It seemed us important to develop analyses able to detect cells which are arising from EMT and spreading into the blood stream. Many arguments reported in recent literature sustained the choice of markers used in our study: PI3Kα, Akt-2, Twist1, ALDH1, Bmi1, CD44 [14
]. In breast cancer, some patients develop metastasis many years after the apparently successful therapy of the primary cancer. The residual cancer disease is based on the notion of occult metastatic tumor cells in bone marrow. Metastasis formation from epithelial tumors progresses through dissemination of tumor cells that can follow two major ways: bloodstream or lymphatic vessels [6
]. Cancer cell invasion seems proceed from EMT program [21
]. The latter is associated and interacts with cellular pathways that confer new characteristics to the cells: apoptosis resistance, chemo and radio resistance. EMT can be viewed as a continuum of a progressive dedifferentiation leading to cells with stem-cell like properties. Overexpression of a few EMT regulators may be sufficient to generate the cancer stem cell phenotype [22
]. A molecular link between EMT and stemness emerged with the finding that Bmi1 is a direct transcriptional target of the EMT inducer Twist1 [23
]. These phenomena establish a hierarchy of cancer cell phenotypes from mesenchymal to stemness status.
Our results led to classify patients as follows. The first phenotype of ddCTC is essentially mesenchymal with markers of EMT. Among the 3 markers, the predominant one is PI3Kα, then in equal proportions Akt-2 and Twist1. The second phenotype is characterized by ALDH1 expression, a stemness marker. The third phenotype is a mixture of the two previous ones and represents 33% of positive patients.
Aktas et al
evaluated 226 blood samples of 39 metastatic breast cancer patients during a follow-up of different therapies for the expression of the stem cell and EMT markers [16
]. In 31% of detected CTC positive samples, 62% were positive for at least one of the EMT markers and 69% for ALDH1. Thus they demonstrated for the first time that a major proportion of CTC of metastatic breast cancer patients shows EMT and tumor stem cell characteristics. In another study, Kallergi et al
identified in patients with early and metastatic breast cancer, CTC expressing Twist and vimentin. Higher incidence of these cells in metastasis disease than in early stage breast cancer supports the hypothesis that EMT is involved in metastatic potential of CTC [24
]. Moreover the detection rates they reported are similar to ours. In our study, 39% of patients who were positive for EMT and ALDH1 markers had Bmi1 and CD44 positive cells in 67% and 33% respectively. The expression of Bmi1 and CD44 arose in 26% and 13% of ddCTC samples. These results support the stemness characters of these positive samples. Indeed it was shown that EMT generates cells with stem like properties [25
] and that the stemness factor Bmi1 is regulated by Twist1 [23
It is the first time that ddCTC presence in the blood is demonstrated with such a high rate (24/61 patients) at least when the primary breast tumor is diagnosed without metastasis. If CTC are considered as a prognostic factor, our results showed clearly that detection of all cell phenotypes should be taken into account to have an idea of the disease progression. Ongoing trials in our laboratory indicate that ddCTC are more frequently detected than CTC characterized by AdnaTestBreastCancerSelect
kits (GA733-2, Muc-1 and Her-2) in primary breast cancer patients. Preliminary results, upon 400 patients showed that 8% of patients are positive for CTC according to this previous basic AdnaGen kit (data not shown). This result is not so far from those of the literature. Banys et al. found 12% of positive patients upon about 209, by using the same analytical test [17
]. Braun et al
demonstrated that incidence of bone marrow micrometastasis (disseminated tumor cells) was similar in patients with lymph node metastasis and those whithout [27
]. Moreover Menard et al
suggested that lymph node metastasis are not necessarily associated with cancer hematogenous spread [28
]. Our results stress that absence of lymph node invasion is not a criteria of non dissemination. Among 45 N-
patients 47% showed a ddCTC positive profile and upon 16 N+
patients only 19% had a positive profile. Nevertheless such a result does not weaken the negative pronostic value of lymph node invasion. These results could be explained as indicated by Chaffer and Weinberg that spread to other sites occurs mostly via the blood stream [6
]. Our results are in agreement with their observations. Albeit statistical results indicated a relationship between axillary node invasion and detection of ddCTC, the significance level had a borderline value (p
= 0.05). These preliminary statistical results have to be confirmed by a larger trial avoiding an inaccurate inference of Khi2.
Some methods for CTC analyses detect entities that by all criteria are cancer cells but which lack the ability to invade, proliferate and cause metastasis [29
]. We can speculate according to these previous publications, that the CTC population is an heterogeneous one. According to these observations, our results showed dedifferentiated EMT and stemness subpopulations in primary breast cancer patients. Such a discrimination had been suggested by Mego et al.
]. Indeed they demonstrated that metastatic breast cancer patients, had substantially different prognosis correlated to CTC subpopulations. A future prospective study is warranted to determine the incidence of ddCTC characterization on the prognosis. We believe that follow-up of ddCTC may facilitate the monitoring of therapeutic agents targeting these cells.