Meta-analyses of studies on bone marrow micrometastases: an independent prognostic impact remains to be substantiated.

Abstract

In 1997, the immunocytologic detection of isolated tumor cells in bone marrow, termed micrometastasis, will be optionally included in the tumor-node-metastasis (TNM) classification indicated M1(i). In the present meta-analyses, 20 studies, which included 2,494 patients, regarding the prognostic influence of a positive bone marrow micrometastases (BMM) status on relapse-free and/or overall survival were analyzed. The literature search included the Medline and Current Contents bibliographic data bases from August 1980 to June 1997. The statistical evaluation considered the prognostic influence of the prevalence of micrometastatic cells in bone marrow on relapse-free and/or overall survival. The comparable effect estimate and its corresponding 95% confidence interval (CI) were calculated with the Mantel-Haenszel method using the originally published data of the retrieved studies. The presence of epithelial cells in bone marrow was detectable in all carcinoma types, with a median prevalence of approximately 35%. Fourteen of 20 studies found a positive correlation between positive BMM status and reduced relapse-free survival by univariate analysis, but only five of 11 studies confirmed positive BMM status as an independent predictor of short disease-free survival. Regarding overall survival, positive BMM status was identified univariately in five of 12 studies, but multivariately in only two studies, as an independent factor of poor survival. Despite the heterogeneity of the studies, calculation of the relative risk (RR) for reduced relapse-free survival was possible for breast cancer, which resulted in a Mantel-Haenszel RR (RR(MH)) of 1.34 (95% CI, 1.27 to 1.42). In conclusion, the results suggest that the prognostic impact of epithelial cells in bone marrow remains to be substantiated by further studies using standardized methodic protocols before its entrance in the TNM classification.