Work package 4: Validation
Lead: University of Tartu
The aim of WP4 is to validate the genetic discoveries in WP2 and prediction and stratification algorithms and systems developed in WP3 by running a pilot intervention using real-world samples. Specific objectives are to:
- Test the validity of the genotype-phenotype associations in independent real-world samples, including RCT samples and yearly updated registry records.
- Perform an intervention for validation of the developed prediction/stratification algorithms in a recall study of real-world patients (reverse-phenotyping).
Description of work
WP4 will operationalise the data and algorithms from WP2, 3 and test the performance in independent samples, including reciprocal validation across RWD-RCT studies. Furthermore, WP4 will perform an intervention to validate the tools, comprising re-phenotyping of individual patients, This WP will also administer a questionnaire to collect feedback from participants relating to their experience of being re-phenotyped and/or receiving of results.
So far, the major known source of variability in drug response caused by differences in the kinetics of drug metabolism (pharmacokinetics) are the polymorphic cytochrome P-450 (CYP) enzymes, the most important family of enzymes catalysing phase I metabolism (102). It has been estimated that six of the CYP enzymes are jointly responsible for 80-90% of the metabolism of all drugs, and CYP2C19 and CYP2D6 are the most relevant for psychiatric drugs. We have, for example, shown that individuals with genotypes causing poor metabolism by CYP2C19 had a 2.68- and 3.3-fold higher serum concentration of sertraline and escitalopram compared to normal metabolisers, with an impact of genotype on the rate of treatment failure (9,103). In another retrospective study (104), we found that the incidence of switching from risperidone to another antipsychotic was 2.94-fold higher among ultrarapid metabolisers and 1.87-fold higher among poor metabolisers, compared to normal metabolizers (104).
Based on known pharmacogenetic associations, we have built an algorithm for the translation of genotype and sequencing data into treatment recommendations (43), and the aim of this WP is to develop these methods further, to combine current pharmacogenetic approaches with the polygenic and AI/ML algorithms developed in WP3, and validate these on RWD from EHRs and drug prescription registries, and data from RCTs.
Key WP4 tasks
- Validation of genetic variants (both rare and common) associated with treatment outcomes in independent samples.
- Testing of the validity of the prediction/stratification algorithm in independent RWD samples from WP1.
- Testing of the validity of the prediction/stratification algorithms in new real-world patients by reverse phenotyping.
- Surveying the response of the participants of the recall study.