Prognostic prediction models can aid clinical decision-making for high-risk non-small cell lung cancer (NSCLC) patients. We conducted a systematic search across multiple databases and evaluated eligible studies using PRISMA and PROBAST checklists. Of 28,833 references screened, 233 studies describing 268 models were included. Among them, 89 underwent external validation; 67 (75.28%) were classified with high risk for bias. Most models were developed in North America (45.06%). The median area under the receiver operating characteristic curve (AUC) for 1-, 3-, and 5-year predictions were 0.759, 0.720, and 0.691, respectively. The most common predictors were age (54.85%) and stage (58.21%). And, 47.01% of models were sex-specific. Recently, easily accessible radiomic features have been increasingly used in statistical modeling compared to molecular omics. Models integrating radiomic and clinical features showed potential for improved performance (1-, 3-, and 5-year AUCs: 0.931, 0.985, 0.942). Model discrimination varied across different studies, and there is a lack of model calibration and clinical utility. This review highlights advances and limitations in NSCLC prognostic models. To enhance model reliability and generalizability, it remains crucial to adhere to the TRIPOD guideline to perform prediction model study, and to emphasize comprehensive model validation and bias-reduction strategies. Following a step-by-step guide to develop and validate clinical prediction models by Efthimiou, et al. (BMJ, 2024), along with a multifaceted, multidisciplinary and multi-regional approach, is the key way to facilitate the development and clinical application of qualified models.

Confounding poses a critical threat to the validity of observational comparative effectiveness research (CER) by distorting the estimated associations between treatments/exposures and outcomes. This challenge is particularly pronounced in real-world data, given the absence of randomization and the prevalence of unmeasured and time-varying confounders. Thus, the observed associations may be attributable to differences other than the treatments/exposures of interest and causality cannot be assumed. By synthesizing theoretical foundations, practical applications, software implementations, and reporting standards, this study provides insights into statistical approaches for identifying and handling confounders in observational CER, in order to strengthen the validity of causal inference in real-world evidence, and ultimately enhance the quality and reliability of observational CER.

Background Artificial light at night (LAN) pollution is a growing public health concern. However, there has been limited epidemiological research on the LAN-lung cancer mortality links. Evidence from large cohorts and causal inference approaches is far more limited. Methods We included 476,328 individuals from 35 randomly selected communities in southern China. A satellite-based approach was used to assess LAN exposure, which was assigned to participants based on their home address. The LAN-lung cancer mortality connections were investigated using marginal structural Cox proportional hazards regression models with inverse probability weightings. The modification effects of sociodemographics on the above relationships were also assessed. Results During a median follow-up of 9.0 years, the incidence of lung cancer mortality was 6.43 per 10,000 person-years. We found a hazard ratio (HR) of 1.055 [95 ​% confidence interval (CI): 1.010–1.102] for lung cancer mortality following each 10 ​nW/cm2/sr increase in long-term LAN exposure. Further stratified analyses revealed that those <65 years, males, and those with primary school and below were more vulnerable to LAN exposure. The LAN-lung cancer mortality relationship was significant in those <65 years (HR ​= ​1.132, 95 %CI: 1.047–1.224) but not in those ≥65 years (HR ​= ​1.041, 95 %CI: 0.984–1.101). The HR for male participants was 1.105 (95 %CI: 1.044–1.169), while the HR for females was 1.033 (95 %CI: 0.958–1.113). A significant positive association was observed only among those with primary school and below. Discussion This large cohort study suggests that long-term LAN exposure was linked to increased lung cancer mortality, particularly in certain vulnerable populations.

