The AAC resulted in the expected increase in capillary ammonia concentrations, together with an increase in subjective sleepiness and changes in both wake and nap EEG characteristics in patients with cirrhosis. Conversely, no significant changes were observed in psychometric performance. In healthy volunteers the AAC was also associated with increased ammonia levels and subjective sleepiness, and with changes in the nap EEG. AAC-related sleep and nap EEG changes had opposite direction in healthy volunteers and patients. The amount of non-REM sleep displayed a trend toward an increase in healthy volunteers but not in patients. In addition, healthy volunteers showed a decrease in fast sleep EEG activity, whereas patients showed a reduction in delta activity, and thus more superficial sleep.
The quality of life, sleep-wake rhythms, and neuropsychiatric performance of the patient population were those expected based on previous studies, and the fact that only relatively well-compensated individuals with no HE and no significant sleep-wake abnormalities were recruited.6, 7, 30 These restrictive selection criteria were aimed at: (1) limiting confounders on sleep-wake variables; (2) avoiding episodes of precipitated, overt HE after the AAC; (3) maximizing the differences between the AAC and the baseline conditions. Nonetheless, the patients' quality of life was impaired, their sleep less efficient, and, albeit only as a trend, delayed compared to healthy volunteers. Neuropsychiatric performance was well preserved but still significantly worse than that of healthy volunteers, with slower reaction times on psychometry and slower wake EEG. The diurnal time-course of subjective sleepiness had not been previously investigated in this patient population and exhibited the expected increase in the early afternoon. Such an increase is known to occur also in healthy subjects,31 and relates to the interaction of the circadian and homeostatic components of sleep regulation.32
The nap EEG characteristics of well-compensated patients with cirrhosis are largely unknown. In this small population they were comparable to those of healthy volunteers, although the patients tended to sleep longer during the nap opportunity. This may relate to the observed, higher baseline ammonia levels and/or increased levels of daytime sleepiness, which have been described in these individuals,6, 7 also in association with blunted melatonin rhythms.33
The AAC led to the expected increase in ammonia levels, with a peak at approximately four hours after administration. Douglass et al.4 reported an earlier peak, at ≈2 hours from the administration of the same mixture. The differences may be due to differences in the patients enrolled, the ones in this study being better compensated, and/or to the fact that hourly ammonia measurements may be insufficient for accurate definition of the peak time. In our study, baseline ammonia concentrations were higher in the patients than healthy volunteers, but showed a similar time course up to the peak. The decrease in ammonia levels after the peak was steeper in the healthy volunteers. This is in agreement with the notions that: (1) hyperammonemia after AAC is largely due to the absorption/oxidation of amino acids; (2) ammonia is a high extraction molecule, which is removed by the liver in a flow-dependent manner,34 thus explaining the reduced clearance in patients.
The observation that the ammonia peak was associated with a quantifiable, transient increase in subjective sleepiness is a completely novel finding. There is some evidence that overt HE is associated with excessive daytime sleepiness,6, 7 and some of the wake EEG features of HE, particularly the anteriorization of the background rhythm, are reminiscent of those observed during the wake-sleep transition.35 The findings in the present study suggest that subjective sleepiness may be increased even for levels of ammonia that do not result in neuropsychiatric alterations. This has relevant clinical implications: (1) measures of sleepiness may be useful as surrogate measures of HE; (2) the relationship between HE and difficulties in complex task execution (i.e., driving) may not lay in specific cognitive deficits36 but in a reduction in vigilance.
The AAC had virtually no effect on paper-and-pencil or computerized psychometric performance, whereas it caused some slowing of the wake EEG in two patients. This is in line with a previous study on AAC4 and with a recent, small series suggesting that a sleep deprivation protocol does not affect cognition in these patients.11 In addition, the tight but necessary exclusion criteria may have led to the selection of a group of subjects who were not particularly prone to develop neuropsychiatric abnormalities, and indeed had excellent baseline psychometric performance despite slightly raised ammonia levels. Finally, it has recently been suggested that the EEG and psychometric alterations associated with HE may have different biochemical correlates, the former being more related to increased concentrations of neurotoxins of intestinal origin, the latter to the activated inflammatory cascade.37
Healthy volunteers and patients had similar nap EEG features at baseline, with comparable ability to generate delta activity, and they both reported subjective sleepiness after the AAC. However, the effect of the AAC on sleep structure and nap EEG was different in the two groups, with non-REM sleep prolongation and fast EEG activity suppression in the healthy volunteers and reduction in delta activity, thus more superficial sleep, in the patients.
Sleep and wakefulness are homeostatically regulated, and the ability to generate restful sleep depends, to some extent, on the quality of the previous waking period.13 Thus, the power of the waking EEG theta band increases as a function of the duration of wakefulness,38 and increased sleep pressure is reflected in an increase in non-REM sleep delta activity in the sleep EEG.13, 39 It is possible that in hyperammonemic/encephalopathic patients with cirrhosis, who show excessive daytime sleepiness and a chronic increase in the waking EEG theta activity,7 wakefulness might be somewhat “inefficient,” thus compromising the build-up of the homoeostatic response, and resulting in an inability to generate deep, restful sleep. Therefore, it can be hypothesized that treatment strategies aimed at reducing daytime sleepiness may also lead to an improvement in night sleep architecture in these patients.
The two case reports confirmed that HE is associated with prominent, reversible changes of both wake and nap EEG structure. Interestingly, in these two cases the HE-related sleep EEG changes were particularly prominent within the sleep spindle range, an area of the spectrum that was only moderately affected by the AAC. Similar findings have been reported once before in a group of patients with overt HE.10 Clearly, differences are to be expected between the electrophysiological profile of full-blown, spontaneous or TIPS-related overt HE and AAC-related hyperammonemia because the latter is only a model of the former, it is not meant to induce severe neuropsychiatric dysfunction and it is not necessarily accompanied by the degree of hepatic failure and/or the precipitants which are associated with spontaneous HE.
In conclusion, profound changes were observed in response to the AAC in clinical (subjective sleepiness), wake and nap EEG indices, suggesting that such techniques are exquisitely sensitive to ammonia levels, which have limited neuropsychiatric/neuropsychological correlates. These findings have important clinical implications: (1) subjective sleepiness may be a useful surrogate marker of HE; (2) correction of excessive daytime sleepiness, either by pharmacological or chronotherapeutic strategies, may also result in improved night sleep.