Quantitative Flow Cytometry Assessment of Feline Circulatory Breast Cancer Stem Cells

| The upsurge need for a rapid and accurate prognostic system for feline mammary neoplasia is crucial, as almost 90% of mammary tumors in cats tend to metastasize. Here, appears the importance of circulatory cancer biomarkers (liquid biopsies) such as circulating tumor cells (CTCs) and their subpopulation circulating cancer stem cells (cCSCs) as an effective, feasible, repeatable, and non-invasive tool for tumor prognosis and treatment monitoring. In the current prototype study, we detected and enumerated the circulatory breast cancer stem cells (BCSCs) for the first time in the feline peripheral blood by using the flow cytometer analysis (FCA) against their specific cellular markers (CD44+/CD24and CD133+). In addition, we calculated Area Under the Curve (AUC), like hood ratios and cutoff points by statistical analysis. The results revealed statistically significant differences (p <0.0001) between healthy and diseased animal groups with an excellent AUC value (0.902 for CD44+/CD24cells and 0.990 for CD133+ population) that supported the use of FCA as a sensitive, specific, and rapid diagnostic and monitoring tool for mammary carcinoma. We also set up a cutoff value of diagnostic significance (>276 for CD44+/CD24populations and >12 for CD133+ cells).

Since 2003, many studies have supported the role of breast cancer stem cells (BCSCs( or tumor-initiating cells (TICs) in metastasis mediation, tumor recurrence, and resistance of chemo and radiotherapies after Al-Hajj et al. used CD44 + /CD24surface markers combination for the first isolation of BCSCs (Al-Hajj et al., 2003;Sayed et al., 2016;Mansoori et al., 2017).
Many previous studies that had been performed on companion animals have reported the identification and isolation of BCSCs. In felines and canines, TICs were identified and isolated by mammosphere formation, decrease ALDH1 enzyme activity, or cell surface markers (CD44, CD24, and CD133) either from spontaneous mammary carcinoma (Barbieri et al., 2012(Barbieri et al., , 2015 or mammary carcinoma cell lines (Michishita et al., 2013;Pang et al., 2013). Similar to humans, the isolated canine and feline BCSCs were invasive with a high potential for mammosphere formation, exhibited epithelial-mesenchymal transition (EMT) phenotype, and showed resistance to chemotherapy drugs and radiation (Pang et al., 2013;Barbieri et al., 2015). These proliferative and aggressive cancer cells showed a promising treatment response by nanoparticles (Gold nanoparticles (GNPs), silver nanoparticles (Ag-NPs)) in many previous studies in humans as well as felines (Ali et al., 2016;Sameen et al., 2020;Jawad et al., 2021).
Because cBCSCs are a rare subpopulation in PB, their isolation and identification represent a real challenge (Yang, Imrali and Heeschen, 2015). As of now, there are two techniques to isolate potential BCSCs depending on their specific surface markers: either by fluorescence-activated cell sorting (FACS) or magnetic-activated cell sorting (MACS) (Akbarzadeh et al., 2019). The Flow Cytometer Analysis (FCA) is the most accurate, sensitive, and definitive laboratory tool for recognizing sporadic cells like BCSCs as it is capable of detecting one epithelial cell in up to 107 peripheral blood mononuclear cells (Hu et al., 2010).
In veterinary oncology, M. Michishita and coworkers (Michishita et al., 2013) were the first to use the FCA to recognize TICs in feline mammary carcinoma cell lines. Also, the diagnostic precision of FCA to discriminate neoplastic from non-neoplastic lymphoproliferative anarchies in cats was reported in a study by Martini and colleagues (Martini et al., 2018).
For the first time in cats, we proposed to detect and enumerate cBCSCs in feline peripheral blood using the FC analysis against their specific cellular markers: CD44 + / CD24combination and CD133 + individually. We aimed to evaluate the sensitivity and specificity of this technique for the early diagnosis of FMC using the receiver operating characteristic (ROC) curve and tried to set up a cutoff point of diagnostic value for each used phenotypic marker/ combination.

