Portal de Periódicos da CAPES

A new approach for rapid and reliable enumeration of circulating endothelial cells in patients

2012; Elsevier BV; Volume: 10; Issue: 5 Linguagem: Inglês

10.1111/j.1538-7836.2012.04681.x

1538-7933

Jaco Kraan, Michiel Strijbos, Anieta M. Sieuwerts, John A. Foekens, Michael A. den Bakker, Cornelis Verhoef, Stefan Sleijfer, Jan W. Gratama,

Single-cell and spatial transcriptomics

Background: Mature circulating endothelial cells (CECs) are surrogate markers of endothelial damage/dysfunction. A lack of standardized assays and consensus on CEC phenotype has resulted in a wide variation of reported CEC numbers (4–1300 per mL). Objectives: Given the need for a quick, reliable, robust and validated CEC assay at an affordable price, we present a novel approach to enumerate CECs using a multi-parameter flow cytometric (FCM) method without immunological pre-enrichment. Methods: CECs were defined as CD34+, CD45neg, CD146+ and DNA+ events based on the immunophenotype of endothelial cells from vein-wall dissections. As CECs express high levels of CD34, we based our assay on absolute CD34 counts after analyzing all CD34 positive events in a total blood volume of 4 mL needed for a precise enumeration of CECs at a frequency of < 1 cell μL−1. Results: The endothelial origin of CECs was confirmed by morphology, immunohistochemistry and gene expression. The new FCM assay was tested in parallel with a validated assay (i.e. CellSearch®). CEC levels ranged from 4 to 79 CEC mL−1 in healthy individuals and were significantly higher in patients with advanced solid malignancies (P = 0.0008) and in patients with hematological malignancies (P < 0.0001). Conclusions: This flow cytometric method should be useful as a fast and economical assay to enumerate and characterize CECs. Current opinion states that circulating endothelial cells (CECs) are mature endothelial cells, which have been shed from the vascular cell lining as a result of vascular damage. As numerous diseases are associated with vascular damage, the enumeration of CECs has for long been considered a promising tool to monitor disease activity with a potential to assess prognosis and response to treatment. Elevated CEC numbers have been found in multiple diseases, including cardiovascular disorders [1, 2], infectious diseases [3-5], immune disorders [6, 7], pulmonary hypertension [8], sickle cell anemia [9, 10], status after organ transplantation [11, 12] and cancer [13-18]. In recent years, several methods have been described to enumerate CECs. However, there is no consensus between CEC phenotypes used in these enumeration techniques. Importantly, most methods have not been properly validated [19]. In combination with the relative rarity of CECs, this situation has resulted in a wide variation of reported CEC numbers, hindering further progress in the field. The great variation in results is exemplified by the reported CEC numbers in healthy humans ranging from 4 to 1300 cells mL−1 [14, 16, 20, 21]. Currently, most CEC enumeration tests rely on an enrichment step by immunomagnetic bead isolation (IB) using magnetic particles coupled to monoclonal antibody (mAb) targeting an endothelial antigen such as CD146, followed by flow cytometry or visual counting with a fluorescent microscope to identify CECs on the basis of morphological or immunophenotypical criteria [21, 22]. A major advantage of this latter approach is that it permits visual identification of CECs, which allows discrimination between CECs and endothelial microparticles or platelets with a strongly overlapping phenotype. However, any enrichment step inevitably leads to cell loss and underestimation of the actual CEC number, whilst manual bead-based isolation procedures are labor intensive and difficult to standardize. A variant of manual IB enrichment techniques is the CellSearch® system (Veridex, Raritan, NJ, USA). Initially designed to detect circulating tumor cells, this system provides a fully automated enrichment procedure that is followed by semi-automated image cytometry. This assay has a high recovery and good reproducibility. Importantly, CECs isolated in this way have been validated by global gene expression [17, 23]. Drawbacks of the CellSearch® system are the costly equipment and reagents, and the assay cannot be customized. Also, the maximum number of eight samples that can be analyzed in a single run, combined with the relatively long duration of a complete run (approximately 4 h), does not allow high throughput analysis. Given the need for a quick, reliable, robust and validated CEC assay that is also affordable, we present a novel and extensively validated approach for enumerating CECs using a multi-parameter flow cytometric method (FCM) without immunological pre-enrichment and the accompanying cell loss causing underestimation of the actual CEC number. The assay is based on absolute CD34 counts after analysis of all CD34-positive events in a total blood volume of 4 mL using a 'live gate' on CD34+ events to exclude most of the cells that are not of interest from the analysis, thus allowing the assessment of CEC as CD34+, CD45neg, CD146+ and DNA+ events. The selected markers were tested for CEC specificity on endothelial cells (ECs) from vein-wall dissections and the endothelial origin of the cells designated as CECs was confirmed by morphology, immunohistochemistry (IHC) and expression of genes thought to occur specifically in CECs. Additionally, the agreement with another validated assay (CellSearch® system) [24] was tested in healthy subjects and patients with pathologies known to be associated with high CEC levels. Peripheral blood (PB) was collected using EDTA-containing or CellSave™ tubes (Veridex) from healthy donors (n = 30), patients with advanced solid malignancies (n = 55) and patients with hematological malignancies in remission prior to cytoreductive therapy as preparation for allogenic stem cell transplantation (n = 60). Samples were stored at room temperature and examined within 8 h after venipuncture. All patients provided written informed consent and the study protocols were approved by the local research and ethics committee. Based on literature [24], we defined CEC as CD34+, CD45neg, CD146+ and DNA+ events. To confirm the CEC specificity of the combination of CD34 positivity, CD146 positivity and CD45 negativity, endothelial cells were isolated from dissections of vein and artery walls using collagenase digestion. Vessel dissections (2–3 cm) were opened with sterile scissors, and the opened vessel was placed in a petri dish with its endothelial side on a sterile cellulose tissue saturated with collagenase solution (Collagenase Type IV, Sigma-Aldrich, St Louis, MO, USA), 2 mg mL−1 in RPMI 1640 (Roswell Park Memorial Institute) medium. After incubation for 60 min at 37 °C, the vessel was placed in a petri dish containing RPMI, after which cells from the endothelial side of the vessel were removed using a cell scraper. After the remainder of the vessel had been discarded, the contents of the Petri dish were transferred to a 50-mL conical tube and the cell suspension centrifuged at 1000 × g for 5 min. The resulting pellet was washed with 50 mL phosphate buffered saline (PBS) and concentrated at 1 × 106 cells mL−1 in PBS for flow cytometry. For CEC analysis, samples were prepared in a lyse-stain wash procedure with minimal sample handling. Four milliliters of PB were transferred into a 50 mL tube and 45 mL of ammonium chloride 0.15 m was added to induce red-cell lysis. After 20 min of lysis at room temperature (RT), the suspension was centrifuged for 5 min at 1000 × g and the supernatant was removed from the sample tube without disturbing the pellet. To reduce Fc receptor-mediated antibody binding, 50 μL of CD32 mAb (clone IV3; Stemcell Technologies, Vancouver, BC, Canada; 10 μg mL−1 in PBS containing 1% bovine serum albumin) was added and the pellet was carefully homogenized using a 100-μL pipette. After 10 min of incubation, cells were stained using 50 μL of a mixture of the following monoclonal antibodies (mAb): fluorescein isothiocyanate (FITC) conjugated CD34 (clone 8G12; BD Biosciences, San Jose, CA, USA); R-phycoerythrin (PE) conjugated CD105 (clone 1G2; IOTest, Marseille, France); peridinin chlorophyll protein (PerCP) conjugated CD45 (clone 2D1; BD Biosciences); allophycocyanin (APC) conjugated CD146 (clone 541-10B2; Miltenyi Biotec GmbH, Bergisch Gladbach, Germany), and 50 μL of the DNA dye 5-bis[2-(di-methylamino) ethyl]amino-4, 8-dihydroxyanthracene-9,10-dione (DRAQ5; Biostatus Ltd, Shepshed, UK). All reagents were diluted in PBS/BSA based on titration (i.e. absence of non-specific staining on negative populations and optimal discriminatory power between negative and positive populations). After 15 min of incubation in darkness at room temperature (RT), 45 mL of PBS were added and the suspension was centrifuged for 5 min at 1000 × g. After