Room: AAPM ePoster Library
Purpose: Early results using chimeric antigen receptor (CAR) T-cell therapy in relapsed/refractory non-Hodgkin lymphoma suggest potential cures in otherwise incurable patients. However, CAR T-cell therapy is associated with significant side effects, namely cytokine release syndrome (CRS) and neurotoxicity. We report predictive models developed to determine the association between 18F-FDG-PET/CT quantitative parameters that predict for CAR T-cell toxicities using the two available commercial CAR T-cell products.
Methods: The metabolic parameters from 18F-FDG-PET/CT data were evaluated by two physicians using our institutional framework to grade treatment outcome for 31 patients. The PET/CT parameters were metabolic tumor volume (MTV) and total lesion glycolysis (TLG). MTV was defined as the sum of metabolic volumes with an uptake = 1.5 × SUVmean?+?2 standard deviations of the liver uptake. TLG was computed as the SUVmean of all active tumor voxels multiplied by the total MTV. Clinical parameters include CAR T-cell product-type and age. A machine learning (RUSEnsemble) model was developed to test the parameters as predictors for CRS and neurotoxicity. The model was validated by ten-fold cross-validation. The prognostic abilities of the parameters were analyzed using ROC curves and several classification metrics.
Results: On univariate analysis, we identified that product-type (AUC = 0.74) and MTV (AUC = 0.86) were correlated to CRS; whereas the occurrence of CAR T-cell therapy toxicity, CRS (AUC = 0.59), age (AUC = 0.65) and product-type (AUC = 0.63) were moderately correlated with neurotoxicity. On multivariate analysis, the combination of all predictive parameters showed a clear relationship with CRS (AUC = 0.70), but no relationship with neurotoxicity (AUC = 0.39).
Conclusion: This study demonstrates the promise of PET/CT and clinical parameters to predict CAR T-cell therapy-related toxicities. Future work should be undertaken to confirm our findings in a larger patient cohort, to further characterize the incidence and management of toxicities.
Funding Support, Disclosures, and Conflict of Interest: Funding Information: 2U24CA180803-06(IROC), 2U10CA180868-06(NRG)
Quality Assurance, Risk, Radiation Therapy
IM/TH- Image Analysis (Single Modality or Multi-Modality): Computer-aided decision support systems (detection, diagnosis, risk prediction, staging, treatment response assessment/monitoring, prognosis prediction)