Background: One neoadjuvant course of intra-arterial high-dose cisplatin (cis-diamminedichloroplatinum [CDDP]) tumor perfusion combined with intravenous sodium thiosulfate (STS) (cisplatin neutralizer) infusion is part of a multimodality concept for treatment of oral cancer. Recently, crystalline cisplatin embolization has been described as a novel treatment variant with increased tumor response rates. Methods: We have compared tumor and plasma concentrations of cisplatin and STS by means of microdialysis in 10 patients with oral cancer treated with intra-arterial cisplatin perfusion (150 mg/m(2) in 500 mL of 0.9% sodium chloride) and 6 patients with oral cancer treated with crystalline cisplatin embolization (150 mg/m(2) in 45-60 mL of 0.9% sodium chloride), respectively. The microdialysis catheter was placed into the tumor, and the intra-arterial catheter into the tumor-feeding artery. Cisplatin was rapidly administered through the intra-arterial catheter and STS (9 g/m(2)) was infused intravenously to reduce the systemic toxicity of cisplatin. STS infusion was started 10 seconds after the cisplatin infusion was started. Results: After embolization, cisplatin tumor maximum concentration (C-max) and tumor area under the concentration-time curves (AUCs) were about 5 times higher than those achieved after intra-arterial perfusion (C-max, 180.3 +/- 62.3 mumol/L versus 37.6 +/- 8.9 mumol/L), whereas the opposite was true for plasma concentrations (C-max, 0.9 +/- 0.2 mumol/L versus 4.7 +/- 0.6 mumol/L). STS plasma levels were about 3 times higher than its tumor concentrations (C-max tumor, 1685 +/- 151 mumol/L; C-max plasma, 5051 +/- 381 mumol/L). After the standard intra-arterial perfusion, the average STS/CDDP AUC ratios for tumor and plasma were 211 +/- 75 and 984 +/- 139, respectively. After cisplatin embolization, the respective ratios were 48.5 +/- 29.5 and 42,966 +/- 26,728. Conclusion: Molar STS/CDDP ratios of greater than 500 are required outside the tumor to neutralize cisplatin, whereas tumor ratios should be lower than 100 to avoid a loss of tumor cell killing. The first goal is achieved with both treatment modalities and the second only with cisplatin embolization, suggesting that crystalline cisplatin embolization is superior to intra-arterial cisplatin perfusion in terms of tumor cisplatin concentrations. Whether this translates into higher tumor response rates needs to be investigated further. (Clin Pharmacol Ther 2003;73:417-26.).