exceeded the threshold including these products was 0.6% using LLPv2, and 8.2% using PLCOm2012. CONCLUSIONS:All tobacco consumption should be considered when determining screening eligibility to avoid missing those who can benefit from LDCT. CLINICAL TRIAL REGISTRATION:Clinicaltrials.gov no: NCT03934866.

New pulmonary lesions after prior cancer present a diagnostic challenge, potentially representing malignancy relapse or new primary lung cancer due to shared risk factors and/or impact of prior oncological therapies. This study evaluated radiologist-defined semantic features for differentiation of second primary lung cancer (SPLC) versus lung metastasis (LM). 651 single-timepoint, pre-treatment CT thorax scans from the multicentre retrospective AI-SONAR biomarker study (IRAS 331656 REC 23/NE/0151) were divided for review by nine thoracic oncology radiologists to evaluate eight semantic features. Logistic regression analysis was undertaken to identify significant features and a developed ‘Second Malignancy Aetiology Recognition Tool’ model (SMART) was compared to real-world clinical reader performance using McNemar’s test. 649 scans were technically usable, 299 SPLC and 350 LM. Emphysema (p < 0.0001, OR 0.20 [95

BACKGROUND AND AIMS:Despite recent progress, advanced non-small cell lung cancer (NSCLC) has poor survival outcomes, necessitating the development of novel therapies. TNF-related apoptosis-inducing ligand (TRAIL) selectively induces cancer cell death and can be delivered to tumors by mesenchymal stromal cells (MSCs) due to the cells' migratory properties. This first-in-human phase I trial assessed safety and dose of umbilical cord-derived MSCs expressing TRAIL (UC-MSCTRAIL) alongside standard NSCLC therapy. METHODS:Participants performance status 0-1 with treatment-naïve, inoperable stage IIIB/IV NSCLC received UC-MSCTRAIL infusions with each cycle of chemotherapy and immunotherapy, up to 3 cycles. A dose de-escalation design was used. Exploratory in vitro and in vivo studies further characterized UC-MSCTRAIL properties. RESULTS:Six participants enrolled; four received 4 × 10⁸ cells/infusion (median 5.4 × 106 cells/kg), and two received 2 × 10⁸ cells/infusion (median 2.4 × 106 cells/kg). Early termination occurred due to asymptomatic pulmonary emboli (N = 5), which included two patients that were anticoagulated as a protocol amendment with prophylactic low molecular weight heparin (enoxaparin 40 mg) and rivaroxaban 20 mg, respectively. Exploratory analyses found no clear pro-coagulant or immunogenic mechanisms, though participants receiving UC-MSCTRAIL had elevated inflammatory markers. CONCLUSION:This first-in-human study of UC-MSCTRAIL in advanced lung cancer was terminated early due to high prevelance of pulmonary embolism. THough exact mechanism remains unclear, UC-MSCTRAIL may have contributed to a pro-inflammatory environment in participants already at elevated risk of thrombosis due to malignancy. Future MSC-based therapies should incorporate close monitoring for asymptomatic thrombosis and follow a dose escalation design for safety. Safety concerns underscore the need for further research.

Lung cancer is a leading cause of ill health and death with early diagnosis key to improving outcomes. The LungIMPACT randomised study found that the use of an artificial intelligence (AI) tool to flag and prioritise chest X-rays (CXR) referred from primary care in the human reporter’s worklist, did not reduce time from CXR to lung cancer (LC) diagnosis. This health economic analysis estimated relative cost-effectiveness of using AI for CXR prioritisation, compared firstly to AI support in interpretation only and, secondly, to current practice using no AI support. The primary analysis estimated incremental cost per day reduction in time from CXR to lung cancer (LC) diagnosis or date of discharge from lung cancer pathway, of AI with vs without prioritisation. Secondary analyses addressed (i) AI with prioritisation vs AI without; (ii) AI with prioritisation vs No AI; and (iii) AI with or without prioritisation vs No AI. AI with prioritisation was not cost-effective vs. AI without prioritisation and secondary analysis showed that no AI was always more cost-effective. Budget impact was £2 million per year in England for AI with prioritisation vs AI without, rising to £16–35 million per year for comparisons to No AI. The results of this analysis show that AI prioritisation of GP-referred CXRs costs the NHS money and evidence must be generated for meaningful impact on the clinical pathway if AI is to be deployed in this setting.

