Radiology and Oncology | Ljubljana | Slovenia | www.radioloncol.com Radiol Oncol 2021; 55(3): 292-304. doi: 10.2478/raon-2021-0033 292 research article Morphological features of breast cancer circulating tumor cells in blood after physical and biological type of isolation Tanja Jesenko1,2, Ziva Modic1,2, Cvetka Grasic Kuhar2,3, Maja Cemazar1,4, Urska Matkovic1, Simona Miceska2,5, Jerneja Varl2,5, Anamarija Kuhar5 and Veronika Kloboves-Prevodnik2,5 1 Department of Experimental Oncology, Institute of Oncology Ljubljana, Ljubljana, Slovenia 2 Faculty of Medicine, University of Ljubljana, Ljubljana, Slovenia 3 Department of Medical Oncology, Institute of Oncology Ljubljana, Ljubljana, Slovenia 4 Faculty of Health Sciences, University of Primorska, Izola, Slovenia. 5 Department of Cytopathology, Institute of Oncology Ljubljana, Ljubljana, Slovenia Radiol Oncol 2021; 55(3): 292-304. Received 5 July 2021 Accepted 9 July 2021 Correspondence to: Assoc. Prof. Veronika Kloboves-Prevodnik, M.D., Ph.D., Department of Cytopathology, Institute of Oncology Ljubljana, Slovenia, email:vkloboves@onko-i.si, and Maja Čemažar, Department of Experimental Oncology, Institute of Oncology Ljubljana, Slovenia. E-mail: mcemazar@onko-i.si Disclosure: No potential conflicts of interest were disclosed. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). Background. Circulating tumor cells (CTCs) have become an important biomarker in breast cancer. Different iso- lation tech-niques based on their biological or physical features were established. Currently, the most widely used methods for visualization after their separation are based on immunofluorescent staining, which does not provide the information on the morphology. Materials and methods. The aim of this study was to evaluate how two different separation techniques affect cell morphology and to analyse cell morphology with techniques used in routine cytopathological laboratory. A direct side-by-side comparison of physical (Parsortix®) and biological (MACS®) separation technique was performed. Results. In the preclinical setting, both isolation techniques retained the viability and antigenic characteristics of MCF7 breast cancer cells. Some signs of degeneration such as cell swelling, cytoplasmic blebs, villous projections and vacuolization were observed. In metastatic breast cancer patient cohort, morphological features of isolated CTCs were dependent on the separation technique. After physical separation, CTCs with preserved cell morphology were detected. After biological separation the majority of the isolated CTCs were so degenerated that their identity was difficult to confirm. Conclusions. Taken together, physical separation is a suitable technique for detection of CTCs with preserved cell morphology for the use in a routine cytopathological laboratory. Key words: CTC; breast cancer; morphology; Parsortix®, MACS®; Giemsa Introduction Circulating tumor cells (CTCs) have become an important biomarker in breast cancer as they can provide critical information about disease progres- sion and response to therapy.1,2 They represent an intermediate part of the metastatic cascade, therefore monitoring CTC levels in the blood has exceptional implications for the management of cancer patients.3 In the blood stream, CTCs are a heterogeneous cell population of tumor cells with different phenotypes. They show high level of epi- thelial-mesenchymal plasticity and can express an epithelial phenotype, mesenchymal phenotype or even a hybrid or partial epithelial/mesenchymal (E/M) phenotype in which cells express proteins of both phenotypes.4,5 Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs 293 Due to their heterogeneity, different isolation techniques from whole blood samples were es- tablished in order to separate them from the sur- rounding hematopoietic cells. Separation tech- niques can be based on CTC biological or physical properties. Biological separation techniques rely on the expression of cell markers and are usually based on epithelial cell markers positive selection or common leukocyte antigen CD45 negative selec- tion. Currently the only FDA approved platform for enumeration of CTCs CELLSEARCH (Menarini Silicon Biosystems) is based on biological proper- ties, which enables the detection of CTCs of epi- thelial origin (CD45-, EpCAM+, and cytokeratins 8, 18+, and/or 19+) in whole blood by fluorescent staining and imaging of cells that are pulled to a single focal depth by a magnetic force.6 An alterna- tive technique that also relies on EpCAM positive selection and magnetic force is Magnetic Activated Cell Sorting (MACS®, Miltenyi Biotec).7 This tech- nique was evaluated in our previous study for the isolation of CTCs in early breast cancer patients.8 It was demonstrated to be a simple and useful meth- od for enrichment of EpCAM expressing cells in a preclinical study, however the isolated CTCs from early breast cancer patients were not morphologi- cally preserved enough for their visualization by methods used in a routine cytopathology labora- tory. The limitation of separation techniques based on biological properties is that only CTCs that ex- press the epithelial cell markers can be detected. Due to this limitation, platforms exploiting physi- cal properties such as cell size, density, electric charge and deformability are gaining more atten- tion. With the Parsortix® system (ANGLE), CTCs are caught in the Parsortix® filtration cassette due to their larger size and lower compressibility than other blood components.9 The isolated cells are vi- able, intact and can be used for further in vitro ex- perimentation and characterization.9 Currently, most methods for visualization of CTCs after separation from whole blood are usu- ally based on staining with fluorescent antibodies and dyes. Therefore, these methods lack the infor- mation on the CTC morphology that is crucial for their identification by light microscopy which is still the gold standard in cytopathological and his- tological examination of tumor cells. The identifi- cation of CTCs by their morphology is challenging as these cells are often severely degenerated due to the combination of