Kevin Smart (PhD), Christophe Boetsch (PharmD) , Francois Mercier (PhD), Valerie Cosson (PhD), Vincent Buchheit (MSc), Oliver Krieter (MD), Alex Phipps (PhD), Benjamin Ribba (PhD)
Roche Innovation Centre Welwyn, UK
Objectives: Longitudinal modelling of tumor lesion diameter data is a useful technique allowing extraction of key metrics of tumor shrinkage, visualization of tumor burden over time and comparison of effect between different treatments. In this work, this technique was applied to platinum (Pt)-resistant/ refractory Ovarian Cancer (PROC) patients treated with Vanucizumab, a novel bi-specific IgG-like antibody directed against both VEGF-A and ANG2, two key factors in tumor angiogenesis.
Methods: Longitudinal data from forty PROC patients treated with vanucizumab (30 mg/kg IV Q2W) were analyzed using non-linear mixed effects modelling techniques. Patients underwent tumor assessment by CT scan at screening, (approximately 8-weeks prior to Cycle 1 Day 1) and then every 8 weeks (wks) while on study. Overall, the data set contained 143 observations (mean 3.6, range 2-6 per patient), and pts received on average 9 cycles of treatment. Data were analyzed using the population model described by Claret et al (2009), using the SAEM algorithm of NonMEM 7.2. The analysis provided quantitative estimation of metrics of efficacy, such as maximum shrinkage and time to tumor regrowth, together with critical kinetic parameters, such as the natural growth rate of the tumor, and resistance to treatment. Resultant model parameters were used to simulate data from 1000 random patients, whose maximum shrinkage and time to regrowth were summarized, allowing comparison with other therapies.
Results: The maximum tumor shrinkage from baseline was calculated to be 49% (95% CI: 47-52%), with a time to re-growth of 5.5 months (95% CI: 5.2-5.7 months). The population estimate of the resistance parameter was 0.02 day-1. The unperturbed tumor growth rate was estimated to be 0.001 day-1.
Conclusions: The Claret model adequately described the data in PROC patients. The estimated values of the tumor size metrics provided refined information on the clinical activity of Vanucizumab, in addition to those supplied by RECIST criteria.
References:
[1] Model-Based Prediction of Phase III Overall Survival in Colorectal Cancer on the Basis of Phase II Tumor Dynamics. Laurent Claret et al, 2013. JCO 27: 4103-4108.
Reference: PAGE 25 (2016) Abstr 5923 [www.page-meeting.org/?abstract=5923]
Poster: Drug/Disease modeling - Oncology