This study is the first to systematically evaluate CECs and CEPs in patients with osteosarcoma. The definition of CECs and CEPs has differed between studies and a consensus definition has not been established [17
]. The current study therefore evaluated a range of markers that have been used to define these cells in previous studies. Despite investigating CECs and CEPs across a range of definitions, cell levels generally did not correlate with osteosarcoma diagnosis or clinical features. Only VEGFR2+ CECs showed differential cell levels between patients and controls, with patients having lower levels than controls. Other groups have also correlated levels of CEC activation and viability with clinical features in patients with cancer [11
]. Evaluation of these same subsets in patients with osteosarcoma in the current study yielded no meaningful associations.
These negative results raise the possibility of methodologic flaws in the current study, though this possibility seems less likely for several reasons. First, the cell counts obtained are within the ranges reported in previous studies [13
]. Second, CEC and CEP counts demonstrated large interpatient variability in both cases and controls, as has been reported by other groups [14
]. Third, the reference laboratory used in the current study serves as the reference laboratory for evaluation of CECs and CEPs as biomarkers in Children's Oncology Group studies of antiangiogenic agents. This laboratory therefore routinely performs this assay on samples obtained from patients across North America. Fourth, while the samples sizes are small and inadequate power may account for these negative results, CEC and CEP levels completely overlapped between each evaluated clinical category. As such, enrollment of additional patients was thought to be unlikely to yield additional meaningful information. Finally, other groups have incorporated CD34 expression into the definition of CEPs [7
]. Our flow cytometry panel did not include CD34, but utilized CD133 to differentiate CECs from CEPs [11
]. This difference from other published reports evaluating CEPs may have also contributed to our negative results.
Another explanation for these negative results may reflect a fundamental difference in the process of angiogenesis between patients with sarcoma and carcinoma. For example, the pathways that regulate VEGF expression appear to differ between ras
-transformed fibroblasts and epithelial cells [21
]. Moreover, microvessels appear to be more diffusely distributed in sarcoma, while more clustered in carcinoma [22
]. In addition, osteosarcoma typically arises from trabecular bone, which has a unique and rich vascular supply.
While most studies of CECs and CEPs have been performed in patients with carcinoma, three reports have included patients with sarcoma. CEC levels were approximately two-fold higher among 15 patients with gastrointestinal stromal tumor compared to healthy controls [23
]. CEP levels appear to be increased in patients with classic Kaposi's sarcoma compared to healthy controls [24
]. Another report evaluated 45 children with a range of solid cancers, including 17 patients with sarcoma (7 with osteosarcoma) [19
]. While CEC levels did not differ between patients and controls, VEGFR2+ CEP levels were increased in patients compared with controls and in patients with metastatic disease compared with patients with localized disease. No additional subset analyses were performed based on histology.
In summary, the current results add to a growing body of literature evaluating CECs and CEPs in patients with cancer. While CEC levels have typically been shown to be increased in patients with cancer, the current study indicates that this finding is not universal across cancer histologies. Additional evaluation of CECs and CEPs in patients with sarcoma will be necessary to determine if these biomarkers are useful in sarcoma histologies.