Abstract
MATERIALS AND METHODS
We collected retrospective data from N = 223 patients (mean age 67 ± 16 years; 151 men) undergoing immunotherapy per standard-of-care, whose treatment decisions did not follow RECIST. Imaging data were gathered for up to two years post-baseline, continuously evaluating RECIST-defined progression types. We analyzed co-occurrence patterns using odds ratios of progression events and assessed prognostic impacts with time-varying Cox regression models.
CONCLUSION
RECIST-defined progression types often co-occur, suggesting they represent variations of a shared treatment failure process rather than independent events. Future developments should focus on quantitative metrics for progression.
RESULTS
Of 223 patients, 115 (52%) experienced at least one type of progression. Most patients (57%) experienced multiple types of progression. Most patients with only one type of progression did not have follow-up scans to determine if other types of progression would have occurred later. In general, all progression types are likely to co-occur, supported by significant positive odds ratios. New lesions were more frequent early in treatment, while other types showed no clear temporal trends. Non-measurable progression had the strongest association with worse survival (HR, 2.33; CI: 1.45-3.74; p < 0.001), followed by new lesions (HR, 1.73; CI: 1.05-2.86; p = 0.03) and target lesion progression (HR, 1.66; CI: 1.07-2.58; p = 0.02). Non-target lesion progression was rare (9%) and not prognostic.
OBJECTIVES
This study investigates whether different types of RECIST-defined tumor progression reflect distinct biological processes or variations of the same progression.