Extract

The enumeration of circulating tumor cells (CTCs) holds great promise for clinical decision making. The CellSearch method has proven prognostic value in advanced breast, colorectal, and prostate cancers ( 1 ). However, this fact does not imply that there is no room for improvement. As mentioned by Hayes and Cristofanilli, CTCs cannot be detected in a substantial number of patients with metastatic breast cancer. Lack of expression of a marker essential for CTC detection may account for this problem. Epithelial cell adhesion molecule (EpCAM) expression on CTCs is crucial for several CTC assays. However, as acknowledged by Hayes and Cristofanilli and by Connelly and colleagues, EpCAM is not a perfect marker. Clinical breast cancer is a highly heterogeneous disease, also in terms of EpCAM expression, as shown in a single invasive breast tumor in Figure 1 (a color version is available online in Supplementary Figure 1 ).

To circumvent the problem of EpCAM heterogeneity in clinical breast cancers, we used homogeneous cell lines ( 2 ). It was by this approach that we were able to demonstrate that cell lines of the intrinsic normal-like breast cancer subtype were not captured by EpCAM-dependent approaches. Accordingly, in their series of 59 human breast tissue samples, Van Laere and coworkers found decreased EpCAM expression in those samples from breast tumors with the normal-like subtype. To further substantiate their findings, we performed a similar analysis on 969 publicly available Affymetrix U133A chip data from five different clinical breast cancer studies ( 3–7 ). We used z -score normalization for data from each study to evaluate differences in EpCAM expression between the different breast cancer subtypes. In analogy with the results of Van Laere and coworkers, we found that the gene encoding EpCAM, TACSTD1 , was expressed statistically significantly lower in the 68 tumors of the normal-like subtype (mean rank  = −0.84, 95% confidence interval [CI]  = −1.22 to −0.46) than in the other subtypes (mean rank  =  0.06, 95% CI  =  0.00 to 0.12; two-sided Mann–Whitney test, P < .001).

You do not currently have access to this article.