The Aggressiveness of African breast cancer: An expose’ on CD44+CD24-/low breast cancer stem cell

Breast cancer (BC) in Africans and people of African descent is generally aggressive, with poorer prognosis and worse clinical outcomes. The molecular basis of this is however not entirely understood. The CD44 + / CD24 -/low BC stem cell is known for its tumourigenic potential, tumour aggressiveness and its association with poor prognosis. This study identies the relationship between CD44 + / CD24 -/low BC stem cells and clinicopathological features of breast cancer in an African population. Methodology A Ghanaian BC cohort (n= 222) was used to assess CD44 and CD24 expression. Tissue microarray was constructed from the cohort samples and Immunohistochemically stained with CD44 and CD24 antibodies. The associations between clinicopathological features and the expression of the individual markers and their combinations were analysed. Results Of the total 222 breast cancer samples, 81.9 % were cytoplasmic CD24 positive and were associated with higher tumour grade (OR-3.623; r=0.199; p=0.004), gender (OR-9.514; p=0.028), clinical prognostic grading (OR-2.357 r= 0.162; p=0.027) and Her2 positivity (OR-0.216; r=-0.155; p=0.026). CD44 was associated with higher tumour grade (OR-3.148; r= 0.145; p-0.037), and increased mitotic count (OR-3.043, r= 0.173; p=0.028). There was no association between CD44 expression and hormone receptor status. Together, CD44 + /CD24 -/low staining was associated with higher tumour grade (OR-3.162; r=0.166; p=0.018), gender (OR- 12.0; p=0.012), and higher clinical prognostic staging (OR- 2.888; r=0.186; p=0.011). An inverse association of CD44 + CD24 + was found with tumour grade (OR-0.220; r=-0.246; p=0.000), mitotic count (OR-0.406; r=-0.190; p=0.017) and clinical prognostic staging (OR-0.486; r=-0.151; p=0.040). There was no association between CD44 - CD24 + and all the clinicopathological features. Conclusion Combined, CD44 + CD24 -/low was associated with poor prognosis and tumour aggression and may contribute to the tumour aggressiveness of African breast cancer. CD24 expression as a stand-alone marker was found to correlate with clinical and pathological indicators of tumour aggressiveness and poor prognosis.


Introduction
A plethora of data con rms the aggressiveness of breast cancers (BC) of African patients and of patients with African descent compared to their Caucasian counterparts. [1][2][3] The molecular basis for this disparity is however not fully understood. The expression of certain BC stem cells has however been suggested to be responsible for enhanced tumour aggressiveness. It is well established that breast cancers in Africans are more likely to be triple negative and basal like. [4][5][6][7] Racial disparity in cancer risk, prevalence and clinical outcomes have also been very well documented. [8][9][10] It is still unclear if breast cancer stem cells contribute to such racial disparity and to what extent they confer tumour aggressiveness remains largely unanswered. Knowledge on breast cancer stem cells in African breast cancers remains scanty.
Cancer stem cells are known to possess the intrinsic ability of self-renewal, have tumourigenic potential and drive tumour progression. 11 They are also known to be responsible for chemotherapy resistance and tumour recurrence. 12 The identi cation of these stem cells in tumours offers important evidences on tumorigenesis, therapeutic resistance and recurrence. These may also serve as important approaches for targeted therapies. Prognostication of tumours may also be based on presence or absence of speci c cancer stem cells. Earlier studies have focused on elucidating the association that exists between cancer stem cells, prognostic markers, and response to adjuvant therapy. It has been reported that triple negativity and basal like phenotype (BLP) have poor prognosis and are associated with candidate stem cells CD44 + / CD24 -. 4,13,14 Since the discovery of CD44 + CD24 -/low phenotype breast stem cells with a tumourigenic potential by Al Hajj et al, various studies continue to elucidate its role in breast cancer. 11 Despite the intense research interest generated on this putative stem cell marker, its role remains puzzling. Do breast cancer stem cells contribute to such racial disparity and to what extent do they confer tumour aggressiveness? This study identi es the role of CD44 + CD24 -/low breast cancer stem cell marker in contributing to the inherent aggressiveness of African breast cancers.