removal of supernatant the cells were resuspended in 500 μL of PBS and transferred to a standard 5-mL flow cytometry tube. The 50-mL tube was rinsed with 500 μL of PBS, and any remaining cells were also transferred to the cytometry tube to assure optimal recovery of cells. Samples were immediately acquired on a FACSCanto II flow cytometer with FACSDiva v6.1. software (BD Biosciences) or stored on ice for a maximum of 1 h. Data acquisition was started by collecting ungated data of 50 000 nucleated cells (DRAQ5+) at a low flow rate (10 μL min−1) (Fig. 1, panel A–F). Data acquisition was completed by acquiring the remainder of the sample at a high flow rate (120 μL min−1) with a threshold (live-gate) on CD34+ events in a separate file (Fig. 1, panel G–I). Gating strategy for circulating endothelial cell (CEC, purple dots) flow cytometric analysis based on CD34 and CD146 expression. Panels A–F: whole blood analysis for absolute counting of CD34-positive cells (red dots). Panel (A) shows the forward and side scatter light characteristics of the leukocytes. Panel (B) describes the gating on lymphocytes (CD45+, SCClow, green dots), which are serving as an internal (negative) control throughout the analysis. Panel (C) shows gate 3 to include all (DRAQ5+) nucleated cells. Panel (D–F) describes the analysis of CD34+ progenitor cells (CD34+, CD45neg/dim, DRAQ5+). Panels G–K: flow cytometric analysis of 4 mL of blood using a CD34 threshold (panel G). CECs are analyzed within the CD34+ gate as a small subset positive for CD146 and negative for CD45 (panel H). Panels H–K: overlay histograms combining the lymphocytes (including a small subset of CD146+ T lymphocytes) from the first 50 000 events acquired without a CD34 threshold and all CD34-positive events using a threshold on the remaining cells. CECs are co-expressing the endothelial marker CD105 (panel I). Panel J represents a patient sample with an increased level of CECs, and panel K a sample of a healthy donor. Panel L shows the morphology and VWF expression of sorted CECs. Absolute cell counting Absolute counts were obtained by multiplying the CEC percentage within the CD34+ population in 4 mL of blood by the simultaneously obtained absolute CD34 counts – obtained on the same blood sample – using the single platform FCM assay according to ISHAGE guidelines [25, 26]. In brief, 200 μL of blood was incubated with CD34-FITC (clone 8G12), CD45-PErCP (clone 2D1) and DRAQ5. After 15-min incubation, 2 mL ammonium chloride lysing solution was added, followed by 100 μL Flow-Count beads. A minimum of 100 CD34+ events were acquired on a FASCCanto II flow cytometer (Fig. 1, panel A–F). Prior to each analytical run, the flow cytometer was cleaned according to the manufacturer's instructions and complete removal of residual cells and particles was checked by running a tube with PBS for 1 min. Carryover between samples was prevented by the use of the FACSCanto SIT Flush device, whilst for each sample the CD34 absolute count tube was performed first, prior to the CEC analysis tube. Cell sorting Cells were sorted using a FACSAria Cell Sorting System (BD Biosciences) equipped with FACSDiva v6.1. software (BD Biosciences). For morphological and immunohistochemical analysis, cells were sorted on glass slides as spots of 50 cells. For quantitative reverse transcriptase polymerase chain reaction (qRT-PCR) studies, 50 cells were sorted in 2-mL Eppendorf tubes, lysed by adding 250 μL of Qiagen AllPrep DNA/RNA Micro Kit Lysis Buffer (RLT+ lysis buffer) (Qiagen BV, Venlo, the Netherlands) and stored immediately at −80 °C until RNA isolation was performed with the AllPrep DNA/RNA Micro Kit (Qiagen) according to the manufacturer's guidelines. Morphology May-Grünwald-Giemsa (MGG) staining was performed using a standard eosin-methylene blue solution and Giemsa solution with an Autostainer XL (Leica Microsystems GmbH, Wetzlar, Germany). All images presented in this study were acquired on a DM2500 microscope equipped with a DC500 digital camera (both from Leica). The acquisition software (Leica IM1000 version 4.0) adjusted focus, brightness and contrast automatically. Immunohistochemistry Two-step horseradish peroxidase (HRP)-based immunohistochemistry (IHC) was performed on sorted endothelial cells and lymphocytes using the ChemMate Envision Detection Kit (Dako, Glostrup, Denmark). Sorted populations were fixed in buffered formalin and stained using mouse-anti-human antibodies against