Summary Several reports have evaluated the correlation between cathepsin-D and overall survival or disease-free survival in node-negative breast cancer patients. Because conflicting data have so far been reported, a meta-analysis was conducted to clarify this problem. Eleven studies were included in our meta-analysis (total of 2690 patients). A specific meta-analytical methodology for censored data was used, and disease-free survival was the primary end point. Patients with low cathepsin-D levels had a significantly better disease-free survival than patients with high cathepsin-D values (meta-analytical odds ratio from 0.59 to 0.60 over the interval from 1 to 7 years). A secondary metaanalysis conducted exclusively on the data from eight studies based on cytosol assay gave substantially similar results. One limitation of our study is that the cut-off values to define high and low cathepsin-D concentrations were not identical in the various studies included in our meta-analysis (range from 20 to 78 pmol mg-' protein), thus introducing a possible bias in the statistical analysis of the data. However, a simulation based on the well-accepted method of the so-called publication bias showed that more than 100 null studies would be required to lead our results to a statistical level of non-significance. Considering the results of our meta-analysis, we conclude that the data presently available confirm a statistically significant association between high cathepsin-D values and poor disease-free survival in node-negative breast cancer patients.
Keywords: meta-analysis; cathepsin D; node-negative breast tumourThe identification of new prognostic factors, more closely related to tumour cell biology, would be of utmost importance for treatment planning in human breast cancer. Improvement in discrimination between low-and high-risk cases is of major concern, particularly in the subset of node-negative patients, 70% of whom are cured by surgery alone and would therefore be spared the cost and potential toxicity of adjuvant chemotherapy (McGuire, 1989;Copper, 1991). To date, several biological factors have been identified and proposed as potential prognostic indexes in human breast cancer. Among these, particular attention has been focused on proteolytic enzymes, such as cathepsin-D and urokinase-type plasminogen activator, which are involved in basement membrane/extracellular matrix degradation and tumour invasiveness and metastasis (Liotta et al, 1991).Cathepsin-D, firstly identified as a 52-kDa oestrogen-regulated glycoprotein (Westley et al, 1970), displays both proteolytic activity in culture and an autocrine mitogenic activity in breast cancer cells (Vignon et al, 1986). The involvement of cathepsin-D in cancer invasion is also supported by the demonstration that transfection of cathepsin-D cDNA into rat tumorigenic cells increases their metastatic potential in nude mice (Garcia et al, 1990). In addition, higher cathepsin-D levels have been found in breast cancer patients with metastatic lymph node involvement than in no...