animals gRoups
This study was carried out in compliance with the ARRIVE guidelines du Sert, Hurst, et al., 2020), and Studies of diagnostic accuracy (STARD)

Advances in Animal and Veterinary Sciences
December 2021 | Volume 9 | Issue 12 | Page 2203 guidelines (Bossuyt et al., 2015). All animals under investigation in the study were handled following the Association for Assessment and Accreditation of Laboratory Animal Care and Office of Laboratory Animal Welfare (AAALAC) guidelines under the direction of the Institutional Animal Care and Use Committee at Cairo University (CU-IACUC) that approved the protocol (code: CU II F 9 16) renewed by (Vet CU28/04/2021/266). Written informed consent was provided by the owners of the cats for recruitment in our research.
Between March 2019 and January 2021, 1ml whole blood samples were collected from each 26 client-owned female cats of different breeds that were admitted to the referral animal hospital of the Faculty of Veterinary Medicine at Cairo University.
We collected nine whole blood samples from 9 healthy queens that have been free of any disease (GA: healthy cats). The other 17 EDTA blood samples were collected from 17 female unspayed cats, diagnosed with mammary tumor (GB: diseased cats), as shown in (Table 1). All samples were stored at 4C° till analysis (within max 2-4hrs. post collection) (Diks et al., 2019). Then, all the diseased cats were introduced to a new treatment research study which will be published soon.
Diagnosis of mammary tumor in GB animals was performed by through physical examination, radiographic imagining for lung metastasis, and histopathological analysis to tumor biopsies by hematoxylin and eosin (H&E) and confirmed by Immunohistochemistry (IHC).

physical and RadiogRaphic Examination and pathological studiEs
In the diseased cases (GB), we measured the dimensions of each tumor using calipers, and for detection of lung metastasis, we used the X-ray machine (Fischer, Berlin, Germany). The radiographic setting factors were as previously described (El-Rasikh et al., 2021). We followed the modified World Health Organization grading system to determine primary tumor size and the presence of lung metastasis (Cassali et al., 2019 (Ali et al., 2018).
Proper negative controls were used according to The Histochemical Society's standards (Hewitt et al., 2014). We could not assess the expression of CD24 because of the absence of antibody cross-reaction with the feline antigen. We implemented the IHC technique according to the kit's manufacture instructions.

Advances in Animal and Veterinary Sciences
December 2021 | Volume 9 | Issue 12 | Page 2204 (ML5), BD Biosciences, USA) according to their manufacture structure. We used CD44v6 antibody as it is nearly absent on most of leukocytes and many studies have approved it as a prognostic marker for BC (Mackay et al., 1994;Liu et al., 2016;Qiao et al., 2018).
The human antibodies used have already been reported to stain feline samples by FCA (Michishita et al., 2013;Pang et al., 2013). One tube served as a negative control (unstained cells) for each sample. Lysis of RBC and fixation of peripheral blood leukocytes after staining with specific antibodies were performed according to the Thermo Fisher Scientific flow cytometry protocol for staining cell surface antigens. Briefly, (1) 10 mL of diluted eBiosci-ence™ 10X RBC Lysis Buffer (Multi-species) (Invitrogen, Thermofisher) were added to each PB sample (1 ml blood), mixed well and incubated at room temperature for 15 min, followed by centrifugation for 10 min. Then, the cell pellets were resuspended and washed three times by 2ml of eBioscience™ Flow Cytometry Staining Buffer (Invitrogen, Thermofisher) and resuspended in 300 µl of the same previous buffer.
(2) For each 100 µl of the previous suspension, we added 20µL of eBioscience™ Fc Receptor Binding Inhibitor Polyclonal Ab, incubated for 15 min at room temperature, washed and resuspended in 300 µl of staining buffer. Then, 20 µl from each specific conjugated antibody (either CD133 + separately or CD44 + / CD24combination) were added to each 100 µl of the previous suspension, vortexed gently, and incubated for 30 min at 4C° in the dark. About 2 mL of diluted eBioscience™ 10X RBC Lysis Buffer (Multi-species) (Invitrogen, Thermofisher) was added to the previous combination, incubated for 15 min in the dark, washed twice in the previously used staining buffer, then resuspended in 500 µl of this same reagent.
(3)10-20 µl of eBioscience™ 7-AAD Viability Staining was added to the final cell suspension 15 min prior to sample acquisition on the cytometer. cBCSCs were counted by recording all events in the whole suspension. BCSCs were identified as CD133 + or CD44 + / CD24combination. The results are represented as mean ± standard deviation (SD), median, and range.

