Dose-finding design using mixed-effect proportional odds model for longitudinal graded toxicity data in phase I oncology clinical trials.

Abstract : Phase I oncology clinical trials are designed to identify the optimal dose that will be recommended for phase II trials. This dose is typically defined as the dose associated with a certain probability of severe toxicity during the first cycle of treatment, although toxicity is repeatedly measured over cycles on an ordinal scale. We propose a new adaptive dose-finding design using longitudinal measurements of ordinal toxic adverse events, with proportional odds mixed-effect models. Likelihood-based inference is implemented. The optimal dose is then the dose producing a target rate of severe toxicity per cycle. This model can also be used to identify cumulative or late toxicities. The performances of this approach were compared with those of the continual reassessment method in a simulation study. Operating characteristics were evaluated in terms of correct identification of the target dose, distribution of the doses allocated and power to detect trends in the risk of toxicities over time. This approach was also used to reanalyse data from a phase I oncology trial. Use of a proportional odds mixed-effect model appears to be feasible in phase I dose-finding trials, increases the ability of selecting the correct dose and provides a tool to detect cumulative effects.
Document type :
Journal articles
Liste complète des métadonnées

https://hal.archives-ouvertes.fr/hal-00936325
Contributor : Adélaîde Doussau <>
Submitted on : Friday, January 24, 2014 - 7:49:07 PM
Last modification on : Thursday, February 7, 2019 - 5:53:52 PM

Links full text

Identifiers

Citation

Adélaïde Doussau, Rodolphe Thiébaut, Xavier Paoletti. Dose-finding design using mixed-effect proportional odds model for longitudinal graded toxicity data in phase I oncology clinical trials.. Statistics in Medicine, Wiley-Blackwell, 2013, 32 (30), pp.5430-47. ⟨10.1002/sim.5960⟩. ⟨hal-00936325⟩

Share

Metrics

Record views

242