Evaluating the use of pharmacogenomics to identify adverse drug reactions: Study design considerations

Increasing concern about safety-based drug withdrawals has prompted renewed research efforts aimed at improving surveillance of approved drugs and reducing adverse drug reactions (ADRs). Pharmacogenomics research is increasingly directed at detecting ADRs.

To evaluate the potential for pharmacogenomics to identify ADRs, our study was aimed at determining 1) the prevalence of the targeted single nucleotide polymorphism (SNP) in a population receiving a drug treatment, 2) the rates of ADRs and non ADRs that occur within this population of patients and 3) the rates of these ADRs in a group of patients not receiving the drug treatment.

Statistical power is calculated to design a case-control pharmacogenomic post-marketing study. We estimated the prevalence of a SNP from the population of patients receiving a specific drug treatment, and used this to determine the association of the SNP with a given ADR. Epidemiologic measures of association are calculated.

For a given study, assuming a 5% prevalence rate of a specific SNP and a sample of 75 patients provided a two-sided 95% confidence interval within 5% of the observed prevalence rate. From this sample, assuming a 25% ADR rate we obtained 19 patients with the ADR and 56 without the ADR, thus detecting a 30% vs. a 5% prevalence of the SNP in patients with ADR vs. no ADR, respectively, with 80% power (two-sided test; &alpha=5%)

Although the science of pharmacogenomics for identifying ADRs is still nascent, we propose that our methodology can be applied in phase IV studies.