CRESTAR aims to develop tools to predict

  • who will NOT respond to usual antipsychotics, indicating treatment with clozapine as early as possible,
  • the 1% of patients who will develop potentially fatal side effects, agranulocytosis, which is the main factor limiting clozapine use, and diabetic ketoacidosis, occurring in up to 2% of patients, and often fatal.
  • We will also predict patients likely to be non-responders to all antipsychotics, i.e. extreme TRS, so that they can be stratified in clinical trials.


CRESTAR will address the research question by examining

  • genome-wide association data,
  • genome sequence,
  • epigenetic biomarkers and
  • epidemiological data

in European patient cohorts characterized for treatment response, and adverse drug reaction. CRESTAR will use data from clozapine therapeutic drug monitoring and linked National population medical and pharmacy databases, alongside existing European projects (e.g. OPTIMISE and EU-GEI national initiatives (e.g. UK10K genome sequencing in schizophrenia to identify predictive factors. In parallel CRESTAR will perform health economic research on potential benefits, and ethics and patient-centered research with stakeholders.

Expected results

The outcome of CRESTAR will be biomarker and genomic tests with associated clinical decision making tools, designed to improve pharmacological treatment of schizophrenia in both efficacy (good clinical response) and safety (reduced risk of side-effects).

  • CRESTAR final summary report

    CRESTAR project officially ended 31st October 2015. The final summary report is available here.
  • Improving the outcome in treatment-
    resistant schizophrenia

    The faster schizophrenia patients receive effective treatment, the better their prognosis. EU-funded CRESTAR researchers are establishing a profile of individuals unlikely to respond to the drugs used in first-line treatment, for whom the last-resort antipsychotic clozapine should be considered sooner. They are also improving the safety of this drug.

    Read more: