Approaches to targeted colorectal cancer screening in Australia: bridging the gap between epidemiology and impact

© 2021 Sibel SayaIn most countries, who should be screened for colorectal cancer (CRC) is determined by age and family history of disease. Risk prediction models add genomic and/or lifestyle risk factors and are able to predict CRC risk more precisely. If these models were implemented within current population screening programs, more cancers could be detected by screening fewer people, as people at elevated risk are more effectively identified. Behavioural interventions for screening participation could also be targeted, increasing uptake of screening and reducing risky health behaviours. Measures of clinical utility evaluate how useful a model is to direct clinical interventions and to assess associated implementation aspects. This PhD sought to answer the research question: what is the clinical utility of two CRC risk prediction models, one incorporating genomic risk factors, the other lifestyle factors, for tailored population CRC screening in Australia? The first study (Chapter 2) compared the risk stratification and potential clinical impact of the two risk predication models, separately and together, across a range of implementation scenarios in the context of Australian population CRC screening. Use of a genomic and/or lifestyle risk prediction model identified additional people at increased risk that would benefit from colonoscopy, and those at lower risk who could potentially delay screening. Category-based scenarios of broad risk groups provided good balance between number of screens and cancers detected. These findings provide a modelling framework for risk-stratified CRC screening to further model impact on mortality and cost-effectiveness. The second study (Chapters 3 to 6) consisted of 150 general practice patients aged 45-74 recruited sequentially from four practices. After a brief discussion that could be replicated in general practice about the implications of a genomic risk test for CRC, each participant was offered the test. Risk results and tailored screening recommendations were given to those who chose to do the test. In-depth interviews were also conducted with a subset of sixteen participants. Uptake of the genomic test was high (84%), indicating acceptability. Qualitative data revealed that the decision to do the genomic test was easy for most and participants had few reservations about taking the test. 73% of participants made an informed choice about the test. A minority of participants did not understand the implications of the test so did not make an informed decision and a few had concerns about its possible implications so chose not to do it. Based on responses to a series of validated questionnaires, there was minimal psychosocial impact of the genomic test. There was limited evidence that genomic testers were more likely to complete their CRC screening compared to non-testers, but some mediators of screening behaviour were altered in genomic testers. The genomic results promoted screening and/or discussion with GPs regarding screening but only for some. Results from this study have been used to strengthen the behavioural impact of genomic results presentation and discussion before trialling its effect on screening behaviour in a newly funded RCT. There is clinical utility of two CRC risk prediction models for targeting CRC population screening in Australia. Genomic testing for this purpose is acceptable and could be feasible within general practice, suggesting patients can make informed choices regarding testing in this setting and with little risk of adverse psychosocial impacts from testing. Further work is underway to refine and test this intervention that aims not only to recommend the right CRC screening based on risk but encourage adherence to those recommendations. This thesis has contributed significantly to the literature on risk-stratified CRC screening and concludes by summarising how it may be implemented in practice