Comparative performance of lung cancer risk models to define lung screening eligibility in the United Kingdom

Abstract

Introduction: The National Health Service England (NHS) classifies individuals as eligible for lung cancer screening using two prediction models, PLCOm2012 and Liverpool Lung Project-v2 (LLPv2). However, no study has compared the performance of lung cancer risk models in the United Kingdom. Methods: We analysed current and former smokers aged 40-80 in the UK Biobank (N¼217,199), EPIC-UK (N¼30,982), and Generations Study (N¼25,849). We quantified model calibration (ratio of expected to observed cases, E/ O) and discrimination (AUC). Results: Risk discrimination in UK Biobank was best for the Lung Cancer Death Risk Assessment Tool (LCDRAT, AUC¼0.82, 95%CI¼0.81-0.84), followed by the LCRAT (AUC¼0.81, 95%CI¼0.79-0.82) and the Bach model (AUC¼0.80, 95%CI¼0.79-0.81) (Figure). Results were similar in EPIC-UK and the Generations Study. All models overestimated risk in all cohorts, with E/O in UK Biobank ranging from 1.30 for PLCOm2012 (95% CI¼1.23-1.36) to 2.16 for LLPv2 (95%CI¼2.05-2.28). Overestimation increased with area-level socioeconomic status. In the combined cohorts, USPSTF criteria classified 50.6% of future cases as screening-eligible. The LCDRAT and LCRAT identified 60.9%, followed by PLCOm2012 (58.3%), Bach (58.1%), and LLPv2 (53.6%). Conclusion: Discrimination of lung cancer risk models in UK cohorts was highest for LCDRAT and LCRAT, and lowest for LLPv2. Our results highlight the importance of context-specific validation for prediction tools.