CD31 and von Willebrand factor (both from Dako). All immunohistochemical stainings were evaluated by a pathologist. qRT-PCR The generation of pre-amplified cDNA from RNA and subsequent Taqman-based qRT-PCR analysis, and the validation procedures to ensure homogeneous amplification, were performed as described previously [27]. RT samples were diluted in nuclease-free ddH2O and analyzed by real-time PCR in a 20-μL reaction volume in a Mx3000PTM Real-Time PCR System (Stratagene, Amsterdam, the Netherlands). Negative controls included samples without reverse transcriptase and samples in which RNA and cDNA had been replaced with PCR-grade H2O or genomic DNA. Quantitative values were obtained from the threshold cycle (Ct) at which the increase in TaqMan probe fluorescent signal associated with an exponential increase of PCR products reached the fixed threshold value of 0.08; in all cases, this was at least 10 times the standard deviation of the background signal. Levels of HMBS, HPRT1 and GUSB were the reference genes used to control sample loading and RNA quality with Taqman assays-on-demand from Applied BioSystems (ABI, Nieuwerkerk a/d IJssel, the Netherlands) as described previously [27]. Protein tyrosine phosphatase receptor type C (PTPRC coding for CD45, ABI: Hs00236304_m1) was the control gene for leukocyte background and CDH5 (coding for CD144, ABI: Hs00174344_m1), MCAM (coding for CD146, ABI: Hs00174838_m1), VEGFR2 (ABI: Hs00176676_m1) and VWF (ABI: Hs00169795_m1) the controls for CEC detection. The delta Ct method was used to evaluate gene transcript levels relative to the reference genes. CECs were enumerated using the CellSearch® Endothelial Cell Kit and CellTracks® AutoPrep® System (Veridex) according to the standard procedure [17]. In brief, a blood sample of 4 mL was mixed with ferrofluid particles coated with anti-CD146 antibodies. After incubation, unlabeled cells and plasma were removed by magnetic separation. Subsequently, the isolated cells were stained with fluorescent monoclonal antibodies for endothelial cells (CD105-PE) and contaminating leucocytes (CD45-APC), and a nuclear staining dye (DAPI). During this sample processing procedure, 4 mL of blood is reduced to approximately 700 μL containing enriched CECs, which is next transferred to a cartridge in a Magnest® device (Veridex). Here, the strong magnetic field moves immunomagnetically labeled cells to the surface of the cartridge in order to render them ready for analysis. All steps were performed fully automatically by the CellTracks® AutoPrep® System. The samples were read with the CellTracks Analyzer®, a semi-automated fluorescence microscope used to scan the entire surface of the cartridge at four different wavelengths in order to detect objects labeled with all fluorochromes used. Captured images that contain objects fulfilling predetermined criteria were automatically presented in a gallery format from which events were classified by a trained operator. A CD146+ cell was classified as a CEC on the basis of its morphologic features and staining patterns (i.e. DAPI positive, CD105 positive and CD45 negative). We used the free FITC marker channel to confirm that CellSearch-isolated CECs are CD34 positive. Statistical analysis was performed using Prism™ software (GraphPad Software, La Jolla, CA, USA). To analyze the degree of agreement between the CD34-based FCM assay and the CellSearch reference assay, Bland-Altman plots were generated to relate the inter-assay difference to the mean CEC count of the two methods. The Mann–Whitney U-test was used to compare healthy control with patient samples. P < 0.05 was considered to be significant. We validated our phenotypic criteria to analyze CECs by FCM on isolated endothelial cells (ECs) from vein (n = 3) and artery (n = 1) dissections. ECs were isolated and stained with the same mAb cocktail as used in the assay. As illustrated in Fig. 2 (panels A–C), isolated ECs showed a high expression of CD34 and CD146 and were negative for CD45, showing that isolated ECs fulfill the phenotypic criteria for CECs (Fig. 1) and can be selected by the Boolean gating strategy (DRAQ5+, CD34+, CD45−,CD146+) used in the FCM assay. Furthermore, we demonstrated that isolated ECs are positive for several other endothelial markers such as the pan-endothelial markers CD31 and CD144, and the subtype-associated markers CD105 and CD309 (VEGFR-2), whilst being negative for the (endothelial) progenitor marker CD133 (Fig. 