For many cancers, treatment options are too numerous to test in clinical trials. Modelling can be used predict the best treatment for patient groups.
A simulation model accurately predicted the clinical outcomes of different drug treatments in groups of patients with metastatic colorectal cancer. The predicted outcomes were overall survival or progression-free survival. Overall survival is the time between cancer diagnosis and death, while progression-free survival is the time between the start of a treatment and the point at which the treatment no longer works.
The model predicted that for 25 per cent of patients, progression-free survival could have been extended by 3 months if they had received a different drug treatment. This does not necessarily mean that those patients received the wrong therapy. Progression-free survival is not the only consideration when choosing a treatment.
The findings show that simulation modelling could supplement clinical trials by giving doctors valuable information about the effectiveness of drug treatments in different patient groups.
The modelling was done by Dr Koen Degeling from the Cancer Health Services Research unit, which is led by Professor Maarten IJzerman. They worked with researchers from WEHI, the Peter MacCallum Cancer Centre, other universities, and hospitals across Australia.
The research team used a discrete-event simulation, which represents a real-life process as a chronological sequence of events. In this case, the events were initial diagnosis, first-line treatment, treatment outcome, further treatment (if relevant), and death.
The model was based on data from patients who received chemotherapy with or without the drug bevacizumab. Chemotherapy kills cancer cells but also damages normal cells. Bevacizumab takes a more targeted approach by blocking the blood supply that helps tumours grow.
Colorectal cancer, also known as bowel cancer, is the third deadliest cancer worldwide. In 2017, it accounted for 896 000 deaths globally. Its incidence varies widely between countries but is increasing overall – particularly in people under 50.
Metastatic, or stage IV, colorectal cancer is cancer that has spread from the initial tumour site to other areas of the body. Doctors have dozens of combinations of drugs to choose from when treating the disease. Many of these drugs work better when used together. But their effectiveness may depend on patient characteristics such as sex or age, and the patient’s desired outcome. For example, the most effective treatment for older men who want to maintain their quality of life may not be the best option for younger women who want to extend their overall survival. Choosing the best treatment is made even more complicated by the fact that drugs used in early treatment may alter the effectiveness of drugs used in later stages of the disease. This complexity means that it is not possible to test all drug combinations and sequences in clinical trials.
The research team designed the simulation model to help solve this problem. Like a clinical trial, the model considers specific outcomes in groups of patients with similar characteristics.
The model used data from the Treatment of Recurrent and Advanced Colorectal Cancer (TRACC) database. TRACC has data from patients with recurrent and advanced colorectal cancer in Australia and Hong Kong. The data includes details of drug treatments, side effects and disease progression. Data on 867 patients who started first-line treatment of doublet chemotherapy (which uses two chemotherapeutic drugs together) with or without bevacizumab between 1 January 2009 and 1 June 2017 was used.
The team trained the model to learn the relationships between the patient characteristics and treatment outcomes. After removing the outcomes from the data, the team tested whether the model could use the remaining information to make predictions about groups of patients. A comparison between the predictions and the real-world data showed that the model was accurate.
To help cancer researchers and clinicians understand the model and its potential, the research team created an online tool. The tool does not use real patient data and cannot be used to make real treatment decisions for patients.
The researchers plan to test the model on data from other patients. This will show whether the model can accurately predict the outcomes for future patients, not just those whose data was used to develop the model. They will then expand the model to include more outcomes, such as quality of life and treatment costs. Finally, they will apply the model to different stages of colorectal cancer and to other cancers.
Netherlands Organisation for Health Research and Development (ZonMw) via the Translational Research program (446001006)
Degeling K et al (2020) Simulating progression-free and overall survival for first-line doublet chemotherapy with or without bevacizumab in metastatic colorectal cancer patients based on real-world registry data. PharmacoEconomics 38: 1263–1275. doi: 10.1007/s40273-020-00951-1
Re-use this text
Please use the text of this article for your own purposes. The text is licensed under the Creative Commons Attribution (CC BY) 4.0 International license. This lets you copy, transform and share the text without restriction. We appreciate appropriate credit and links back to this website. Other content on this page (such as images, videos and logos) is not covered by the CC BY license and may not be used without permission from the copyright holder. If you have any questions about using this text, please contact the research web team.
First published on 30 March 2022.
Share this article
Illumina-University of Melbourne Partnership
Partnering to provide the infrastructure, expertise, systems and analysis to translate and implement genomics into routine clinical care.
Pursuit: Better cancer treatments from better data analysis
Traditional clinical trials have been vital for improving cancer treatments for the last 70 years, but new data-driven approaches can enhance treatment for individuals and better allocate healthcare resources for everyone.
Pursuit: Testing, testing, testing: Can blood tests help detect cancers early?
With the COVID-19 pandemic highlighting the importance of testing, and our capacity to undertake it at large scale, can we use these testing strategies for earlier diagnosis of cancer?