The lack of standardization of which genes or alleles should be included in pharmacogenetic testing panels is a major barrier to the full implementation of pharmacogenomics in the clinic. In an effort to help clinicians select an appropriate pharmacogenetic test, Dr. Chad Bousman and Dr. Abdullah Al Maruf of the University of Calgary and Dr. Daniel Mueller of the University of Toronto gathered available pharmacogenetic evidence from a number of sources, including PharmGKB, CPIC and the Pharmacogene Variation Consortium(PharmVar). This evidence informed the authors’ recommendations of which genes and variants should be included in a minimum pharmacogenetic testing panel for psychiatry.
For inclusion on the panel, a variant drug interaction needed to be associated with a Level 1A or Level 1B Clinical Annotation in PharmGKB, an FDA or EMA drug label which specifically recommended or required pharmacogenetic testing prior to administering the drug and/or a CPIC or DPWG dosing guideline recommendation. Variants also had to be present at a frequency of >1% in at least two of the seven major Human Genome Diversity Project - Centre d'Etude du Polymorphisme Humain (HGDP-CEPH) populations, which are used in the PharmGKB gene information tables. To ensure that the panel could be easily validated, each variant had to have reference material available through the Genetic TestingReference Materials Coordination (GeT-RM) Program.
Starting from a list of 91 drugs which are used in psychiatry, 448 initial gene-drug interactions were identified. Following visualization of the data using a network map showing the strength of evidence for each interaction, 31 drug-gene pairs were determined to have sufficient evidence to warrant their inclusion on the minimum testing panel. The resulting panel, published in Current Opinion in Psychiatry, contains 16 alleles in the genes CYP2D6, CYP2C9, CYP2C19, HLA-A and HLA-B.
The authors emphasize that this panel should be considered a minimum standard for pharmacogenetic testing in psychiatry and that potential gene-gene interactions are not covered. The panel will also need to be regularly updated to include new genes and variants as new guidelines and published evidence become available.
We would like to note a few relevant points about the paper. Given the reliance on our Level 1A/1B Clinical Annotations, it is important to state that some genes with a high level pharmacogenomic association may have variants or alleles which are not covered by a Level 1A or 1B Clinical Annotation, due to a lack of published evidence specifically studying these alleles. As an example, CYP2D6 has over 100 documented variant alleles. While PharmGKB has multiple Level 1 Clinical Annotations summarizing associations between CYP2D6 alleles and drugs, not every allele in every diplotype that CPIC provides recommendations for will be represented in our clinical annotations.
It is unclear how exactly information from PharmVar was used in the process of designing this panel. While PharmVar provides a wealth of information about pharmacogenetic allele definition and nomenclature, it does not assign a level of evidence to pharmacogenetic variants in the way that PharmGKB assigns a level of evidence in our Clinical Annotations. The paper also does not explicitly state whether drug labels which recommend or require testing were identified using the PharmGKB drug label annotations or by another method. In any case, we would like to emphasize that the PGx level of drug labels annotated in PharmGKB is determined by PharmGKB curators rather than any regulatory body.
Readers should also note that using allele frequencies as a condition for inclusion on a testing panel will, by definition, result in rare pharmacogenetic alleles remaining undetected in patients.