We read with great interest the letter of Steinert and colleagues (Gut 2005;54:1045–6). They described disseminated epithelial cells in the bone marrow of patients with colorectal adenomas. As adenomas are non-cancerous, it is questionable whether these epithelial-like cells really represent disseminated cancer cells. If so, the benign nature of intraepithelial neoplasia is basically challenged.

The search for micrometastatic disease in the bone marrow of cancer patients is an intriguing field in which many different diagnostic molecular modalities are being explored. However, the clinical impact of bone marrow micrometastatic disease is not clear. Patients with breast cancer frequently develop clinically manifest bone marrow metastases and in these patients detection of this type of micrometastatic disease has been shown to have prognostic impact.^1,2 Gastrointestinal cancers, however, relatively rarely develop overt bone marrow metastases and also the clinical impact of the detection of bone marrow micrometastases in these patients is less clear.

We studied the impact of bone marrow micrometastases in patients with oesophageal adenocarcinoma and encountered a concern which might be relevant for the interpretation of the conflicting findings, as described by Steinert et al. We obtained bone marrow samples of 20 patients who underwent a potentially curative oesophagectomy for oesophageal adenocarcinoma. Cytospins of these samples were stained with Ber-EP4 (Dako, Hamburg, Germany) and A45-B/B3 (Micromet, Munich, Germany) using the alkaline phosphatase-antialkaline phosphatase technique. In this series we found positive staining for these epithelial markers in four (20%) and five (25%) of the patients for Ber-EP4 and A45-B/B3, respectively. As a negative control group we obtained a bone marrow sample in 20 patients who underwent cardiac surgery and had no known cancerous disease. In this control group, to our surprise, we found positive stainings in three (15%) and four (20%) patients for Ber-EP4 and A45-B/B3, respectively. Positive staining in the bone marrow of patients with no known cancerous disease indicates that such results should be interpreted with caution and that these cells do not necessarily represent micrometastatic disease.

Many groups have studied micrometastatic disease for which many different detection methods have been used. Studies that have examined negative control patients have also reported positive bone marrow staining for Ber-EP4 cells in up to 10% of patients.^3,4 One study focused on the methodology of the alkaline phosphatase-antialkaline phosphatase staining and reported positive staining in up to 42% of patients without cancer. Based on these results it was suggested that plasma cells can directly react to alkaline phosphatase.⁵ Morphology of the cells is another important factor in the evaluation of possible micrometastases. In fig 1 of the letter Steinert et al, the cells do not have the morphological aspect of a tumour cell and also in other papers on micrometastases, morphology of the cells is not taken into account. In our data set, we could not detect cells in the bone marrow cytospins which had the morphological aspect of a tumour cell.

With current techniques, epithelial-like cells can be detected in the bone marrow of many patients not known to have cancer. Probably these cells have no clinical relevance in patients with adenomas. In future studies, more attention should be paid to the morphological criteria of the cells.