We read with interest the article by Steindl and colleagues [1]. The authors thoroughly assessed 24-hour plasma melatonin levels and showed that significantly higher levels persisted throughout daytime hours in patients with liver cirrhosis.

Several points deserve further consideration. We agree that disruption of the diurnal rhythm of melatonin probably has clinical implications and may in part explain the high incidence of sleep disturbances in cirrhotic patients. However, Steindl and colleagues' attempt to correlate plasma melatonin profile to sleep patterns by using sleep diaries seems unreliable. The subjective quality of sleep may not correlate with objective measures of sleep. Furthermore, the authors claim that their cirrhotic patients had subclinical hepatic encephalopathy, a clinical condition associated with decreased cognition and judgment. This claim makes the reliability of their patients' diaries even more questionable. We believe that a much better approach would be the use of wrist actigraphy. This simple and sensitive test enables objective monitoring of several sleep measures [2]. We have been using this method successfully in elderly persons, who, like patients with liver cirrhosis, may have decreased compliance with polysomnography [3].

Steindl and colleagues suggested that the high daytime melatonin levels phase shifted the endogenous clock; however, this may also have lead to the apparent phase-delay melatonin onset. Steindl and coworkers offered two mechanisms to explain their findings: 1) a direct effect induced by ammonia [or other toxic metabolites that may accumulate in patients with cirrhosis] on melatonin production in the pineal gland or on tryptophan accumulation in other areas of the brain or 2) decreased hepatic elimination of the hormone [1]. We believe the latter explanation is the correct one, but this issue could have been resolved by simultaneous evaluation of urinary levels of 6-sulphatoxymelatonin. This is the main liver metabolite of melatonin; in normal persons, its urinary profile parallels that of plasma melatonin [4]. If low urinary concentrations of this metabolite were found in cirrhotic patients who had high plasma melatonin levels, the latter finding would be the result of impaired melatonin metabolism. We plan to monitor serum melatonin and urinary 6-sulphatoxymelatonin levels in cirrhotic patients and suggest that Steindl and colleagues consider doing likewise in their patients.