Materials And Methods
This is a hospital based retrospective study of breast biopsies and reports of patients (n=222) presenting with breast cancer at the departments of pathology, Korle-Bu Teaching Hospital, Accra and the Cape Coast Teaching Hospital, Cape Coast, Ghana from 2012 to 2018. The Korle-Bu Teaching Hospital department of pathology is the largest in Ghana, receiving specimens from the Korle-Bu Teaching Hospital, the largest referral hospital in Ghana and from other health facilities within the Greater Accra Region. The department also receives specimen from all other regions of Ghana. The Cape Coast Teaching Hospital's pathology department also receives specimen from mainly the Central and Western regions of Ghana. Clinical and demographic data regarding age, gender, and clinical information were obtained from the histopathology request forms and registry. Histopathology slides of selected cases within the study period (2012-2018) were reviewed by two pathologists (PKA & LDK). Archival blocks of primary breast carcinoma from the Departments of Pathology, Korle-Bu and Cape Coast Teaching Hospitals were retrieved. Additional clinical information and histopathological features were obtained from the histopathology reports of patients. The information included the mean age of presentation, duration of symptoms, tumour grade, (based on mitotic count, nuclear grade, tubule formation). All cases were reviewed histopathologically and classi ed according to the recent WHO classi cation for breast tumours and histopathological grading done in accordance with the Nottingham criteria 15 .

Tissue Microarray (TMA) Construction
Archival formalin xed para n embedded (FFPE) blocks' and Hematoxylin and Eosin (H&E) stained slides were reviewed and areas of tumour selected. Areas of normal tissue, necrosis, and haemorrhage were ignored. Three cores 1mm each (2 peripheral tumour and 1 central) were punched out from the representative selected areas and arrayed into a new recipient para n block using TMA Grand Master® (3D HISTECH®, Budapest, Hungary). Four micrometre thickness of TMA sections were cut and mounted on Superfrost slides.

Immunohistochemistry (IHC)
TMA were stained using CD24 Monoclonal antibody (SN3), Thermo sher and CD44 monoclonal antibodies (156-3C11), Thermo sher at dilutions of 1:200 and 1:750 respectively. Immunohistochemical antibody labelling was done using the NOVOLINK polymer detection system. To enhance tissue adhesion to the slide, tissue microarrays were pre-heated at 60 o C on a hot plate for 20 minutes and cooled. Depara nisation was done on tissue sections in xylene and rehydrated through a series of graded alcohols and rinsed in distilled water. Antigen retrieval was enhanced by boiling slides in citrate buffer (27ml of citrate in 123 ml disodium citrate and made up to 1.5L with ddH2O) at pH-6.0 and microwaved at full power for 20 minutes. Endogenous peroxidase activity was blocked using the Peroxidase blocking reagent from the NOVOLINK® kit for 5 minutes and rinsed with PBS for 15 minutes. Protein blocking was done for 5 minutes to minimize nonspeci c binding and rinsed thoroughly with PBS for 15 minutes. Primary antibodies were added in the following dilutions: CD24-1:200 and CD44-1:750 and incubated in a black box for 1hour at room temperature. A thorough rinse was done for 15 minutes with PBS tween and then incubated with Post Primary Novolink reagent for 30 minutes in a black box. After a 15-minute thorough rinse, a Polymer was added and incubated for 30 minutes. Peroxidase was then developed by incubating 3,3'-diaminobenzidine chromogen solution (DAB) made up to in 1:20 dilution with DAB substrate buffer and incubated for 5 minutes. Counter staining with hematoxylin was done and incubated for 6 minutes. Dehydration and clearing were done using the Leica autostainer. Sections were then mounted with DPX. Breast cancer cases known to be positive for the markers being studied were used as positive controls. Evaluation of staining was done for both markers.
The semi-quantitative H scoring system was employed in scoring. The intensity of CD24 and CD44 expression were scored as 0 (no expression), 1 (weak), 2 (moderate) and 3 (strong). The total score was calculated as the percentage of positive cells multiplied by the intensity giving a range of 0-300. A cut point of ≤40% score was designated as negative and >40% as positive in keeping with Ahmed et al's study 16 . Statistical Analysis IBM SPSS version 24.0 package program (SPSS inc., Chicago, IL, USA) was used in the statistical analysis. The association between the markers and clinicopathological features were done with cross tables using chi-square test and odd ratios. Correlations were done with Pearson's correlation test. Statistical signi cance was put at the α level.

