A method to summarize toxicity in cancer randomized clinical trials

Background/Aims: Despite development of clinical "value frameworks" by national and international groups, there remains no generally accepted method to summarize toxicity in cancer clinical trials. We explored ways to simplify toxicity data of an arm of a cancer clinical trial to a single value, termed a weighted toxicity score (WTS). Methods: We compiled 58 randomized clinical trials of FDA-approved kinase-directed inhibitors. We generated 5 models, each of which assigned different weights for each observed grade 1 to 4 toxicity. For each model, we calculated WTS values as different weighted averages of the sum of the toxicities. We correlated each WTS with the dose reduction rate in each trial, using the dose reduction rate as a clinically relevant surrogate of treatment that is too toxic. The WTS method yielding the strongest correlation with frequency of dose reduction was declared the best model. Results: Nineteen of 58 trials were placebo-controlled and had complete data. Of the 5 models examined, differences in dose reduction rates correlated best with differences in WTS using a model employing a clinician-weighted scale for toxicities (model M5). The WTS difference thus serves as a surrogate for a desired dose reduction rate difference, and could be used to adjust dose/schedule as patients are accrued to a clinical trial. Conclusion: The WTS distills a tabular listing of toxicities of a treatment into a single value, and provides a simple method that can be incorporated into value frameworks, or used to guide discussion of the risks and benefits of systemic therapy.

https://ift.tt/2I7Sr5o