Cancer has become the second leading cause of death, the global cancer burden is rapidly increasing, and there are marked disparities between and within countries worldwide. Population-based cancer registries systematically collect data on cancer patients in defined populations, which play a crucial role in planning and assessing cancer prevention and control strategies. While the development of cancer registration has been marked by increasing standardization of definitions and methods and the electronic processing of data, the advent of artificial intelligence (AI) offers opportunities to further reduce the labor-intensive nature of registry operations, particularly where registry resources are scarce. These include enabling the processing of large datasets, extracting complex or unstructured data patterns to support cancer registration data abstraction, and facilitating data quality and control. The analysis and dissemination of registry data are also increasingly integrating AI methodologies. This paper provides a comprehensive overview of the application of AI in cancer registration. We investigate the challenges associated with integrating AI into existing cancer registry structures, with a particular emphasis on network and computational constraints, uneven resource allocation, and potential biases and limitations within AI systems. We propose a forward-looking AI-enhanced framework for cancer registration, highlighting AI’s potential to optimize efficiency in cancer registration and the use of registry data for cancer control and cancer research.

Predictive models in healthcare are widely published, yet few achieve routine clinical use due to gaps in methodological rigor, workflow integration, and governance. Existing guidelines primarily focus on clinical settings, with few addressing broader healthcare delivery contexts. We propose a practical framework for translating code to continuous care: Development, Implementation, And MONitoring for Dependable AI prediction model (DIAMOND). Models must be built for explicit clinical use cases, supported by interoperable data, standardized predictors, and rigorous validation with prospective designs. Translation into practice requires workflow integration, proportionate regulatory oversight of intended use, transparency, uncertainty, bias, and accountability, and continued post-deployment evaluation—testing transportability across settings, monitoring and updating for data shift and performance degradation, and assessing health-economic impact to inform iterative refinement. By systematically linking these stages, the DIAMOND framework provides a structured pathway for advancing AI predictive models from promising algorithms to dependable clinical tools.

RATIONALE:Genetically predicted molecular traits provide a cost-effective approach for identifying biomarkers and uncovering underlying biological mechanisms. We extended this framework to investigate gene-smoking interactions in lung cancer susceptibility. OBJECTIVES:To identify trans-omics gene-smoking interactions affecting lung cancer risk and to assess how biomarkers modify effect of smoking. METHODS:We conducted the first trans-omics gene-smoking interaction study of lung cancer by integrating consortium-scale individual genotype data (27 737 cases vs 449 910 noncases) from the International Lung Cancer OncoArray Consortium (ILCCO-OncoArray), Transdisciplinary Research Into Cancer of the Lung (TRICL), Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial (PLCO), and the UK Biobank (UKB) with alliance-based summary-level molecular quantitative trait loci (xQTL) data, involving DNA methylation, gene expression, protein, and metabolite. Based on the identified biomarkers, we developed a molecular modifying score (MMS) to delineate gene-smoking interaction patterns and stratify smokers at high risk of lung cancer. MEASUREMENTS AND MAIN RESULTS:Eight biomarkers showing significant interactions with smoking were identified through a 2-phase analytic strategy, comprising CpG sites in the nicotinic acetylcholine receptor region and gene RP11-326C3.14. The MMS, constructed by integrating these biomarkers with their effect estimates derived from meta-analysis of all available datasets, effectively stratified lung cancer risk among smokers. Trans-omics integrative analysis revealed functional relationships across molecular layers, particularly implicating the NELFE gene in smoking-related carcinogenesis pathways. CONCLUSIONS:The trans-omics association study (xWAS) framework enables systematic discovery of trans-omics gene-environment interactions. The MMS effectively delineates the patterns of the interaction effects and facilitates risk stratification. Additionally, we launched a free online platform, LungCancer-xWAS-GxE (http://bigdata.njmu.edu.cn/LungCancer-xWAS-GxE/).

BACKGROUND & AIMS:Serum lipids, including lipoproteins, cholesterol, and triglycerides, are important modifiable factors influencing human health. However, the associations among different serum lipid profiles and mortality remain insufficiently understood, particularly regarding potential causality and population heterogeneity. This prospective study aims to systematically investigate the relationships between serum lipid concentrations of different densities and sizes with all-cause and cause-specific mortality. METHODS:Cox proportional and Fine-Gray subdistribution hazard models were applied to investigate the associations of 54 lipid concentrations with all-cause and cause-specific mortality (including cardiovascular disease (CVD), cancer, and respiratory disease) in the UK Biobank cohort of 441,448 individuals with 17-year follow-up. Cohorts of 120,967 and 44,168 individuals from the Women's Health Initiative (WHI) with 16-year follow-up and a large-scale meta-analysis were utilized for external replication. We further assessed the underlying causality using Mendelian randomization (MR) and possible modifiers using multiple subgroup analyses. RESULTS:During a median follow-up of 13.8 years, 39,290 deaths occurred, including 7399 from CVD, 18,928 from cancer, and 2707 from respiratory disease. We identified 160 significant associations between lipid concentrations and all-cause and cause-specific mortality. Importantly, most were inverse, with decreased lipid levels linked to increased risk of premature death [hazard ratios (HRs): 0.70-0.98 per standard deviation (SD)]. In contrast, positives were observed for HDL (large/very large) and triglyceride concentrations [HRs: 1.02-1.25 per SD], indicating increased mortality risk with higher levels. Most lipoproteins and cholesterol exhibited nonlinearly correlations with mortality, especially the significant U-shaped in total/HDL. However, MR showed that elevations in several lipids were associated with increased all-cause and CVD-specific mortality risk. Multiple subgroup analyses revealed that age, sex, and lipid-modifying drugs modified the lipid-mortality relationship; specifically, higher lipid concentrations increased mortality risk in younger adults not taking lipid-modifying drugs, but decreased mortality in older adults taking lipid-modifying drugs. The majority of associations were replicated in the WHI and external cohorts. CONCLUSION:Our study systematically reported a large number of associations between serum lipid concentrations and mortality. Subgroup-based population heterogeneity analysis suggests that age, sex, and lipid-modifying drugs could be modifiers for the lipid-mortality relationship. These findings provide more guidance for lipid management and individualized prevention.

Emerging evidence highlights the role of thyroid hormones in cancer, although findings are controversial. Research on thyroid-related traits in lung carcinogenesis is limited. Using UK Biobank data, we performed bidirectional Mendelian randomization (MR) to assess causal associations between lung cancer risk and thyroid dysfunction (hypothyroidism and hyperthyroidism) or functional traits (free thyroxine [FT4] and normal-range thyroid-stimulating hormone [TSH]). Furthermore, in the smoking-behavior-stratified MR analysis, we evaluated the mediating effect of thyroid-related phenotypes on the association between smoking behaviors and lung cancer. We demonstrated significant associations between lung cancer risk and hypothyroidism (hazard ratio [HR] = 1.14, 95% confidence interval [CI] = 1.03-1.26, P = 0.009) and hyperthyroidism (HR = 1.55, 95% CI = 1.29-1.87, P = 1.90 × 10 -6) in the UKB. Moreover, the MR analysis indicated a causal effect of thyroid dysfunction on lung cancer risk (OR inverse variance weighted [IVW] = 1.09, 95% CI = 1.05-1.13, P = 3.12 × 10 -6 for hypothyroidism; OR IVW = 1.08, 95% CI = 1.04-1.12, P = 8.14 × 10 -5 for hyperthyroidism). We found that FT4 levels were protective against lung cancer risk (OR IVW = 0.93, 95% CI = 0.87-0.99, P = 0.030). Additionally, the stratified MR analysis demonstrated distinct causal effects of thyroid dysfunction on lung cancer risk among smokers. Hyperthyroidism mediated the effect of smoking behaviors, especially the age of smoking initiation (17.66% mediated), on lung cancer risk. Thus, thyroid dysfunction phenotypes play causal roles in lung cancer development exclusively among smokers and act as mediators in the causal pathway from smoking to lung cancer.