statistical analysis
The data were checked for errors. Normality was checked using descriptive statistics, plots (histogram and box plot), and the Shapiro Wilk test. Comparison of the CD44 + /CD24and CD133 + absolute number of cells was implemented using the Mann Whitney U test. We used a binary logistic regression to define the cancer probability using the enumerated cellular markers. Area Under the Curve, sensitivity, specificity, and likelihood ratios, and cutoff points were calculated. Sensitivity values were plotted against complementary specificity values in the receiver operating characteristic (ROC) curve. The significance level was set at a p-value of 0.05 (Younus et al., 2019). For data analysis, we used MedCalc Statistical Software version 19.2.6 (MedCalc Software Ltd, Ostend, Belgium; https://www.medcalc.org;2020).

physical and RadiogRaphic Examination
Physical examination of GB revealed about 39 tumors of different size categories and sites ( Fig.1; A-C). Data considering the incidence of cats' tumor numbers, sites, and sizes are shown in (Table 2). The radiographic examination diagnosed seven cats with lung metastasis (41.18%) among the diseased groups ( Fig.1; D).

Advances in Animal and Veterinary Sciences
December 2021 | Volume 9 | Issue 12 | Page 2205

pathological studiEs
All the examined tumor biopsies revealed features of mammary gland carcinoma (either ductal, tubular, or adenocarcinoma) when stained with H&E ( Fig.2; A-C). In IHC, CD44 and CD133 demonstrated positive peroxidase reaction in all examined tissues either in the cell membrane of the lining epithelium (CD44) alone or with cytoplasm (CD133) (Fig.2; D&E).

FlowcytomEtRy Evaluation
The cells with the phenotypic expression of (CD44 + / CD24 -) combination and CD133 + that correspond with breast cancer stem cells were detected and enumerated in each PB sample in this study (GA and GB) by using a flow cytometer (Fig.3). The count of (CD44 + /CD24 -) cells and CD133 + populations were significantly prevalent in cats diagnosed with mammary tumor (GB; n= 17) vs apparently healthy cats (GA; n=9) (p <0.0001). Mean ± standard

Advances in Animal and Veterinary Sciences
December 2021 | Volume 9 | Issue 12 | Page 2206  deviation (SD), median, and range of CD44 + /CD24cells and CD133 + phenotypes of control and diseased groups are shown in (Table 3). Queens that diagnosed with lung metastasis (n:7) showed remarkable increase in the (CD44 + / CD24 -) cell populations over the rest of the GB (n:10). On the contrary, the CD133 + cells count was relatively higher in female cats without metastasis than those showed lung metastasis (Table 4).

