Isolated organ perfusion with hyperthermia and melphalan with or without tumor necrosis factor-of both the limb and liver. Many patients, however, will eventually relapse at distant sites. We used everse transcription-polymerase chain reaction (RT-PCR) to determine whether significant tumor microembolization occurs in patients undergoing isolated limb perfusion (ILP), isolated hepatic perfusion (IHP), or hepatic resection. Methods: designed for RT-PCR to screen melanoma or colon adenocarcinoma, respectively. RNA from human melanoma lines (Pmel and 1286) and human colon adenocarcinoma lines (H508 and HT29) were used to generate positive control cDNA. Normal human blood was inoculated with tumor cells at concentrations that ranged from 10 Systemic and perfusate blood samples were drawn from 15 patients (8 patients underwent IHP, 5 patients underwent ILP, and 2 patients underwent resection) before the start of the operation, immediately before and during the perfusion, and postoperatively. Mononuclear cell fractions were separated from the blood samples and RNA was extracted for the RT-PCR assay. Standard primers for human Results: the 8 IHP patients, 6 had colon metastases and 2 had ocular melanoma metastases to the liver. All 5 ILP patients had in transit melanoma of the extremity. Two patients with colon metastases to the liver were found to have resectable disease. There were no detectable circulating tumor cells in the systemic circulation either preoperatively or postoperatively in all 15 patients that were screened. Conclusions: Manipulation of the limb or liver followed by resection or isolated hyperthermic perfusion does not cause detectable release of circulating tumor cells. The late development of distant metastases observed in many of these patients does not correlate with the ability to measure circulating tumor cells during regional therapy.