The hype, hope and reality of pharmacogenetic tests

Pharmaceutical Technology Europe

The overarching aim of personalised medicines is to individualise and stratify medicines to maximise benefit, minimise harm and optimise allocation of resources (e.g., for expensive drugs).

The overarching aim of personalised medicines is to individualise and stratify medicines to maximise benefit, minimise harm and optimise allocation of resources (e.g., for expensive drugs). There are many markers that can predict drug response, including clinical features, non-genetic blood markers (e.g., LDL-cholesterol), proteomic and metabolic profiling, genomic variation and somatic variation in cancer cells. The latter two constitute pharmacogenetics.

Pharmacogenetics is the study of human genetic variation that influences the inter-individual response to drug therapy and was initially recognised more than 50 years ago.1

Presently, few pharmacogenetic tests have been validated; however there is much expectation.2Table 1 lists those pharmacogenetic tests that are in current clinical use, while Table 2 describes those in which a robust association between gene and drug response exists, but for which clinical utility remains to be validated.

Unfulfilled promises
To try to understand why only a handful of pharmacogenetic tests are presently in use despite the hype and hope, we performed a systematic review of pharmacogenetics and identified more than 1600 primary research articles over more than 20 years.3 We found several reasons that could account for the unfulfilled promise of pharmacogenetics:

  • There were 25 times more review articles than primary research articles, which may fuel expectations in this field.
  • There was a lack of focussed research on any particular gene-drug combination with a mile-wide inch deep focus (common to other biomarkers4).
  • Many of the >500 genes had only one study (Figure 1) with a small number of participants, meaning the synthesis of meta-analyses was hindered. This would also hamper the summation of data to robust findings.
  • The predominance of nominally significant P values suggested the presence of reporting bias.
  • Most studies investigated the intended outcomes of drugs (Figure 2) rather than adverse reactions — it is this latter end-point in which pharmacogenetics is most likely to come to fruition.
  • The alleles investigated were common, which means the expected effect size on response to drug would be small, translating to lower impact.
  • The outcomes of the studies were most often surrogate markers of the more clinically meaningful hard end-points, muddying the translation of research findings.
  • Individuals were most often from European populations (Figure 3), meaning that findings may not be applicable to individuals of differing ancestry because of the inheritance pattern of genetic markers, which differs across populations of non-similar ancestry.

Several of the above issues — particularly the small sample size and reporting bias — could be helped if guidance was developed for publishing pharmacogenetic research, such as those for the reporting of gene association studies (STREGA).5

The reality of pharmacogenetics
Reliability can be considered in 3 ways: validity of genotyping (assay), validity of association and clinical validity.

Genotyping validityGenotyping is of high fidelity, and is not influenced by operator characteristics associated with some phenotypic variables (e.g., flow-mediated dilatation, a measure of endothelial function).

Validity of associationWith the exception of a few recent pharmacogenetic tests, much data on pharmacogenetics arises from single studies using small numbers of participants that remain to be replicated. There is a paucity of meta analyses and, thus, conclusive data on strengths of association between gene and drug interaction are absent.

Clinical validityThe clinical validity of a pharmacogenetic test is an assessment of how consistently and reliably it can predict a drug response, the effect size of a drug-gene interaction and the prevalence of an allele under investigation. Information on clinical validity can be derived from modelling and/or from clinical trials in which participants are randomised to treatment with and without pharmacogenetic knowledge. To date, few randomised clinical trials or modelling studies assessing clinical validity have been performed, with notable exceptions in CYP2C9/VKORC1 with warfarin initiation6 and HLA-B5701 with abacavir hypersensitivity.7

A few words of advice
Our role is not to advise companies. Rather, the intention of our appraisal was to inform on the current state of pharmacogenetics with the aim of guiding future research to improve population health. From a broader healthcare perspective, however, we make the following recommendations:

