A new systematic review of risk prediction models for colorectal cancer in asymptomatic individuals has ‘set a new bar for cancer prevention and control approaches’ according to Frank L Meyskens Jr in Cancer Prevention Research (1). The review, which was published in Cancer Prevention Research by Juliet A. Usher-Smith, Fiona M.Walter, Jon D. Emery, Aung K.Win, and Simon J. Griffin at the Primary Care Unit (Institute of Public Health, University of Cambridge), may provide a strong basis for a similar assessment in other common cancers as well.
Risk prediction tools have the potential to improve patient outcomes through enhancing the consistency and quality of clinical decision-making, facilitating equitable and cost-effective distribution of resources such as screening tests or preventive interventions, and encouraging behaviour change. Risk scores have been part of routine clinical practice for many years in the field of cardiovascular disease but have been less widely used in the field of cancer. Nevertheless, their potential has been recognised by the National Cancer Institute as an ‘area of extraordinary opportunity’ (National Cancer Institute, 2006) and an increasing number of risk prediction models continue to be developed. In the near future, risk prediction models are also likely to incorporate genomic data and could contribute to the introduction of more personalised medicine, including individually tailored screening programmes, into clinical practice.
However, there remains uncertainty about the clinical utility of risk tools and how to implement them to maximise benefits and minimise harms such as over-medicalisation, anxiety and false reassurance.
Simple risk prediction models may perform as well as complex ones for bowel cancer
In the systematic review focusing on risk prediction tools for bowel cancer – the first comprehensive review of this type – the authors identified 52 published models predicting the risk of developing bowel cancer in individuals without symptoms. By grouping the models according to the type and number of variables included they showed that adding increasing numbers of risk factors collected from questionnaires and blood tests, does not clearly improve the risk models. Simple risk models based on readily available information, such as age, sex and BMI, could potentially be used to stratify the population into risk groups. Characteristics of the screening programme (such as the test, age of first invitation and interval between tests) could then be varied according to individuals’ level of risk thereby reducing unnecessary testing and improving the safety and efficiency of screening for bowel cancer. The authors identified a few risk models which include genetic information and while they may improve risk prediction compared with simple scores they require further testing in population-based samples.
Making the most of risk prediction tools in primary care
In a separate review published in the British Journal of Cancer by Juliet Usher-Smith, Fiona Walter and Simon Griffin from the Primary Care Unit (Institute of Public Health, University of Cambridge), the authors provide a systematic overview of the types of available risk prediction models, their potential uses, the existing evidence around their use, the challenges to implementation and the key issues for future research. They found data on predictive utility that suggests that many models have potential for clinical application. However, there has been little work on implementation and further research is needed to assess the acceptability, clinical impact and economic implications of incorporating them in practice.
Read the papers:
Juliet A. Usher-Smith, Fiona M.Walter, Jon D. Emery, Aung K.Win, and Simon J. Griffin Risk Prediction Models for Colorectal Cancer: A Systematic Review. Cancer Prevention Research 2016; 9; 13 http://cancerpreventionresearch.aacrjournals.org/content/9/1/13.full.pdf+html
Juliet Usher-Smith, Jon Emery, Willie Hamilton, Simon J Griffin and Fiona M Walter Risk prediction tools for cancer in primary care. British Journal of Cancer 2015; 113; 1645
See the editorial:
(1) Frank L. Meyskens Jr Risk Factor Models and Personalized Health: Opportunities and Challenges for Asymptomatic Individuals. Cancer Prevention Research 2016; 9:11