We appreciate the interest and points raised by Albillos and colleagues concerning our recent publication. We are very familiar with the body of literature identifying an increase in circulating TNF-α levels and numbers of TNF-α-producing monocytes in the setting of cirrhosis (and fully recognize the important contributions made by these authors in this area).1, 2 Although we feel that our identification of similar findings in mice without cirrhosis is important, we do not feel that this is the “key finding” of our studies. We have identified that circulating TNF-α-producing monocytes are capable of interacting via specific adhesion molecules with activated endothelial cells, and subsequently infiltrating a tissue other than the liver, in the setting of experimental liver disease. Moreover, we have demonstrated in cholestatic mice that infiltration of the brain by TNF-α-secreting monocytes is associated with resident tissue macrophage (i.e., microglia) activation and TNF-α production within the brain, and suggest that cytokines produced by these activated immune cells may be able to alter the normal function of the tissue in which they are produced (i.e., change neurotransmission to induce fatigue). We do recognize, however, that cytokines released within the circulation are able to communicate with the brain (either directly or indirectly) to induce behavioral changes, including fatigue.3 Interestingly, in contrast to the setting of cirrhosis, we were unable to identify increased circulating TNF-α levels in cholestatic mice without cirrhosis; possibly reflecting a more profound inflammatory response in the setting of cirrhosis. Furthermore, in patients with cirrhosis TNF-α production by monocytes has been reported to correlate significantly with circulating lipopolysaccharide-binding protein levels,1, 2 potentially implicating circulating endotoxin in the observed increase in monocyte TNF-α production. However, we did not find a reduction in the increased production of TNF-α in circulating monocytes in cholestatic mice without cirrhosis which were unable to respond to endotoxin (i.e., TLR4 deficient), compared to TLR4 deficient non-cholestatic controls (MG Swain, unpublished observation, June 2005). This finding suggests that increased monocyte TNF-α production in the setting of liver disease may not be the result of direct endotoxin activation of monocytes (either within the circulation or within mesenteric lymph nodes) but instead may be mediated by other Toll-like receptors (e.g., stimulation of TLR2 on monocytes in cirrhotic patients mediated by gram positive microbial products, as suggested by Riordan et al.).4
We agree that our findings may very well have important implications for the development of both peripheral (e.g., muscle) and central (i.e., neurotransmission) fatigue in cirrhosis. However, as these authors suggest, cirrhosis (especially with associated ascites) is a significantly more complex clinical situation than liver disease without cirrhosis, with a more profound systemic inflammatory response, hematological and renal changes, possible superimposed hepatic encephalopathy, and muscle wasting; all of which may actively contribute to the development of and/or clinical expression of fatigue. This is one of the reasons why we purposely chose to do our current series of experiments in animals without cirrhosis and why we have previously performed our fatigue-related behavioral studies in animals without cirrhosis.5, 6 In fact, a recent editorial has highlighted the complex interplay between infection, the systemic inflammatory response, and the development of hepatic encephalopathy in patients with cirrhosis,7 a discussion which has direct relevance with regards to factors which are likely to play a role in the development of fatigue in the setting of liver disease. Obviously this is an important and fertile area for future investigation, which may have significant potential for therapeutic intervention in patients.