Keywords: Bayesian Network Meta-analysis, Dose-response
Network meta-analysis (NMA) has been increasingly used to integrate available information from randomized controlled trial (RCT) and to make comparison about on relative efficacy among different treatments in lieuack of direct evidence. Trial design and goon/no- go decisions are always often informed inferred by the outputs from of NMA. In NMA, different doses from the same compound are usually considered as independent treatments and ignore dose-response relationship. Missing theOmitting dose-response relationship will result in larger variance in estimation and disconnected network. In this poster, we extended and applied the Emax Bayesian NMA model proposed by Mawdsley (2016) to biologics with different classes on treating Psoriasis. The endpoint is PASI90 (at least a 90% reduction in PASI scoring), which is a binary measure. Prior information for the PK parameter about each compound can be incorporated into the model. We allow each class to have a common Emax, which is the maximum effect, while different compound in each class has its own ED50 (dose at which half maximum effect obtained). The results derive useful insights into dose-response relationship among biologics, improve model-fitting, and help making prediction about efficacy for new doses introduced in later trial stage.