Participant selection in lung cancer screening: a validation of current risk-prediction criteria and why cancers are missed

Abstract: 
Background: Low-dose computed tomography (LDCT) screening in high-risk populations reduces lung cancer mortality. Current criteria for defining high-risk populations are mainly based on age and smoking status. This study aimed to evaluate the performance of age/smoking criteria and existing risk-prediction models identifying lung cancer cases within a Western general population.
Methods: Baseline data from the Dutch population-based Lifelines cohort, collected between 2006-2013 were linked to the Dutch cancer registry to confirm lung cancer diagnoses. Five-year lung cancer risk was estimated based on age/smoking selection criteria and published risk-prediction models. The number of individuals eligible for screening and lung cancer cases were determined, given the selection strategies applied. 
Results: Of 139,120 adult Lifelines participants, 218 (0.16%) developed lung cancer within 5 years of follow-up. Age/smoking criteria identified 4.9% to 7.4% participants as eligible for screening, but missed more than 53% of lung cancer cases. Risk-prediction models reduced the number of individuals eligible for screening by 30.8% to 57.3% relative to age/smoking criteria, at the cost of missing over 61% lung cancers. Of the missed cases, 44%-61% occurred in former smokers, 25%-40% in current smokers, and 41%-67% in individuals younger than 50. Among former smokers, 43%-70.1% had quit smoking >15 years prior to diagnosis. 
Conclusion: Current criteria for defining high-risk populations eligible for lung cancer screening miss 50-60% of individuals who develop lung cancer within 5-years. The challenge remains in identifying lung cancers that arise in individuals not considered as high risk by current eligibility criteria.
Keywords: Mass screening, lung neoplasms, risk assessment, prediction model, targeted screening.