Background and aims Inflammation may play a role in driving interstitial lung diseases (ILD). Radiological ground-glass opacity (GGO) may not reliably distinguish fine intralobular fibrosis from inflammatory processes in fibrotic ILD. We therefore investigated the relationship between GGO, fibrosis and leukocytes in bronchoalveolar lavage (BAL).Methods We recruited patients with fibrotic ILD at a single centre between May 2014 and February 2018. The extent of GGO and fibrosis was evaluated by two radiologists. Linear regression examined the association between leucocyte numbers in BAL obtained from the right middle lobe and GGO/fibrosis extent in whole lung, adjusting for age, sex and smoking. A Z-test was used to compare the association between BAL and GGO/fibrosis.Results 316 patients were included. Adjusting analyses for covariates, only BAL eosinophil and eosinophil-to-macrophage ratio were positively associated with GGO involvement (0.23 (95% CI 0.03 to 0.42) p=0.023 and 11.21 (95% CI 1.33 to 21.08) p=0.026). Lymphocyte percentages (fibrosis −0.17 vs GGO −0.02 p=0.046); neutrophil percentages (fibrosis 0.38 vs GGO 0.06 p=0.002); neutrophil-to-lymphocyte ratio (fibrosis 0.63 vs GGO −0.05 p=0.027); neutrophil-to-macrophage ratio (fibrosis 14.08 vs GGO 2.57 p=0.015) and neutrophilia (fibrosis 6.81 vs GGO −0.31 p=0.002) all demonstrated a significantly stronger association with fibrosis than GGO.Conclusions Lack of relationships between radiological GGO and BAL leucocyte counts in fibrotic lung disease indicates that GGO may not always be inflammatory in nature. Higher levels of neutrophils were associated with more extensive fibrosis.

BACKGROUND:Lung cancer screening is now widely adopted across healthcare systems. Screening typically occurs in asymptomatic individuals. As symptoms of lung cancer overlap with those of chronic conditions, defining asymptomatic screening is challenging. This study outlines the frequency of symptomatic participants in a lung cancer screening trial. RESEARCH QUESTION:How common are respiratory and red flag symptoms in lung cancer screening participants and how do these symptoms impact screening outcomes? STUDY DESIGN AND METHODS:SUMMIT is a prospective observational cohort study to assess the implementation of Low-Dose Computed Tomography (LDCT) screening for lung cancer. Baseline clinical assessments collected self-reported symptoms, medical history, demographics, and spirometry. Haemoptysis (in last year) and unintentional weight loss (≥5kg in 3 months) were classified as red flag symptoms, and cough (acute: onset less than six weeks, or chronic: >6 weeks) and dyspnoea (modified Medical Research Council scale≥1) classified as non-specific symptoms. Lung cancer diagnoses within one year were ascertained. Multivariable logistic regression was used to assess associations between symptoms and lung cancer. RESULTS:Among 13,035 participants eligible for a baseline LDCT 76% (N=9,859) reported at least one symptom. Cough was present in 36% (N=4,707) and 66% of participants reported dyspnoea. Only 6% of participants reported red flag symptoms, including hemoptysis and weight loss. The presence of any of these symptoms was associated with higher likelihood of lung cancer diagnosis in the year following assessment (OR 1.45, p=0.03, adjusted for other baseline factors). INTERPRETATION:Symptoms are commonly reported in those undergoing lung cancer screening and those with symptoms are more likely to be diagnosed with lung cancer.

