Testosterone concentrations (total and/or free) in serum are decreased in male patients with alcoholic and non-alcoholic liver cirrhosis (1). Accordingly, liver disease influences testosterone metabolism. The opposite effectthat of testosterone on liver diseasehas attracted less attention.Based on extensive animal studies, performed during the 1930s by Korenchevsky and coworkers, Hall & Korenchevsky (2) were the first to suggest a possible practical value of testosterone in the therapy of the liver. In 1946, Samuels et al.(3) were the first to administer anabolic-androgenic steroids (AAS) to liver patients in order to study the effect of 17-methyl testosterone on creatine formation. Rosenak et al. (4) therefore introduced AAS to liver patients for therapeutic reasons in 1947. Administration of testosterone propionate (25-1 00 mg intramuscularly three times a week) to 12 male patients with alcoholic cirrhosis was claimed to induce objective and subjective improvement, but liver function did not change significantly (4). Since then, AAS treatment has been claimed to exert positive effects on survival, complications of liver disease, clinical course and impotence in both uncontrolled and controlled studies. Textbooks on liver diseases, however, either do not mention AAS treatment (5, 6) or state AAS to be without efficacy in patients with liver cirrhosis (7).Because of these diverging opinions and the lack of effective treatment of most chronic liver diseases, the pertinent literature on A A S treat-ment of liver patients has been analyzed. The findings are discussed and compared to the effects of AAS on the liver in man and animals. The aim has been to draw conclusions where data offer a sufficient basis or to seek the explanation of discrepant findings.
Material and methodsTo provide the relevant literature, several sources have been utilized. For the years 1979-1982, a MEDLARS survey was undertaken, searching publications dealing with effects of AAS on the liver. Reports published during 1982 were further assured from Index Medicus. Finally, a cross-bibliographic check was made. Publications were included provided that they were: 1) Dealing with original data on effects of AAS in pa-2) Published in English or German for uncontrolled 3) Published in any language for controlled studies.Rouble publications were excluded. The publications included were subdivided into those dealing with cirrhotic patients and those comprising non-cirrhotic patients. The latter included patients with fatty liver, alcoholic hepatitis, acute viral hepatitis and chronic hepatitis.Concerning AAS, trivial names have been used with systemic names given in parentheses when first mentioned.For statistical analysis, Fischer's exact test, the t-test for paired data and the Chi-square test have been used. The relative mortality risk of controlled studies was calculated according to Armitage (8), using the Mantel-Haenszel pooled estimate. The type I error was fixed at 5%. tients with liver diseases, studies, or 160 GLUUD