Hepatocellular carcinoma (HCC) is the most common type of primary liver cancer. However, no previous studies systematically provided global HCC burden and population attributable fractions (PAFs) of major risk factors for HCC at the global, regional, subregional, and national levels. We conducted a population-based study to assess the global burden of HCC and the contribution of nine modifiable risk factors. We used data from GLOBOCAN 2022, CI5, GBD 2021, and other large-scale data sources. We categorized nine major modifiable risk factors as infections (hepatitis B or C virus [HBV, HCV], or C. sinensis), metabolic factors (obesity, type 2 diabetes mellitus, and metabolic dysfunction-associated steatotic liver disease [MASLD]), and behavioral/toxic factors (high alcohol use, smoking, and aflatoxin B1). In 2022, globally, there were 684,659 new HCC cases and 597,434 deaths, with the highest age-standardized rates observed in Eastern Asia and Northern Africa. An estimated 78.4% (536,571 /684,659) of global HCC cases were attributable to the evaluated risk factors, with 82.5% in Asia and 60.4% in America. Infections contributed most worldwide (65.9%), followed by behavioral/toxic risk factors (22.4%) and metabolic factors (19.7%). Region-specific predominant risk factors included HBV in Eastern Asia (72.5%), HCV in Northern Africa (43.9%), smoking in Northern America (24.5%), and high alcohol use in Western Europe (24.9%). Between 1990 and 2022, infection and behavioral/toxic factors declined globally, whereas metabolic factors steadily increased. The substantial variations in HCC burden and PAF across regions highlight the importance of tailored, region-specific preventive interventions to address the varying modifiable risk factors.

Background: Spicy food consumption has been reported to be inversely associated with mortality from multiple diseases. However, the effect of spicy food intake on the incidence of vascular diseases in the Chinese population remains unclear. This study was conducted to explore this association. Methods: This study was performed using the large-scale China Kadoorie Biobank (CKB) prospective cohort of 486,335 participants. The primary outcomes were vascular disease, ischemic heart disease (IHD), major coronary events (MCEs), cerebrovascular disease, stroke, and non-stroke cerebrovascular disease. A Cox proportional hazards regression model was used to assess the association between spicy food consumption and incident vascular diseases. Subgroup analysis was also performed to evaluate the heterogeneity of the association between spicy food consumption and the risk of vascular disease stratified by several basic characteristics. In addition, the joint effects of spicy food consumption and the healthy lifestyle score on the risk of vascular disease were also evaluated, and sensitivity analyses were performed to assess the reliability of the association results. Results: During a median follow-up time of 12.1 years, a total of 136,125 patients with vascular disease, 46,689 patients with IHD, 10,097 patients with MCEs, 80,114 patients with cerebrovascular disease, 56,726 patients with stroke, and 40,098 patients with non-stroke cerebrovascular disease were identified. Participants who consumed spicy food 1–2 days/week (hazard ratio [HR] = 0.95, 95% confidence interval [95% CI] = [0.93, 0.97], P <0.001), 3–5 days/week (HR = 0.96, 95% CI = [0.94, 0.99], P = 0.003), and 6–7 days/week (HR = 0.97, 95% CI = [0.95, 0.99], P = 0.002) had a significantly lower risk of vascular disease than those who consumed spicy food less than once a week ( P trend <0.001), especially in those who were younger and living in rural areas. Notably, the disease-based subgroup analysis indicated that the inverse associations remained in IHD ( P trend = 0.011) and MCEs ( P trend = 0.002) risk. Intriguingly, there was an interaction effect between spicy food consumption and the healthy lifestyle score on the risk of IHD ( P interaction = 0.037). Conclusions: Our findings support an inverse association between spicy food consumption and vascular disease in the Chinese population, which may provide additional dietary guidance for the prevention of vascular diseases.

This study systematically evaluated the spatial distribution, health risks, and regulation of per- and polyfluoroalkyl substances (PFAS) in global drinking water using the PubMed and Web of Science databases (January 1, 2000 to February 25, 2025). Among the 122 studies reviewed, perfluorooctanoic acid (PFOA) and perfluorooctane sulfonate (PFOS) received the greatest research attention (detected in 102 and 100 studies, respectively) and showed the highest detection rates (64.69% and 60.72%, respectively). Several other compounds, including perfluorooctane sulfonamide, perfluorobutanesulfonamide, and perfluoropropane sulfonate, also exhibited high detection rates but remain underregulated, underscoring the need for further research and regulatory oversight. The three countries with the highest concentrations of [Formula: see text] were the Republic of Korea, the United States, and China. Risk assessments indicated that perfluorohexanoic acid, perfluorobutanoic acid, and perfluorobutanesulfonic acid posed negligible health risks, while perfluorohexane sulfonic acid (PFHxS), PFOA, PFOS, and perfluorononanoic acid (PFNA) showed descending levels of health risk (PFHxS > PFOA > PFOS > PFNA). Regulatory approaches are shifting from compound-specific standards to integrated mixture-based frameworks, reinforced by progressively stringent limits.

