Abstract
Pharmacogenetics focuses on heritable factors that influence drug responses. For a given drug and patient, pharmacogenetics aims to optimise the intended therapeutic benefits, while minimising adverse drug reactions (ADR). Important pharmacogenes include those for drug metabolising enzymes CYP2D6, CYP2C19, and CYP2C9, and the transporter encoding SLCO1B1 and ABCG2 genes. The Understanding Adverse Drug Reactions Using Genomic Sequencing (UDRUGS) project was initiated in 2012 (Chua, 2015; Maggo et al., 2017). As a New Zealand biobank of clinical cases who have experienced a wide range of ADR or ineffectiveness towards drugs, UDRUGS provides a useful database for drug response phenotypes, contributing significantly to the samples used for analysis herein.
The aim of this thesis was to study the application of pharmacogenetics in understanding two main drug response phenotypes: drug ineffectiveness and ADR, in the prescribing of antidepressants, statins, and proton-pump inhibitors (PPI). The role of pharmacogenetics in drug response phenotypes for antidepressants and statins was analysed using existing cohorts drawn from ongoing UDRUGS recruitment, while that of PPI, were new recruits.
Heterogeneity in drug response is a well-recognised challenge in mental health, with high rates of ineffectiveness and incidence of ADR. A total of 52 cases were eligible, who experienced either ADR or ineffectiveness or both towards antidepressants. The allele and genotype-inferred phenotype frequencies of CYP2D6 and CYP2C19 in the cases were determined, and were compared with a reference population (McInnes et al., 2021).
In order to predict the clinical utility of CYP2D6 and CYP2C19 pharmacogenetics in antidepressant response, 45 cases from the cohort which involved gene-drug-response pairs with Clinical Pharmacogenetics Implementation Consortium (CPIC) evidence levels A, A/B, or B were selected for clinical analysis. The cases were assigned as ‘actionable’ if the CYP phenotypes potentially contributed to the observed responses. 79 (CYP2D6-37, CYP2C19-42) gene-antidepressant-response pairs were identified, and out of these, 41% (15/37) of CYP2D6-antidepressant-response pairs and 36% (15/42) of CYP2C19-antidepressant-response pairs were actionable. Overall, a total of 38% pairs, consisting of 48% ADR and 21% ineffectiveness pairs were actionable based on the available CYP2D6 and CYP2C19 genotypes information. This highlighted the relevance of pharmacogenetics in antidepressant prescribing.
The next section of this thesis focussed on statins, which are one of the most commonly prescribed drugs in cardiovascular health, indicated for hypercholesterolemia and secondary prevention of ischaemic events. Statin-induced myotoxicity (SIM) is a common ADR with phenotypes ranging from mild myalgia to fatal rhabdomyolysis. A total of 33 SIM cases were identified from UDRUGS, including eight cases which were presented with well-defined SIM phenotype, and with available whole exome sequencing (WES) data. Allele and genotype-inferred phenotype frequencies of 27 genetic variants shortlisted from literature review (Kee et al., 2020) were determined, and were compared with the reference populations. The comparison showed a significant (p<0.05) over-representation of CYP2C19 intermediate metabolisers (IM) and CYP2C19*2 allele, and under-representation of CYP2C19 rapid metabolisers (RM) and ABCC2 rs717620. However, none were significant after Bonferroni correction.
To date, the association between SIM and prior candidate genes remains unclear including the striking finding of SNP SLCO1B1*5 rs4149056 and the risk of simvastatin-induced myopathy, which have not been replicated consistently throughout the literature. Therefore, the possibility of having rare or undiscovered genetic variants predisposing to the risk of SIM remains. Using WES data from the eight well-defined SIM cases, we explored the role of missense variants using two approaches. First, variants which are rare or low frequency were examined; second, pre-selected variants from a candidate gene list for genes or genetic variants involved in the phenotype of interest such as pharmacokinetics of statins and the muscle structure. The variants were filtered using several pre-defined criteria such as dbNSFP Functional Prediction 3.0 and ClinVar classification. The frequency of the extracted variants in cases were compared with the frequency for non-Finnish Europeans in gnomAD exome.
Taken together, this case cohort highlighted the potential role of genes CYP2C19, ABCC2, ABCA4, RYR1, VKORC1 and MLYCD in predisposing to the risk of SIM. While there are no clear associations between these genes with the pharmacokinetics of statins, and the pathogenesis of SIM, the possibility of these genes and the respective SNP exerting their effects through unknown mechanisms remains, thus, they represent reasonable candidates for future study.
The last section of this thesis studied drug ineffectiveness phenotypes for the PPI omeprazole, in treatment of acid-related disorders including gastro-oesophageal reflux disease (GORD). PPI are the mainstay treatment for GORD, but omeprazole treatment failure in refractory cases remains a clinical challenge. Omeprazole is primarily metabolised by CYP2C19, and two variants lead to increased enzyme activity (rapid or ultrarapid metaboliser status). CYP2C19*17 has long been recognised to increase CYP2C19 activity, and the recently discovered CYP2C:TG haplotype is also associated with this metabolic phenotype. Therefore, I tested the hypothesis that refractory GORD cases which responded poorly to omeprazole may result from a suboptimal omeprazole concentration associated with genetically derived ultrarapid metabolism of CYP2C19. A total of 55 GORD cases who either did not respond to omeprazole or experienced breakthrough heartburn symptoms despite at least eight weeks of omeprazole (≥ 40 mg/day) were recruited. Cases were assigned as ‘with objective GORD evidence’ if they were presented with typical GORD symptoms in addition to having at least one positive pathological GORD test (either gastroscopy or 24-hour oesophageal pH/impedance tests).
Frequency comparison between cases and reference populations were carried out for various combinations of these alleles and inferred phenotypes. No significant differences (p>0.05 for all comparisons) were observed, when the cases were collectively assessed. However, CYP2C:TG homozygotes (CYP2C:TG/TG) were significantly enriched in the subgroup with objective GORD evidence (N=39). This finding suggested that omeprazole treatment failure in GORD is associated with CYP2C:TG/TG, but not with the CYP2C19*17 allele.
For a complex trait such as drug response, pharmacogenetics may be useful to provide insights into the pharmacokinetics and pharmacodynamics of drugs for a given patient. The work of this thesis has provided a further insight into the clinical application of pharmacogenetics in understanding responses associated with the use of antidepressants, statins, and omeprazole. At the same time, this work also highlighted the challenges in elucidating the clinical utility of pharmacogenetics, which is one of the main barriers for implementation in standard clinical practice.