REM sleep effects as a biomarker for the effects of antidepressants in healthy volunteers.

Rijnbeek B, de Visser SJ, Franson KL, Cohen AF, van Gerven JM

The potential use of rapid eye movement (REM) sleep effects as a biomarker for the therapeutic effects of antidepressants in healthy volunteers is reviewed. A literature search was performed to select studies investigating the effects of antidepressants on REM sleep. To assess the specificity of REM sleep effects as a biomarker, the effects of other central nervous system drugs on REM sleep were also investigated. A significant REM sleep reduction was shown for 16 of 21 investigated antidepressants after single-dose (mean reduction 34.1%) and for 11/13 drugs after multiple-dose administration (mean reduction 29.2%). The median increase in REM latency was approximatety 60% after single- or multiple-dose administration. REM sleep effects were linearly normalized to therapeutic doses, by dividing the REM sleep effect by the investigated dose and multiplying by the therapeutic dose. Normalized REM sleep effects were highly variable (range -27.0% to 81.8% for REM sleep; range -17.0% to 266.3% for REM latency) and demonstrated no relationship with relevant pharmacological properties of the investigated drugs. No quantifiable dose-response relationship could be constructed after single and multiple dose administration. REM sleep effects were not specific for antidepressants. Benzodiazepines, for example, caused an average dose normalized REM sleep reduction of 8.7% and a median 8.6% increase of REM latency. This review demonstrates that although REM sleep effects occur with most of the antidepressants, it is by itself of limited value as a biomarker for antidepressant action. The specificity for antidepressants is limited, and it does not show a quantitative dose-response relationship to antidepressant agents. This is at least partly due to the complex relationships between drug pharmacokinetics and the variable time course of REM and other sleep stages throughout the night. Models that take these complex relationships into account may provide more comprehensive and quantifiable results.