Document Type
Article
Publication Date
2015
Abstract
Most dose-finding clinical trials in oncology aim to find the highest dose yielding an acceptable toxicity profile for patients. The conventional dose-finding framework utilizes a binary toxicity endpoint that treats low to moderate toxicities as irrelevant, ignoring potentially harmful combinations of such toxicities. A handful of novel dose- finding methods have been introduced that combine multiple toxicities across varying grades into a composite toxicity severity score. Toxicity scores provide the advantage of accounting for all toxicity information in a patient profile, but calculation of such scores require prior specification of toxicity severity weights to represent the relative toxicity burden each toxicity type of each grade adds to a toxicity profile if observed. Elicitation of severity weights generally rely on subjective specification, and resulting continuous scores may be confusing in clinical settings. In a statistical framework, we propose a novel method of estimating toxicity weights via a cumulative logit model, assuming there to be a latent continuous toxicity score characterized by the set of observed toxicity types and grades a patient exhibits. Toxicity scores are directly associated with an ordinal outcome assigned to toxicity profiles by clinicians, which correspond to simple dose escalation decisions. The toxicity score elicitation method (TSEM) produces an accurate toxicity scoring system through evaluation of a balanced subset of toxicity profiles in terms of severity, and we present an adaptive weight finding algorithm to facilitate this. This approach bridges the gap between relating continuous toxicity scores to clinically logical ordinal outcomes akin to traditional toxicity grades, and provides an objective method for determining toxicity weights and scores.
Recommended Citation
O'Connell, Nathaniel S. and Garrett-Mayer, Elizabeth, "Latent Variable Approach to Elicit Continuous Toxicity Scores and Severity Weights for Multiple Toxicities in Dose-Finding Oncology Trials" (2015). MUSC Department of Public Health Sciences Working Papers. 9.
https://medica-musc.researchcommons.org/workingpapers/9
Comments
Manuscript for an article written by researchers for the Medical University of South Carolina Department of Public Health Sciences. Includes abstract, references, and tables.