A critical assessment of the melatonin effect on sleep in humans.

Abstract

Melatonin is synthesized and secreted during the dark period of the light-dark cycle. The rhythmic nocturnal melatonin secretion is directly generated by the circadian clock, located in mammals within the suprachiasmatic nucleus (SCN), and is entrained to a 24-hour period by the light-dark cycle. The periodic secretion of melatonin may be used as a circadian mediator to any system that can 'read' the message. In addition, direct effects of the hormone on the SCN could explain some of the melatonin effects on the circadian system. Duration of the melatonin nocturnal secretion is directly proportional to the length of the night and it has experimentally been demonstrated to be the critical parameter for photoperiod integration. The sites and mechanisms of action of melatonin for circadian and photoperiodic responses are far from being elucidated, but action through specific membrane receptor sites starts to emerge. A possible bicompartmental model of distribution for melatonin, the first compartment in plasma acting on peripheral organs and the second in the cerebrospinal fluid affecting neurally mediated functions at a much higher concentration, has recently been proposed. From earlier studies it was concluded that melatonin administration to humans reduces sleep latency and induces sleepiness and fatigue. More recently, the effect of lower pharmacologic or physiologic doses of melatonin was examined in different laboratories. These studies included young normal volunteers and patients with chronic insomnia, as well as dementia patients exhibiting sundowning syndrome. Irrespective of the method of assessment, melatonin showed effects in insomniac patients in most studies. With some exceptions, melatonin administration reduced sleep latency and/or increased total sleep time and sleep efficiency. Furthermore, melatonin was more effective when given to elderly insomniacs, or Alzheimer disease patients, although sleep improvement was not strictly correlated with prior levels of the hormone.