Results
A total of 222 breast carcinoma samples were used in the study. The study population were predominantly females with 219/222 (98.6%) being females and 3/222 (1.4%) males. The mean age of patients presenting with breast carcinoma in the cohort was 51.4 ±12.5. A total of 110 (49.5%) involved the right breast, while 107(48.2%) involved the left breast and only 1(0.5) were bilateral. There was no indication of laterality in 4 of the patients. Of the total, 177(79.7%) presented with solitary masses and 41(18.5%) with multiple masses with 4 missing data. Table 1 shows details of clinicopathological features of patients. Cytoplasmic CD24 expression and its association with clinicopathological features Eighty-one-point one percent of tumours were cytoplasmic CD24 positive. Tumour grade (p=0.004) and clinical prognostic staging (p=0.027) were the only clinicopathological features which had signi cant association with CD24 cytoplasmic expression. Age did not have an association with CD24 cytoplasmic expression (OR-1.9, p=0.064). Grade 2 & 3 tumours had higher CD24 cytoplasmic expression with between 3 to 4-fold increase (OR-3.6, p=0.004). Females predominantly had higher CD24 cytoplasmic expression (OR-9.5, p=0.028). Although marginally non-signi cant (p=0.085), it appeared tumours with >10 mitotic features per 10 high power eld (/10hpf) had a 2-fold increased CD24 cytoplasmic expression compared with tumours <10/10hpf from the odd ratio. Clinical Prognostic staging (stage III and above) was associated with about 2-fold increased cytoplasmic CD24 expression compared with stage I&II tumours (OR-2.3, p=0.027). Table 2 summarises the cytoplasmic expression of CD24 and its association with clinicopathological features. Figure 1 shows membranous staining of CD44 and homogenous cytoplasmic staining of CD24  Table 3 indicates the cytoplasmic CD24 expression pattern in relation to hormone receptor status. Only Her2 had a signi cant association with CD24 cytoplasmic expression. Her2 negative tumours had 5 times higher CD24 expression in comparison with Her2 positive tumours (OR-4.626, p=0.026). There was no association between the molecular subtypes and CD24 expression (Table 3).

