Evaluation of Tumor Size Metrics to Predict Survival in Advanced Gastric Cancer
AL. Quartino (1), L. Claret (2), J. Li (1), B. Lum (1), J. Visich (1), R. Bruno (2), J. Jin (1)
(1) Genentech Research and Early Development (gRED), Roche, South San Francisco, CA (2) Pharsight Consulting Services, Pharsight, a CertaraTM Company, Marseille, France
Objectives: A disease model framework has been successfully applied to predict overall survival (OS) in cancer patients based on observed longitudinal tumor size data (1-4) to aid early clinical development decision-making (4-6). The aim of this project is to evaluate metrics of tumor size response and prognostic factors to predict OS in patients with HER2 positive advanced gastric cancer (AGC).
Methods: The change in tumor size in AGC patients following treatment with trastuzumab plus chemotherapy (n=228) or chemotherapy (n=228) in the Phase III study ToGA was described by longitudinal tumor size models; the simplified tumor growth inhibition model (sTGI) (1) or the empirical tumor size model (7). Model-predicted metrics of tumor response (e.g. time to tumor growth (TTG) derived from the sTGI model and the tumor growth rate parameter (G) in the empirical tumor size model), patients characteristics and drug effect were evaluated as predictors for OS in a parametric survival model assuming a log-logistic density function for the survival time distribution. The predictors were explored in multivariate analysis: backward elimination (p<0.01) of the covariates significant (p<0.05) in univariate non-parametric Cox regression. The models were assessed by posterior predictive checks where the OS and hazard ratios (HR) of trastuzumab plus chemotherapy vs. chemotherapy were simulated in multiple replicates (n=1000) of the original study.
Results: The best predictor for OS was G, followed by TTG (ΔOFV 7.5). The trastuzumab effect on G fully captured the trastuzumab effect on survival. In addition to G, survival was associated with baseline ECOG performance status, number of metastatic sites, HER2 expression, Asian origin and serum albumin. Simulations showed that the model accurately predicted the OS distribution in each study arm and subpopulation as well as trastuzumab HRs (e.g. model prediction [95% prediction interval]: 0.71 [0.58 - 0.86] vs. 0.65 for OS in trastuzumab plus chemotherapy).
Conclusions: This analysis propose that the metrics (G and TTG) of longitudinal tumor size response models are good predictors of OS and fully captured the effect of trastuzumab on survival in AGC patients including the shorter survival observed in patients with low trastuzumab exposure (Cmin) (8). The identified prognostic baseline factors for survival are in line with literature (8, 9). The developed disease model for AGC patients is drug-independent and thus is a useful tool in design and evaluation of clinical trials of also new investigational agents under development for treatment of HER2 positive AGC and allows early prediction of OS.
References:
[1] Claret L, Girard P, O'Shaughnessy J et al. Model-based predictions of expected anti-tumor response and survival in Phase III studies based on phase II data of an investigational agent. J. Clin. Oncol. 24, 307s (suppl, abstract 2530), 2006.
[2] Claret L, Girard P, Hoff PM et al. Model-based prediction of Phase III overall survival in colorectal cancer based on Phase II tumor dynamics. J. Clin. Oncol. 27, 4103-4108, 2009.
[3] Wang Y, Sung YC, Dartois C et al. Tumor size-survival relationship in non-small cell lung cancer patients to aid early clinical development decision making. Clin. Pharmacol. Ther. 86, 167-174, 2009.
[4] Bruno R, Jonsson F, Zaki M et al. Simulation of clinical outcome for pomalidomide plus low-dose dexamethasone in patients with refractory multiple myeloma based on week 8 M-protein response. Blood, 118, 1881 (abstract), 2011.
[5] Bruno R, Claret L. On the use of change in tumor size to predict survival in clinical oncology studies: Toward a new paradigm to design and evaluate Phase II studies. Clin. Pharmacol. Ther. 86, 136-138, 2009.
[6] Claret L, Lu J-F, Bruno R et al. Simulations using a drug-disease modeling framework and phase II data predict phase III survival outcome in first-line non-small-cell-lung cancer. Clin. Pharmacol. Ther. 92, 631-634, 2012
[7] Stein WD, Gulley JL, Schlom J, et al. Tumor regression and growth rates determined in fixe intramural NCI prostate cancer trials: the grothe rate constant as an indicatior of therapeutic efficacy. Clin. Cancer Res. 17, 907-917, 2011.
[8] Yang J, Zhao H, Garnett C et al. The combination of exposure-response and case-control analysis in regulatory decision making. J Clin Pharmacol May 2012, DOI:10.1177/0091270012445206
[9] Bang YJ, Cutsem EV, Feyereislova A et al. Trastuzumab in combination with chemotherapy versus chemotherapy alone for treatment of HER2-positive advanced gastric or gastro-oesophageal junction cancer (ToGA): a phase 3, open-label, randomized controlled trial. The Lancet, 376, 687-697, 2010