physical stress (shear forces), immune surveillance and the lack of growth factors in the blood stream.3,10 Furthermore, separation methods could also induce some additional dam- age that can result in degeneration and cell death which change morphological features of CTCs and influence their proper identification.11 Probably for these reasons, data regarding basic CTC mor- phology in breast cancer are limited and thus the selection of the appropriate separation method is of outmost importance for proper cytopathological identification.11–14 There are few published data how separation techniques affect CTCs morphology.13–16 Therefore, we designed a study aimed to select the separa- tion method that would allow the identification of CTCs based on their morphology that could be in- tegrated in a routine cytopathology laboratory. The advantage of cytopathological analysis is that it is easily accessible in the clinical environment as cy- topathologists are an integral part of cancer patient care. Two separation techniques were first evalu- ated in the preclinical setting by spiking of blood of healthy volunteers with MCF7 breast cancer cell line, which is the most studied human breast can- cer cell line.17 The preclinical evaluation was fol- lowed by a prospective clinical trial in metastatic breast cancer patients with the primary objective being a side-by-side comparison of both separation techniques. The aims of this study were i) to evalu- ate how two different separation techniques affect cell morphology and ii) to analyze cell morphology with techniques used in routine cytopathological laboratory. Materials and methods Ethic statement The study was conducted at the Institute of Oncology Ljubljana, Slovenia and was reviewed and approved by the Institutional Ethical Review Committee (ref. nb. ERID-KSOPKR-0071/2020) and National Medical Ethics Committee at the Slovenian Ministry of Health (ref. nb. 0120-150- 2019/4). All enrolled patients and healthy volun- teers signed an informed consent. The study was conducted in accordance with the Declaration of Helsinki. Blood samples were collected during the patients’ routine blood draw, which is a minimally invasive procedure. Patients and healthy volunteers Thirty-three patients were recruited in the study, however only 30 patients donated their blood sam- ples, therefore their identification numbers are in the range P1 to P33. Three healthy volunteers were Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs294 also included in the study. Each patient or healthy volunteer donated two blood samples in a 10 ml EDTA collection tube (BD, Franklin Lakes, NJ, USA). The blood samples were processed within 1 h after collection to ensure the highest viability of the CTCs. One tube was used for Parsortix® and the other for MACS® separation. After separation, cytological slides were prepared as described be- low. Spiking of MCF7 cell line Human epithelial breast cancer cell line MCF7 was obtained from ATCC (ATCC® HTB-22, ATCC, Manassas, VA, USA) and was cultured in Advanced MEM medium (Gibco, Thermo Fisher Scientific, Waltham, MA, USA) supplemented with 5% fetal bovine serum (Gibco), GlutaMAX (100x, Gibco) and Penicillin-Streptomycin solu- tion (Sigma Aldrich, Merck, Darmstadt, Germany). The cells were cultured in a 5% CO2 humidified incubator at 37°C until they reached 80% conflu- ence. Afterwards, the medium was removed, the cells were washed with phosphate buffered saline (PBS) and detached from the surface with 0.25% trypsin/EDTA in Hank’s buffer (Gibco). After col- lection, the cells were counted and 5x105 cells were spiked into each blood collection tube and sepa- rated by Parsortix® or MACS®. For the purpose of retaining control non-separated cells, the same amount of MCF7 cells was seeded into a 24-well ultra-low attachment plate and incubated in a 5% CO2 humidified incubator at 37°C until the end of Parsortix® and MACS® separation. Separation techniques Parsortix® separation was conducted by care- fully following the manufacturer’s instructions. Parsortix® separation cassette contains a stepped structure, gradually narrowing in diameter un- til reaching a final gap of 6.5 µm, therefore all of the cells that are larger than 6.5 µm are retained and isolated, while all smaller cells continue to flow through the cassette into a waste container (Figure 1A).18 After the separation, retained cells were harvested into a 5 ml plain red-top BD va- cutainer tube without a pre-harvest flush and re- FIGURE 1. Schematic presentation of both separation techniques. (A) Parsortix® separation cassette contain a stepped structure, gradually narrowing in diameter until reaching a final gap of 6.5 µm, therefore all of the cells that are larger than 6.5 µm are retained and isolated while all smaller cells continue to flow through the cassette into a waste container. The technique enables the isolation of CTCs with different phenotypes. After the whole sample has been processed, the liquid flow is reversed and CTCs can be harvested. (B) MACS® separation column uses magnetic beads covered with anti-epithelial cell adhesion molecule (EpCAM) antibodies for positive selection of CTCs with epithelial origin. When magnetic bead labeled CTCs are passed through a dense magnetic column, they are retained by a strong magnetic field. Other non-labeled cells are passed through the column into the waste tube. After the whole sample has been passed through the column, the column is removed from the magnet and retained CTCs can be eluted. A B Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs 295 suspended in an in-house cell medium: 20% fetal bovine serum (Gibco), 5% EDTA (Sigma Aldrich) in PBS. MACS® separation was performed as pre- viously described (Figure 1B) and harvested in MACS® Whole Blood Column Elution Buffer 8. MACS® separation is not limited by the cell size. Preparation of cytological slides For preclinical part of the study, three cytospins were prepared using a cytocentrifuge (Thermo Scientific Shandon Cytospin R 4 Cytocentrifuge, Waltham, MA, USA) by centrifugation at 700 rpm for 4 min at room temperature (RT). The first cyto- spin was stained with Giemsa (Sigma Aldrich, Merck) for microscopic evaluation, the second one was fixed in Delaunay (2500mL acetone, 2500 mL absolute ethanol and 2,5 mL 1 M trichloroacetic acid) and stained with Papanicolaou (PAP), which was subsequently stained for pancytokeratin AE1/ AE3. The third one was fixed in methanol for es- trogen receptor immunocytochemical staining. For clinical part of the study two cytospins were made, one for Giemsa staining for morphological evalu- ation and the second for PAP staining and subse- quent dual AE1/AE3 and vimentin immunocyto- chemical staining. Staining For Giemsa staining, the slide was air dried for at least 30 minutes at room temperature and stained with Giemsa. The second cytospin was fixed (Delaunay, 2 hours or overnight) and stained PAP using an automated stainer Leica Multistainer ST5020 (Leica Microsystems, Buffalo Grove, IL, United States). Immunocytochemical staining of CK AE1/AE3 and estrogen receptor (ER) was performed with our routine immunocytochemical staining proto- cols. All protocols included manually performed endogenous peroxidase activity inhibition with 5% H2O2-methanol solution for 10 min at RT. After the incubation was done, slides were washed once 1x PBS and once with Reaction Buffer (Ventana, Roche Diagnostics). The presence of the anti- gens was detected with iView detection kit on Ultra autostainer (Ventana, Roche Diagnostics). Enzymatic detection was accomplished when a streptavidin enzyme (streptavidin-HRP) conju- gates with the biotin-bound secondary antibody. Chromogen was deposited by a reaction with hy- drogen peroxide in the presence of diaminobenzi- dine (DAB) and copper sulfate, producing brown precipitate. External positive controls were used in all batches. Immunocytochemical staining of the CK AE1/ AE3 and vimentin was performed as dual staining on the same PAP stained cytospin. The CK AE1/ AE3 staining was performed with anti CK mono- clonal antibody (CK AE1/AE3, ref. nr. M3515, dilu- tion 1:100, Agilent, Santa Clara, CA, United States. The vimentin immune staining was performed af- ter the CK AE1/AE3 staining was done, with an- ti-vimentin antibody (V9, ref. nr. M0725, dilution 1:500, Agilent, Santa Clara, CA, United States). The presence of vimentin was detected with ultraView Universal Alkaline Phosphatase Red Detection Kit (Ventana, Roche Diagnostics), enzymatic detection was accomplished with an alkaline phosphatase and chromogen Fast Red, producing red precipi- tate. For ER staining, methanol fixed cytospin was stained with anti ER monoclonal antibody (ER, ref. nr. NCL-L-ER-6F11, dilution 1:25, Novocastra, Leica Biosystems, United Kingdom). ER protocol included additional manual antigen retrieval step with previously boiled 1x TRIS-EDTA buffer solu- tion (pH 9) for 3 min. Evaluation of cytological slides Giemsa slides were evaluated by an experienced cytopathologist (VKP) and images of the cells on slides were captured with a DP72 CCD camera con- nected to a BX-51 microscope (Olympus, Hamburg, Germany. In the preclinical part, morphological features were quantified in 100 cells from each ex- perimental group. Cell diameter, nuclear diameter and thickness of the cytoplasm were measured using ImageJ software.19 Other morphological fea- tures that were analyzed were cytoplasmic and nu- clear chromasia, degeneration characteristics (cyto- plasmic blebs, cytoplasmic villous projections and vacuolization), regularity of plasma membrane and nuclear membrane and chromatin features. These morphological features were quantified by count- ing the cells that displayed these characteristics. In the clinical part of the study, the slides were evalu- ated by the cytopathologist and CTCs were identi- fied by their cytomorphological features. Based on their morphological appearance, they were catego- rized as morphologically “preserved” or “unpre- served”. Criteria for the preserved CTCs were: cells with morphological features of malignancy such as large nuclei, high nuclear to cytoplasmic (N/C) ra- tio, scant cytoplasm, visible chromatin structure or presence of mitotic figures. Criteria for identifica- Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs296 tion of morphologically unpreserved CTCs were: cells with morphological features of malignancy such as large nuclei, high N/C ratio, scant cyto- plasm and loss of chromatin structure and nuclear membrane integrity. All of the identified CTCs were imaged and their diameter was measured. Flow cytometry Cell viability and type of cell death of MCF7 cells was determined by FITC Annexin V Apoptosis Detection Kit with 7-AAD (BioLegend, San Diego, CA, USA) according to manufacturer’s instruc- tions. Within 1 hour after separation, Annexin V and 7AAD were added to the cells and the meas- urements were performed on at least 100,000 cells per sample using FACSCanto II flow cytometer (BD Biosciences, San Jose, CA). Data were analyzed using FlowJo software. Statistical analysis The values in this study are represented by median with interquartile range, mean ± standard error of the mean (SE) or mean with 95% confidence in- terval and are defined in the figure legends. The patient’s categorical characteristics were presented as frequencies and percentages. Age was presented as median and range. The comparison of means of more than two groups was statistically evalu- ated by one-way ANOVA followed by a Dunnett’s multiple comparisons test. The comparison of medians of more than two groups was evaluated by Kruskal–Wallis one-way ANOVA followed by Dunn’s multiple comparisons test. Fisher’s exact test was used to determine the significance be- tween two categorical variables. A P-value of <0.05 was considered to be statistically significant. A sample size (n) for each experiment is stated in the figure legend. For statistical analysis and prepara- tion of graphs, GraphPad Prism 8 (La Jolla, CA, USA) was used. Results Hydropic degeneration after Parsortix® and MACS® separation In the first part of the study, the effect of both separation techniques on cell morphology was evaluated on Giemsa stained cytological slides of intact (non-separated) MCF7 cells vs. Parsortix® or MACS® isolated MCF7 cells. In comparison to the intact MCF7 cells (Figure 2A, No separation), the Parsortix® and MACS® separated MCF7 cells showed signs of cell swelling i.e. hydropic degen- eration (Figure 2A, Parsortix® and MACS®). Also high background of blood cells was observed after MACS® separation protocol (Figure 2A, MACS®). Preservation of antigenic epitopes for immuno- cytochemical staining was determined by immu- nocytochemical staining of AE1/AE3 cytokeratin and estrogen receptor, as these stainings are usual- ly used in the cytopathological detection of breast cancer cells. Both separation techniques retained the epitopes for immunocytochemical staining. All MCF7 cells presented with strong cytoplasmic AE1/ AE3 cytokeratin positive staining (Figure 2B) in all experimental groups. Nuclear estrogen receptor staining was positive in around 70% of MCF7 cells either in non-separated cells or after Parsortix® or MACS® separation (Figure 2C). Cell swelling was quantified based on the measurements of cell diameter, nuclear diam- eter and thickness of the cytoplasm (Figure 2D). Both separation techniques induced similar sta- tistically significant enlargement of the cell di- ameter (Figure 2E), nuclear diameter (Figure 2F) and thickness of the cytoplasm (Figure 2G). Due to the increase in the thickness of cytoplasm and size of nucleus, cytoplasmic staining was signifi- cantly altered. Dark basophilic staining was sta- tistically significantly reduced after both separa- tion techniques (Figure 2H). Contrary, pale baso- philic staining increased in the opposite manner, with MACS® separated cells having the highest fraction of pale basophilic cytoplasmic staining (Figure 2H). The plasma membrane was intact after both separation techniques (Figure 2A). The nuclear features such as chromatin structure and nucleoli were preserved after both separation tech- niques (Figure 1A). Nuclear staining was mostly hyperchromatic (Figure 2A). Chromatin was most- ly coarsely granular (Figure 2D). The nuclear mem- brane was intact with some irregularities observed after both separation techniques (Figure 2A). Degenerative cytoplasmic changes and cell viability after Parsortix® and MACS® separation Increased fraction of cells with degenerative cy- toplasmic changes were observed in Giemsa stained MCF7 cells after both separation tech- niques. Morphologically non-degenerated cells were presented with homogenous cytoplasm and intact smooth plasma membrane (Figure 3A). Degenerative changes that were observed more Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs 297 A B C D G H E F FIGURE 2. Parsortix® and MACS® separation techniques induced hydropic degeneration of cells. (A) Giemsa stained, (B) AE1/ AE3 cytokeratin stained and (C) estrogen receptor stained non-separated MCF7 cells (No separation) and separated MCF7 cells by Parsortix® (Parsortix®) or MACS® (MACS®). Scale bar represents 50 μm. Note that estrogen receptor stain is nuclear therefore cells appear smaller due to the poor contrasting of the cytoplasm. (D) Presentation of measured and evaluated cell sizes. Non- separated and Parsortix® or MACS® separated (E) cell diameter of, n=100 cells, (F) nuclear diameter, n=100 cells and (G) thickness of cytoplasm, n=100 cells. (H) Cytoplasmic staining of non-separated MCF7 cells and Parsortix® or MACS® separated cells, n= 100 cells. Values in (E, F, G) represent median with interquartile range and (H) mean with 95% confidence interval. Statistical significance was determined by Kruskal–Wallis one-way ANOVA followed by Dunn’s multiple comparisons test for (E, F, G) and by Fisher’s exact test for (H). ** = p< 0.01; *** = p<0.001; **** = p< 0.0001; ns = not significant Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs298 frequently after both separation protocols were cy- toplasmic villous projections (Figure 3B, black ar- row), membrane blebbing (Figure 3C, black arrow) and cytoplasmic vacuolization (Figure 3C, white arrow). Quantification of the observed changes demonstrated that MACS® separation induced the highest increase in the fraction of cells with cytoplasmic blebs (Figure 3D). Villous projections were more frequent in Parsortix® separated cells (Figure 3D). Increased fraction of cells with cyto- plasmic vacuoles was observed after both separa- tion techniques (Figure 3D). However, although degenerative changes were observed after both separation techniques, the sep- arated cells retained their viability. Flow cytomet- ric analysis of Annexin V (apoptosis) and 7AAD (necrosis) demonstrated that MCF7 cells remained their viability after both separation techniques (Figure 3E). The quantification demonstrated that the percentage of live cells (Annexin V-, 7AAD-) remained unchanged following both separation techniques (Figure 3F). No significant changes were observed in the percentage of cells in early apoptosis (Annexin V+, 7AAD), late apoptosis (Annexin V+, 7AAD+) and necrosis (Annexin V-, 7AAD+) among the groups (Figure 3E). Evaluation of both separation techniques for isolation of CTCs in metastatic breast cancer Both separation techniques were also evaluated in clinical setting in metastatic breast cancer patients. Patient characteristics are presented in Table 1. The median age of the patients was 58.1 years (min 39.5, max 79.3). CTCs were isolated side-by-side by Parsortix® and MACS® separation, each from 10 ml of whole A B C D E F FIGURE 3. Parsortix® and MACS® separation techniques induced degenerative cytoplasmic changes but retained the viability of MCF7 cells. (A) Morphologically non-degenerated cells presented with homogenous cytoplasm and intact smooth plasma membrane. Scale bar represents 10 μm. (B) Cytoplasmic villous projections as observed degenerative changes are indicated by black arrow. Scale represents 10 μm. (C) Membrane blebbing (indicated by black arrow) and cytoplasmic vacuolization (indicate by white arrow) as observed degenerative changes. Scale represents 10 μm. (D) Fraction of cells with villous projections, blebs or vacuolization. Values represent mean with 95% confidence interval, n = 100 cells. Statistical significance was determined by Fisher’s exact test. * = p<0.05, ** = p< 0.01, *** = p<0.001, **** = p< 0.0001, ns = not significant. (E) Representative dot plots of Annexin V and 7AAD staining in non-separated cells and Parsortix and MACS separated cells (F) Percent of live cells (Annexin V-, 7AAD-), n=3. The values represent mean ± (SE), n = number of biological replicates. Statistical significance was determined by one-way ANOVA followed by a Dunnett’s multiple comparisons test; ns- not significant. Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs 299 blood collected through the same blood draw. CTCs were identified and enumerated based on their cytomorphological features such as their size (CTCs larger than 6.5 µm), round or oval shape, large nucleus, scant cytoplasm and a high N/C ratio. In some cases, mitoses were observed which were suggestive of CTCs since mitosis is not typically observed in normal blood cells. After Parsortix® separation, higher numbers of CTCs were identified. In total, Parsortix® separation re- sulted in identification of 221 CTCs and MACS® separation in 120 CTCs altogether (Figure 4A). In 58% of the patients, more CTCs were identified af- ter Parsortix® separation, in 27% of patients more CTCs were identified after MACS® separation and in 15% equal numbers of CTCs were identified after both separation techniques (Figure 4B). The iden- tified CTCs were diverse in their size. CTC mean diameter was significantly larger in CTCs after Parsortix® separation (Figure 4C). Histograms of CTC sizes demonstrating that smaller CTCs were identified after MACS® separation are presented in Figure 4D and Figure 4E. Besides cell size, different separation techniques also affected the morphology of CTCs. The iden- tified CTCs were classified as morphologically “preserved” and “unpreserved” based on their morphological characteristics. After Parsortix® separation, the majority of identified CTCs were presented with preserved morphology with only 2 CTCs being identified as unpreserved (Figure 5A). Contrary, the majority of CTCs after MACS® separation were presented with unpreserved morphology, only one preserved CTC was iden- tified after this separation technique (Figure 5B). The preserved CTCs were presented as cells with morphological features of malignancy and blast morphology such as large nuclei, high N/C ratio, A B C D E FIGURE 4. Parsortix® separation enabled identification of more CTCs and their size was larger compared to MACS® separation. (A) Total number of identified CTCs in the patient cohort. (B) Number of identified CTCs for each individual patient after Parsortix® or MACS® separation. (C) Mean cell diameter of CTCs after Parsortix® or MACS® separation. (D) Histogram showing CTC size distribution after Parsortix® separation and (E) MACS® separation. **** = p< 0.0001; n = total number of cells Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs300 scant basophilic cytoplasm and finely granular chromatin structure (Figure 5C). Nucleoli were not visible. CTCs also presented signs of degeneration such as cytoplasmic blebs and cytoplasmic villous projections (Figure 5C). Some of the cells were in a state of active mitosis, which is also a character- istic of tumor cells (Figure 5C). After Parsortix® separation routine cytokeratin AE1/AE3 (CK) im- munocytochemical staining was positive in 2 out of 30 patients that were morphologically identi- fied as preserved CTCs (Figure 5C, CK). Vimentin staining was not confirmed in any of the samples due to the suboptimal ICC reaction. The external controls displayed a strong positive reaction, how- ever internal controls (lymphocytes) did not stain properly. Therefore, the staining was not consid- ered valid and further optimization steps are war- ranted. Morphologically unpreserved CTCs were presented as cells with morphological features of malignancy such as scant eosinophilic cytoplasm with vacuoles and eosinophilic inclusions, large nuclei, high N/C ratio, and irregular nuclear con- tours with loss of chromatin structure (Figure 5C). After MACS® isolation, unpreserved CTCs were positive for routine cytokeratin AE1/AE3 staining in 2 out of 30 patients. CTC polymorphism in metastatic breast cancer is dependent on the isolation protocol In this metastatic breast cancer patient cohort, highly polymorphic CTCs with blast morphology were identified. CTCs isolated by Parsortix® from the same blood sample varied in size (Figure 6). Numerous CTCs presented villous projections or cytoplasmic blebs, however, plasma membrane of some CTCs was smooth (Figure 6). Cytoplasm was predominantly scant and in some cells, it was almost not visible (Figure 6). MACS® separa- tion from a parallel blood sample collected in the same blood draw resulted in predominantly un- preserved CTCs that also varied in size (Figure 6). Their size was generally smaller (Figure 4C and Figure 6), sometimes the size of a lymphocyte. Discussion In the preclinical part of study, we demonstrated that Parsortix® and MACS® isolation techniques retained the viability and antigenic characteristics of MCF7 breast cancer cells. Hydropic and some other signs of degeneration such as cytoplasmic blebs, villous projections and vacuolization were observed after separation with both techniques, however these changes were not severe, and the cells retained their viability. Parsortix® separation induced lower levels of degeneration compared to MACS® in some of the evaluated morphologi- cal features. MACS® samples also contained high background of remaining hematopoietic cells. In the clinical setting in metastatic breast cancer pa- tient cohort, we demonstrated that morphologi- cally preserved CTCs were detected by Parsortix® method only. On the contrary, after MACS® sep- aration the majority of detected CTCs were mor- phologically unpreserved. Furthermore, we dem- onstrate that CTCs from metastatic breast cancer show a high degree of polymorphism, even within the same patient. Taken together, morphological features of isolated CTCs are dependent on the separation technique and also to some degree to the in vivo degeneration in the blood stream. Parsortix® separation was demonstrated to be a TABLE 1. Patient characteristics Characteristics