Between- and within-patient n-level response surface pathway design in dose-finding studies

Abstract

Sagita Dewi,1 Veronica Kristiansen,2 Steen Lindkær-Jensen,3,4 Stig Larsen1 1Centre for Epidemiology and Biostatistics, Faculty of Veterinary Medicine and Biosciences, Norwegian University of Life Sciences, Ås, Norway; 2Department of Companion Animal Sciences, Faculty of Veterinary Medicine and Biosciences, Norwegian University of Life Sciences, Oslo, Norway; 3Department of Surgery and Cancer, Hammersmith Hospital Campus, Imperial College, Du Cane Road, London, UK; 4Meabco A/S, Copenhagen, Denmark Background: The most commonly used escalation methods in dose-finding studies have obvious weaknesses, and the Bayesian approach is difficult for clinicians to understand and to apply. The study aims were to introduce and assess the performance of clinically based response surface pathway (RSP) design for dose-finding studies, exemplified by one between-patient study, one within-patient study, and simulation studies. Methods: The between-patient study consisted of 15 women suffering from stage IV breast cancer, while the within-patient study consisted of seven female dogs with metastatic mammary cancer. The studies were conducted to determine the maximum tolerated dose (MTD) of a new anticancer agent named BP-C1 using three-level RSP designs. Adjustment of the dose from one design level to the next was based on a k-adjustment factor estimated to ensure coverage of the entire predefined dose window. Patient sequences with an equal number of patients as design levels were included in the between-patient design, whereas the same patients were included in all the design levels in the within-patient design. Results: Four of the five patient sequences in the between-patient study and all seven dogs in the within-patient study reached the upper limit of the dose windows without any increase in toxicity. The MTD of BP-C1 was thus found to be higher than the predefined cumulative dose window for both patient groups. In all three scenarios, the RSP design estimated MTDs better than the traditional 3+3 design; however, the toxicity rates were found to be higher when the target MTD was under the starting dose. Conclusion: The RSP designs do not need an assumed statistical model, and may be useful in estimating MTD, using a minimal sample size. The k-adjustment factor ensures complete dose window coverage and the design utilizes more information by allowing multinomial outcomes. Keywords: dose escalation procedure, maximum tolerated dose, MTD, phase I clinical trial, RSP