Micrometastases within bone marrow have been shown to indicate a poor prognosis in patients with epithelial tumours. However, the degree to which micrometastases represent true residual disease or cell shedding and metastatic potential, is unclear. To explore whether micrometastases represent residual disease, bone marrow taken from carefully staged patients before and after (> 6 months) "curative" resection of a primary gastrointestinal cancer was studied prospectively. Seventy two consecutive patients were studied; the only exclusions were patients with known overt metastatic disease at the time of surgery. Micrometastatic cells were quantified per 10(5) marrow cells by flow cytometry after staining for contaminant cytokeratin-18 positive cells. Micrometastases were detected preoperatively in 22% (16/72) of all patients, comprising 11 (23%) of 48 with colorectal cancer, five (33%) of 15 with gastric adenocarcinoma and none (0%) of nine with oesophageal squamous cancer. Although fewer metastatic cells were detected in postoperative bone marrow, and clearance of marrow deposits was evident in most patients, the persistence of micrometastases in five of 16 patients after resection, without evidence of tumour recurrence, indicates a subset with true residual disease. Detection of micrometastases postoperatively (persistent or newly developed) was significantly associated with development of overt metastases during the subsequent 12-18 months of follow up (nine of 19 patients) when compared with patients testing negative for micrometastases (eight of 53; p < 0.01). Preoperative detection of micrometastases may reflect either transient shedding of cells, metastatic potential or residual disease, but postoperative micrometastases indicate minimal residual disease. Identification of these patients is important because they may benefit from adjuvant therapy.