Frequency N (%) Histology Invasive ductal carcinoma Invasive lobular carcinoma 28 (93.3) 2 (6.7) Tumor stage T1 T2 T3 5 (16.7) 18 (60.0) 7 (23.3) N stage N0 N1 N2 N3 Unknown 5 (16.7) 9 (30.0) 3 (10.0) 7 (23.3) 6 (20.0) Grade Grade I Grade II Grade III Unknown 1 (3.3) 9 (30.0) 19 (63.3) 1 (3.3) Hormone receptor Estrogene receptor positive Estrogene receptor negative Progesterone receptor positive Progesterone receptor negative 24 (80.0) 6 (20.0) 21 (70.0) 9 (30.0) HER2 status Positive Negative 4 (13.3) 26 (86.7) Molecular subtype Luminal A-like Luminal-B like HER2 negative Luminal-B like HER2 positive HER2 positive Triple negative 8 (25.8) 13 (43.3) 3 (10.0) 1 (3.3) 5 (16.7) Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs 301 suitable technique for detection of CTCs with pre- served cell morphology in metastatic breast cancer for their identification by routine cytopathological techniques. In our metastatic breast cancer patient cohort, Parsortix® separation was in general superior to MACS® in terms of the quantity and quality of identified CTCs. However, in some samples, MACS samples contained higher numbers of CTCs. When comparing both methods, it should be kept in mind that the successful CTC isolation is not dependent only on the separation technique, but also on the blood sample. In our case, both blood samples were taken during the same blood draw, however in two separate 10 ml blood collec- tion tubes. Therefore, due to the fast blood flow it cannot be expected that both samples will contain equal numbers of CTCs, as CTC numbers in blood can depend on dynamics of CTC shedding from the tumor and also their distribution in the blood stream. Similar discrepancies in CTC numbers were observed also in the study of Xu et al., com- paring three isolation platforms.20 Higher number of identified CTCs by Parsortix® separation can be to some extent attributed to its higher sensitivity. The sizes of CTCs isolated in our study were all larger than 6.5 µm (gap size in the Parsortix® cassette), therefore the Parsortix® iso- lation capacity was retained. High recovery rates of different cell lines were reported also when directly compared to EpCAM positive selection techniques.9,18,21 In the study of Maertens et al., the median recovery rate of the Parsortix® system was 66% compared to 23% for EpCAM based separa- tion (Epithelial Enrich Dynabeads, Invitrogen) af- ter spiking of different cell lines.21 The advantage of Parsortix® size exclusion separation is also that it enables the capture of CTCs of different pheno- types, therefore higher capture capacity can be ex- pected. Parsortix® separation was demonstrated to identify significantly more mesenchymal hu- man CTCs compared to the clinical CellSearch system.22 Contrary, MACS® separation is limited by the capture of CTCs of epithelial phenotype, as was already confirmed in our previous study in spiking experiments of human fibroblasts into the buffy coat, demonstrating 0% recovery rate for mesenchymal cell line.8 CTCs found in the blood of cancer patients are a heterogeneous population as epithelial, mesenchymal and hybrid E/M pheno- types were detected in different cancer types.23–25 Detection of mesenchymal phenotype was shown to be associated with distant metastasis in breast cancer patients, therefore detection of CTCs with this phenotype seems to be of outmost impor- tance.24 In our study we detected only 2 out of 30 patients with advanced metastatic disease with epithelial phenotype. CTCs presented with blast morphology, however unfortunately we could not confirm their mesenchymal phenotype due to the problems with our routine vimentin ICC staining, which seems to be suboptimal for CTC slides ob- FIGURE 5. Effect of separation techniques on morphology of breast cancer CTCs. (A) Number of morphologically preserved or unpreserved CTCs after Parsortix® separation. (B) Number of morphologically preserved or unpreserved CTCs after MACS® separation. (C) Images of CTCs with preserved or unpreserved morphology after both separation techniques. Scale bar represents 20 μm. Cells were stained with Giemsa (G) and cytokeratin AE1/AE3 (CK). A B C Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs302 tained in our study protocol. Further optimization steps for fixation and ICC staining are warranted. Parsortix® separation was demonstrated to pre- serve the morphological properties of CTCs much better than MACS® separation. Both blood sam- ples were taken during the same blood draw in the same type of collection tube and both separations were started at the same time, therefore the CTCs were retained at the same environmental condi- tions until the separation. Separation that followed was different. Parsortix® separation is based on microfluidic system that retain larger cells in the cassette with 6.5 µm size gap. In MACS® separa- tion, blood is passed through a dense magnetic col- umn and cells are retained by a strong magnetic field. After both separations, cells were harvested and cytological slides were prepared by the same protocol. For Parsortix®, harvest (elution) was performed directly from the instrument into the in-house washing buffer. For MACS®, elution of cells from the column was made by MACS® Whole Blood Column Elution Buffer of undisclosed com- position. Therefore, the differences in cytomor- phology of separated cells can arise from the sepa- ration technique itself and also the use of different recommended buffers for separation and elution. A majority of patient’ CTCs presented with the signs of morphological degeneration such as cytoplasmic blebs and villous projections after Parsortix® separation. Compared to our preclini- cal morphological evaluation of MCF7 cells, CTCs isolated from patient samples were more degener- ated, displayed loss of chromatin structure with invisible nucleoli. The higher level of degeneration could be attributed to in vivo degeneration in the blood stream. After MACS® separation, degenera- tion was more severe and the cells displayed the characteristics of cell death such as loss of plasma membrane and nuclear membrane integrity, loss of chromatin structure, vacuoles and cytoplasmic eosinophilic