Background: A number of lung cancer prediction models have been developed worldwide. However, few validation studies have been conducted on Chinese populations. The objective of this study is to evaluate the feasibility and efficacy of 17 global lung cancer risk prediction models when applied to Chinese healthcare big data. Methods: The study included individuals with information recorded in the Yinzhou regional health care database from January 1, 2010 to December 31, 2021. Seventeen lung cancer risk prediction models (Bach, Spitz, Hoggart, PLCOm2012, Korean Men, PLCOall2014, Pittsburgh Predictor, LLPi, LCRAT, HUNT, JPHC, Reduced HUNT, LLPv3, LCRS, OWL, UCL-I, Shanghai-LCM) were evaluated for their performance in overall population and subgroups. The discrimination of the 17 models was assessed using the Harrell's C-index and time-dependent area under the curve (AUC) as metrics. The calibration of the models was evaluated using the expected-to-observed ratio (EOR) and calibration curves. Moreover, the models were recalibrated in the Yinzhou population, and the calibration of the recalibrated models was evaluated. Findings: A total of 907,200 study participants were included in the analysis, comprising 69,263 smokers and 837,937 non-smokers. Of the 17 models initially considered, only 6 (Bach, Hoggart, Pittsburgh Predictor, JPHC, Reduced HUNT, UCL-I) were available in the Yinzhou regional health care database with complete predictor data. Models that predicted risk over a ten-year period (Bach, JPHC, LCRS, and Shanghai-LCM) exhibited C-indices and AUCs of 0.75 or greater in the ever smokers. The majority of models demonstrated an overestimation of incidence risk in the ever smokers and an underestimation in the never smokers. The JPHC and LCRS models exhibited the most optimal calibration curves and the best EOR, whereas the other prediction models had suboptimal calibration. After recalibration, all models showed improved calibration; meanwhile, the JPHC and LCRS models retained the highest level of calibration. Interpretation: Only six models can be directly applied to the Yinzhou regional health care database. The JPHC model developed for the Japanese population and the LCRS model developed based on the China Kadoorie Biobank (CKB) performed better in the Chinese population than other models. Funding: This work was supported by the National Natural Science Foundation of China (82473728 to Y.W.) and Medical and Health Science and Technology Project of Zhejiang Province, China.

The subjectivity of morphological assessment and the overlapping pathological features of different subtypes of myeloproliferative neoplasms (MPNs) make accurate diagnosis challenging. To improve the pathological assessment of MPNs, we developed a diagnosis model (fusion model) based on the combination of bone marrow whole-slide images (deep learning [DL] model) and clinical parameters (clinical model). Thousand and fifty-one MPN and non-MPN patients were divided into the training, internal testing and one internal and two external validation cohorts (the combined validation cohort). In the combined validation cohort, fusion model achieved higher areas under curve (AUCs) than clinical or DL model or both for MPNs and subtype identification. Compared with haematopathologists with different experience, clinical model achieved AUC which was comparable to seniors and higher than juniors ( p  = 0.0208) for polycythaemia vera. The AUCs of fusion model were comparable to seniors and higher than juniors for essential thrombocytosis ( p  = 0.0141), prefibrotic primary myelofibrosis ( p  = 0.0085) and overt primary myelofibrosis ( p  = 0.0330) identification. In conclusion, the performances of our proposed models are equivalent to senior haematopathologists and better than juniors, providing a new perspective on the utilization of DL algorithms in MPN morphological assessment.