2, panels D–I). Flow cytometric analysis of endothelial cells isolated from a dissected vein. Firstly, nucleated cells were selected based on DNA content using DRAQ5 expression (panel A). Secondly, endothelial cells were identified within the CD34-positive population (panel B, purple dots) expressing the pan-endothelial markers CD146, CD144 and CD31 (panel C, D, H and I, respectively), and subtype-associated markers CD105, CD309 (VEGFR-2) and CD133 (panels E–G). Residual lymphocytes (CD45+, green dots) are serving as an internal (negative) control. First, CECs were sorted on glass slides and stained for MGG and, by IHC, the endothelial specific markers CD31 and VWF. More than 95% of all sorted events showed specific EC morphology (nucleated cuboidal or elongated cells) and were brightly positive for VWF (Fig. 1, panel L) and CD31 (not shown). In parallel, lymphocytes and human hematopoietic stem cells (HSCs) were sorted to serve as control cells. Lymphocytes were negative for VWF and partially weakly positive for CD31. HSCs were negative for VWF and partly positive for CD31 (data not shown). Second, sorted CECs were analyzed for mRNA expression by qRT-PCR of both endothelial-specific genes and leukocyte-specific genes in duplicate assays. This RNA profiling by qRT-PCR showed the presence of endothelial specific markers (CDH5, MCAM, VWF) and absence of a hematopoietic marker (PTPRC coding for CD45) in our CEC-sorted preparations (Table 1). This result indicates an endothelial-specific gene expression profile for the cells meeting the phenotypic criteria of the FCM assay. To measure the variability of CEC counts obtained from two 4-mL aliquots derived from a single blood draw, CECs were enumerated in duplicate using the blood in a single draw from 13 different samples. Regression analysis showed a best-fit line with a slope of 1.006 (95% confidence interval = 0.8702–1.141), a Y-intercept of 1.511 (95% confidence interval = −7.854 to 6.206) and an R2 of 0.96 (Fig. 3). The average coefficient of variation for duplicate samples was 14.4% (range, 0–35%). Assay reproducibility for the CD34-based FCM assay. Correlation between duplicate samples (n = 13) assessed in two independently performed experiments. Regression analysis showed a best fit line with R2 of 0.96, a slope of 1.01 (95% confidence interval [CI] = 0.87–1.14) and a Y-intercept of 1.51 (95% CI = −7.85 to 6.21). We studied a total of 28 subjects (10 healthy controls and 18 patients with different metastatic carcinomas) for comparison of the two techniques. Figure 4 shows the correlation between the number of events (red dots) and the obtained absolute number (blue dots) by the FCM method compared with the CellSearch reference assay. Regression analysis for the event count analysis (red dots) showed a best fit line with a slope of 0.66 (95% confidence interval = 0.61–0.71), an intercept of 4.16 (95% confidence interval = −1.95 to 10.26) and an R2 of 0.96. Regression analysis for the CD34-based absolute count analysis (blue dots) showed a best fit line with a slope of 2.60 (95% confidence interval = 1.92–3.27), an intercept of 72.21 (95% confidence interval = −7.88 to 152.3) and an R2 of 0.71. Figure 5 shows the corresponding Bland–Altman plots. The median number of CECs in the FCM event count analysis was significantly lower compared with the CellSearch reference assay (median 20 vs. 26, P = 0.0065), but significantly higher when expressing CECs as absolute numbers by multiplying the CEC percentage within the CD34+ population in 4 mL of blood by the absolute CD34 counts obtained on the same blood sample (median 129 vs. 26, P = 0.0002). This suggests a slightly lower recovery for event counts as a result of limited cell loss during sample preparation and FCM dead volume but a significantly higher recovery when corrected for cell loss using the CD34+ cell-based FCM absolute number method. Correlation of CEC values comparing the CellSearch assay with the CD34-based FCM event counts (red dots; R2 of 0.96 with a slope of 0.66 [95% CI = 0.61–0.71] and an intercept of 4.16 [95% CI = −1.95 to 10.26]) as well as the obtained FCM absolute number (blue dots; R2 of 0.71 with a slope of 2.60 [95% CI = 1.92–3.27] and an intercept of 72.21 [95% CI = −7.88 to 152.3]). Bland–Altman plots comparing the CellSearch assay with the CD34-based FCM event counts and FCM absolute count assay for the enumeration of CECs. The number of CECs using CD34-based event count analysis