Prognostic Value of CD133 Caused by Mutant K-Ras and B-Raf—Letter

Kemper and colleagues (1) provided evidence that the expression of the cancer stem cell (CSC) marker CD133 is upregulated in colorectal tumors that have a hyperactivated Ras–Raf–MEK–ERK pathway and is, therefore, related to mutations in K-Ras or B-Raf. Thus, CD133 expression is not indicative for CSC numbers but rather for the mutation or activity status of the Ras–Raf pathway.

In our opinion, some key points should be discussed in more detail. First of all, the choice to focus on stage II colon cancer is valuable, as adjuvant treatment is a matter of debate in this subset. However, Kemper and colleagues do not include all the clinical variables with already known prognostic value, particularly lymphovascular invasion and the number of examined nodes. Therefore, it is difficult to argue for the additive value of CD133 evaluation in the prognostic stratification of stage II patients beyond conventional clinicopathologic features.

More importantly, CD133 levels seem to retain more prognostic information than that enclosed in the K-Ras/B-Raf mutational status. In fact, neither K-Ras nor B-Raf mutations [as well as microsatellite instability (MSI) status] are associated with relapse-free survival (RFS) at univariate analysis, whereas an analysis comprising all the examined variables is not provided. As reviewed in the work of Sinicrope and Sargent (2), K-Ras and B-Raf mutations do not predict RFS, whereas MSI status is convincingly associated with tumor recurrence in stage II colon cancer. The prognostic value of B-Raf mutation in colorectal cancer seems confined to the metastatic setting, as shown by the association with overall survival rather than RFS in larger series of stage II–III cases (2). Unfortunately, Kemper and colleagues do not evaluate the role of MSI in B-Raf mutant patients, as the literature suggests that prognosis may be particularly poor in MSI-low or stable B-Raf mutant tumors (2).

From a biologic perspective, presented data indicate that K-Ras and B-Raf mutations contribute to the regulation of CD133 expression, but they are probably not the only actors. Additional investigation, therefore, is needed to reach the conclusion that the link between CD133 levels and prognosis is only dependent on the Ras–Raf axis. Moreover, factors other than somatic mutations contribute to the regulation of the Ras–Raf pathway, as proven by the recent identification of a subpopulation of B-Raf wild-type tumors with poor prognosis and gene expression patterns similar to those of B-Raf mutant tumors (independently of the K-Ras status; ref. 3). Furthermore, it is unlikely that a single marker (such as CD133) covers all the complexity and dynamics of CSCs (4). For example, Huang and colleagues showed that only the CD133⁺/CD44⁺ fraction predicts metastatic risk in colorectal cancer (5). We suggest that CSC-specific signaling pathways, rather than putative CSC surface markers, may be more useful prognostic and predictive biomarkers.

In conclusion, factors other than K-Ras/B-Raf mutational status may be involved in CD133 regulation and, more intriguingly, additional CSC parameters may strengthen the value of CD133 in clinics beyond already available prognostic factors. We are therefore indebted to the authors for their report, but this is just a glimpse into the dark mysteries of CSCs in colorectal cancer.

See the Response, p. 4474