CTC Technologies and Tools

Cancer cells can enter the blood stream. Some of these circulating tumor cells (CTC) extravasate and form distant metastases which ultimately lead to the death of the patient. CTC have been observed quite some time ago 1-3. The major difficulty for the identification of CTC is their extremely low frequency (~1/ml in metastatic patients). When no CTC were detected in a tube of blood of a patient with metastases one always wonders whether they were not present or whether they were missed by the detection technology. Another hurdle is the extensive heterogeneity of the physical and immunological appearance of CTC within and between patients. Tumor cell lines play a pivotal role in both informing technology developers of the probable physical and immunological properties of CTC, as well as in the determination of the analytical accuracy, reproducibility, and linearity of any developed technology. However, tumor cell lines are not CTC. Excellent performance on tumor cell lines does not warrant that 1 the developed technology will actually identify CTC in blood samples of cancer patients nor that 2 these CTC relate to clinical outcome nor that 3 information relevant for treatment can be extracted. At the start of the trajectory toward the FDA cleared CellSearch system very little was known about the CTC frequency and their physical and immunological appearance. CTC enrichment in 10 ml of blood was performed by manual immunomagnetic enrichment targeting EpCAM using the GA73.3 antibody. The enriched cell suspension was fluorescently labeled with the PE-labeled CAM5.2 antibody recognizing cytokeratins 7 and 8, a PerCP-labeled antibody recognizing CD45 and a nucleic acid dye that could be measured in the FITC channel of a flowcytometer. Using this approach, it was first shown that CTC indeed could be detected in 10 ml blood of cancer patients with metastatic disease and at a lesser frequency in patients with no detectable metastasis while few “CTC” were detected in healthy controls 4-8. These results were sufficiently encouraging to further develop the CTC technology and design and conduct clinical studies to explore the clinical utility of CTC. The commercial CellSearch system that was ultimately developed included (1) a blood collection tube to preserve the fragile CTC for a period of up to 96 h, (2) an automated sample preparation system in which (3) the EpCAM antibody was replaced by VU1D9, (4) CAM5.2 cytokeratin antibody was replaced by C11 and A.53B/A2 to increase the range of cytokeratins, (5) the CD45-PerCP was switched to CD45-APC, (6) the nucleic acid dye was switched to DAPI, and (7) the flow cytometer replaced by a semi-automated fluorescence microscope with dedicated software to present CTC candidates to a reviewer. Thanks to an excellent team of reagent developers and engineers the system was reliable and provided accurate results over a large range of spiked tumor cells in blood with a high sensitivity and specificity. Whereas few “CTC” were detected in healthy donors and patients with benign disease a large range of CTC were detected in cancer patients 9. The real excitement arose when the results of the prospective studies became available which showed that CTC were superior to serum tumor markers and/or radiographic imaging in predicting efficacy of ongoing therapy 10-14. By now the original results have been confirmed and validated in numerous studies and expanded to different cancers and earlier disease stages 15-20. Since the introduction of CellSearch a large number of new technologies have been introduced to improve CTC detection and extract information relevant for treatment from CTC. However, the definitions used for CTC, as well as the level of evidence that the detected CTC are indeed of importance varies greatly, complicating technology assessment. The aim of this special Cytometry issue is to obtain an overview of CTC technologies in which the authors provide a succinct description of the working principle of their devices and provide information that helps the community assess the status and potential of the technologies. One of the challenges the “CTC field” is facing is the extensive heterogeneity of CTC leading to extreme large ranges of reported numbers. To overcome this challenge as part of the European program CANCER-ID (www.cancer-id.eu) open source image analysis tools (ACCEPT) are being developed and introduced to aid in the evaluation of new CTC technologies and arrive at a common CTC definition 21-23. In this issue, we compared CTC scoring between operators and show that a computer-generated definition is most likely preferable to trying to have operators agree on a definition 24. The observation that all circulating EpCAM + CK + CD45-objects predict overall survival in castration-resistant prostate cancer 25 may make the “morphological defintion” less critical. Still for reporting “CTC numbers” in relation to the monitoring of therapy the assignment to a certain class becomes very important 26. To illustrate the challenge of defining what is and what is not a CTC the ACCEPT analysis of two blood samples from metastatic prostate cancer patients processed with the CellSearch system is illustrated in Figure 1. The standard CTC count of the sample in Panel A was 1,477 and for the sample in Panel B it was 7. A 50% misclassification of CTC will not affect the assignment of Patient A as one with a relatively poor prognosis, but for Patient B it surely will as the “cut-off” defined in the FDA cleared system is 5 CTC. Eight objects with both CK and DAPI staining can be observed in Patient B two of these objects will most likely be classified as CTC by the majority of reviewers but the other six objects can be subject to debate. In both patients depicted in Figure 1 many CK+ objects are small and may not have any nuclear staining. We baptized them tumor derived Extra Cellular Vesicles (tdEVs). These tdEVs can easily be quantified with ACCEPT and their presence is indeed strongly associated with poor clinical outcome 27. Cancer cells can be considered the smallest functional unit and thus will be preferable for in depth characterization. Nevertheless, valuable information can be extracted from tumor derived free plasma DNA 28, circulating miRNAs 29, 30, tdEV, and tumor educated platelets 31. Disadvantage of the traditional plasma proteins used for cancer screening and monitoring