Aminoglycoside Antibiotics and Renal Dysfunction associated with a greater incideice of renal dysfunction in patients with liver disease than in patients without liver disease? The answer is obviously important to those who must choose the appropriate antibiotics for patients with liver disease. The subject is controversial, as judged from the article by Camps et al. in this issue of Hepatology (1). A review of the evidence from investigations in both man and animals should shed light on this point and also should provide a perspective on the contributions that can come from animal studies and those that can come only from investigations in man.

The evidence suggests that aminoglycoside antibiotics are associated with a greater incidence of renal dysfunction in patients with liver disease and that this interaction is clinically important. Moreover, a knowledge of the interaction could contribute to our understanding of the pathophysiology of both aminoglycoside nephrotoxicity and the hepatorenal syndrome. The evidence in favor of liver disease as a risk factor for renal dysfunction in patients receiving aminoglycoside antibiotics must ultimately come from investigations in man since no animal model can provide proof that such an association does or does not occur in man. The clinical evidence supporting an association among liver disease, aminoglycoside antibiotics and renal dysfunction has come from our group at Johns Hopkins (Moore, Smith, Lietman and colleagues) and other investigators from Northwestern University (Desai and Tsang). Our interest was generated by observations of our hepatology colleagues whose clinical impression was that their patients had an unusually frequent decrease in renal function when aminoglycoside antibiotics such as gentamicin, tobramycin or amikacin were used. Liver disease was, therefore, added to a rather long list of possibilities that were analyzed by Moore et al.(2, 3), Smith et al.(4) and Sawyers et al.(5) to identify risk factors for aminoglycoside nephrotoxicity using a large data base generated from two prospective, randomized, controlled and double-blinded clinical trials comparing tobramycin to gentamicin (6, 7). Liver disease was clearly a risk factor for aminogly coside nephrotoxicity when a simple univariate analysis was performed, and continued to be an important risk factor when a multivariate stepwise discriminant analysis was added to assess the ability of each independent risk factor to appropriately assign patients to groups with and without nephrotoxicity in the pres-