Liverpool Lung Project lung cancer risk prediction model Free

54

Introduction: Lung cancer is today the most frequent cause of cancer deaths in the Western world. An accurate, objective lung cancer risk prediction model could help clinicians assess patients’ risks and improve the decisions about screening. Previous lung cancer risk prediction models have been based on smoking and age. Using these and additional lifestyle risk factors, we have developed the LLP Risk Model to calculate an individual’s absolute risk for lung cancer within a 5-year period.

Method: The case-control data used to develop the LLP Risk Model were derived from an ongoing molecular-epidemiological study of lung cancer in Liverpool, The Liverpool Lung Project (LLP), of which 579 cases and 1159 controls were used in this analysis. Each potential risk factor for lung cancer was examined first by conditional logistic regression and only statistically significant covariates (p<0.05) were included in the multivariate model. Significant risk factors include smoking duration, family history of lung cancer, history of non-pulmonary malignant tumour, history of pneumonia and occupational exposure to asbestos. The components used to project absolute risk of lung cancer over a defined period were the relative risk model, the baseline incidence rate for lung cancer and the corresponding risk of mortality from non-lung cancer causes.

Results: Risk Scenarios(Percentage risk (%); confidence interval (95% CI)): (i) Male, 60y, Family history not<60, 40y smoking (3.73% CI 1.85-7.38); (ii) Male 60y, Family history not <60y, Never-smoker (0.31% CI 0.19-0.52); (iii) Male 60y, 40 y smoking (3.17% CI 1.71-5.81); (iv) Male 60y, Family history <60y, 40y smoking, asbestos exposure, Pneumonia, Previous other malignancy (30.95 % CI 19.64-45.12); (iiv) Male 60 years Family history <60y, Never smoker, & Asbestos exposure, Pneumonia, Previous other malignancy (3.52% CI 1.9-6.45).

Conclusions: In comparison with previous lung cancer prediction models, the LLP risk model has distinctive strengths. First the predictor variables are all explicitly defined and can be readily assessed at the time of patient presentation. Second, patients can be assigned to their appropriate risk class on the basis of information from the initial history alone. The LLP Risk Model will require rigorous validation in a separate population. One of the model’s potential applications is in the Primary Care setting to facilitate the referral of high risk individuals for bronchoscopic and spiral CT investigations for lung cancer early detection.

Funded by the Roy Castle Lung Cancer Foundation UK.