Beneath the surface. Endophenotypes sound great on paper. As the actual disease phenotype is difficult to classify and complicated, a biomarker is used for genetic analysis that is assumed to be closer to the genetic underpinnings of a disease than the disease itself. A recent study in Cell now investigates the hunt for risk factors for Alzheimer’s disease using a well-established endophenotype. The results demonstrate the complexity of endophenotypes and the difficulties in interpretation.
Endophenotypes. Particularly many neuropsychological or neurodevelopmental disorder can manifest in various ways, even in cases with an identical genetic basis. To get closer to a possible genetic basis of these disorders, Gottesman suggested that endophenotypes rather than phenotypes should be the target of genetic studies. Endophenotypes are considered measurable, unseen components of the disease process that are thought to be closer to the biology than the disease itself. For the epilepsies, EEG patterns or blood biomarkers might be endophenotypes, even though gene identification based on these markers in epilepsy is in its infancy.
Alzheimer’s disease. In contrast to the lack of established biomarkers for epilepsy, there are several biomarkers for Alzheimer’s disease including the concentration of certain proteins in the cerebrospinal fluid. There is solid evidence that the level of CSF tau and phosphorylated tau (Ptau) as well as the level of A-beta-42 are risk factors for AD. Hyperphosphorylated tau protein is the main component of the so-called neurofibrillary tangles, the histological feature of Alzheimer’s disease. A-beta-42 is the cleavage product of the amyloid precursor protein (APP) and the main component of the so-called beta amyloid, which is also a histological hallmark of Alzheimer’s disease. In their current study, Cruchaga and colleagues looked for common genetic variants predisposing to elevated tau and Ptau level in ~1300 patients with Alzheimer’s disease.
Three novel variants, but no genes. Cruchaga and colleagues identified four variants that were associated with CSF tau and Ptau levels. Common variants in APOE have previously been described and showed the strongest association. In addition, they identified three loci on chromsomes 3, 6 and 9. None of these variants fell into the coding region of a gene. This demonstrates the difficulties in handling association findings that do not make immediate sense, but require further work-up.
From endophenotype to disease. Ideally, the endophenotype shortens the way to discovering a risk factor for the disease. However, with the different factors involved, the study by Cruchaga and colleagues resulted in a complex network of interactions and interdependicies. For example, much of the association of the CSF tau levels with APOE is driven by A-beta-42 levels, while the other variants are largely unaffected. In addition, for the chr6 locus, rare variants rather than the reported common variants are associated with the disease. For one variant (chr9) found to be associated with CSF tau, no association with the actual disease was found. This leaves us with a complicated puzzle for the endophenotype genetics in AD.
Lessons for epilepsy. Endophenotypes are useful markers for identifying a homogenous patient population when phenotypes vary widely. However, in contrast to other neurological disorders, the number of biomarkers for epilepsy is rather small and none of these markers has been investigated in genetic studies. The paper by Cruchaga and colleagues now raises the question whether results from endophenotype studies may come with their own set of problems and can virtually take on a life of their own.
Pingback: Genetic imaging in Dravet Syndrome – variation on a theme? | Beyond the Ion Channel
Pingback: Guilt by association: SCN1A in Temporal Lobe Epilepsy | Beyond the Ion Channel
Pingback: The genetics of treatment response in newly diagnosed epilepsy | Beyond the Ion Channel
Pingback: Mining GWAS mountains for missing heritability | Beyond the Ion Channel
Pingback: The age of mega-genomics, type 2 diabetes, and protective variants in SLC30A8 | Beyond the Ion Channel