Emilie Schindler (1), Michael A. Amantea (2), Lena E. Friberg (1)
(1) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden; (2) Pfizer Global Research and Development
Objectives: To investigate the predictive ability of drug exposure, circulating biomarkers (the vascular endothelial growth factor VEGF and its soluble receptors sVEGFR-1, sVEGFR-2, sVEGFR-3) and tumor sum of longest diameters (SLD) on overall survival (OS) in patients with metastatic renal cell carcinoma.
Methods: OS data were available from a phase II study including 64 Japanese mRCC patients treated with oral axitinib at a starting dose of 5 mg b.i.d continuously and followed up for a median time of 65 weeks [1]. Individual parameter estimates from previously-developed models were used to predict daily area under the curve (AUC) and the time-course of biomarkers and SLD [2]. Patients were assumed to stay on treatment until time of death or censoring. Parametric time-to-event models were fitted to the OS and to the censoring data. Baseline hazard distribution was investigated as constant, Weibull, Gompertz, log-normal and log-logistic. Baseline Eastern Cooperative Oncology Group (ECOG) status, daily AUC, the model-predicted time course and absolute/relative change from baseline over time in the biomarkers and in SLD, and the relative SLD change from baseline at week 6 were evaluated as predictors of OS, one by one and in combination.
Results: A log-logistic model with a shape factor greater than 1 best described the baseline hazard for OS which first rises, then decreases monotonically. A competing log-logistic function described the hazard of being censored. In the univariate analysis SLD time-course (dOFV=-20.9), baseline SLD (dOFV=-13.8) and sVEGFR-2 time-course (dOFV=-7.9) were statistically significant predictors of OS (p=0.01). When SLD time-course was included in the model none of the other predictors further improved the fit. Larger SLD were associated with a lower survival probability.
Conclusions: The SLD time-course best predicted OS in axitinib–treated mRCC patients. These results differ from the ones in sunitinib-treated gastro-intestinal stromal tumor patients, where sVEGFR-3 was a better predictor of OS than SLD [3]. In a next step, safety biomarkers including diastolic blood pressure will be modelled and investigated as predictors of OS
Acknowledgements: This work was supported by the DDMoRe project (www.ddmore.eu) and the Swedish Cancer Society.References:
[1] Tomita et al. Eur J Cancer (2011); 47: 2592-2602.
[2] Schindler et al. PAGE (2013); Abstr 2859.
[3] Hansson et al. CPT:PSP (2013); 2: e84.
Reference: PAGE 25 () Abstr 5877 [www.page-meeting.org/?abstract=5877]
Poster: Drug/Disease modeling - Oncology