Statistical issues and approaches in endophenotype research

The endophenotype concept was initially proposed to enhance the power of genetic studies of complex disorders. It is closely related to the genetic component in a liability-threshold model; a perfect endophenotype should have a correlation of 1 with the genetic component of the liability to disease. In reality, a putative endophenotype is unlikely to be a perfect representation of the genetic component of disease liability. The magnitude of the correlation between a putative endophenotype and the genetic component of disease liability can be estimated by fitting multivariate genetic models to twin data. A number of statistical methods have been developed for incorporating endophenotypes in genetic linkage and association analyses with the aim of improving statistical power. The most recent of such methods can handle multiple endophenotypes simultaneously for the greatest increase in power. In addition to increasing statistical power, endophenotype research plays an important role in helping to understand the mechanisms which connect the associated genetic variants with disease occurrence. Novel statistical approaches may be required for the analysis of the complex relationships between endophenotypes at different levels and how they converge to cause the occurrence of disease.