The clinical utility of circulating tumour cell enumeration as a prognostic indicator has been demonstrated in numerous carcinomas. However, poor quality images and limited characterisation available on the main platform used for CTC enumeration hamper the exploitation of CTCs in other areas of clinical oncology. Additionally, the reliance on the expression of EpCAM for the detection of CTCs limits detection of cells that have undergone epithelial/mesenchymal transition, and may therefore not directly detect metastatic precursor cells. In contrast, flow cytometric methods of CTC enumeration which allow multiple markers to be used simultaneously, are more sensitive than microscopy methods, but carry the caveat that cellular debris may not be resolvable from cells with a high degree of confidence. Our aim was to establish the sensitivity of detection of CTCs by imaging flow cytometry.


ML-1 follicular thyroid carcinoma cells were spiked into 4 ml whole blood from healthy volunteers collected into EDTA vacutainers to give concentrations of 500, 50 or 5 cells per ml. The samples were subjected to red blood cell lysis and fixation. CD45 positive leukocytes were then depleted by immuno-magnetism using a CD45 antibody. The remaining cells were permeabilised and stained with flurochrome conjugated primary antibodies to thyroglobulin, sodium iodide symporter, EpCam, cytokeratin, vimentin and CD45 as well as DAPI as a nuclear stain. The stained cells were processed using an imaging flow cytometer.


Preliminary experiments to validate the specificity of the antibodies indicated the potential importance of heterogeneity as revealed by multi parameter staining. Only 8 ± 3.6% of ML-1 cells with a mesenchymal phenotype (EpCAM -ve/Vimentin+ve) were in G2 compared with 32 ± 8.7% cells from the same culture with a more epithelial (EpCAM +ve/Vimentin-ve) phenotype (mean and SD of three independent experiments, p = 0.012 T-test).

Recovery experiments following red cell lysis and CD45 depletion of spiked cell samples gave >90% depletion of leukocytes and 48 ± 5%, 56 ± 15% and 78 ± 32% recovery at spiking densities of 500, 50 and 5 cells per ml respectively (mean and SD of three independent experiments).


The method described allows the capture of multi-parameter high quality fluorescent images of extremely rare cell populations in a whole blood matrix and we are currently applying it to clinical samples derived from cancer patients and healthy volunteers. The method is also readily transferable to the detection of CTCs from other tumours by the substitution of tumour/tissue specific antibodies and optical space remains for the introduction of two additional probes. We will exploit this by incorporating probes for pharmacodynamic biomarkers specific to compounds in early phase clinical trials, thus allowing simultaneous PD assays in CTCs as a liquid biopsy and residual haematopoietic cells as a surrogate tissue.

Citation Format: Barry Dent, Rachel L. O'Donnell, Laura F. Ogle, Emma D. Rourke, Hamsavardhini P. Ramesh, Maddie Moat, Nick Hayes, Ujjal K. Mallick, Felicity EB May, Helen L. Reeves, Nicola J. Curtain, Richard J. Edmondson, Alan V. Boddy, Ruth Plummer, David Jamieson. Detection and characterization of circulating tumor cells by imaging flow cytometry. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 3059. doi:10.1158/1538-7445.AM2014-3059