Advances in Animal and Veterinary Sciences
December 2021 | Volume 9 | Issue 12 | Page 2207 The Odds Ratios (OR) (
Herein, we document the first time the flow cytometry quantitative analysis of BCSCs in the PB of cats diagnosed with the mammary tumor. The results showed that BCSCs could be detected and counted in the PB of feline depending on their definite cellular markers ((CD44 + / CD24 -) and CD133 + ); FCA is a sensitive, specific, and rapid diagnostic tool for mammary carcinoma and we also established a cutoff value of diagnostic significance for both (CD44 + /CD24 -) and CD133 + populations.
To overcome the paucity of BCSCs, we imitated the flow cytometry analysis technique previously reported (Al-Hajj et al., 2003;Martini et al., 2020). This strategy included all events in the whole suspension, verified by a back-gating method based on cellular morphologic properties ( In the current prototype, (CD44 + /CD24 − ) and CD133 + cell populations were detected in PB of all animals under investigation with a statistically significant difference between diseased group (n=17) and healthy queens (n=9) (p <0.0001).
These flow cytometry outcomes were utterly matched with our results of the initial clinical and IHC diagnosis. The same diseased positive groups gave statistically significant higher values in the flow cytometry analysis. Our findings are similar in humans as many studies validate the BCSC expression as a diagnostic and prognostic marker for aggressive breast cancers (Matsuoka and Yashiro, 2015;Sayed et al., 2016;Mansoori et al., 2017;Lee et al., 2019;Elbaiomy et al., 2020).
Previous studies have proposed flow cytometry as the best laboratory tool for detecting rare cell populations like BCSCs in blood as it can identify cells at frequencies as low as 0.0001% (Goodale et al., 2009;Watanabe et al., 2014). Our ROC curve analysis strongly supported this conclusion by a sensitivity of 88.24% for (CD44 + /CD24 − ) cell population and 100% for CD133 + cell population. While specificity was 100% and 88.89%, respectively. Furthermore, the ROC curve gave an excellent AUC value of 0.902 (95% CI: 0.721-0.983) and 0.990 (95% CI: 0.850 -1.00) for (CD44 + /CD24 − ) and CD133 + expressions, respectively (Table 5). These presented findings confirm the former conclusion about the diagnostic accuracy of FCA to distinguish tumorigenic from non-tumorigenic lymphoproliferative disorders in cats (Martini et al., 2018).
For the first time in cats, we set up a diagnostic significant cutoff value for the measurements of cBCSCs in 1 mL of feline peripheral blood, which is >276 for (CD44 + /CD24 -) cells and >12 for CD133 + cells that generated a Youden index of 0.88 and 0.89, respectively. In humans, no estimated diagnostic cut-off value for cBCSCs was established yet. However, prognostic cutoff points of 2 CTCs/4ml blood in one study and 3CTCs/1ml blood in another have been reported (Lopresti et al., 2019;Jin et al., 2020).
The same is also noticed among canine investigations; there is no clear estimated diagnostic cutoff point for cBCSCs. However, a value of zero CTCs in the healthy negative control dogs was reported in a study by L. Marconato. et al (Marconato et al., 2019).
The stress of handling and speed during the venipuncture and the procedure site might affect hematology results by shifting up the RBC and elevating the WBC count, which might explain the relatively high values of the established cutoff points in the feline. So, it is recommended to restrict sampling to one person with standardized procedures (O'Brien et al., 1998). Upon this conclusion, we cannot exclude such an effect on CTCs and their subpopulation (cBCSCs). Also, on luminal epithelial cells of the feline malignant mammary tumor, the upregulation of P-Cadherin and loss of expression or abnormal function of E-cadherin increase tumor cell proliferation, motility, invasiveness, high infiltrative growth, and the presence of neoplastic emboli that consequently increase CTCs and cCSCs populations in queens with mammary tumor (Figueira et al., 2014).
We also applied Binary Logistic Regression analysis for the first time in queens to assess the strength of association between CD44 + /CD24or CD133 + counts and having breast cancer as independent factors. The odds of having a mammary tumor were 1.004 times as likely to have one CD44 + /CD24cell and 2.240 times as likely to have one CD133 + cell. Previous studies have measured the strength of the association between BCSCs and disease prognosis