Background and aims:Hyperthyroidism is a clinical syndrome caused by the excessive production of thyroid hormones, which can have a broad impact on overall health. We systematically investigated the subsequent multisystem comorbidities associated with hyperthyroidism and the progression of these conditions. Methods:After a 1:4 propensity score matching, a total of 5,832 hyperthyroidism patients and 22,579 controls from the UK Biobank were included in this study. Phenome-wide association study was conducted to explore the associations between hyperthyroidism and a broad range of subsequent diseases, supplemented by landmark analysis to depict the time-varying effects. Disease trajectory analysis was used to explore the sequential pattern of comorbidity progression of hyperthyroidism. Results:Patients with prior diagnosed hyperthyroidism were observed to have an elevated risk of developing 110 subsequent diseases across multiple systems and all-cause mortality and four causes of death, with particularly marked short-term adverse effects. Disease trajectory analysis demonstrated that the three disease clusters most affected by hyperthyroidism were cardiovascular disease cluster, gastrointestinal inflammatory disease cluster and diabetes-mediated disease cluster. Conclusion:Hyperthyroidism is associated with an elevated risk of subsequent multisystem diseases and mortality. Disease trajectory analysis has elucidated critical sequential patterns of disease progression, offering valuable insights for the management of comorbidities in patients with hyperthyroidism.

Rheumatoid arthritis (RA) is a chronic autoimmune disease characterized by symmetric joint swelling, pain, and progressive bone destruction. Although the advent of biologic therapies has significantly improved treatment outcomes, challenges remain in early detection and timely intervention. This study utilizes a nested case-cohort design from the UK Biobank (UKB), integrating proteomics, genome-wide association studies (GWAS), and single-cell RNA sequencing data from the GEO database to systematically evaluate proteins associated with RA risk and identify novel therapeutic targets. Through Cox analysis of proteomic data from 706 RA patients and 1410 controls, we identified 440 plasma proteins. Mendelian randomization analysis further narrowed down 35 plasma proteins, and colocalization analysis ultimately confirmed strong associations and colocalization for ICAM3, CTSV, and RNASET2 in the UKB-PPP dataset. Additionally, we developed an RA risk prediction model based on plasma proteins using the XGBoost algorithm, which demonstrated moderate performance (AUC = 0.74) with a prediction window of up to 5 years in advance. Furthermore, through functional enrichment analysis, protein-protein interaction (PPI) networks, and single-cell RNA sequencing, we elucidated the biological roles and mechanisms of these proteins in RA pathogenesis, providing new strategies for identifying biomarkers and developing targeted therapies for rheumatoid arthritis.

