Gene Definition Exceptions
Genotype determination is based on CPIC gene definition tables, with modifications for the following genes:
SLCO1B1
CPIC provides recommendations based on the SLCO1B1 star allele genotype. The CPIC guideline for statins and SLCO1B1, ABCG2, and CYP2C9 (PMID:35152405) includes the following excerpt:
The most common and well-studied variant in SLCO1B1 is c.521T>C (rs4149056), and can be genotyped alone (e.g., PCR-based single SNV assay) or multiplexed on a variety of array-based platforms. All SLCO1B1 genetic tests should interrogate c.521T>C; however, while other less common variants in this gene may have limited evidence to guide action, they may also be important.
PharmCAT attempts to determine the star allele genotype for SLCO1B1, but in cases where no call can be determined it provides the CPIC recommendation based on the rs4149056 variant genotype.
DPYD
The CPIC DPYD allele definition file includes variants that have been assigned normal function (activity value 1), decreased function (activity value 0.5), or no function (activity value 0). According to the CPIC guideline for fluoropyrimidines and DPYD (PMID:29152729), the DPYD phenotype is assigned using a gene activity score, which is calculated as the sum of the activity scores of the two DPYD variants with the lowest variant activity score.
If two different decreased/no function variants are present, they are presumed to be on different gene copies. Irrespective of the presence of decreased/no function variants, patients may carry multiple normal function variants. Common normal function variants may be located on the same gene copy as other normal function variants or decreased/no function variants.
More details are available on the PharmGKB's CPIC DPYD reference page. The DPYD Allele Functionality Table has function assignment information and the DPYD Diplotype-Phenotype Table includes example translations considering one or two variants.
Calling DPYD named alleles
Note: the combinations research flag is ignored when calling DPYD.
Effectively phased data is unphased data that is homozygous at all positions or is heterozygous at a single a position. Since we can effectively predict the alleles on each chromosome in this situation, we can treat this data as we would phased data.
Phased or effectively phased data
If phased or effectively phased data is provided in the VCF file, the Named Allele Matcher produces an output that lists all detected variants per allele. For example: [c.498G>A + c.2582A>G]/[c.2846A>T + c.2933A>G] . If no variants are found on an allele, the Named Allele Matcher returns Reference for that allele.
Unphased data
If unphased data (that cannot be considered effectively phased) is provided in the VCF file, and the data are not homozygous at all positions, the Named Allele Matcher will not attempt to call a diplotype. Instead, it produces a list of all detected DPYD variants in the sample. It will, however, check if variants can be called on both strands. If so, it will call the variant twice. For example: c.1627A>G (*5), c.1905+1G>A (*2A), c.1905+1G>A (*2A). If the sample doesn't contain variants at the positions from the allele definition file, and/or if those positions are omitted from the vcf file, the Named Allele Matcher returns Reference.
Calling DPYD allele functionality and phenotype
The report lists the respective allele functionality for each variant and for Reference. If a diplotype was called from phased/all-homozygous data, the lowest function variants on each strand will be used to determine the gene activity score and DPYD phenotype. Otherwise, the two lowest function variants found are used to determine the gene activity score and DPYD phenotype. The phenotype and gene activity score are utilized to retrieve the corresponding drug recommendations.
Calling the HapB3 allele
DPYD variants included in the CPIC and DPWG guidelines are single variants except for c.1129-5923C>G, c.1236G>A (HapB3). HapB3 consists of an exonic SNP at rs56038477 and an intronic SNP at rs75017182. Both variants have been found in high linkage, however this is not the case for every sample. HapB3's decreased function is thought to be due to the rs75017182 (intron) SNP, while the exon SNP is a synonymous substitution (p.Glu412=).
The Named Allele Matcher will rely on the intronic SNP (rs75017182) to call HapB3 if it's available, and only use the exonic SNP (rs56038477) when it is not. If the VCF file input is conflicting between the two variants, or if the exonic SNP is missing, the Named Allele Matcher will use the input for the intronic SNP to determine if HapB3 is present and include a warning. If the intronic SNP is missing in the VCF file, the Named Allele Matcher will use the input for the exonic SNP to determine if HapB3 is present and include a warning that the intronic SNP is missing and should be genotyped to confirm the presence or absence of HapB3 and decreased function.