Lung cancer screening with low-dose computed tomography has been proven to reduce lung-cancer-specific and all-cause mortality. The UK launched the NHS England Targeted Lung Health Check Programme in 2019, which has now become the national Lung Cancer Screening Programme, with full coverage expected by 2030. Here we present the progress and outcomes of the program. People aged 55-74 were offered low-dose computed tomography of the thorax if they had ever smoked and if risk thresholds, as determined by multivariable models, were met. Delivery of the program is through regionally federated clinical infrastructure and leadership, with national strategic, clinical and economic frameworks. The program has invited over two million people, with 7,193 lung cancers diagnosed-63.1% at tumor, node, metastasis stage 1 and 12.6% stage 2-to March 2025. This has increased the early-stage proportion of lung cancer in England over 5 years, particularly in socioeconomically deprived regions. The NHS England Programme exemplifies how large-scale implementation can be achieved at speed through centralized protocols and effective project management. The program has demonstrated feasibility and scalability in reaching high-risk and underserved populations, but needs to further address inequalities in participation. These findings support adoption of lung cancer screening across the UK and globally, and offer practical tools for international adaptation.

Background Lung cancer screening (LCS) participants have a high competing risk of non-lung cancer related death, which limits screening benefit. We hypothesized that imaging-based biological age can identify individuals at higher risk of non-lung cancer mortality. Methods Low dose computed tomography scans from a large LCS trial were retrospectively analyzed. Bone density loss, muscular fat infiltration, vascular calcification, and visceral fat mass were calculated using segmentations from a nnU-Net based deep learning segmentation model. Each participant’s age gap, defined as the difference between their estimated biological age and chronological age, was estimated with disease course mapping using a Bayesian mixed-effects model. Disease course map construction used all available longitudinal scan data, whilst individual biological age prediction used baseline scan data only. The performance of age gap as a parameter in competing risk models for non-lung cancer related death versus lung cancer diagnosis was evaluated, following the TRIPOD + AI reporting guideline. Findings : We included 29,745 scans from 12,478 participants in the final analysis. Median follow-up was 5 years, representing 58,284 person-years at risk. Multivariable survival models of competing risks were constructed, including the estimated age gap. The subdistribution hazard ratio (sHR) for each decade of age gap for non-lung cancer related death was higher (sHR = 2.0, 95% CI 1.8–2.3), compared to lung cancer diagnosis (sHR = 1.3, 95% CI 1.2–1.5). Addition of age gap improved model concordance of non-lung cancer death compared to a model with clinical variables alone (pooled C-index = 0.69 vs 0.73, difference + 0.04, 95% CI 0.03–0.04, p < 0.001). Interpretation : Biological age estimation using an imaging-based disease progression model improves prediction of non-lung cancer related death in lung cancer screening.

RATIONALE:Lung cancer screening regularly identifies participants with interstitial lung abnormalities (ILA). Existing classification methods may underestimate the prevalence of clinically relevant ILA phenotypes. OBJECTIVES:Can a classification system for ILAs developed in a lung cancer screening setting identify clinically relevant phenotypes? METHODS:Classification criteria based on the presence and lobar extent of traction bronchiolectasis (TBe) were developed internally by expert consensus. Categories included: no ILA, non-fibrotic ILA (NF-ILA), fibrotic ILA (F-ILA), and undiagnosed fibrotic ILD (U-ILD). Interobserver agreement was calculated between two readers. Clinical characteristics, respiratory hospitalizations, and survival were compared between participants of different ILA grades. MEASUREMENTS AND MAIN RESULTS:Eight thousand, one hundred sixty-nine participants were included in the final analysis. TBe showed improved interobserver agreement compared to the American Thoracic Society (ATS) classification, identifying 344 participants (4%) with U-ILD, 86% more than the ATS classification. An additional 405 had F-ILA (5%) and 667 had NF-ILA (8%). Compared to participants without ILA, participants with U-ILD had a higher rate of respiratory hospitalization (IRR = 4.4, 95% CI 2.7-7.5, P < .001) and increased risk of death (aHR = 2.4, 95% CI 1.9-3.0, P < .001). Increasing ILA grade was associated with higher modified Medical Research Council dyspnea scores (OR = 1.1, 95% CI 1.0-1.1, P = .02). CONCLUSIONS:In a lung cancer screening setting, an ILA scoring system focused on lobar TBe identifies more high-risk participants and demonstrates improved interobserver concordance than the ATS classification. TBe identifies participants with a respiratory phenotype who may warrant further investigation and follow-up.

