Developing a biomarker for restless leg syndrome using genome wide DNA methylation data.

This study reports on an epigenetic biomarker for restless leg syndrome (RLS) developed using whole genome DNA methylation data. Lymphocyte-derived DNA methylation was examined in 15 subjects with and without RLS (discovery cohort). T-tests and linear regressions were used followed by a principal component analysis (PCA). The principal component model from the discovery cohort was used to predict RLS status in a peripheral blood (N = 24; including 12 cases and 12 controls) and a post-mortem neural tissue (N = 71; including 36 cases and 35 controls) replication cohort as well as iron deficiency anemia status in a publicly available dataset (N = 71, 59 cases with iron deficiency anemia, 12 controls). Using receiver-operating characteristic analysis the optimum biomarker model - that included 49 probes - predicted RLS status in the blood-based replication cohort with an area under the curve (AUC) of 87.5% (confidence interval = 71.9%-100%). In the neural tissue samples, the model predicted RLS status with an AUC of 73.4% (confidence interval = 61.5%-85.3%). An AUC of 83% was found for predictions of iron deficiency anemia. Thus, the blood-based biomarker model reported here and built with epigenome-wide data showed reasonable replicability in lymphocytes and neural tissue samples. A limitation of this study is that we could not determine the metabolic or neurobiological pathways linking epigenetic changes with RLS. Further research is needed to fine-tune this model for prospective predictions of RLS and to enable translation for clinical use.