A new AI-powered MRI analysis shows that changes in body composition during treatment, such as decreased muscle and increased visceral fat, can predict how well multiple myeloma patients will respond to therapy and survive progression-free.
- Researchers developed an automated AI pipeline to measure body composition from routine whole-body MRI scans in multiple myeloma patients.
- Patients with greater abdominal muscle and subcutaneous fat reserves at baseline showed better progression-free survival outcomes.
- Increases in visceral fatty tissue over treatment were linked to worse progression-free survival (hazard ratio 2.89).
- The automated analysis process takes only seconds to minutes, eliminating manual annotation work and patient burden.
- Body composition monitoring could enable timely interventions to prevent disease-related frailty and improve quality of life in myeloma patients.
Routine whole-body MRI scans could reveal much more about multiple myeloma patients’ health, suggest findings published recently in Blood Advances.
Researchers led by Christina Messiou, MD, from the The Royal Marsden NHS Foundation Trust in London used a deep-learning pipeline and reported that body composition at diagnosis and changes during treatment are tied to patient outcomes.
“These findings support opportunistic body composition phenotyping from diagnostic whole-body MRI as a scalable biomarker for risk stratification and mechanistic study,” the Messiou team wrote.
Drug treatments for patients with multiple myeloma can affect overall physical health, and current biomarkers are limited in early diagnosis and prevention. But advancements in whole-body imaging and AI for automated organ and tissue segmentation could help analyze body composition metrics over time.
Messiou and colleagues developed an AI-based pipeline for automated quantitative image-derived phenotypes in non-diseased tissue from routine whole-breast MRI in patients with multiple myeloma. For the study, they performed baseline and longitudinal measurements and explored associations with patient outcomes.
The team used 20 MRI scans for model training and 19 for validation. For the study, the researchers included 69 patients.
The researchers reported the following findings:
The team observed longitudinal changes throughout treatment (p < 0.001), which included a decrease in abdominal skeletal muscle and transient increases in abdominal subcutaneous and visceral fatty tissue.
Greater reserves of abdominal subcutaneous (hazard ratio [HR] = 0.60) and abdominal subcutaneous adipose tissue (HR = 0.67) at baseline are tied to better progression free survival.
Increases in visceral fatty tissue over time are linked to inferior progression free survival (HR = 2.89).
The study authors highlighted that the automated pipeline speeds up the annotating process to “seconds or minutes” and frees up operator time. And they reiterated imaging’s importance in analyzing longitudinal changes they observed, as well as patient comfort.
“Imaging used for diagnosis and imaging-based minimal residual disease and response assessment are increasingly central to myeloma care, and adding body composition metrics to whole-body MRI workflow leverages the same acquisition without additional burden to patients,” they wrote.
The authors added that monitoring body composition could make way for timely interventions, potentially slowing disease-related frailty progression and improving quality of life.
“Measures of body composition will become increasingly relevant as evidence for nutritional support, prehabilitation and rehabilitation grows,” they wrote.
Read the full study here.