Advances in Animal and Veterinary Sciences
December 2021 | Volume 9 | Issue 12 | Page 2208 in humans (Elbaiomy et al., 2020), and canines (Marconato et al., 2019). Nevertheless, this association between cBCSCs and having mammary tumors is not statistically significant (P<0.05) (Table 6), which may be related to the relatively low number of animals used in this study.
Previous studies have proposed the role of Epithelial-Mesenchymal Transition (EMT) during the starting of the metastasis process (Yang et al., 2015). EMT enables carcinoma cells to gain migratory and invasive abilities together with obtaining stemness characteristics. This transition helps tumor cells escape from the primary tumor lesions, intravasate into the circulation, and disseminate at distant secondary tumor sites (Mani et al., 2008;Yilmaz and Christofori, 2009). Tumor epithelial cells that undergo EMT show upregulation of CD44 down-regulation of CD24 and other stem cell markers (Mani et al., 2008). These events could explain the remarkable increase of (CD44 + /CD24 -) cell population in the diseased cats with lung metastasis more than the rest of the diseased group. Thus, our results came compatible with previous studies which confirmed that the (CD44 + /CD24 -/low ) cell population is associated with boosted invasion and metastasis (Baba and Câtoi, 2007), and so we are suggesting that (CD44 + /CD24 -/low ) cell enumeration could be used as a prognostic tool for metastasis initiation and treatment monitoring in the feline.
Several former articles demonstrated the ability of neoplastic cells to adapt to the state of low oxygen accessibility (hypoxia). Almost half of breast cancer lesions, enfold many hypoxic regions varying in amount and size (Bhandari et al., 2019). These hypoxic regions are reported to be associated with clinically aggressive tumor behavior (Walsh et al., 2014). The expression of CD133 was informed to be induced by low oxygen availability. Therefore, an increase in CD133expression can be considered an aggressive tumor evolution marker with prognostic and predictive values (Brugnoli et al., 2019), and might explain why, in our investigation, the CD133 + cells count was relatively higher in female cats without metastasis than those with metastasis.

concluSIon
In conclusion, during this study, we assessed the diagnostic significance of the flow cytometer enumeration of cBCSCs for the first time in the feline PB. The (ROC) analysis revealed excellent AUC rates (0.902 and 0.990) for both (CD44 +/ CD24 − ) and CD133 + phenotypes respectively. Also, we established a cutoff point of diagnostic values for both (CD44 + /CD24 -/low ) and CD133 + cell populations (>276 and >12), respectively. In addition, our team used the Odds Ratio to measure the strength of the asso-ciation between (CD44 + /CD24 -/low ) and CD133 + cells detection and mammary tumor occurrence that was (1.004) OR for (CD44 +/ CD24 − ) population and (2.240) OR for CD133 + cells.
Since our investigation was based on a relatively low number of subjects, we recommend applying the mentioned FCA diagnostic technique to a large-scale animal population and investigating any correlation between tumor size, sites, numbers, and circulatory BCSCs count. However, we noticed a nonsignificant correlation between lung metastasis and upregulation of (CD44 + /CD24 -/low ) cells in queen's circulation, further separate studies about the relation of FC enumeration of the cBCSCs and feline mammary tumor metastasis are recommended.
In addition, we endorse including cases with benign mammary conditions such as benign tumors or hyperplasia to evaluate the ability of FCA to discriminate between benign and malignant diseases. We also suggest running a comparative analysis between the number of cCSC and their number in the solid tumor besides applying the previous diagnostic regime as a model in human breast oncology studies.
One of this study's limitations; is that we can't run a single flow panel with both CD44 + , CD24 -, and CD133 + as the available CD44 and the CD133 antibodies were of the same fluorophore (PE).

AcKnowlEDGMEntS
We thank Prof. Dr. Naglaa M. Alkalamawy (Pathology Department at Animal Health Research Institute (AHRI), Egypt) for examination and imaging the pathological slides. We thank Assistant lecturer; Hams Abdelrahman (Dental Public Health Department at Alexandria University, Egypt) for performance of all statistical analysis. We thank Christopher P Cook and Aymen Khalifa (Department of Dermatology, University of California, San Francisco, San Francisco CA, USA) for their generous contribution in editing of flow cytometry analysis.

conFlIct oF IntErESt
No potential competing of interest was reported by authors.