Model-Based Meta-Analysis

What is Model-Based Meta-Analysis?

Model-based meta-analysis (MBMA) combines meta-analysis that uses statistical methods to pool data across disparate studies along with a pharmacological or PK/PD model to compare the efficacy or safety profiles of different competitor drugs. By integrating a thorough review of the available data/literature along with an appropriate pharmacological or therapeutic hypothesis, MBMA allows the comparison of drugs that work through different modalities and/or different targets but in a competitive therapeutic space.

When to use Model-Based Meta-Analysis

MBMA allows a pharmaceutical organization to map the competitive landscape, identify minimum thresholds for efficacy and safety, identify best-in-class properties and benchmark its therapy against its competitors.

Example questions addressed by MBMA:

  • Is there a relationship between a class of mechanism and efficacy/safety?
  • Can a relationship be derived between model derived parameters and clinical endpoints?
  • Based upon my compound’s PK properties, can I benchmark its efficacy and potential for tox to other competitors?
  • Is the tox that's being seen in trials related only to specific therapeutics or is it a function of the therapeutic class?

Model-Based Meta-Analysis Example

Challenge: Our partner's drug had an unintended effect of penetrating the blood brain barrier and bounding to the serotonin receptor. They were concerned whether this would lead to the unintended side effect of serotonin syndrome.

Solution: The PK and binding properties of several CNS drugs of different classes that bound to the serotonin receptor were collected along with the associated tremor incidence which is a common symptom of serotonin syndrome. A PK model was used to determine the brain exposure of each drug and the ratio of exposure to its IC50 in a serotonin receptor assay was computed. When the tremor incidence was plotted as a function of this ratio, an Emax model could be fitted relating the exposure metric (ratio) to the tremor incidence. Bootstrapping was performed to determine the 95% confidence interval of this relationship. Based upon this, the probability of tremor incidence for the customer’s drug was estimated.

Free Consultation

What scientific questions are you trying to answer? We'll put you in touch with the right person on our team. 

Contact us