DPWG recommendation
As of October 2022, DPWG recommendations are available for 4 DPYD variation:
c.1129-5923C>G, c.1236G>A (HapB3)2846A>Tc.1905+1G>A (*2A)1679T>G (*13)
When inferring gene activity score and phenotype from the two variants with the lowest activity value (unphased data) or the lowest per strand (phased data), PharmCAT uses the variants that are included in both CPIC and DPWG if more than one variant with the same activity value is found.
For example, if a sample has been called with a diplotype of [c.1905+1G>A (*2A) + c.2933A>G]/c.498G>A, which is composed of c.1905+1G>A (*2A) (no function), c.2933A>G (no function, unknown to DPWG), c.498G>A (normal function), the inferred diplotype used to look up the DPYD phenotype and recommendation will be c.1905+1G>A (*2A)/c.498G>A rather than c.2933A>G/c.498G>A.
Furthermore, to increase the likelihood of a match with DPWG, PharmCAT treats any variant that is unknown to DPWG but has a normal function in CPIC as Reference (i.e. normal function).
For example, in the above sample, the inferred genotype c.1905+1G>A (*2A)/c.498G>A will be further translated to c.1905+1G>A (*2A)/Reference and used to query DPWG data. Since c.1905+1G>A (*2A) is a no function variant included in the DPWG data, DPWG guidance for c.1905+1G>A (*2A)/Reference will be included in the report.
CYP3A4
PharmGKB annotates PGx-based drug dosing guidelines published by the Royal Dutch Association for the Advancement of Pharmacy - Pharmacogenetics Working Group (DPWG). PharmGKB curates allele function assignments and phenotype mappings from the DPWG to provide genotype specific DPWG guideline recommendations. Where possible, PharmGKB maps DPWG terms to CPIC terms, as outlined on PharmGKB.
CYP3A4 is currently not part of a CPIC guideline. Since the DPWG CYP3A4 documentation includes limit variant notations for the included alleles (only *16, *20, and *22 have variant positions specified, document from March 2022) PharmCAT relies on PharmVar CYP3A4 allele definitions. The CYP3A4 *16, *20 and *22 definitions are the same in the DPWG CYP3A4 gene document and PharmVar.
G6PD
As of June 2023, PharmCAT does not call the following two G6PD alleles:
- Mediterranean Haplotype (different from the Mediterranean, Dallas, Panama, Sassari, Cagliari, Birmingham)
- Villeurbanne
Mediterranean Haplotype is defined by rs5030868 (shared with Mediterranean, Dallas, Panama, Sassari, Cagliari, Birmingham allele) and rs2230037 G>A (exclusive to Mediterranean Haplotype). PharmCAT excluded rs2230037 as the reference genotype in the Mediterranean Haplotype (G) does not align with the GRCh38 reference base (A). As a result, Mediterranean Haplotype is currently not called in PharmCAT.
Instead, individuals with the Mediterranean Haplotype will be recognized as having Mediterranean, Dallas, Panama, Sassari, Cagliari, Birmingham. Both alleles have the same function (deficient); therefore, the phenotype assignment and subsequent drug prescribing recommendations are not affected.
It remains challenging to properly represent SNPs and INDELs that share the same genetic positions in a VCF file. This affects NC_000023.11:g.154532991154532993delGGT and rs5030869 in the _G6PD allele definitions. Left-aligned, normalized NC_000023.11:g.154532991154532993delGGT is located at the same genomic position as rs5030869. This poses nontrivial challenges to properly format variant representations for both SNP and INDEL in a VCF file and for PharmCAT to accurately recognize any genetic variants of this kind at any allele-defining position. Due to this challenge, NC_000023.11:g.154532991_154532993delGGT, which defines Villeurbanne, is currently excluded from the _G6PD allele definitions in PharmCAT.