Tre-P-03 - Retrospective analysis of standard-of-care imaging in patients with cutaneous T-cell lymphoma

Abstract

Lymph node involvement is associated with a worse prognosis for patients with CTCL. However, current radiologic criteria, including RECIST and WHO criteria, may not provide adequate assessment of nodal involvement or response to therapy. Standard radiologic techniques using unidimensional measurement changes to assess lymph node status and response to therapy in CTCL are prone to inter-observer variability. These shortcomings underscore the need for better methodology for staging and assessment of responses in the lymph nodes of CTCL patients. Using a novel algorithm developed by the Department of Radiology at Columbia University for robust and reliable segmentation of lymph nodes on volumetric CT images, we analyzed the CT scans of a cohort of 40 subjects enrolled in the MAVORIC clinical trial. This algorithm combines volumetric, radiomic, and kinetic modeling and was shown to improve accuracy and reproducibility of response assessments in other cancers. We then explored relationships between measured lymph node changes and observed clinical responses. Our radiologists were blinded to the clinical outcomes. In the MAVORIC trial, an investigator and independent reviewer characterized clinical responses in the skin, blood, and lymph nodes. In our cohort, 50% of lymph node response assessments differed between the investigator and independent reviewer. Using our novel CT volumetric technique, we were able to demonstrate remarkable consistency in nodal assessments. Decrease in lymph node volume showed significant correlation with clinical responses in the skin and blood but increase in lymph node volume was not predictive of skin and blood responses or disease progression. For the first time, we were able to provide a novel validated tool for accurate staging and response assessment, applicable to future clinical trials and clinical practice.