Background:A number of lung cancer prediction models have been developed worldwide. However, there have been limited validation studies conducted specifically on Chinese populations. The objective of this study is to evaluate the feasibility and performance of 17 global lung cancer risk prediction models when applied to Chinese healthcare big data. Methods:The study encompassed individuals whose information was recorded in the Yinzhou Regional Health Care Database (YRHCD) between January 1, 2010 and December 31, 2021. The 17 lung cancer risk prediction models, which comprised the Bach, the Prostate, Lung, Colorectal, and Ovarian Cancer Screening Trial 2012 model (PLCOm2012), the Korean Men, the PLCOall2014, the Pittsburgh Predictor, Liverpool Lung Project Risk Prediction Model for Lung Cancer Incidence (LLPi), the Lung Cancer Risk Assessment Tool (LCRAT), Constrained LCRAT, the Nord-Trøndelag Health Study (HUNT), the Japan Public Health Center-based study (JPHC), Reduced HUNT, the PLCOm2012 without information of race (PLCOm2012-norace), the Liverpool Lung Project version 3 (LLPv3), Lung Cancer Risk Score (LCRS), the Optimized Early Warning Model for Lung Cancer Risk (OWL), the University College London-Incidence (UCL-I), the Shanghai Lung Cancer incidence Model (Shanghai-LCM), were evaluated for their performance in overall population and subgroups stratified by age and sex. The discrimination of the 17 models was assessed using Harrell's C-index and time-dependent area under the curve (AUC). The calibration of the models was evaluated using the expected-to-observed ratio (EOR) and calibration curves. Moreover, the models were recalibrated in the Yinzhou population, and the calibration of the recalibrated models was evaluated. For each model before and after recalibration, we redefined risk thresholds that would select the same number of individuals as the China National Lung Cancer Screening Guideline with Low-dose Computed Tomography 2023 Version (CNLCS 2023) could screen out. The Kaplan-Meier method was used to estimate the incidence and number of cases of lung cancer in individuals screened according to different criteria or models over a five-year follow-up period, and Kaplan-Meier survival curves were plotted. Findings:A total of 904,667 study participants were included in the analysis, comprising 66,730 ever smokers and 837,937 never smokers. Among the 17 models initially considered, only six (Bach, Pittsburgh Predictor, JPHC, Reduced HUNT, Constrained LCRAT, UCL-I) had complete information of predictive variables available in the YRHCD. Most models showed similar levels of discrimination, with C-indices ranging from 0.78 (95% CI 0.74-0.82) to 0.88 (0.87-0.89) and time-dependent AUCs ranging from 0.74 (95% CI 0.73-0.75) to 0.88 (0.87-0.89). The majority of models showed an overestimation of incidence risk among ever smokers, with EORs ranging from 1.10 (95% CI 1.02-1.19) to 4.37 (4.16-4.58), and an underestimation among never smokers with a few models showing exceptions - EORs ranging from 0.12 (95% CI 0.11-0.14) to 1.30 (1.26-1.35). After recalibration, all models showed improved accuracy of predicted probability. The five-year incidence rates observed in the model-selected population, ranging from 0.81% (95% CI 0.64%-0.96%) to 1.29% (1.08%-1.48%), were consistently higher than that observed in the criteria-selected population (0.75%, 95% CI 0.59%-0.90%). Following recalibration, the five-year incidence rates in the model-selected population improved, ranging from 0.81% (95% CI 0.64%-0.96%) to 1.60% (1.36%-1.82%). Interpretation:The majority of recalibrated models demonstrated comparable and favorable discrimination and calibration capability, and were capable of identifying individuals at an elevated risk of lung cancer with greater precision than the criteria. Models designed for the general population (such as LLPv3, LLPi, Korean Men, JPHC, and LCRS) are more appropriate for identifying high-risk groups compared to those exclusively for smokers. Funding:National Natural Science Foundation of China, General Project of Zhejiang Provincial Medical and Health Technology Plan for the Year 2024, Natural Science Foundation of Zhejiang Province.

Aims:The IVUS-ACS trial demonstrated that intravascular ultrasound (IVUS) guidance reduces target-vessel failure (TVF) in patients with acute coronary syndromes (ACSs) undergoing percutaneous coronary intervention (PCI). Whether this benefit applies to all ACS patients across the spectrum of risk is unknown. We sought to develop a new risk score for 1-year TVF after PCI in ACS and determine whether IVUS guidance compared with angiography guidance improves outcomes in both high- and low-risk patients. Methods and results:From the angiography-guided group of the IVUS-ACS trial (n = 1743), the TVF-ACS risk score was developed using the least absolute shrinkage and selection operator method in a derivation group (n = 1288), and its robustness was assessed in an internal validation group (n = 455). External validation was then performed separately in the IVUS-XPL and ULTIMATE trials. Outcomes in high- and low-risk patients randomized to IVUS guidance vs. angiography guidance were then examined. Ten readily available clinical, laboratory, and angiographic variables were selected for inclusion in the TVF-ACS risk score. A cut-off value of 15.64 discriminated angiography-guided PCI patients at high-risk vs. low risk [area under the curve (AUC) 0.715, 95% confidence interval (CI) 0.653-0.777]. The AUC was similar in the validation group [0.709 (95% CI 0.630-0.788)]. High-risk patients exhibited a higher 1-year rate of TVF compared with low-risk patients [19.8 vs. 5.7%, hazard ratio (HR) 3.81, 95% CI 2.06-7.02, P = 0.00002]. Among 3486 randomized patients, IVUS guidance compared with angiography guidance reduced 1-year TVF in high-risk patients (6.9 vs. 17.6%; HR 0.38, 95% CI 0.24-0.59) with a lesser effect in low-risk patients (3.2 vs. 4.3%; HR 0.75, 95% CI 0.51-1.11; P interaction = 0.02). External validation in the IVUS-XPL and ULTIMATE trials confirmed these benefits but with consistent effects in high- and low-risk patients (P interactions = 0.49 and 0.92, respectively). Conclusion:The TVF-ACS risk score reliably stratifies ACS patients undergoing PCI into high- and low-risk groups. The benefits of IVUS guidance during PCI are most pronounced in high-risk ACS patients, although all ACS patients are likely to benefit.

