City of Hope National Medical Center Duarte, CA, United States
W. T. Watkins1, C. Hao1, E. E. Kibreab2, B. Nguyen2, A. Liu1, and S. V. Dandapani1; 1Department of Radiation Oncology, City of Hope National Medical Center, Duarte, CA, 2University of California Riverside, Riverside, CA
Purpose/Objective(s): Estimating immune system plasticity remains a significant challenge due to varying concentrations of immune cells in bone marrow, blood, lymph nodes, and other tissue. Longitudinal Immuno-Positron Emission Tomography (ImmunoPET) has the potential to determine the spatial distribution of T-cells within different human anatomy and estimate immune reactions to treatment. This work introduces a multi-compartment model to estimate CD8+ T-cell distributions and response in humans undergoing radiation therapy (RT). Materials/
Methods: Compartments including the circulating immune system (heart, liver, and spleen), the reactive immune system (lymph nodes and thymus), and the intact immune system (bone and marrow), were defined via artificial intelligence (AI) segmentation. The AI segmentation also included kidneys, bowel, bladder, and rectum (composing an exit compartment). Together with injected activity, the exit compartment was used to normalize CD8+ T-cell concentrations within the three immune compartments. Five patients from an ongoing clinical trial (ELIXR, NCT05371132) were used to validate the model. Patients received repeated imaging of 89Zr crefmirlimab berdoxam (ImaginAb, Inglewood, CA), a Zr-89 tagged minibody with a high affinity for CD8+ T-cell binding, before, during, and after RT. We estimated the patient-specific CD8+ T-cell concentrations in total and in each of the immune compartments and describe inter-patient variations and changes after RT with ranges and mean± s. Results: AI segmentations required human edits to spleen and liver based on ImmunoPET blurring due to breathing motion. Exit concentrations ranged from 1.5%-20.5% (14%±6%) across patients and timepoints. The total model immune cell concentrations after correcting for activity decay and the exit compartment ranged from 15.3-27.9 MBq/ml, showing significant inter-patient variation. For individual patients, all patients showed a reduction in total CD8+ T-cell binding within the immune compartments after RT, but the variation was highly patient dependent with range 0.1-10.2 (4.8±3.6) MBq/ml. The distribution of T-cells among different compartments was also significantly different among different patients, with circulating compartment making up 57%±8%, bone marrow 39%±8%, and nodes 4%±2%. Shifts between compartments for different patients was primarily between the circulating and bone marrow compartments, including two patients with >5% shifts from circulating blood into bone marrow. Conclusion: The proposed multi-compartment model shows potential for describing immune system plasticity and shifts in immune cells between anatomic regions, which may be important to understanding the immune response to radiation therapy. Future directions include improved AI-segmentation which incorporate ImmunoPET images and more detailed regional PET analysis (e.g. radiation dose levels and regional bone marrow) may be important to assess treatment response.
