1921

Objective: The discriminatory power of previous risk prediction models for all lung cancers generally ranged from 0.57 to 0.72. We constructed individual risk prediction models for all lung cancers and its specific histological subtypes among Hong Kong general male population. Methods: Epidemiological data of 1069 male lung cancer cases confirmed by histology and 1208 community controls were included in this analysis. Community controls were randomly selected from the general male population frequency-matched by age and sex. Annual concentration of ambient PM2.5 and PM10 air pollution were retrospectively reconstructed based on individual household location and the nearest monitoring station data. LASSO-model was used to select optimal risk factors for each specific prediction model. Receiver-operator characteristic curves (ROC) and the area under the curve (AUC) was used to demonstrate the model performance and the ability to differentiate cases from non-cases. Results: AUC for all lung cancers was 0.782 (95%CI: 0.762-0.801) and the discriminatory power increased to 0.824 (95%CI: 0.807-0.841) after PM10 was included into the prediction model, and this improvement was consistently shown in prediction models stratified by smoking status. When the prediction models included ambient PM10 and were further specified by histological subtypes, a notably higher AUC was demonstrated for squamous cell lung cancer (0.878, 95%CI: 0.857-0.898) than that of the adenocarcinoma (0.778, 95%CI: 0.751-0.804). Conclusion: Lung cancer risk assessment tool in Hong Kong Chinese men based on LASSO selection is promising, which shows a relatively higher discriminative accuracy than those developed in many other populations. Risk model is improved after including PM10 air pollution, indicating the importance of addressing ambient PM air pollution in risk prediction. However, external validation of this model in an independent population is recommended to be the next necessary step. [Funding source: Research Grant Council, HKSAR, Project no. CUHK4460/03M]