Population analysis of the DCE-MRI response of liver metastases to a single dose of bevacizumab in CRC patients
G.Z. Ferl(1), J.P. O'Connor(2), R.E. Port(1)
(1)Early Development PKPD, Genentech, South San Francisco, California, USA; (2)Imaging Science and Biomedical Engineering, School of Cancer and Imaging Sciences, University of Manchester
Objectives: Our objective is to analyze the time course of the Ktrans response to a single dose of bevacizumab (bev) based on Dynamic Contrast Enhanced MRI data from a previously published phase II clinical trial [1], where Ktrans is a measure of vascular permeability to contrast agent, surface area and rate of tissue perfusion. Decrease in Ktrans is observed shortly after dosing, however, the subsequent dynamics of this parameter over a 12 day period are not entirely clear due to inter-patient variability.
Methods: 2 baseline (3 and 2 days predose) plus 4 subsequent DCE-MRI scans following a single 10 mg/kg dose of bev (4hr, 2d, 8d and 12d) were obtained from 10 patients, each with between 1-6 colorectal liver metastases (26 lesions total). Two patients had missing data points; the first (4 lesions) missed the 8d scan while the other (3 lesions) missed the 8d and 12d scans. Ktrans values for each scan were estimated [1] using the extended Tofts version of the Kety compartmental model [2]. Ktrans dynamics plus inter-individual (IIV) and inter-lesion (ILV) variability were described using an indirect response model with feedback, implemented in NONMEM. True individual baseline Ktrans was estimated for each lesion by assuming that it varies around the observed average baseline with variance (sigma^2)/2 [3]. Post-hoc analysis provided parameter estimates for each lesion and simulated Ktrans profiles were produced using the estimated population mean and variance parameters.
Results: ILV was not identifiable and was assumed to be zero, except for inter-lesion variability in baseline Ktrans. Inter-patient variability was significant. IIV of model parameters is identifiable and estimated to be larger for parameters in the indirect response equation than for the feedback equation (66% CV vs. 48% CV). The predicted population response describes a rapid decrease of Ktrans to 57% of baseline followed by a slower return to baseline within 12 days after a single dose of bev, with rebound over baseline in some patients.
Conclusions: There is a significant amount of inter-patient variability (IIV) in Ktrans response to a single dose of bev; inter-lesion variability (ILV) is not detectable within these data. However, a population trend of fast decrease of Ktrans to 57% of baseline followed by a return to baseline within 12 days is clear from this analysis. Based on data from a previous PK analysis, this does not appear to be driven by a decline in plasma bevacizumab concentration.
References:
[1] O'Connor et al. Quantifying antivascular effects of monoclonal antibodies to vascular endothelial growth factor: insights from imaging. Clin Cancer Res. 2009 Nov 1;15(21):6674-82.
[2] Tofts PS, Kermode AG. Measurement of the blood-brain barrier permeability and leakage space using dynamic MR imaging. 1. Fundamental concepts. Magn Reson Med. 1991 Feb;17(2):357-67.
[3] Port RE et al. Predicting the time course of haemoglobin in children treated with erythropoietin for renal anaemia. Br J Clin Pharmacol. Br J Clin Pharmacol. 1998 Nov;46(5):461-6.