To determine if a pragmatic, conservative approach to managing < 3 cm anterior mediastinal lesions in lung cancer screening (LCS) is safe. 55- to 77-year-old current or former smokers underwent low-dose computed tomography (LDCT) screening. Anterior mediastinal lesions < 3 cm at baseline were managed conservatively with annual LDCT follow-up for up to 2 years. Lesions ≥ 3 cm at baseline, growing during follow-up (based on visual assessment), or demonstrating concerning radiological characteristics were referred for further assessment. Outcomes for all anterior mediastinal lesions were assessed using follow-up LDCT images, electronic health records and the national cancer registry. Descriptive frequencies were calculated for all reported outcomes. The baseline prevalence of anterior mediastinal lesions was 0.7

INTRODUCTION:Low-dose CT lung cancer screening (LCS) inevitably detects incidental findings (IFs) beyond its primary purpose; while some offer clinical benefit, they also risk overdiagnosis, overtreatment, and increased healthcare costs. This study evaluated the clinical urgency of IFs during the first year of the 4-IN-THE-LUNG-RUN LCS program. METHODS:Participants in the Dutch arm of the 4-IN-THE-LUNG-RUN program who underwent baseline and annual CT screening between December-2022 and May-2025 were included. Clinical urgency was assessed by the proportion of radiologist-described IFs that, following centralized expert adjudication, were reported to primary care, referred for further work-up, or required (urgent) referral or intervention. At annual screening, all previously reported IFs were classified as new, stable, progressing, or surgically intervened. RESULTS:Among 4122 participants (56.5% male, median age 68 years), IFs were described in 291 participants (7.1%) across both screening rounds. Following centralized adjudication, 75 participants (1.8%) had an IF reported to primary care, of whom 19 (0.5%) were referred for further work-up. At baseline, 67 IFs in 58 participants (1.4%) were reported to primary care, with 10 referred for clinical work-up, including one urgent referral. Four received surgical intervention, two cardiovascular IFs were non-symptomatic and thus screening-driven. At annual screening, the remaining baseline IFs were predominantly stable (59/63), only four showed progression and were subsequently re-reported. New IFs at annual screening were rare. CONCLUSION:While lung cancer screening inevitably reveals findings beyond lung cancer, the majority are not clinically relevant, rarely require urgent intervention, and can be effectively managed within the screening program.

INTRODUCTION:Lung cancer is the leading cause of cancer-related death worldwide. Low-dose CT (LDCT) screening improves outcomes by detecting early-stage cancers as pulmonary nodules. As most are benign, diagnosing these nodules is challenging and often requires surveillance imaging. The aim of this study is to assess the frequency of cancer progression in a lung cancer screening study as measured by tumour stage (T-stage). METHODS:SUMMIT is a prospective cohort study assessing implementation of lung cancer screening with LDCT in a high-risk population. Screen-detected lung cancers with clinical tumour stage cT1a-c at time of referral were included. Upstaging was defined as an increase in T-stage from referral to treatment. The date of death was obtained from the National Cancer Registration and Analysis Service. Cox proportional hazards analysis with adjustment for age, sex, Charlson Comorbidity Index and pack years was used to assess mortality between groups. RESULTS:390 screen-detected cancers were stage cT1a-c at time of referral. Upstaging occurred more frequently in cT1a (n=48, 56%) compared with cT1b (n=83, 38%) or cT1c (n=34, 40%), p=0.01. The proportion of part-solid nodules was similar between groups (upstaged-N=47, 27%, vs not upstaged-N=45, 21%, p=0.19). 43% of tumours increased T-stage from referral to first treatment (n=165). In participants upstaged, time from referral to treatment was longer (upstaged: 84 days, 46-298 vs not upstaged: 72 days, 43-211, p=0.04). Adjusted overall survival analyses showed an association between upstaging and mortality (HR 1.68, 95% CI 1.13 to 2.51, p=0.01). CONCLUSION:Tumour upstaging occurred in nearly half of early-stage cases in a lung cancer screening population. Tumour upstaging was associated with longer time to treatment and poorer outcomes in this population. TRIAL REGISTRATION NUMBER:NCT03934866.