was significantly lower compared with the CellSearch reference assay (median; 20 vs. 26, P = 0.0065), suggesting a slightly lower recovery for event counts as a result of limited cell loss during sample preparation and FCM dead volume but a significantly higher recovery when corrected for cell loss using the CD34+ cell-based FCM absolute number method (median; 129 vs. 26, P < 0.0001). Although the investigation of CEC levels in different types of cancer other than for validation purposes is beyond the scope of this paper, Fig. 6 shows that CEC levels in healthy individuals ranged from 4 to 79 cells per 4 mL (n = 30, median 18) and were significantly higher in patients with advanced solid malignancies (ranging from 5 to 354 [n = 51, median 35, P < 0.005]) and in patients with hematological malignancies in remission (range, 4–653 [n = 66, median 37, P < 0.0005]). Prevalence of CEC using the CD34-based FCM assay in 4 mL of blood from 30 healthy individuals (HD) and significant increase in 51 patients with advanced solid malignancies and 60 patients with a hematological malignancy in remission (prior to SCT). Solid lines indicate the median value for each group. The aims of this study were to (i) develop a quick, reliable, robust and validated FCM method based on CD34 absolute counts without immunological pre-enrichment; (ii) confirm the phenotypic criteria for analyzing CECs by FCM on isolated endothelial cells from vein wall dissections; (iii) confirm the endothelial origin of all cells designated as CECs by morphology, immunohistochemistry and gene expression; and (iv) test the agreement between the novel FCM method with the validated CellSearch® assay. We here describe a novel two-step approach using a standardized absolute CD34 count in 200 μL whole blood [28] followed by analyzing all CD34-positive cells in a total of 4 mL blood to reliably count the CECs within this population without immunological pre-enrichment and with minimal sample handling. After a mild erythrocyte lysing step, a single wash step and a detection threshold on CD34-positive events to remove most cells not of interest from the analysis, a complete sample could be run within 10 min on a flow cytometer without abort counts or impractically large data-files. In contrast to the enrichment procedures reported [16, 17, 21], which inevitably lead to cell loss and underestimation of the actual CEC number, one of the advantages of this approach is that only a small degree of random cell loss will occur during the lyse/stain/wash procedure. Additionally, the absolute number of the CEC population can be precisely calculated on the basis of the total absolute CD34 count. Another advantage compared with conventional FCM assays is that we made it possible to overcome the reported obstacle of variability [29, 30] in (extremely) rare events by processing a larger sample volume (i.e. 4 mL instead of 50–200 μL). A volume of more than 1 mL will be needed for a precise enumeration of CECs at a frequency of < 1 cell μL−1 blood to obtain a population of at least 100 cells fulfilling the flow cytometric criteria with a coefficient of variation of 10% [26]. Flow cytometric assays are at risk of overestimating rare events such as CECs by enumerating false-positive events caused by non-specific antibody binding. Non-specific binding can be a result of Fc receptor binding, binding to dead cells, or improper use of reagents. In a lyse-stain-wash procedure, the blocking of Fc receptor binding and the use of a 'dump' channel is crucial because these binding sites are no longer saturated by the free plasma immunoglobulins [31]. Next to the interference by non-specific antibody binding, false positivity can also result from specific binding to soluble forms of the antigen. Many endothelial surface antigens are secreted, and their uptake by platelet aggregates can result in false-positive signals. An effective strategy for excluding platelets, aggregates and endothelial microparticles from analysis is by using a DNA-specific stain, such as DAPI or DRAQ5. As DRAQ5 is a membrane-permeant dye, no additional fixation and permeabilization is needed. Dyes also staining RNA such as PI or LDS751 will fail to exclude platelets by the relatively high expression of RNA by these cells [19]. The difficulties in CEC enumeration is perhaps best illustrated by the enormous range of CEC numbers in healthy controls (4–1300 cells mL−1) in previous studies. This is most certainly a