In Reply: Does ammonia really disrupt brain oxygen homeostasis?

We are grateful for Drs. Sørensen and Vilstrup’s interest and thoughtful consideration of our recent study. We are encouraged that they agree with our substantive conclusions and welcome the opportunity to reply to their concerns regarding our specific interpretation that the decreased tissue pO₂ could be contributing to the metabolic disruption seen in hepatic encephalopathy (HE).

We understand they argue that observations of metabolic disruption, such a decreased cerebral blood flow (CBF), are secondary to a decrease in metabolic demand caused by ammonia reducing CNS activity. We agree that a reduction in metabolic demand is a significant component contributing to the neuronal dysfunction seen in patients with HE. We ask Drs. Sørensen, Vilstrup and your readers to consider the possibility that we are observing a complex problem on a 1-dimensional manifold but when the full dimensionality is revealed, our hypotheses could be fully compatible. We propose that a negative feedback loop is constructed between the supply and demand of all metabolic substrates in the brain during the development of HE as a result of the maladaptation of a number of homeostatic processes that would normally closely match energy demand and supply. Whether it is demand or supply that is first to drop, it still feeds into the same negative feedback loop that sees the brain less able to respond to the metabolic demands of neurons leading to neurological dysfunction.

The first pillar of their concern relates to the baseline tissue pO₂ observed in our model. Specifically, that it is not sufficient to induce significant metabolic disruption as some of the cohort display a pO₂ above a proposed ‘critical’ range of 6.7-8.8mmHg in healthy rats. We posit that the absolute concentration of oxygen is immaterial to our hypothesis and our observation simply provides evidence that a major metabolic component (oxygen) is significantly lower in our model of HE when compared to sham subjects. Any reduction in pO2 would impair the brain’s ability to maintain metabolic sufficiency.

We argue that the absolute concentration of tissue oxygen is not a good indication of the potential for neuronal dysfunction. Firstly, pO2 measurements are not generalizable given the lack of homogeneity of tissue oxygenation within the brain and differences in experimental conditions (such as anesthetics and sampling locations) that would affect baseline tissue pO2 values. Additionally, the level of oxygen tension used by several studies to mimic hypoxia is in the range of 20–25 mmHg , which is very close to normal brain parenchymal pO₂. However, astrocytes do respond to decreases in pO₂ below ∼17mmHg with elevations in intracellular calcium and release of ATP by exocytosis. Further, in astrocytes, hypoxia leads to inhibition of mitochondrial respiration, facilitated mitochondrial ROS production, increased rate of lipid peroxidation and others. This is further supported by the observation that Ca²⁺-dependent release of vasoactive substances by astrocytes have a significant effect on cerebral vasculature. Such astrocytic activation may be important for local control of cerebral microcirculation when pO₂ in a particular microdomain of the brain decreases. Chronic activation of this pathway is expected to have detrimental effects on metabolic supply matching, and consequently neuronal function. Our conclusions merely speculate on the possible effects that a constraint of the oxygen supply seen in our model may have on neuronal function given that the brain is an extremely energy intensive organ with several systems having evolved to ensure constant energy supply. Impairment of any one of these mechanisms would, at the very least, put pressure on the reserve of the others to effectively manage brain metabolic demand.

As for our interpretation of the observed low lactate concentration, we argue that the brain would reduce the production of lactate in response to a gradual downward pressure on both metabolic demand and supply in the brain. Increases in lactate would be indicative of extreme metabolic distress as there is sufficient tissue oxygenation even at 6.7mmHg for some level of metabolism to be maintained. It is worth noting here that our model is one of mild HE. We agree that CBF data is essential to our understanding of this complex system, and we are in the process of addressing this unmet need.

Finally, although potentially contradictory to Drs. Sørensen and Vilstrup data, recent in vitro (coculture of astrocytes and neurons) and ex vivo studies (brain slices from bile duct ligated rats), have shown that concentrations of ammonia as low as 5 μM induce mitochondrial hyperpolarisation, lipid peroxidation, increase in ROS production, as well as profound neuronal death in the hippocampus.

To summarize, we consider metabolic demand and supply closely coupled in the pathophysiology of HE. Simultaneous intervention in both would be the best approach to breaking the reinforcement of our proposed negative feedback loop to maximize the possibility of complete reversal of neurological impairment after resolution of hyperammonemia.