Comprehensive Summary
This commentary explores how precision exercise medicine can be advanced for patients with cancer via the aggregation of individual-level patient data from exercise oncology trials. The POLARIS initiative aggregates individual-level data from 52 international randomized controlled trials and enables researchers to consider how exercise outcomes differ for individual patients and the characteristics of interventions. The findings show that exercising benefits patient physical fitness, fatigue, quality of life, cognition, sleep, and anxiety and depression, with the largest effect of exercise being with supervised and outcome-based targeted interventions or patients with worse pre-treatment values. Likewise, exercise outcome also varies by patient characteristics such as age, marital status or educational engagement and exercise characteristics such as type, intensity and frequency of exercise intervention. Importantly, older patients and patients classified as very unfit seemed to receive less benefit from the modeled interventions indicating a current gap for them that requires direct focus and new approaches. The authors suggested increasing focus on other under-studied populations, clinical outcomes and their biological markers and leveraging machine learning and AI as tools to refine precision exercise prescriptions in cancer care.
Outcomes and Implications
This research is relevant because it demonstrates that exercise is not only broadly beneficial for individuals with cancer but that the benefit is specific to who the patient is and how the intervention is administered. The POLARIS initiative combined individual-level data in this way, which brings the field closer to precision exercise medicine, whereby exercise prescriptions can be individualized to a similar degree as drug therapies. Clinically, the results provide further context and support for supervised and clinically informed exercise programs, which were associated with the greatest improvements in fitness, quality of life and mental health. Based on pooled RCT data, there is also evidence to suggest that the current methods of exercise prescription may not be effective for older and unfit patients, highlighting the need for interventions to be tailored to patients with cancer. While this work will require ongoing validation within new populations, and integration of clinical outcomes and biomarker data, the potential use of machine learning and AI analytical techniques in pooled datasets offers a credible pathway to refining individually tailored, evidence-based exercise prescriptions for routine cancer care.