Power and Type 1 Error of Tumour Size Metrics Used to Predict Survival
Nick Holford
University of Auckland
Objectives: A quantitative disease progress model for non-small cell lung cancer growth and response to treatment was described in 2008 [1]. Subsequent reports have described a link of tumour size with survival but the link between drug exposure and response has either not been considered [2] or has not been based on plausible pharmacological mechanisms [3,4]. It is expected that mechanism based pharmacological models applied to effects on tumour growth will be more robust and better suited for prediction of suitable doses and dosing schedules.
Metrics based on tumour size to baseline ratio at 6 weeks (TSR6 [2]) or baseline plus fractional change from baseline at 8 weeks (S0FC8 [5]) and the time to tumour growth with Weibull baseline hazard (WTTG [6]) have been proposed to detect treatment effects on survival. A hazard based approach to prediction of survival would suggest that consideration of the time course of disease progress would be more likely to be informative [7].
The objectives of this study were to determine the tumour metric which had acceptable Type 1 error when there was no treatment effect on survival and which had sufficient power to detect a treatment effect on survival.
Methods: A mixed effect modelling approach using simulation has been used to examine the Type 1 error associated with using EBE derived tumour response metrics and to estimate the power of suitable metrics to detect treatment effects. 1000 trials with 100 subjects randomized to placebo or 3 active doses were simulated under the assumption that survival hazard was either not related to tumour size or was proportional to tumour size.
Results: Type 1 error rates for TSR6 (90%), WTTG (29%) and S0FC8 (12%) were greater than nominal 5% when the mean survival time was 1 year. The full time course of predicted tumour size did not have inflated Type 1 error and was the most powerful metric to detect a treatment effect on survival. It is no more complex to compute than other size based EBE metrics.
Conclusions: The full predicted time course of tumour size is recommended for detection of treatment effects on survival that are linked to changes in tumour size.
References:
[1] Tham LS, Wang L, Soo RA, Lee SC, Lee HS, Yong WP, et al. A pharmacodynamic model for the time course of tumor shrinkage by gemcitabine + carboplatin in non-small cell lung cancer patients. Clin Cancer Res. 2008;14(13):4213-8.
[2] Wang Y, Sung C, Dartois C, Ramchandani R, Booth BP, Rock E, et al. Elucidation of relationship between tumor size and survival in non-small-cell lung cancer patients can aid early decision making in clinical drug development. Clin Pharmacol Ther. 2009;86(2):167-74.
[3] Ribba B, Holford N, Magni P, Trocóniz I, Gueorguieva I, Girard P, et al. A review of mixed-effects models of tumor growth and effects of anticancer drug treatment for population analysis. CPT: pharmacometrics & systems pharmacology. 2014;Accepted 15-Mar-2014.
[4] 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. 2009;86(2):136-8.
[5] Claret L, Girard P, Hoff PM, Van Cutsem E, Zuideveld KP, Jorga K, et al. Model-Based Prediction of Phase III Overall Survival in Colorectal Cancer on the Basis of Phase II Tumor Dynamics. J Clin Oncol. 2009:1-7.
[6] Claret L, Gupta M, Han K, Joshi A, Sarapa N, He J, et al. Evaluation of tumor-size response metrics to predict overall survival in Western and chinese patients with first-line metastatic colorectal cancer. J Clin Oncol. 2013;31(17):2110-4.
[7] Holford N. A Time to Event Tutorial for Pharmacometricians. CPT: pharmacomet syst pharmacol. 2013;2:e43 doi:10.1038/psp.2013.18.
