Baseline sarcopenia and treatment-emergent muscle loss are orthogonal risk factors for adverse outcomes in DLBCL, with sarcopenia strongly predicting nonrelapse mortality and hematologic toxicity rather than lymphoma-specific death.
Key Findings
Results
Patients in the lowest tertile of normalized skeletal muscle mass exhibited inferior survival after adjustment for established risk factors.
Analysis was performed on patients with newly diagnosed DLBCL enrolled in the prospective phase 3 PETAL trial
Machine learning-supported body composition analysis (BCA) was applied to computed tomography imaging
Skeletal muscle mass was normalized and patients were stratified into tertiles
Survival was assessed after first-line immunochemotherapy
Results
Sarcopenia was not associated with lymphoma-specific death but strongly predicted nonrelapse mortality.
Cause-specific time-to-event analyses were used to distinguish between lymphoma-specific death and nonrelapse mortality
This pattern indicates a potential role of sarcopenia as a biomarker of host vulnerability rather than disease aggressiveness
The finding suggests sarcopenia reflects impaired host tolerance to treatment rather than tumor biology
Results
Sarcopenic patients had a higher probability of experiencing hematologic toxicity during immunochemotherapy.
Sarcopenia was the only independent risk factor for higher-grade hematotoxicity in multivariable analyses
Treatment-related hematologic toxicities were assessed in relation to BCA results
This association further supports sarcopenia as a marker of host vulnerability to treatment-related adverse effects
Results
Longitudinal BCA revealed inferior survival in patients with early muscle loss during therapy.
Muscle mass was assessed both at baseline and during treatment to capture treatment-emergent muscle loss
Early muscle loss during therapy was associated with worse survival outcomes
Longitudinal BCA was used to track changes in body composition over the course of immunochemotherapy
Results
Baseline sarcopenia and treatment-emergent muscle loss were not correlated, representing distinct biological phenomena.
The two phenotypes — baseline sarcopenia and treatment-emergent muscle loss — were identified as orthogonal risk factors
Only a small fraction of the interindividual variability in muscle mass could be attributed to age and lymphoma burden
The lack of correlation suggests independent underlying biological mechanisms for each phenotype
Results
Both sarcopenia phenotypes were independent of DLBCL molecular clusters, and no recurrently mutated gene was associated with lower skeletal muscle mass.
Baseline sarcopenia was not correlated with DLBCL molecular clusters
Treatment-emergent muscle loss was also independent of DLBCL molecular classification
No recurrently mutated gene was found to be associated with lower skeletal muscle mass
This suggests that sarcopenia reflects host rather than tumor-intrinsic factors
Methods
Machine learning-supported BCA was applied to CT imaging to quantify radiologic sarcopenia in DLBCL patients.
Body composition analysis provided an objective assessment of metabolic states
The approach was applied to computed tomography imaging from the prospective phase 3 PETAL trial
BCA results were assessed in relation to survival, hematologic toxicities, and molecular disease features
The method enabled both baseline and longitudinal quantification of skeletal muscle mass
What This Means
This research suggests that measuring muscle mass from CT scans using machine learning can identify DLBCL lymphoma patients who are at higher risk for poor outcomes — not because their cancer is more aggressive, but because their bodies are less able to tolerate intensive treatment. Patients with low muscle mass (sarcopenia) at the time of diagnosis were more likely to die from causes other than their lymphoma, such as treatment-related complications, and were more likely to suffer severe blood-related side effects from chemotherapy. Separately, patients who lost significant muscle mass early during treatment also had worse survival, even though this early muscle loss was not related to how much muscle they had at the start.
Importantly, the study found that these two risk factors — low baseline muscle mass and muscle loss during treatment — appear to be biologically distinct and independent of each other, as well as independent of the genetic makeup of the lymphoma itself. Only a small part of the variation in muscle mass between patients could be explained by age or how advanced the lymphoma was, suggesting other factors are at play that are not yet fully understood.
This research suggests that routinely measuring body composition from CT scans — which are already performed as part of standard lymphoma care — could help doctors better identify patients who need extra support or modified treatment approaches to reduce the risk of serious side effects. Incorporating this type of analysis into risk stratification tools could support more personalized treatment planning for lymphoma patients.
Check Your Own Numbers
Upload your bloodwork. We'll cross-reference your results against this study and 4,700 others.
Ullrich F, Hosch R, Kocakavuk E, Zieger H, Hotz P, Freund N, et al.. (2026). A machine learning-derived sarcopenia index is associated with survival and nonrelapse mortality in DLBCL.. Blood advances. https://doi.org/10.1182/bloodadvances.2026020504