result of using different phenotypic definition of CECs. A phenotypic overlap between CECs and other cell populations may lead to an overestimation due to endothelial microparticles and platelets [19, 32], whilst the use of markers that are not constantly expressed by all CECs can led to underestimation. To overcome this problem we confirmed the phenotypic criteria, based on literature and previous experiments, in our assay on endothelial cells isolated from vein-wall dissections. We confirmed the bright expression of CD34 [33-36] and CD146 [24] on these cells as well as the expression of the endothelial-associated markers CD31 and CD105 and the endothelial-specific marker CD144 [30]. On the basis of these findings, the combination of (i) the brightly expressed pan-endothelial markers CD34 and CD146, (ii) the negative selection marker CD45 and (iii) the DNA-specific stain DRAQ5 to exclude microparticles and platelets, was chosen for the immunophenotypic definition that covers the vast majority of mature CECs in our assay [30, 37]. We could also demonstrate that more than 98% of CECs isolated with CD146 capture antibodies in multiple CellSearch assays were CD34 positive using the free FITC channel (data not shown). Clinical specimens were also used to test the agreement with the validated and automated CellSearch assay, showing a high level of agreement. The number of CECs using CD34-based event count analysis was significantly lower compared with the CellSearch reference assay. This difference may have been caused by a slightly lower recovery for event counts as a result of limited cell loss during sample preparation and a residual or 'dead' volume during flow cytometer acquisition. However, when corrected for cell loss using the approach based on CD34+ cell-based FCM absolute counts, a significantly higher recovery was found compared with the CellSearch assay. The latter discrepancy is most likely due to the relatively low recovery for this assay of ≤ 70% as measured by spiking HUVEC cells into peripheral blood [17]. Our findings of significantly higher levels in patients with metastatic carcinomas confirm previous publications [16, 17, 38]. We also found significantly elevated CEC levels in patients with hematological malignancies in remission prior to SCT and a further increase after myeloablative conditioning regimens prior to hematopoietic stem cell transplantation (data not shown). This suggests that these increased CEC counts are merely the result of therapy-induced damage to the vasculature rather than tumor growth or bone marrow reconstitution. An advantage of our flow cytometric approach is the possibility to study additional antigens within this assay format. Assessment of viability and activation markers of CECs may provide further insight into their pathophysiology, whilst expression of tumor-associated antigens on CECs may provide clues to their putative tumor origin [39, 40]. To have clinical utility, the method in use must have a low variability and a validation of the true endothelial origin of cells designated as CECs by the assay. Variation in CEC counts taken from the same tube was minimal and more than 95% of designated events could be confirmed as endothelial-specific by morphology, immunohistochemistry and gene expression profile. These findings are important as they may clarify the ongoing debate over estimation of CEC values in other (conventional) flow cytometry assays lacking this kind of extensive validation. Clinical specimens were also used to test the agreement with the validated and automated CellSearch assay, showing a high level of agreement. CEC values for healthy subjects for both methods are similar and also in line with other recently reported immunomagnetic bead enrichment assays [16, 21]. In conclusion, our study demonstrates that the new CD34-based FCM assay is able to enumerate a well-defined CEC population in peripheral blood. We envisage that this flow cytometric method can be used as a relatively cheap, high-throughput assay to accurately enumerate and further explore the characteristics of circulating endothelial cells. This will be particularly attractive for large-scale clinical studies to establish the clinical value of CECs in disease with endothelial damage or to monitor angiogenesis-inhibiting therapies. We are grateful to P. van der Spoel and J. Bolt-de Vries for excellent technical assistance. The authors state that they have no conflict of interest.