O-014 Evaluation of depth of response within a volumetric model in patients with metastatic colorectal cancer: results of the SIRFLOX study

Abstract

Introduction: Depth of Response (DpR) has been proposed as a measure of efficacy that predicts the long-term treatment outcome for patients with metastatic colorectal cancer (mCRC). DpR is a continuous measure that defines the nadir of tumour response, typically by the sum of the longest diameters of all target lesions compared to baseline. There are limited data on the value of applying DpR to clinical trial results, and there is a lack of data relating DpR to baseline tumour load. In the present analysis we applied a volumetric model to the independent blinded reader RECIST data from SIRFLOX. This study was a multicentre randomised controlled trial of 530 patients with non-resectable, liver-only or liver-dominant mCRC that compared first-line mFOLFOX6 (+ bevacizumab at the investigators' discretion) plus selective internal radiation therapy (SIRT) using Y-90 resin microspheres (SIR-Spheres; Sirtex, Sydney, Australia) to mFOLFOX6 (± bev) alone (Control). Methods: Spherical tumour volume was estimated from the longest unidimensional length for ≤5 target hepatic lesions (RECIST v1.0) in the SIRFLOX ITT population. Finite mixture modelling was used to assess baseline tumour distribution and identify the optimal cut-point for subpopulations within which any potential predictors of DpR could be described. Results: There was a 7% greater decrease in median liver tumour volume in SIRT vs. Control (-75% vs. -68%; p = 0.036). Deepest response (median time to nadir) occurred 60 days later in SIRT than Control (266 vs. 206 days; p < 0.001). A larger treatment effect and longer time to nadir were observed in patients with >12% hepatic tumour burden (median DpR -90% in SIRT vs. -77% in Control, p = 0.003; median nadir 298 vs. 196 days, p < 0.001), compared to those with ≤12% tumour burden (median DpR -93% vs. -94%, p = 0.763; median nadir 243.5 vs. 220 days, p = 0.152). PFS in the liver by Competing Risk analysis was significantly longer in patients with >12% hepatic tumour burden receiving SIRT vs. Control (median 27.2 vs. 13.1 months, p = 0.003), whereas complete response (CR) for SIRT vs. Control was more common in patients with ≤12% liver tumour burden (12.9% vs. 3.5%; p = 0.001). Conclusion: The present study integrates baseline tumour volume into the analysis of volumetric DpR modelled from existing unidimensional target lesion assessment. In SIRFLOX, the addition of SIRT to standard first-line chemotherapy significantly increased hepatic DpR. The impact of SIRT on PFS in the liver was greatest in patients with a baseline tumour burden >12%, whereas the impact on the CR rate was greater where tumour burden was <12%.