We have clinically demonstrated the dynamic pattern of CEC and CEP during the course of chemotherapy treatment. The mean levels of CEC, V-CEC, and CEP all significantly decreased during the first week of chemotherapy. Then the levels of CEC and V-CEC returned gradually after one week whereas the level of CEP remained significantly reduced. After six cycles of chemotherapy, the level of CEC decreased significantly, V-CEC trended toward a decrease, and CEP increased significantly. These results were not significantly affected by clinical factors including the existence of a tumor, the chemotherapy regimen, and the use of GCSF.
The sole antigen to distinct CEPs with CECs is CD133 at present. However, human CD133 is not only expressed in endothelial progenitor cells but also in haematopoietic progenitor cells and bi-potential haemangioblasts. Currently, a phenotypic distinction between these three kinds of cells is not feasible and controversial . In our six-color flow cytometry, CD45 expression can be used to exclude haematopoietic cells from the analysis. Theoretically, CD146 is expressed in endothelial cells but not platelet, and CD133 expression can be used to differentiate platelet and endothelial microparticles from CEP. However, lack of DNA staining is at a risk since a nuclear-staining molecule can be useful to exclude aggregated platelets and/or endothelial microparticles (which share some surface markers with CECs and CEPs) from the CEC count .
Longitudinal follow-up found that the levels of CEC and V-CEC increased from the baseline before chemotherapy to the first day of the first cycle of chemotherapy, but this did not occur with the CEPs (Figure 2). Current thought is that CECs are mature endothelial cells that have detached from their basement membrane in response to a blood vessel injury [24, 42]. The increase in CEC and V-CEC levels we observed may be due to vessel damage caused by implanting port-A catheters or by tumor resection surgery, since all of the patients were treated with one of these two procedures before the first day of the first cycle of chemotherapy. The CEPs are progenitor cells that are recruited from bone marrow rather than detached from the vessel wall, and their level did not change in a similar manner.
Since several studies have indicated that levels of CEC and CEP correlate with tumor size and grade, [20, 40, 43–46] we assumed that tumor-bearing patients would have higher CEC and CEP levels than patients without tumors, but our analysis found no significant differences. Perhaps CEC and CEP levels correlate with not just the presence of a tumor, but with many factors, including tumor size, tumor grade, vascular invasion, and lymphatic invasion. Although patients who received adjuvant chemotherapy mostly had no gross residual tumor, but vascular or lymphatic invasion, even cancer related cytokines was possible and may explain for the CEC and CEP kinetics. Since our analysis of CEC and CEP levels in tumor–bearing and non–tumor bearing patients included a mix of many variables, we cannot make a conclusion about which variables affect their levels. Here we provided some representative figures of the CEC and CEP kinetics after administration of chemotherapy in tumor-bearing and non-tumor bearing patients (Additional file 6: Figure S6). The CEC and CEP kinetics consistently showed similar wave pattern and had no obvious differences between patients with and without tumor bearing. This explanation is also consistent with a study that found that CEC and CEP levels did not significantly change before and after surgery in 15 breast cancer patients, although vascular invasion and tumor size were independently associated with the CEC levels .
Administering GCSF to mice has been found to elevate CEP and accelerate tumor growth [47–49]. A human study analyzing CEC levels in four patients (one with seminoma, two with nonseminoma testis cancer, and one with small-cell lung cancer) receiving chemotherapy with the support of GCSF found that three out of four patients had CEC amounts increased 8- to 9-fold 3–8 days after GCSF administration . In the same study, the CEC level was almost undetectable in leukopenic patient. This implies that the GCSF effect on CEC levels may be markedly related to bone marrow recovery after chemotherapy. When we compared the seven patients who received GCSF during chemotherapy courses to the eight who did not, we found no differences in the dynamic trends in the CEC and CEP levels between these groups. Since our patients received different chemotherapy regimens and may have different grades of bone marrow suppression which lead to different timing of bone marrow recovery. We suggest that in our study, the effects of chemotherapy on CEC and CEP may mask the GCSF effect.
We wanted to understand whether CEC and CEP levels are affected differently by different chemotherapy agents. Certain chemotherapy drugs, such as taxane and fluorouracil, have been demonstrated to rapidly induce CEP mobilization and subsequent tumor homing, while others, such as gemcitabine, cisplatin, and doxorubicin, do not . Our analysis of 29 cycles of taxane-based chemotherapy and 26 cycles of non-taxane-based chemotherapy found no significant differences in the dynamic trends of these two groups. However, this is not conclusive, since all of the chemotherapy regimens in this study used multiple chemotherapy agents and it is difficult to evaluate the effect of single chemotherapy agent on CEC/CEP kinetics. We provided some representative figures of the CEC and CEP kinetics after administration of different types of chemotherapy regimens in five different patients (Additional file 7: Figure S7). Their CEC and CEP kinetics consistently show similar wave pattern. It suggests that dynamic changes of CEC and CEP induced by chemotherapy may have more significant effect than using different drugs. It supports our conclusion that timing of measurement of CEC and CEP after chemotherapy should be seriously considered and unified in the future studies using CEC and CEP as endpoint.
The levels of CEC and V-CEC significantly decreased as the number of chemotherapy courses progressed (P = 0.02), but the CEP levels significantly increased during the first three courses of chemotherapy (P = 0.02) and then declined during later treatment courses. Although this study was restricted by the small number of patients and chemotherapy cycles, the trends we observed were consistent with the analysis of Furstenberger et al. . They analyzed the levels of CEC and CEP before and after neoadjuvant chemotherapy in breast cancer patients and found that, after two cycles of chemotherapy, the CEC levels decreased and the CEP levels increased. This phenomenon may be explained by the CEPs behaving as progenitor cells that could be mobilized from the bone marrow by a regular dose of chemotherapy. Although the CEPs would gradually differentiate into CECs, the next course of chemotherapy might destroy some cells that were in transition before they could fully differentiate into CECs. This may explain why CEC levels decreased and CEP levels increased after cycles of chemotherapy.
Many preclinical studies have indicated that CEPs contribute to tumor growth [23, 40, 43], which warrants further concern that chemotherapy may remobilize CEPs and trigger additional tumor growth. We demonstrated that the CEC and CEP levels decreased in the first week of chemotherapy, CEC increased after one week with each cycle of chemotherapy, and CEP rebounded even more slowly than CEC. This suggests to us that the rate of tumor control achieved by chemotherapy may theoretically be helped by adding metronomic chemotherapy or an antiangiogenic agent one week after chemotherapy to suppress the rebound of CEC and CEP. This hypothesis warrants further prospective study.
Our observation that CEC rebounds earlier than CEP has two possible explanations: 1) CEP is not the only source of CEC, which has three main sources: tumor vasculature, normal vasculature, and CEP. The three sources could explain why CEC is not fully synchronized with CEP. 2) The CEPs are a kind of progenitor cell where one CEP may differentiate into a population of CECs, which would explain the earlier rise in the slope of the CECs as compared to the CEPs.