Artificial intelligence analysis of routine whole-body MRI scans can measure muscle and body fat distributions in multiple myeloma patients, linking specific tissue changes directly to progression-free survival rates, according to a study published in Blood Advances by researchers at The Royal Marsden NHS Foundation Trust and The Institute of Cancer Research in London.
AI Body Composition Measurements and Myeloma Outcomes
Researchers at The Royal Marsden developed a deep-learning pipeline to automatically extract body composition metrics from non-diseased tissue captured during standard whole-body MRI scans, according to findings from the prospective iTIMM study. The analysis evaluated 70 patients with newly diagnosed or first-relapse multiple myeloma who underwent induction treatment and stem cell transplantation. Unlike traditional body mass index (BMI) calculations that fail to distinguish between different tissue types, the AI tool measured abdominal skeletal muscle, subcutaneous fat beneath the skin, and visceral fat surrounding internal organs in just minutes per scan, according to study data.
The findings indicate that baseline muscle and fat metrics correlate with how long patients remain free from disease progression. According to the study data, patients with higher levels of abdominal skeletal muscle at baseline experienced better progression-free survival, registering a hazard ratio of 0.60. Similarly, higher levels of subcutaneous abdominal fat were associated with better progression-free survival, with a hazard ratio of 0.67. Conversely, researchers observed that increases in visceral fat during treatment carried a hazard ratio of 2.89, indicating a substantially greater risk of disease progression.
Clinical Implications for Personalized Cancer Care
When combined with routine clinical information, the AI-derived measurements achieved a concordance index of 0.725 for predicting progression-free survival, demonstrating potential value for risk assessment, according to the research team. Professor Christina Messiou, Consultant Radiologist at The Royal Marsden and Professor in Imaging for Personalised Oncology at The Institute of Cancer Research, stated that the technology extracts broader physical health insights from imaging already performed during standard care. “By using AI to understand changes in body composition over the course of treatment, we hope to build a more complete picture of a patient’s overall health, not just how their cancer is responding,” Professor Messiou said.
While manual segmentation of whole-body imaging traditionally requires several hours per patient, the automated deep-learning pipeline performs the analysis in minutes, making large-scale clinical application more practical. However, investigators emphasize that the technology remains at the research stage. The study did not establish whether altering treatment or supportive care based on these AI measurements directly improves patient outcomes, meaning larger, multicentre studies are required before clinical implementation.
Future Directions in Supportive Oncology
If validated in future clinical trials, automated body composition analysis could help clinicians identify vulnerable patients who might benefit from earlier supportive interventions. According to The Royal Marsden, these interventions could include tailored exercise programmes, nutritional advice, or physiotherapy as part of a holistic approach to managing multiple myeloma, a cancer of plasma cells that affects approximately 6,000 people in the UK each year. Funding for the research was provided by The Royal Marsden Cancer Charity, the National Institute for Health and Care Research (NIHR) Biomedical Research Centre at The Royal Marsden NHS Foundation Trust and The Institute of Cancer Research, and Cancer Research UK’s National Cancer Imaging Translational Accelerator (NCITA).
Related reading