Abstract
Circadian rhythmicity and non-visual sensitivity to light can be assessed, in healthy subjects, by measuring the rhythm of the urinary melatonin metabolite 6-sulphatoxymelatonin (aMT6s) and by determining the response of plasma melatonin to nocturnal retinal light exposure, respectively. However, the validity of these techniques has not been assessed in disease states in which disruption of the circadian rhythm is known or suspected to occur. Thus, the aims of this study were as follows: (i) to assess the reliability of circadian aMT6s profile estimates derived from 36 h versus 56 h urine collections and (ii) to test different models for calculating melatonin suppression in response to light in healthy volunteers and patients with cirrhosis. Twenty patients with biopsy-proven cirrhosis and 10 matched healthy volunteers undertook: (i) separate 36 - and 56-h urine collections, under controlled conditions, for cosinor analysis of the urinary aMT6s profile; (ii) a melatonin suppression test, comprising of a baseline night, during which subjects were woken and asked to sit in front of a switched off light sphere, and an experimental night, identically executed, except that the light sphere was switched on and the subjects were exposed to white light (4.1 × 10(14) photons/cm(2)/s) for 30 min. Alternative approaches to the calculation of melatonin suppression were taken, with/without inclusion of the baseline night. Eighteen patients and eight healthy volunteers had matched analysable 36 - and 56-h urinary samples. Cosinor analysis showed a significant fit in 88% of the remaining 56 h collections, and 48% of the remaining 36-h collections. Thus, eight patients and five healthy volunteers had matched analysable samples for cosinor analysis. In the healthy volunteers, aMT6s profile indices obtained using the 36 - and the 56-h collections did not differ significantly. In contrast, considerably more variability was observed in patients [i.e. the difference in the aMT6s peak time was 0.5 ± 1.7 h (limits of agreement: -3.9; +2.9 h)]. No difficulties were encountered in obtaining suppression estimates by use of the experimental night only. In contrast, suppression estimates obtained by use of both nights were considered inaccurate in one (11%) healthy volunteer and in 5 (28%) patients, primarily because: (i) melatonin concentrations at the beginning of light administration were significantly different on baseline and experimental night; (ii) the rise in melatonin was inconsistent on baseline night; and (iii) the shape of the rising phase of melatonin was different on baseline and experimental night. In conclusion, shorter urine collections lead to a higher number of profiles with no significant cosinor fit, and differences in cosinor indices obtained from the 36 - and 56-h collections were considerable, especially in patients. Thus, 56-h collections are probably advisable. Use of both baseline and experimental nights to calculate melatonin suppression often resulted in increased variation and confounding, due to point oscillations in melatonin concentration and lack of repeatability of the melatonin profiles on the two nights. Thus, use of the experimental night only is probably advisable.