Antitumor activity of ZD1694 (tomudex) against human head and neck cancer in nude mouse models: role of dosing schedule and plasma thymidine.

We studied the antitumor activity and toxicity of ZD1694 (tomudex), a specific inhibitor of thymidylate synthase (TS), in nude mice bearing human head and neck squamous cell carcinoma A253 and FaDu xenografts. Mice were treated by single i.v. push (i.v. x 1), i.v. push once a week for 3 weeks (weekly x 3), and i.v. push once a day for 5 days (daily x 5), and the maximum tolerated doses (MTDs) of ZD1694 were 300 mg/kg, 60 mg/kg/week, and 30 mg/kg/day, respectively. ZD1694 was moderately active against both A253 and FaDu xenografts. Antitumor activity was schedule-dependent in both tumors: weekly x 3 > or = i.v. x 1 >> daily x 5. In contrast, the rank order of toxicity was daily x 5 >> weekly x 3 > or = i.v. x 1. ZD1694 at the MTD produced 20% complete tumor regression and 20% partial tumor regression (PR) with i.v. x 1 and weekly x 3 schedules and 12-day tumor growth delay with daily x 5 schedule against FaDu xenografts. No complete tumor regression was achieved with ZD1694 with any schedule against A253; a 20% PR, 40% PR, and 10-day tumor growth delay were observed with i.v. x 1, weekly x 3, and daily x 5 schedules, respectively. The data indicate that ZD1694 was slightly more effective against FaDu than against A253. Of interest and potential clinical importance was the observation that ZD1694 was still active at doses lower than the MTD (> or =1/3 MTD), which showed a high therapeutic index and wide safety margin. Study of ZD1694 compared with 5-fluorouracil and 5-fluoro-2'-deoxyuridine at the MTD revealed that the antitumor activity of ZD1694 was comparable with or superior to 5-fluorouracil and 5-fluoro-2'-deoxyuridine against both A253 and FaDu xenografts, with less toxicity. High plasma thymidine in mouse relative to human (approximately 1.3 microM and <0.1 microM, respectively) may complicate the study of antitumor activity and toxicity of TS inhibitors with human tumor xenografts grown in the mouse. To test this hypothesis, we preadministered methoxypolyethyleneglycol-conjugated thymidine phosphorylase (MPEG-TPase; 2500 units/kg/dose) to reduce mouse plasma thymidine, then treated with various doses of ZD1694 using the daily x 5 or i.v. x 1 schedules in the A253 tumor model. MPEG-TPase significantly increased the toxicity of ZD1694; the MTD of ZD1694 plus MPEG-TPase was reduced 3- and 10-fold compared with ZD1694 alone for i.v x 1 and daily x 5 schedules, respectively. However, preadministration of MPEG-TPase did not potentiate the antitumor activity of ZD1694 with either schedule. The data indicate that the study of TS inhibitors in rodent models may not be suitable for predicting a safe dose for clinical study. However, rodent models, particularly human tumor xenografts, are still useful models for evaluation of antitumor activity and schedule selection for TS inhibitors.