inclusions. CTCs with the similar morphological characteristics were identified in our previous study in early breast cancer patient cohort and were presented as canonical and non- canonical.8 CTCs were termed canonical if malig- nant morphological features as well as cytokera- tin positivity were observed and non-canonical if malignant morphological features were observed but there was no cytokeratin expression.8 In the current study, we were unable to use the same criteria as Parsortix® separation enables the isola- tion of CTCs with various phenotypes, therefore cytokeratin positive staining could not be consid- ered as an identification criterion. Therefore, we focused exclusively on morphological features of cells, which are still a gold standard in cytopathol- ogy.26 For more precise characterization of CTCs further immunochemical studies of their pheno- type are required. Our routine cytokeratin AE1/ AE3 staining resulted in positive staining in 2 out of 30 patients and routine vimentin staining was not positive in neither of the patients. Despite the fact that this staining performs well in the routine work, the staining is most probably not optimal for the CTCs from clinical samples as the signal of the staining was poor. Same was confirmed for vimen- tin staining, which could not be properly detected. The external controls displayed positive reaction, however internal controls (lymphocytes) did not FIGURE 6. Morphological polymorphism of CTCs. Images of identified CTCs on a single patient level in four patients (patient 18, 23, 30, 31) are presented following Parsortix® or MACS® separation from blood obtained in a single blood draw. CTCs varied in size and their morphological characteristics. Scale bar represents 10 μm. Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs 303 stain properly, therefore the staining was not con- sidered valid. Therefore, we believe that optimiza- tion of the routine immunocytochemical staining of CTCs is needed in order to stain these delicate group of cells. Therefore, our next study is aimed at optimization of immunocytochemical and im- munofluorescent staining protocols for breast can- cer CTCs taking in mind the distinct phenotypes and other markers used for identification in breast cancer cells. The discrepancy in cytomorphological changes after both separation techniques in preclinical in clinical setting could imply on the fact that pa- tient’ CTCs are more sensitive and susceptible to degeneration as cell lines. CTCs in the blood are exposed to the shear stress of blood flow, lack of growth factors and immune surveillance, which can degenerate a large proportion of CTCs.10 These in vivo degenerated CTCs can also be more suscep- tible to subsequent stress during the separation. Furthermore, in our protocol, their degeneration can be enhanced also by centrifugation in a cyto- centrifuge when preparing the cytological slides. To the best of our knowledge, our study is one of the few studies investigating detailed morphologi- cal features of breast cancer CTCs with routine cy- topathological techniques. A direct comparison to other studies of CTC morphology in distinct cancer types cannot be made as the protocols for separa- tion and visualization are different. Therefore, a comparison can be made only based on the pub- lished images. In the study of Hattori et al. in breast cancer and Tsutsuyama et al. in colorectal patients, morphologically preserved single or clustered CTCs were obtained.14,15 The cells showed no sign of hydropic degeneration and the nuclear struc- ture was preserved, as the samples were fixed in formalin prior to cytological slides preparation. Both studies utilized Optnics Precision Co. filtra- tion device, which therefore also seems suitable for cytology-based detection of CTCs. In a case study of Marrinnuci et al., 659 CTCs were identified in a single 10 ml blood sample in metastatic breast cancer patient following the fiber-optic array scan- ning technology (FAST) cytometry.13 Similar to our study, they found that the patient’s CTCs exhibit a high degree of polymorphism with CTCs exhibit- ing early and late apoptotic changes.13 In the study of Kuvendjiska et al. in non-metastatic esophageal adenocarcinoma, different types of CTCs were identified based on their morphological features after ISET filtration.16 Small and large single-CTCs, cluster CTCs and circulating cancer-associated macrophage-like cells (CAML) were identified. CTCs presented with increased N/C ratio, enlarged and hyperchromatic nuclei, loss of chromatin structure, irregular nuclear borders and sometimes multilobulated nuclei, therefore, their morphology was to our opinion not sufficiently preserved.16 This study was a preliminary study, aimed at side-by-side comparison of two available separa- tion techniques in order to identify which of the two methods is more appropriate for integration into our routine clinical cytopathology laboratory. The integration of this method to our laboratory will allow us to conduct subsequent clinical stud- ies, which will be aimed at investigation of the clinical relevance of the CTC number and their pre- served morphology in different tumor types. In conclusion, Parsortix® technology is straight- forward technology for CTC isolation, which ena- bles the preservation of cell morphology and can be easily integrated into a routine cytopathology laboratory. Cytopathological analysis and micro- scopic examination of cells is still the gold standard in cytopathology and histology in cancer manage- ment. The advantage of cytopathological analysis is that it is easily accessible in the clinical envi- ronment as cytopathologists are an integral part of cancer patient care. Therefore, the analysis can be done without the use of special techniques and instruments such as flow cytometry or single cell next generation sequencing (NGS), which require substantial financial and human resources. Acknowledgments We would like to thank all patients participating in this study. We would also like to thank ANGLE plc. for providing the Parsortix® instrument for evalu- ation in our lab with special thanks to Adam Giles, Michael