such as PSA, CA15.3, and CA-125 is that their elevation is not necessarily associated with cancer neither their increases and decreases with treatment response. To interpret results obtained with the CellSearch system one has to keep in mind that whole blood is centrifuged at 800g for 10 min and only the cell fraction is processed on the CellSearch system. This implies that only the relatively large tdEVs will end up in the CellSearch system and a large portion of smaller tdEVs will be missed. By centrifugation of a blood sample the majority of the cells will settle at 300g, dead cells and apoptotic bodies (50–50.000 nm) at 2,000g, cell debris & microvesicles (100–1,000 nm) at 10,000g and exosomes (50–100 nm) at 100,000g. Note that the smallest apoptotic bodies require ultracentrifugation speeds to sediment. Whereas all CTC will be present in the bottom fraction after centrifugation at 800g and the tumor derived free plasma DNA in the top fraction, tdEVs, circulating miRNAs, and tumor educated platelets will be present in both fractions and some will be lost in the interface. The proportion of circulating miRNAs and tumor derived free plasma DNA which is contained within EVs is still not completely understood nor is the size distribution of the EVs containing this miRNA and DNA. The influence of centrifugation on the purity that can be obtained for different biomarkers is illustrated by the model described in this issue 32. Perhaps the largest challenge for the CTC field is their extremely low frequency. Using the CellSearch system one cannot detect at least one CTC in 7.5 ml of blood in ~20% of metastatic prostate cancer, ~25% of metastatic breast, and ~50% of metastatic colon cancer patients 33. Extrapolation of the frequency distribution of CTC in patients with metastatic prostate cancer, breast, and colorectal cancer showed that an increase of blood volume from 7.5 ml to 600 ml >90% of patients with metastatic cancers will have CTC detected 33. The use of apheresis was introduced to enable processing of the mononuclear cell fraction of large blood volumes 34. This procedure baptized Diagnostic Leukapheresis indeed allows for the isolation of a substantial larger number of CTC and enables a true “liquid biopsy.” However, at present the available technologies are not yet capable of isolating CTC from the complete DLA volume 34-36. In this issue, clinical results and recommendations for standardized reporting of DLA results are presented 37. To provide direction to the CTC and tdEV field we show in figure 2 the frequency, size and density of the different cells detected in blood. This includes red blood cells, platelets, granulocytes, lymphocytes, monocytes, hematopoietic progenitor cells (CD34+), circulating endothelial cells (CEC), CTC and tdEV. The latter is divided into small (sm tdEV) and large (lg tdEV) Extracellular Vesicles. The lg tdEV are > 1μm in size and obtained after centrifugation at 800g and EpCAM immunomagnetic selection the sm tdEV are less than 1 μm in size and can be obtained by EpCAM immunomagnetic selection but data obtained is insufficient to provide their actual frequency. Hematopoietic EVs are the most probable source of background in tdEV detection, however, their properties are insufficiently characterized to include in this figure. With the introduction of new methods for CTC isolation many have argued that the lack of detection of CTC in cancer patients with the CellSearch system may be contributed to the CTC definition used as it detects only cells expressing both EpCAM and Cytokeratin. A description of several of these EpCAM independent isolation methods can be found in this issue 39-45. However, for these different CTC phenotypes, clinical confirmation of the correlation of CTC to patient survival will be necessary. Observation in NSCLC, breast and prostate cancer showed that frequently both EpCAM+ and EpCAM− CTC populations are present and rarely only one of the two are detected. Moreover, in patients with metastatic NSCLC and prostate cancer we found that the EpCAM+, CK+ CTC, and not the EpCAM−, CK+ are correlated with poor outcome 46-48. Hopefully research will be conducted with the newly available technology that can provide in depth information on the frequency of different subpopulations of CTC, other (rare) cell types and their relation with the disease. The most exciting promise of CTC is the potential to extract information that enables the selection of the most effective treatment and the ability to identify resistance to therapy early and switch to an alternative treatment. Addition of fluorescently labeled antibodies recognizing treatment targets to the staining cocktail is relatively easy, although, one is limited by the number of fluorochromes that can be determined simultaneously. Reports on the expression of treatment targets on CTC showed extensive intra and inter patient heterogeneity of expression 49-51. Probing for gene aberrations of CTC by fluorescence in situ hybridization has also been demonstrated and again reiterated their extensive intra- and inter-patient heterogeneity 52-55. For extensive interrogation of the CTC the individual cells will have to be isolated and for molecular characterization the nucleic acids will have to be amplified. Tools to isolate the single cells from CTC enriched cell suspensions are available 39, 41, 42, 56, 57. The steps needed to get from blood to single CTC surely will be associated with cell losses and the different steps will have to be optimized to minimize these losses. Still when CTC losses are kept to a minimum one has to deal with the state the CTC are in. As many CTC are undergoing apoptosis, the nucleic acids may not be of a sufficient quality for thorough interrogation. Therefore, it would be great if tools were available to provide a likelihood that the nucleic acids are of sufficient quality after isolation of the single CTC 58, 59. This may save effort and cost involved with sequencing, fluorescence in situ hybridization, etc. The processing of larger blood volumes combined with a technology for single CTC sorting and subsequent phenotyping or genotyping will result in a sufficient number of CTC from individual patients for further characterization. This will improve our understanding of the heterogeneity of CTC in individual patients, and will certainly improve our understanding of the role of CTC in the progression of cancer.