Recruitment and selection of cases and

controls is a methodological issue that equally applies to both CGAS and GWAS. Although 10 cases may be sufficient for a genetic association study when effects are large and the number of controls is high,42 inclusion of 100 cases or more will generally be necessary for the study of complex diseases like DILI in order to identify low-risk variants. Such studies are therefore dependent on large networks that collect and evaluate DILI cases with standardized criteria.38, 43, 44 Although a sufficient number of cases is crucial in order to detect true associations with reasonable power, it is even more important to subject potential cases to a rigorous selection process in order to avoid misclassification. When standardized assessment tools are PF-02341066 solubility dmso used,45, 46 patients who fall into categories of questionable causality should not be included as cases, because misclassified cases will inevitably dilute risk estimates toward a null effect and therefore lead to an underestimation of true associations or even to false-negative results. It is important to realize that such a loss of power to detect a true association is related to the study design and will neither be reflected in statistical 3-deazaneplanocin A power calculations nor in the confidence intervals of risk estimates. When the number of cases is limited, more controls, usually up to four

times the number of cases, can provide additional power. The answer to the question whether controls should have been exposed to the drug under study without development of DILI depends on the likely absolute risk of DILI in the control population. Whenever the risk is very

low, which is typical for idiosyncratic hepatotoxicity, it is perfectly acceptable to use unexposed subjects.38 However, if the risk of the outcome is close to 10% or higher, as it is in, for example, the case of increased aminotransferases under isoniazid or ximelagatran, one should recruit exposed cases or use a cohort design where exposed patients serve as the control group.14, 47 If unexposed controls MCE公司 are a suitable solution, the selection of controls from genotyped standard populations may be an efficient and attractive option.37, 48, 49 In this case, appropriate control for the use of different technology and population stratification has to be considered.38, 48 Based on the role of drug metabolism for both toxification and detoxification, metabolizing enzymes have long been a prime target for research relating to DILI.5, 6 A recent investigation reported that drugs with more than 50% hepatic metabolism are more likely to cause DILI than those with lesser hepatic metabolism.50 Furthermore, another possible mechanism how metabolizing enzymes could contribute to hepatotoxicity is the formation of reactive oxygen species (ROS) from endogenous substrates.