Association between CD44 cytoplasmic expression and hormone receptor status
This study revealed no association between CD44 cytoplasmic expression and ER, PR & Her2 status. A summary of CD44 cytoplasmic expression and hormonal status can be found in table 5. Triple negative tumours predominantly expressed CD44 (45%). Predominantly, CD44 + CD24 + was the most occurring combined phenotype (76%) with CD44 + CD24as the second most common phenotype (15%). CD44 -CD24 -(3%) was however the least occurring phenotype Fig.  2.
Association between CD44/CD24 combined phenotypes and clinicopathological features Figure 3 shows the various combinations of CD44 and CD24 using same cases as reference.
CD44 + CD24 -/low CD44 + CD24 -/low was found to be signi cantly associated with tumour grade (OR-3. There is no association between CD44 -CD24 + and all the clinicopathological features. Table 6 summarises the CD44CD24 combinatorial phenotypes with clinicopathological features. No signi cant association is realised between CD44/CD24 combined phenotypes and hormonal status ( Table 7).  In keeping with literature 1-3 , tumours from our cohort (African population) were aggressive as evidence by patients presenting with higher clinical prognostic stage (about 6 out of 10 having a clinical prognostic stage of III), higher tumour grade (9 out of 10 with tumour grade of II and above) and poor NPI (about 70% with NPI>5.4). Furthermore, over 90% of the tumours were >2cm in size with a total lymph node involvement of 77% at the time of presentation. The tumours also had a very high triple negative prevalence (44.3%) consistent with previous studies in African populations. [4][5][6][7] To answer the question of whether breast cancer stem cells contribute to the inherent tumour aggressiveness in this cohort, we sort to nd out the expression pattern of one of the well characterized breast cancer stem cell markers CD44 + CD24 -/low across our cohort. The clinicopathological association of the expression of the individual markers (CD44 and CD24) as well as their combined phenotypes were studied.
We found a high cytoplasmic CD24 expression associated with tumour aggressiveness and poor prognosis comparable to earlier studies. 16 Similarly, the absence of either CD44 and CD24 expression was also associated with some features of tumour aggression and poor prognosis comparable to Giatromanolaki et al's study. 13 Tumours with CD44 -CD24 -/low phenotype were 10 times more likely to have a mitotic count ≥10 and a 7-8 times likelihood of having a poor NPI >5.41.
In keeping with Ahmed et al's study 16 , CD44 + CD24 + was the most occurring combination phenotype (76%) however, a low CD44 + CD24 -/low phenotype expression (6.1%) was recorded in their cohort as opposed to 15% in this current study. Their cohort rather expressed higher CD44 -CD24 + phenotype (31.5% vs 5.6%). 16 This disparity may be as a result of different molecular signatures between African and Caucasian breast cancers underlying the racial disparity in tumour prognosis, aggressiveness and clinical outcomes favoring the later. Their study was on a larger series of early primary invasive breast carcinoma (1046) as opposed to the more heterogenous and smaller cohort of advance breast carcinoma (222) in this current study.
These factors may also explain the difference in the phenotypic expressions.
This study corroborates other studies that reported the involvement of CD44 + CD24 -/low phenotype in tumour aggressiveness and poor prognosis. 12,13,21,22 Its increased expression was signi cantly associated with higher tumour grade and higher clinical prognostic staging. Ahmed et al and Mylona et al however had the exact opposite conclusion with CD44 + CD24 -/low cancers showing favourable outcomes. 16,23 Several studies have already linked the CD44 + CD24 -/low phenotype to triple negative breast cancer 6,[12][13][14]20,[24][25][26] and hence it is not surprising that our cohort with a high triple negative prevalence records high CD44 + CD24 -/low . This breast cancer stem cell marker may be related to tumour aggression in our cohort despite its non-association with lymph node involvement. The high tumour aggressiveness of our cohort may favour hematological dissemination rather than lymphatic spread explaining this lymph node nonassociation. 21 Conversely, the presence of the CD44 + CD24 + phenotype was associated with good prognosis and less tumour aggression a nding contrary to an earlier Asian study 12 . Our study reported a signi cant inverse association existing between high CD44 + CD24 + expression and parameters such as tumour grade, mitotic count and clinical prognostic staging. Chen et al however had no association with any of the clinicopathological parameters 12 .
CD44 -CD24 + phenotype on the other hand did not have any association with the clinicopathological features in keeping with Chen et al 12 . Ahmed et al however recorded positive association between tumour grade, Nottingham prognostic index and CD44 -CD24 + phenotype. 16 The combination phenotypes similarly did not show any signi cant association with the hormone receptor status.

Conclusion
CD44 + CD24 -/low stem cell marker is associated with poor tumour prognosis, and aggressiveness and may be partly responsible for the inherent African breast cancer aggressiveness. This study corroborates the clinicopathological signi cance and prognostic value of the putative stem cell CD44 + CD24 -/low in breast cancer. It is to be best of our knowledge the rst study of CD44 + CD24 -/low in an African population. It is recommended that future studies look at the association between CD44 + CD24 -/low and clinical outcomes such as Disease-Free Survival (DFS), and Overall Survival (OS) in African populations.