Several recent studies showed that the phenotypic and genotypic characterization of CTCs may provide valuable information of clinical relevance [16–18]. However, unbiased CTC isolation is a crucial initial step for their subsequent characterization.
Different methods have been routinely employed for CTC enrichment and detection. The CellSearch System is a semi-automated enrichment and immunocytochemical detection system approved by the FDA, using EpCAM expression as its primary mechanism of selection of CTCs. In a cohort of metastatic breast cancer patients, an average recovery of 74.9% was obtained . Enrichment by MACS columns is another technique used. This system involves tumor cells coupled with specific microbeads that are enriched by removing unlabeled cells via washing, using a column placed in a magnetic device. Recovery rates ranging 60%-80% have been reported . More recently, the development of a microchip technology based on EpCAM-coated microposts capture of epithelial cancer cells allowed recoveries over 65%, and purity of over 50% . All the enrichment methods mentioned above are based on the expression of surface markers on CTCs, in particular, EpCAM.
We tested three different enrichment methods (positive selection, CD45 depletion and the combination of both) in a spiking experiment model using a cell line known to be positive for EpCAM, and CK 7 and 8. We observed the highest recovery in sole CD45 depletion. In the case of EpCAM-positive selection, the recovery rate was lower compared to many other studies published. In order to evaluate if cancer cells might be lost in the non-enriched fraction, both the fractions (enriched for EpCAM-positive cells and non-enriched) were analyzed by flow cytometry. In the non-enriched fraction, we were able to find a few cells that tested CK and EpCAM positive. The mean fluorescence intensity of the EpCAM-positive cells in the non-enriched fraction resulted to be lower in comparison to the mean fluorescence of the SW620 cells and considerably lower in comparison to the fluorescence of the SW620 cells we were able to detect in the enriched fraction (data not shown). Excluding the possibility of EpCAM down-regulation after antibody binding , the relatively low fluorescence signal due either to inferior EpCAM surface expression, or to the weakening of the Fitc-staining (the lapse of time between staining and FACS analysis in case of positive enrichment is of at least 90 minutes compared to 25 minutes when CD45-depletion was performed) might be an explanation to the low recovery rate obtained after EpCAM-based immunoselection in accordance to the fact that the cells' recovery would increase with increasing fluorescence of the Fitc-labelled cells. Consequently, CTCs that do not express EpCAM at sufficient levels could be missed by these assays, which may limit the sensitivity, and could potentially lead to a loss of particular cell subpopulations. Indeed, heterogeneous expression of epithelial surface markers has been previously reported in different tumor entities at tissue level [23, 24], as well as the loss of EpCAM expression in the case of epithelial-mesenchymal transformation [25, 26].
Only a few studies applied negative enrichment for CTCs detection [27–32]. Lara et al. reported 46% average recovery rate and depletion efficiency up to 5.7 Log by enriching cells by means of a flow-through system . A similar recovery rate was obtained by Zigeuner et al., who compared in spiking experiments positive selection of epithelial cells with the antiepithelial antibody BER-EP4 with CD45 depletion. Furthermore, when a single tumor cell was spiked in 30 ml, CD45 depletion revealed epithelial cells in all 14 cases, whereas positive selection in 12 of 14 cases . Higher recovery rates found to be comparable to ours were obtained by Meye et al.  by applying CD45 autoMACS depletion. The same group also observed a significant correlation between presence of CTCs and lymph node status, and occurrence of synchronous metastases in a cohort of patients affected with renal cell carcinoma .
We detected CTCs after CD45 depletion in 48 epithelial cancer patients and 22 melanoma patients. The 64% of melanoma patients resulted to be positive for CTCs which is in accordance to results of a previous study from our group . The median count of CTCs in melanoma patients was significantly higher than the median count of CTCs (defined as CD45-EpCAM+CK+) in carcinomas signifying that either hematogenous spread of melanoma is somehow easier, or that the definition of CTCs in carcinoma is too restrictive leading to an underestimation of CTCs when the common definition of EpCAM CK double positive is applied. However, when defining cells as EpCAM+ and CK+, our data showed similar or slightly higher detection rates compared to data reported by other authors who detected CTCs in comparable cohorts of patients (56% in metastastic breast cancer , 64.7% in NSCLC , 38% in ovarian cancer and 31% in gastric cancer ).
We used antibodies against CK7 and CK8 for cytokeratin detections. We chose CK7 and CK8 (always associated to expression of CK18) because they resulted to be the most expressed CKs in carcinomas along with CK19 . In particular, CK8 is expressed by a variety of carcinomas. Since CK expression pattern in carcinoma is heterogeneous, addition of further anti-CK antibodies might increase the sensitivity of the detection method [38, 39], but congruently the false positive rate. In our preliminary experiments, the use of CK19 as an additional antibody resulted in a higher background in healthy controls (data not shown).
We analyzed CTCs in peripheral blood and in matched pleural effusion or ascites specimens of seven patients. In five out of seven cases a correspondence of EpCAM and CK expressions was observed between CTCs, and tumor cells in ascites or pleural effusion samples. This result is consistent with the present understanding that CTCs and disseminated tumor cells released from the primary tumor tissue, i.e, with the same origin, or might re-circulate between metastatic sites . However, in two cases, although EpCAM+ CK+ cells were detected in peripheral blood, CK positive cells were detected in ascites, which may be due to the fact that circulating cells with different phenotypic characteristics may specifically colonize an organ [41–43] or an anatomical space. Ascitic fluid may in this case represent a reservoir for naturally enriched, disseminated tumor cells bearing specific features as it has been shown to occur in other compartments . In the two NSCLC patients, only CK- positive cells could be detected both in blood and pleural effusion. Cells obtained from pleural effusion could be passaged in culture several times, supporting the hypothesis of their neoplastic origin. An enrichment method based on EpCAM-positive selection would therefore not have been able to detect this fraction of cells. Consequently, the definition of CTCs as CD45- and EpCAM and CK double positive might be too restrictive. Loss of epithelial markers like EpCAM and CK is a common phenomenon which typically occurs in cells which undergo the epithelial-mesenchymal transition (EMT), a process that has been linked to the generation of cells with properties of stem cells, and to the ability of tumor cells to enter the circulation and seed metastases. EpCAM-CK double positive CTC might represent only a subpopulation of the whole pool of CTCs. Establishment of new assays based on EMT or stem cells markers are therefore necessary.