O’Bri-en and Lara Stevenato for a detailed training and their valuable advice. We acknowl- edge the financial support of Slovenian Research Agency (grant no. P3-0003, P3-0289 and P3-0321). References 1. Bilani N, Elson L, Liang H, Elimimian EB, Arteta-Bulos R, Nahleh Z. Prognostic and predictive value of circulating and disseminated tumor cells in breast cancer: A National Cancer Database (NCDB) analysis. Technol Cancer Res Treat 2020; 19: 1533033820980107. doi: 10.1177/1533033820980107 2. Theil G, Fornara P, Bialek J. Position of circulating tumor cells in the clinical routine in prostate cancer and breast cancer patients. Cancers 2020; 12: 3782. doi: 10.3390/cancers12123782 3. Lozar T, Gersak K, Cemazar M, Grasic Kuhar C, Jesenko T. The biology and clinical potential of circulating tumor cells. Radiol Oncol 2019; 53: 131–47. doi: 10.2478/raon-2019-0024 Radiol Oncol 2021; 55(3): 292-304. Jesenko T et al. / Morphological features of CTCs304 4. Grosse-Wilde A, Fouquier d’Hérouël A, McIntosh E, Ertaylan G, Skupin A, Kuestner RE, et al. Stemness of the hybrid epithelial/mesenchymal state in breast cancer and its association with poor survival. PLoS One 2015; 10: e0126522. doi: 10.1371/journal.pone.0126522 5. Jolly MK, Boareto M, Huang B, Jia D, Lu M, Ben-Jacob E, et al. Implications of the hybrid epithelial/mesenchymal phenotype in metastasis. Front Oncol 2015; 5:155. doi: 10.3389/fonc.2015.00155 6. Riethdorf S, Fritsche H, Muller V, Rau T, Schindlbeck C, Rack B, et al. Detection of circulating tumor cells in peripheral blood of patients with metastatic breast cancer: a validation study of the CellSearch System. Clin Cancer Res 2007; 13: 920–8. doi: 10.1158/1078-0432.CCR-06-1695 7. Hu XC, Wang Y, Shi DR, Loo TY, Chow LWC. Immunomagnetic tumor cell en- richment is promising in detecting circulating breast cancer cells. Oncology 2003; 64: 160–5. doi: 10.1159/000067776 8. Lozar T, Jesenko T, Kloboves Prevodnik V, Cemazar M, Hosta V, Jericevic A, et al. Preclinical and clinical evaluation of magnetic-activated cell separation technology for CTC isolation in breast cancer. Front Oncol 2020; 10: 554554. doi: 10.3389/fonc.2020.554554. 9. Miller MC, Robinson PS, Wagner C, O’Shannessy DJ. The ParsortixTM cell separation system — a versatile liquid biopsy platform. Cytometry A 2018; 93: 1234–9. doi: 10.1002/cyto.a.23571 10. Huang Q, Hu X, He W, Zhao Y, Hao S, Wu Q, et al. Fluid shear stress and tumor metastasis. Am J Cancer Res 2018; 8: 763–777. PMID: 29888101 11. Lowe AC, Pignon JC, Carvo I, Drage MG, Constantine NM, Jones N, et al. Young investigator challenge: application of cytologic techniques to circulat- ing tumor cell specimens: detecting activation of the oncogenic transcrip- tion factor STAT3. Cancer Cytopathol 2015; 123: 696–705. doi: 10.1002/ cncy.21640 12. Kamal M, Macall L, Horton C, Hills N, Davis R, Nguyen R, et al. Cytopathologic identification of circulating tumor cells (CTCs) in breast cancer: Application of size-based enrichment. Clin Diagnostic Pathol 2019; 3: 1–5. doi: 10.15761/ CDP.1000S1001 13. Marrinucci D, Bethel K, Bruce RH, Curry DN, Hsieh B, Humphrey M, et al. Case study of the morphologic variation of circulating tumor cells. Hum Pathol 2007; 38: 514–9. doi: 10.1016/j.humpath.2006.08.027 14. Hattori M, Nakanishi H, Yoshimura M, Iwase M, Yoshimura A, Adachi Y, et al. Circulating tumor cells detection in tumor draining vein of breast cancer patients. Sci Rep 2019; 9: 18195. doi: 10.1038/s41598-019-54839-y 15. Tsutsuyama M, Nakanishi H, Yoshimura M, Oshiro T, Kinoshita T, Komori K, et al. Detection of circulating tumor cells in drainage venous blood from colorectal cancer patients using a new filtration and cytology-based automated platform. PLoS One 2019; 14: e0212221. doi: 10.1371/journal. pone.0212221 16. Kuvendjiska J, Pitman MB, Martini V, Braun C, Grebe K, Timme S, et al. Cytopathological heterogeneity of circulating tumor cells in non-metastatic esophageal adenocarcinoma. Anticancer Res 2020; 40: 5679–85. doi: 10.21873/anticanres.14582 17. Lee AV, Oesterreich S, Davidson NE. MCF-7 cells - changing the course of breast cancer research and care for 45 years. J Natl Cancer Inst 2015; 107: djv073. doi: 10.1093/jnci/djv073 18. Koch C, Joosse SA, Schneegans S, Wilken OJW, Janning M, Loreth D, et al. Pre-analytical and analytical variables of label-independent enrichment and automated detection of circulating tumor cells in cancer patients. Cancers 2020; 12: 442. doi: 10.3390/cancers12020442 19. Schneider CA, Rasband WS, Eliceiri KW. NIH Image to ImageJ: 25 years of image analysis. Nature Methods 2012; 9: 671–5. doi: 10.1038/nmeth.2089 20. Xu L, Mao X, Imrali A, Syed F, Mutsvangwa K, Berney D, et al. Optimization and evaluation of a novel size based circulating tumor cell isolation system. PLoS One 2015; 10: e0138032. doi: 10.1371/journal.pone.0138032 21. Maertens Y, Humberg V, Erlmeier F, Steffens S, Steinestel J, Bögemann M, et al. Comparison of isolation platforms for detection of circulating renal cell carcinoma cells. Oncotarget, 2017; 8: 87710–7. doi: 10.18632/onco- target.21197 22. Kitz J, Goodale D, Postenka C, Lowes LE, Allan AL. EMT-independent detection of circulating tumor cells in human blood samples and pre- clinical mouse models of metastasis. Clin Exp Metastasis 2021; 38: 97-108. doi:10.1007/s10585-020-10070-y. 23. Pan L, Yan G, Chen W, Sun L, Wang J, Yang J. Distribution of circulating tumor cell phenotype in early cervical cancer. Cancer Manag Res 2019; 11: 5531–5536. doi: 10.2147/CMAR.S19839. 24. Zhang S, Wu T, Peng X, Liu J, Liu F, Wu S, et al. Mesenchymal phenotype of circulating tumor cells is associated with distant metastasis in breast cancer patients. Cancer Manag Res 2017; 9: 691–700. doi: 10.2147/CMAR. S149801 25. Messaritakis I, Politaki E, Kotsakis A, Dermitzaki EK, Koinis F, Lagoudaki E, et al. Phenotypic characterization of circulating tumor cells in the peripheral blood of patients with small cell lung cancer. PLoS One 2017; 12: e0181211. doi: 10.1371/journal.pone.0181211 26. Dey P. Routine staining in cytology laboratory. In: Basic and advanced laboratory techniques in histopathology and cytology. Singapore: Springer; 2018. p. 133–8.