Conjugates between chemotherapeutic agents and antibodies, linked by a dextran bridge, were previously shown to be effective in suppression of hepatoma growth in uitm and in uiuo. However, scaling up of production of such conjugates may lead to a high degree of variation in molar ratios of drug to antibody in different batches. In this study, an alternative link between drug and antibody was evaluated. A conjugate between adriamycin and murine IgGl monoclonal antibodies to human a-fetoprotein was prepared using a polyglutamic-acid bridge. The simple and reproducible method of linking adriamycin to a specific site on the antibody enabled the binding of the drug to a-fetoprotein with a high yield (63% to 68%); the molar ratio of drughntibody was in the range of 110: 1 to 120: 1. The conjugate retained its capacity to bind to purified a-fetoprotein. Incorporation of ['HI-thymidine or ['HI-leucine into hepatoma cells, which express a-fetoprotein, was inhibited by the conjugate, compared with unconjugated antibody. Furthermore, 90% of this pharmacological activity was preserved, compared with free adriamycin. In uitro, the inhibitory activity of the polyglutamic acid conjugate was higher than that of a conjugate in which dextran was used as the linker between drug and antibody. In uiuo, both conjugates were equally effective in suppression of hepatoma growth transplanted subcutaneously in athymic mice. However, this effect lasted only during the treatment period of 2 to 3 wk. Six days after discontinuation of therapy, reacceleration of tumor growth was observed regardless of the conjugate used. These preliminary results suggest that polyglutamic acid may be used as an alternative link to dextran for coqjugation of anti-a-fetoprotein to adriamycin. However, use of monoclonal anti-cu-fetoprotein for delivery of adriamycin to the hepatocellular carcinoma transplanted in athymic mice causes only partial remission of tumor growth. Therefore conju-