Machine learning reveals sources of accelerated brain aging in childhood cancer survivors

Survivors of childhood cancer experience unique long-term health effects from their cancer and treatment. These effects include accelerated aging, where they develop health concerns much earlier than their chronological age would suggest. A study led by Nicholas Phillips, MD, PhD, Department of Psychology & Biobehavioral Sciences and Kevin Krull, PhD, Department of Psychology & Biobehavioral Sciences chair, used a machine learning method, BrainAGE, to estimate the observed brain age of survivors compared to the expected chronological brain age. All survivors have older-appearing brains (higher BrainAGE scores) compared to healthy controls.

Higher BrainAGE scores are associated with worse neurocognitive function in survivors, increased plasma biomarkers of oxidative stress, neuroinflammation, and cardiovascular health, as well as exposure to central nervous system–directed therapy. The uncovered biomarkers, many related to vascular disease, suggest future research should explore targeted interventions to reduce vascular disease to potentially protect their neurocognitive health as they age. The findings were published in JAMA Network Open.

“This study validates what we have been hearing from our survivors,” Phillips said. “They are experiencing age-related memory loss and processing speed problems at a much younger age than their siblings. It also suggests that interventions that often support cerebrovascular health (e.g., physical exercise, healthy nutrition, sleep hygiene, stress management) may reduce this risk and promote healthy brain aging.”

Read more