PCN95 - A Multi-State Model of Metatstatic Colorectal Cancer

Abstract

The aim of this study is to develop and validate a decision-analytic model describing the current course of disease, including treatment, in metastatic colorectal cancer. This baseline model will serve as the comparator in analyses of the (cost-) effectiveness of new treatment strategies. An individual-based micro-simulation model was constructed based on the disease states a patient may experience after a diagnosis of metastatic colorectal cancer. The states include first-line second-line and third-line treatment, as well as states of progression of disease after first-, second- or third-line, finally a death state is included. Time spent in each disease state was predicted using log-logistic, log-normal or weibull survival models, each dependent on a number of patient characteristics. All survival models and patient characteristics were based on patient-level data, provided by the CAIRO trial (NCT00312000). Two oncologists evaluated the model for face validity, the model was further validated by comparing various model outcomes with the original data, the national cancer registry and a population based study. There were no significant differences in patient and treatment characteristics, nor intermediate and overall survival estimates between the simulated and original patient-level data. External validation with national cancer registry data showed few differences in survival with the simulated data. Additionally the simulated survival did not significantly differ from the survival as recorded in a pilot oxaliplatin study of 119 patients who were observed in the same timeframe as the RCT. The micro-simulation decision model described in this article underwent an internal and external validation and can be used to evaluate new possibilities for research and treatment in metastatic colorectal cancer.