Methylomic biomarkers of lithium response in bipolar disorder: a clinical utility study

Abstract

Background Response to lithium (Li) is highly variable in bipolar disorders (BD). Despite decades of research, no clinical predictor(s) of response to Li prophylaxis have been consistently identified. Recently, we developed epigenetic Methylation Specific High-Resolution Melting (MS-HRM) assays able to discriminate good responders (GR) from non-responders (NR) to Li in individuals with BD type 1 (BD-I). This study examined whether a combination of clinical and epigenetic markers can distinguish NR from other types of Li responders. Methods We recorded clinical variables that are potentially associated with Li response in 64 individuals with BD-I. MS-HRM assays were performed on DNA isolated from peripheral blood. We used backward stepwise logistic regression analyses, followed by receiver operating characteristic (ROC) curve analysis to estimate the performance of the clinical variables, alone then in combination with the epigenetic biomarkers, to identify GR and partial responders (PaR) vs NR. Results Polarity at onset, psychotic symptoms at onset and family history of BD classified correctly 70% of individuals according to their Li response (PaR + GR = 86%; NR = 35%). When combined with the epigenetic biomarkers, these three clinical variables plus alcohol misuse (and one DMR: Differentially Methylated Region) correctly classified 86% of individuals, improving the prediction of PaR + GR (93%) and of NR (70%). The ROC analysis demonstrated an improvement in the area under the curve from 0.75 (clinical variables alone) to 0.87 (combination of clinical and epigenetic markers). Conclusions Combining clinical predictors and DNA methylation markers of Li response may have greater utility in clinical practice than relying on clinical characteristics alone.

Blood epigenome-wide association studies of suicide attempt in adults with bipolar disorder

“…Overall, the evidence provides a precedent for integrating patterns in DNA methylation to existing risk prediction models in order to improve their performance. The multiple-levels-of-analysis approach to understanding SA risk in BD promises improvements to prevention and possible intervention [ 21 ].…”

“…Results from EWAS of SA in BD could provide insights to the underlying biology, as well as provide possible diagnostic aids. Critically, this information from the biological domain, instead of being considered independently from the clinical domain, could be integrated with existing measures, such as clinical interviews [ 21 ].…”