The article by Koch et al. [1] is another piece of work for our understanding of the potential of colorectal cancer to metastasize through detection of peripheral tumour cells both in the blood stream as well as in the bone marrow.Primary aim of the study was a correlation of the long-term recurrence-free survival of curatively resected patients with colorectal cancer liver metastases (CRCLM) with their status of detected peripheral cancer cells during surgery. Thirty-seven patients were prospectively studied over a period of three years for the detection of disseminated cancer cells by reverse transcriptionpolymerase chain reaction (RT-PCR) both in blood and bone marrow perioperatively. All patients underwent a curative liver resection for their CRCLM and had blood samples taken after induction of anaesthesia, immediately following liver resection and 24 hours post resection. Bone marrow samples were obtained at the beginning of the operative procedure. Positivity of cancer cells detected was correlated with recurrence-free survival (RFS). Clinical and pathological characteristics of the 37 patients were heterogeneous but were analyzed both in univariate and multivariate analyses for their importance in regards to RFS.Tumour cells were detected in blood samples in 30% of the patients preoperatively, in 46% intra-operatively and in 22% postoperatively. Bone marrow was found positive for tumour cells in 16% of the investigated patients (25). 25 patients (68%) developed tumour recurrence after a median follow-up of 38 months with nearly half of them solitary in the liver. Interestingly a low incidence of well established negative prognostic factors would not assume such a high rate of recurrences (51% positive lymph node primary, 22% more than 2 CRCLM, 19% tumour size larger 5 cm). Univariate RFS analysis revealed significant differences for positive bone marrow samples, intraoperative blood samples and the number of positive blood samples. Although not statistically significant in univariate RFS analysis, all patients with positive postoperative blood samples developed tumour recurrence as did patients with positive bone marrow samples. Worth mentioning is the fact that up to 65% of the patients with blood sample negative for the detection of tumour cells developed recurrence. In a multivariate recurrence-free survival analysis, both intraoperative positive blood samples and positive bone marrow were identified as being significantly correlated with a worse outcome, although analyzed in two different models. A wellknown prognostic index established by Yuman Fong's group [2] in over 1000 resected patients with CRCLM did not reach significance in this model.The authors discuss the value of their findings in regards to adjuvant treatment concepts and surgical approaches to minimize intraoperative tumour cell dissemination. They conclude that the detection of tumour cells both in peripheral blood samples and in bone marrow samples has been identified as new prognostic marker which may change surgical technique and ma...