Background/Objectives: This study evaluated the impact of influenza vaccination on mortality using real-world data and compared the effect of current-season-only vaccination versus continuous two-season vaccination. Methods: The 2017–2019 data from the Center for Disease Control and Prevention of Shenzhen, Guangdong, China, included 880,119 individuals aged ≥65 years. The participants were divided into vaccinated and unvaccinated groups and matched using propensity scores with a 1:4 nearest-neighbor approach. Vaccinated individuals were further divided into current-season-only and continuous two-season vaccination groups, matched 1:1. Cox’s multivariable proportional hazards regression models were used to assess the effect of vaccination on all-cause mortality, with Firth’s penalized likelihood method applied to correct for a few events. The Fine–Gray competing risk models were used to assess the effect of vaccination on cardio-cerebral vascular disease (CCVD) mortality. Sensitivity analyses, including caliper matching, a nested case–control design, and Poisson’s regression, were performed to test the robustness of the results. Results: Influenza vaccination reduced all-cause mortality by 39% (HR = 0.61, 95% CI: 0.47–0.80) and 55% (HR = 0.45, 95% CI: 0.33–0.60) in 2017–2018 and 2018–2019, respectively. Current-season-only vaccination showed stronger protective effects than continuous two-season vaccination (HR = 0.56, 95% CI: 0.31–0.99). Influenza vaccination reduced CCVD mortality by 46% (HR = 0.54, 95% CI: 0.34–0.84) in 2018–2019. The results were consistent across the sensitivity analyses. Conclusions: Influenza vaccination was associated with a reduced risk of all-cause and CCVD mortality in older adults, underscoring the importance of routine influenza vaccination in older populations. Stronger effects were observed for current-season-only vaccination, warranting further research to confirm the association and explore mechanisms.

BACKGROUND:Numerous lung cancer risk prediction models have been developed and validated worldwide. It is imperative to offer a comprehensive overview and comparative analysis of their performances. METHODS:We conducted an extensive literature search to identify studies developing and/or validating lung cancer risk prediction models. Then we summarised and compared the external performance of these models, focusing on discriminative accuracy (C-index) and calibration performance (E:O ratio). RESULTS:After an initial screening of 10 210 articles, 35 studies on 21 distinct prediction models were identified, which used 42 different types of predictors spanning seven categories. Notable performance variations were observed in external validations. In North American cohorts, the C-index ranged from 0.60 to 0.87, with E:O ratios from 0.62 to 3.70. Among the European cohorts, the Trøndelag health study HUNT and CanPredict exhibited C-indices surpassing 0.870. Conversely, the Bach, lung cancer risk assessment tool (LCRAT), prostate, lung, colorectal and ovarian cancer screening (PLCO)m2012 and PLCOall2014 performed poorly in electronic health records of the Qresearch database subgroup, with C-indices falling below 0.60. PLCOm2012 reached the best E:O ratio of 1.00 (95% CI: 0.93 to 1.08) in the UK Biobank subgroup. In Asian cohorts, the C-index ranged from 0.54 to 0.87. Only three models, Korean Men, LCRAT and Liverpool lung project incidence risk model (LLPi), achieved a C-index exceeding 0.80. LCRAT demonstrated the best calibration, while Hoggart performed the worst. CONCLUSIONS:Performance of lung cancer risk prediction models, despite being well developed and validated, varies in diverse populations. Significant regional imbalance persists in the development of these models. Rigorous external validation or recalibration study in the target population is crucial in accordance with the guidance prior to model implementation. PROSPERO REGISTRATION NUMBER:CRD42022324602.