  • Primary research in pharmacogenetics should give due emphasis both to adverse and intended effects of drugs (the former is likely to be of greater clinical utility).
  • Studies should have sufficient participants to ensure they are adequately powered to detect gene-drug interactions.
  • The end-points under investigation should be clinically relevant and preferably not surrogate markers (although such markers do play an important role in investigative studies).
  • The paucity of studies conducted in individuals of non-European ancestry should be redressed so that knowledge of gene-drug interactions has broader worldwide applications. Similarly, currently neglected drugs and disease areas should be focussed upon (e.g., communicable diseases).
  • In circumstances where mechanisms of drug metabolism/action are unknown, whole genome analyses should be performed.
  • To prevent reporting bias, post-hoc subgroup analysis should be avoided, except where justified and powered, with findings reported with due caution. Evidence of independent replication should also be sought, as is the prerequisite for gene disease association studies.
  • Mechanisms should exist to encourage the reporting of null findings from high-quality studies to minimise publication bias.
  • Systems should exist for the systematic and comprehensive collating, archiving and dissemination of reports of pharmacogenetic research, to highlight continuing gaps in knowledge and promote successes.
  • High quality, systematic reviews and meta-analyses should be routinely performed and keep abreast of new studies in the same gene-drug combinations.
  • For genotype-based predictive tests that do show promise, we recommend that these be re-evaluated in independent prospective studies using clinically relevant (non-surrogate) outcomes. They should be evaluated using the appropriate metrics for diagnostic, screening and predictive tests and, where appropriate, tested in randomised trials or using statistical modelling.

The future
Pharmacogenetics is likely to play a role in the future management of pharmacological treatments of common disease. So far, and until recently, much research has been spread across many candidate genes with small sample sizes. For pharmacogenetics to be clinically useful, gene alleles of large predictive function are required, which will most likely be those associated with adverse drug reactions (versus genes that predict an intended therapeutic response). Prior to testing for clinical utility, the robust verification of gene-drug interactions needs to be acquired through the systematic reporting of pharmacogenetic studies, as has been advocated for other biomarkers, followed by systems to comprehensively collate data and perform meta analyses (Figure 4).

Once a robust gene-drug interaction has been identified, utility may need to be demonstrated through randomised controlled clinical trials or other high-quality evaluative studies. Cost-effectiveness analyses must then follow prior to the adoption of a pharmacogenetic test into routine clinical use.

References
1. Meyer UA. Pharmacogenetics - five decades of therapeutic lessons from genetic diversity. Nature reviews 2004;5(9):669-76.
2. Wolf CR, Smith G, Smith RL. Science, medicine, and the future: Pharmacogenetics. BMJ (Clinical research ed 2000;320(7240):987-90.
3. Holmes MV, Shah T, Vickery C, Smeeth L, Hingorani AD, Casas JP. Fulfilling the promise of personalized medicine? Systematic review and field synopsis of pharmacogenetic studies. PLoS One 2009;4(12):e7960. Available open-access at: www.plosone.org/article/info:doi/10.1371/journal.pone.0007960
4. Hemingway H, Riley RD, Altman DG. Ten steps towards improving prognosis research. BMJ (Clinical research ed 2009;339:b4184.
5. Little J, Higgins JP, Ioannidis JP, Moher D, Gagnon F, von Elm E, et al. STrengthening the REporting of Genetic Association Studies (STREGA): an extension of the STROBE statement. PLoS medicine 2009;6(2):e22.
6. Klein TE, Altman RB, Eriksson N, Gage BF, Kimmel SE, Lee MT, et al. Estimation of the warfarin dose with clinical and pharmacogenetic data. N Engl J Med 2009;360(8):753-64.
7. Mallal S, Phillips E, Carosi G, Molina JM, Workman C, Tomazic J, et al. HLA-B*5701 screening for hypersensitivity to abacavir. N Engl J Med 2008;358(6):568-79.

Based on contributions by Michael Holmes, Aroon Hingorani andJuan Casas from the University College London and London School of Hygiene and Tropical Medicine (UK).