P1.11-08 Are Risk Prediction Models Superior Over Standard Criteria for Lung Cancer Screening in Europe? Macroscale Simulation on Large Polish Cohort

Abstract

Inauguration of the national lung cancer screening programme in Poland is to take place in 2019. Yet, issues such as optimal selection criteria remain unresolved. A computational macroscale simulation of lung cancer risk prediction models’ implementation and comparison in a large lung cancer screening cohort of 5,534 individuals from a single, experienced European center was performed. A total of 5,534 healthy volunteers (aged 50-79, smoking history ≥30 pack-years) were enrolled in the Moltest Bis Programme (Moltest) between 2016 and 2017. Inclusion criteria were based on the Lung Cancer Screening National Comprehensive Cancer Network Clinical Practice Guidelines. Each participant underwent a low-dose computed chest tomography scan and selected participants underwent a further, diagnostic work-up. A computational macroscale simulation of Tammemagi PLCOm2012, Liverpool Lung Project (LLP) and Bach risk models’ implementation was applied. Jupyter notebook v.1.0 scientific environment was used to calculate lung cancer probability of all Moltest participants. Patients i) with 6-year lung cancer probability ≥1.3% were considered as high risk in PLCOm2012 model, ii) in LLP model with 5-year lung cancer probability ≥5.0%, and iii) in Bach model with 1-year lung cancer probability ≥2.0%. Such selected patients were eligible for the inclusion to the simulated lung cancer screening programme. Boolean functions were created and data frames containing patients’ epidemiological characteristics were joined using Pandas Python Library v.0.23 for Python v.3.7. In a computational macroscale simulation 3,897 (70.4%), 3,118 (56.3%) and 925 (16.7%) out of 5,534 Moltest participants met the threshold criteria of lung cancer probability in PLCOm2012, LLP and Bach models, respectively. With 199 (3.6%) Moltest individuals initially referred for diagnostic work-up in the programme, lung cancer was confirmed in 105 (1.9%) cases. Contrarily, among high-risk individuals selected based on PLCOm2012, LLP and Bach models, respectively, 103 (2.6%), 56 (1.8%) and 24 (2.6%) constituted the lung cancer cases primarily detected in the Moltest programme. Thus, in PLCOm2012, LLP and Bach models the proportions of screen-detected lung cancer cases were 98.1%, 53.3% and 22.9%, respectively. Risk prediction models provide a vast disparity in selecting lung cancer high-risk individuals. Lung cancer screening enrollment based on Tammemagi’s PLCOm2012 risk prediction model is superior over LLP, Bach models and standard selection criteria based on age and pack-years.