Skip to navigation Skip to main content Skip to footer

Approved research

Performance and improvement of lung cancer risk prediction models

Principal Investigator: Dr David Muller
Approved Research ID: 4669
Approval date: December 1st 2013

Lay summary

Lung cancer is the most commonly diagnosed cancer in the world. Overall, survival from lung cancer is low, but there is substantial variability in survival rates by stage at presentation. This suggests that screening for lung cancer might be beneficial in order to identify tumours earlier, when the potential for effective treatment is greatest. While screening for lung cancer could possibly provide a benefit in terms of reducing mortality, there is also the potential for harm and needless expense in following up false positives detected by the screening program. Thus, any screening program must target individuals at high risk of lung cancer in order to maximise its potential benefits, while minimising potential harm. Numerous statistical models have been developed for the purposes of predicting individual risk of lung cancer based on characteristics such as tobacco use, family history of lung cancer, and prior diagnosis of lung diseases (see references in section 2). Before these models can be used to inform screening guidelines, they must be externally validated and calibrated. That is, the models must be able to discriminate between those at high and low risk of lung cancer, and must provide accurate estimates of absolute risk when applied to different populations. We propose to compare the performance of several established lung cancer risk prediction models using data from UK Biobank. In addition, we aim to assess whether measures of lung function, such as the forced expiratory volume in one second, might improve the discriminatory power of lung cancer risk prediction models. These models will be applied to all UK Biobank participants for whom spirometry data are available. In addition to spirometry data, this research requires questionnaire data and incident cancer diagnoses. No biological specimens are required. This proposal is consistent with UK Biobank's mission of health-related research in the public interest.