Colorectal cancer liver metastases (CRLM) present a significant clinical challenge, with approximately 80% of patients initially diagnosed as having unresectable disease due to extensive tumor burden or critical lesion locations. Systemic induction chemotherapy aims to downsize tumors, enabling potential secondary local interventions, or to prolong survival in palliative contexts. Currently, tumor response assessment largely relies on the RECIST1.1 criteria, which measure changes in the longest diameter of up to two target lesions per organ. However, RECIST1.1 has notable shortcomings, including limited reproducibility and insufficient representation of overall tumor burden, as it neglects volumetric changes across all metastases. Advances in imaging and deep learning have made volumetric assessment of total tumor volume (TTV) feasible, with preliminary data suggesting prognostic superiority to RECIST1.1. Yet, the clinical validity, generalizability, and predictive utility of TTV in patients undergoing systemic induction therapy remain to be elucidated.
Study Design and Methods
This ancillary study analyzed data from the phase 3 CAIRO5 trial (NCT02162563), including 425 patients with initially unresectable liver-only CRLM who received induction systemic chemotherapy. Patients were randomized based on primary tumor location and RAS/BRAF mutation status to receive systemic regimens including FOLFOX-/FOLFIRI-bevacizumab, FOLFOXIRI-bevacizumab, or FOLFOX-/FOLFIRI-panitumumab. Contrast-enhanced CT scans prior to treatment and at first follow-up were used to quantify baseline TTV and changes post-treatment using a semi-automatic volumetric segmentation approach validated by expert radiologists. RECIST1.1 assessments, including sum of diameters and categorization of response, were performed by an expert panel. The main endpoint was overall survival (OS), with statistical analyses employing multivariable Cox regression models adjusted for baseline clinical and molecular variables, and flexible parametric survival modeling to assess prognostic and predictive capacities of TTV relative to RECIST1.1.
