It is well known that the response to medicinal products can significantly vary from patient to patient: a drug, given at the same dose, can vary in efficacy and safety in different patients. Our genotype, the collective term for our whole set of genes and genetic variants, is estimated to account for 15-20% of these inter-individual differences, but for some medicinal products, they are responsible for up to 95% of these differences.
The ISoP pharmacogenomics special interest group
In 2020 the International Society of Pharmacovigilance (ISoP) established the Pharmacogenomics Special Interest Group (SIG) to establish a network of people with an interest in pharmacogenomics to share information, to network, and to build on existing knowledge in this area to better inform therapeutic decisions. If you enjoy this article, you may also be interested in following the work of that SIG on the ISoP website.
For example, we drink coffee for its stimulant effects, but we don’t all react to its effects in the same way. Some may take it in the evening with limited or no effect on sleep, while others can’t take it in the afternoon, or they will suffer from insomnia. The reason for this variability lies in alleles that code for the enzymes that metabolise caffeine.
In the simplest scenario, a single gene contributes to an observable physical characteristic, known as a phenotype (in this case, caffeine metabolism). Humans have two copies of each gene, one inherited from the mother and one from the father, known as alleles. These genes code for the different proteins in the body. In this case, the gene codes for an enzyme that metabolises caffeine. One of the alleles codes for an enzyme with low activity (named A1), while the other codes for an enzyme with high activity (A2). The possible combinations of the alleles in the two genes (A1/A1, A1/A2, A2/A2) will result in three possible phenotypes: low, intermediate, and high enzyme activity, respectively. In the case of caffeine metabolism, individuals with enzymes that rapidly metabolise caffeine will sleep well regardless of when they drink their last cup of coffee, while slow metabolisers are at risk of insomnia.
More often, multiple genes can contribute to a physical trait and a gene can have multiple alleles, with any range of enzyme activity between high and low.
The genetic differences that are most relevant to medicines safety depend upon which protein a gene codes for. These differences can be grouped into three broad categories:
1) Alleles coding for proteins involved in the medicine’s pharmacokinetics (e.g. how the drug is metabolised in the body) can affect the efficacy of the medicine, resulting in overdosing or underdosing which increases the risk of adverse drug reactions (ADRs). Dose modification may be required.
2) Alleles coding for proteins involved in biological pathways related to the drug’s effect can increase or decrease the primary and secondary pharmacological effects of the drug. A change in the drug’s dose may also be required.
3) Alleles coding for proteins in the immune system can result in possible immune reactions to a medicine. Avoidance of the medicine or further patient monitoring may be needed.
Tests to identify these specific genetic variants (genotypes) do exist, however, it is necessary to understand when pharmacogenetic testing is required. One must consider the clinical utility of testing, which is the balance of the risks and benefits associated with using a genetic test in routine clinical practice, including its ability to support making clinical decisions, preventing, and predicting an important ADR. The seriousness and frequency of the ADRs associated with the genetic variant must also be considered since it might not be worthwhile to conduct a genetic test to identify patients at risk of experiencing an adverse reaction that is not life-threatening.
The probability that a genetic test will correctly indicate genetic variants is also important. Furthermore, it should be realised that while a specific genetic variant may be necessary for an ADR to occur (it has a high negative predictive value in that, without this variant, the ADR risk is low), this genetic variant alone may not be sufficient to cause the reaction (it has a low positive predictive value). Other known or unknown genetic and/or non-genetic factors, (such as alcohol intake, smoking, age, gender, diet, exercise, concomitant diseases, etc) may contribute to ADR risk as well.
To better understand this, let’s take two specific examples:
Abacavir, a medicine used to treat Human Immunodeficiency Virus (HIV) can cause serious hypersensitivity in three to nine percent of patients. Not administering abacavir to patients with a genetic variant (called HLA-B*57:01), eliminates the risk of experiencing this serious ADR. Therefore, the negative predictive value of the genetic test for this variant is 100% (i.e. all patients without the variant will not experience this ADR), while the positive predictive value of the test is around 48% (i.e. 48% of patients with the variant will experience the ADR). This means that, in addition to HLA-B*57:01, there are other factors that together with this variant may cause abacavir hypersensitivity. Also important to know is that the screening of about 25 patients is sufficient to prevent a case of this serious reaction. For this reason, the abacavir label states that before taking this medication, all patients should undergo testing for HLA-B*57:01 and, if they have this variant, they should not take the medicine.
The second example regards the antibiotic flucloxacillin. Patients carrying the HLA-B*57:01 variant are 80 times more likely to experience drug-induced liver injury (DILI) compared to those who don’t have the variant. However, the incidence of flucloxacillin-induced DILI is around 8.5 for every 100,000 patients who take the antibiotic, and the positive predictive value of testing for HLA-B*57:01 is 0.12%. Given a prevalence of this variant of around 7% in Caucasians, it has been calculated that around 13,500 patients need to be screened to prevent one case of flucloxacillin-induced livery injury by not administering this drug to carriers. Therefore, the clinical utility of this test is considered low since so many patients need to be screened to avoid the otherwise effective medicine-induced case of DILI. As a result, flucloxacillin’s label mentions the risk of experiencing DILI in patients with HLA-B*57:01, but it also specifies that routine screening for this allele is not recommended.
In conclusion, the effect of a genetic variant on an ADR can vary from being the most important risk factor to being only a minor contributor. The different prevalences of an allele in different ethnicities affect the probability of experiencing an ADR; this together with the possibility of clinical monitoring of serious ADRs, adds complexity to decision-making for routine genotyping. We hope that, in the future, the ever-decreasing cost of genetic testing will result in every patient having their whole genome genotyped. With the help of artificial intelligence, more information about the interplay between genetic variants and non-genetic factors will guide healthcare professionals and individual patients on whether and how a medicine should be used safely in their treatment.
Disclaimer: In addition to the standard Uppsala Reports disclaimer, it should be noted that Dr Qun-Ying Yue serves as coordinator of the ISoP Pharmacogenomics SIG and co-authored this article in her private capacity, outside of her role at Uppsala Monitoring Centre (UMC). Any opinions or recommendations expressed in this article are those of the authors only and do not necessarily reflect the views of UMC. In particular, the field of pharmacogenomics lies outside UMC’s core activities and UMC makes no recommendations as to issues such as genome screening or other related practices.
