Pharmacogenomics in 2023: Big studies, big results, big implications, big responsibilities: Editorial

Abstract

This themed issue is concerned with recent developments in pharmacogenomics and originates from the 19th IUPHAR World Congress of Basic and Clinical Pharmacology 2023 (WCP2023) event in Glasgow, Scotland, where a well-attended symposium featured presentations from Masaru Koido (Japan), Volker Lauschke (Germany and Sweden) and Erika Cecchin (Italy), all early career scientists with rising reputations in the field. These speakers have provided invited review articles for this issue.

The term pharmacogenomics now encompasses the area of pharmacogenetics, which had its origins in the 1950s¹ and developed subsequently as an important branch of clinical pharmacology. British Journal of Clinical Pharmacology has been a major contributor to publications of key research in this field since the 1970s and 1980s when some of the initial reports on the debrisoquine polymorphism and the later identification of CYP2D6 as the relevant enzyme involved appeared.^{2, 3} Follow-up studies on the relevance of CYP2D6 phenotype and genotype to metabolism of drugs such as beta-adrenoreceptor antagonists and opioids subsequently appeared in the journal^4-6 with coverage also extending to other highly polymorphic P450s of clinical pharmacology importance such as CYP2C9 and CYP2C19.^{7, 8} In spite of the extensive knowledge and understanding achieved in this field during the 50 years of British Journal of Clinical Pharmacology,⁹ clinical implementation of pharmacogenomics is still limited. However, appreciation of the importance of personalized prescribing based on patient genotype is now widespread, as reported recently by Turner and colleagues who reviewed a joint report from Royal College of Physicians and British Pharmacological Society on using pharmacogenomics to improve patient outcomes.¹⁰

In parallel with development of pharmacogenomics, population studies worldwide, including the UK Biobank, FinnGen and All of Us, feature extremely large sample sizes while parallel developments in genomics in relation to performing and analysing whole genome sequencing as well as the availability of complementary ‘omics’ data such as serum proteomics now provide a huge amount of data. Dealing with these data will require use of approaches such as Machine Learning (ML), but this should further increase pharmacogenomics knowledge and facilitate personalized prescribing.

In this issue, Tremmel and colleagues¹¹ discuss the challenges that still exist in genome sequencing, especially in relation to some pharmacogenes such as CYP2D6. They also consider the range of bioinformatic tools that can be applied to sequencing data to extract pharmacogenomic data and use of artificial intelligence (AI) and ML to predict pharmacological effects and potentially therapeutic response (efficacy-adverse). The need to confirm computational predictions by using experimental approaches is also considered, including the use of animal models ranging from zebrafish to mice.

Koido¹² focusses more specifically on the continuing value of genome-wide association study (GWAS) data and the possibility of using ML and other advanced statistical approaches to understand complex pharmacogenomic problems such as identification of those at risk of drug-induced liver injury (DILI) using complex polygenic risk scores. The use of novel modelling on the GWAS data followed by use of cell-based models to confirm predictions on DILI risk factors is described. This approach is very much in line with the recommendations above from Tremmel et al.¹¹ A further use of GWAS data more generally to understand the processes of transcriptional regulation by variants in noncoding regions of the genome by use of transcriptomic and epigenomic combined with genomics, complemented by high throughput methods such as massively parallel reporter assays on cultured cells is described. This issue of noncoding variants is relevant both to pharmacogenomics and susceptibility to common diseases.

Peruzzi and colleagues¹³ move to consider implementation of pre-emptive genotyping using a panel approach, following up on the recent successful implementation of a 12 gene panel Europewide in the U-PGX study to guide prescribing and prevent adverse events. While summarizing the very positive findings from U-PGX,¹⁴ the article also mentions the need to perform pharmacoeconomic analysis of the pre-emptive gene panel approach and importance of ethnicity/ancestry in further implementation as well as the importance of rare variants not detectable by the panel approach in overall pharmacogenomic response. The example of dihydropyrimidine dehydrogenase (DPD) in relation to fluoropyrimidine treatment is discussed in detail as a good example of a pharmacogenomic test introduced recently in Europe following an EMA recommendation with coverage also of the need to improve DPD testing to include African-specific variants.

To complement the three wide-ranging review articles above, we also highlight three Original Articles. Two are concerned with opioids and nicely highlight the value of applying pharmacogenomics to its two therapeutic areas. Healy et al.¹⁵ consider the use of tramadol in paediatrics for chronic pain relief. This drug is being used off-label in young children. The study had plasma concentration and genotype data (CYP2D6 and OCT1) available for a group of neonates and also used drug concentration data from older children and adults to aid in simulation in clinical settings. A two-compartment pharmacokinetic model was developed showing relevance of genotype to clearance of both tramadol and the mu opioid active metabolite M1 by CYP2D6 and M1 clearance only by OCT1. The authors concluded that genotyping for the relevant variants in both genes was useful in setting dose or that, if genotyping could not be performed, a low initial dose could be used with dose escalation.

The other opioid-related article concerns methadone and involves an African population where pharmacogenomic relationships with metabolism of this drug have not been studied previously.¹⁶ Tanzanian patients receiving methadone maintenance treatment were genotyped for a range of pharmacogenetic polymorphisms, and plasma drug and metabolite concentrations were determined. The main finding was a strong association between CYP2B6 genotype and drug and metabolite concentrations with parent drug concentrations higher in those with defective variant alleles. A secondary association with genotype for the ABCB1 transporter gene 3435 C/T variant was also detected. The CYP2B6 findings overall are in line with previous reports for European populations, but importantly, CYP2B6 poor metabolizers are more common in the Tanzanians, and overall allele distributions are also different. The article nicely illustrates the importance of implementing pharmacogenomics in Africa and considering African-specific genotypes, as also highlighted by Peruzzi et al.¹³

Cancer treatment remains the area where pharmacogenomic testing is implemented most widely. A recent study on the tyrosine kinase inhibitor sunitinib, which is subject to wide interindividual variation in metabolism, aimed to identify genetic variants relevant to this variation using GWAS.¹⁷ The analysis showed genome-wide significance for a single nonsynonymous variant in the GLP1R gene (encoding glucagon-like peptide 1 receptor) but could not find associations with genes implicated in sunitinib disposition, even just above the genome-wide significance threshold. The authors concluded that genomic factors do not explain the considerable interindividual variation in pharmacokinetics for this drug. An alternative explanation might be that studying 69 patients suffering toxicity due to high plasma drug concentrations is not adequate to detect such factors. The GLP1R association does show some biological plausibility but only with respect to drug absorption but needs further study.

The articles highlighted in this themed issue all illustrate the value of pharmacogenomics implementation in different ways. While Tremmel et al.¹¹ and Koido¹² describe complex and highly sophisticated bioinformatic examples using big data that are currently remote from the clinic but are likely to serve as an important basis for future advances, the remaining highlighted articles^{13, 15-17} are closer to patient care and provide a good rationale for pharmacogenomics implementation without further delay worldwide.

The authors declare no conflicts of interest.