Aims In breast cancer models, the functional roles of CD147 in proliferation, invasion and treatment resistance have been widely reported. However, there are only a few studies examining the clinicopathological correlation and prognostic relevance of CD147 in breast cancer, especially in relation to breast cancer molecular subtypes.
Methods In this study, we analysed CD147 expression in a large cohort of breast cancers, correlating with clinicopathological features and the expression of a comprehensive panel of biomarkers in triple-negative breast cancer (TNBC) and non-TNBC subsets. Its relationship with patients’ survival was also analysed.
Results CD147 was expressed in 11.9%(140/1174) of all cases and in 23.8% (40/168) of TNBC. The expression was associated with tumour histological subtypes (p=0.01) and most commonly seen in carcinoma with medullary features (26.0%). CD147 expression correlated with high tumour grade, presence of necrosis and basal-like breast cancer (BLBC) subtype, high Ki67 and expression of some other basal markers and stem-like markers. CD147 expression was also associated with poor overall survival in chemotherapy treated patients with TNBC.
Conclusions CD147 is a putative marker in identifying TNBC or BLBC, and may be useful as a prognosis indicator for patients with TNBC or BLBC post chemotherapy.
- breast cancer
Statistics from Altmetric.com
CD147, a transmembrane glycoprotein, is a member of the human immunoglobulin superfamily, also known as extracellular matrix metalloproteinase inducer (EMMPRIN). It is overexpressed in many tumours, including breast cancer.1 As an inducer of matrix metalloproteinase (MMP), CD147 plays a critical role in cancer proliferation, invasion and metastasis by remodelling the extracellular matrix through activation of MMP. In addition to its regulation on MMP, it can induce vascular endothelial growth factor production, thus driving tumour angiogenesis.2 CD147 is also involved in trafficking of monocarboxylate transporters (MCTs) to the cell surface, important for removing the excessive lactate produced during anaerobic glycolysis in cancer cells.3 Upregulation of CD147 has been observed in several multidrug-resistant cancer cell lines.4 Recent studies suggested the underlying mechanisms for chemoresistance conferred by CD147 was related to the cancer stem pathways and the interactions of CD147 with ATP-binding cassette transporter 2 and epidermal growth factor receptor (EGFR).4–6 CD147 also interacts with hyaluronan (HA) receptors, such as CD44, and upregulate HA synthesis.7 CD147–CD44 axis may regulate epithelial mesenchymal transition (EMT), invasion and chemoresistance.8
In breast cancer models, the functional roles of CD147 in proliferation, invasion and treatment resistance have been widely reported.6 9–12 Given its involvement in the multiple processes during cancer progression, it could be a promising therapeutic target or biomarker for diagnosis or prognosis in breast cancers. However, few studies examined the clinicopathological correlation and prognosis of CD147 in breast cancer. These are limited to small-sized cohorts1 13 or focused on triple-negative breast cancers (TNBCs).1 14 Even in studies with larger cohorts,15 16 the association of CD147 had not been analysed according to different breast cancer subtypes. Indeed, CD147 showed differential expression in different subtypes, with the highest expression in TNBC,13 thus the prognostic impact of CD147 in breast cancer could be related to its differentially high expression in TNBC. Hence, the precise implications and use of CD147 in clinical breast cancer remains to be investigated. In this study, we analysed CD147 expression in a large cohort of breast cancers, correlating with the clinicopathological features and the expression of a comprehensive panel of biomarkers in TNBC and non-TNBC subsets. Its relationship with patients’ survival was also analysed.
Patients and methods
Invasive breast cancer excision specimens were collected from three of the involved institutions. All specimens were formalin fixed, paraffin embedded and routinely processed. The histological diagnoses were confirmed using WHO criteria.17 Tumour grade and other histological features including apocrine morphology, necrosis, lymphovascular invasion (LVI), extensive in situ component (EIC), tumour-infiltrating lymphocytes (TILs) as well as fibrotic focus (FF) were all evaluated as previously stated.18 TILs were evaluated basing on the latest recommendations19 and a cut-off point of 20%, which corresponded to the top one-fifth of cases with the highest TIL was used. Patients’ age, tumour size, pT stage, pN stage, follow-up data and treatment information were retrieved from the medical records. For the outcome data, overall survival (OS) was defined as the time interval from the date of initial diagnosis to the date of breast cancer–related death. Disease-free survival (DFS) was defined as the time interval from the date of initial diagnosis to the first detection of breast cancer–specific relapse or death.
Tissue microarray (TMA) construction
Representative areas of the tumours were chosen on the slides and the corresponding areas were taken from the paraffin blocks for TMA construction. Two 0.6mm tissue cores from each case were obtained. The TMAs were assembled with a tissue arrayer (Beecher Instruments, Silver Springs, Maryland, USA). Serial 4 µm sections were cut and one section from each tissue array block was stained with H&E to confirm the presence of representative tumours in the TMA blocks.
Immunohistochemistry and scoring
The TMA slides were stained using a BenchMark XT autostainer. The biomarkers examined were as follows: CD147, oestrogen receptor (ER), progesterone receptor (PR), androgen receptor (AR), human epidermal growth factor receptor 2 (HER2), Ki67, p63, c-kit, EGFR, nerve growth factor receptor (NGFR), CK14, CK5/6, α–Bcrystalline (A-bc), aldehyde dehydrogenase (ALDH), CD44, SOX2, E-cadherin, p-cadherin and vimentin. After deparaffinisation, rehydration, antigen retrieval and primary antibody incubation, the sections were incubated with secondary antibody for 30min at room temperature, and then developed with 3,3'-diaminobenzidine (DAB) and counterstained with haematoxylin.
For scoring, different assessment criteria were applied to different biomarkers. (Details of immunohistochemical (IHC) staining are described in online supplementary table 1.) CD147 was assessed by the intensity and percentage of positively membranous stained cells. The staining intensity was graded from 0to3, while the proportion of stained cells was scored on the percentage of positively stained cells. An immunoscore was obtained by multiplying the intensity score with percentage of stained cells, giving a range of 0–300. CD147 expression was scored as negative (score 0), low (score 1–20) and high (score>20). Cut-offs applied for other markers are described in online supplementary table 1.
Using IHC results as surrogate, the breast cancers were also classified into five different molecular subtypes as follows:
Luminal A: ER+, PR≥20%, HER2−, CK5/6± and Ki67<20%.
Luminal B: ER+and CK5/6±; PR <20%, HER2+or Ki67 ≥20%.
HER2 overexpressed (HER2-OE): ER−, PR−, HER2+ and CK5/6±.
BLBC: ER−, PR−, HER2− (triple negative), CK5/6+ and/or EGFR+.
Non-basal: ER−, PR−, HER2− (triple negative), CK5/6− and EGFR−.
The findings were analysed using the statistical software SPSS for Windows, V.3. Chi-square (χ2) analysis or Fisher’s exact test was used to test for the association of
CD147 expression with most of the clinicopathological features. Mann-Whitney U test was used to analyse the differences in patients’ age and tumour size with CD147 expression. CD147 and other biomarkers’ expression was examined as continuous variable and their association evaluated by Spearman’s correlation. Survival data were evaluated by Kaplan-Meier analysis. Statistical significance was defined at p<0.05.
Totally 1174 cases of breast cancer were included in this cohort. The age of the patients ranged from 22to97years with a mean age of 54.5years. Tumour size ranged from 0.1to13.0cm (mean 2.69cm). This cohort included infiltrating ductal carcinoma of no special type (87.2% of all tumours), carcinoma with medullary features (50/1174), infiltrating lobular carcinoma (ILC, 35/1174), metaplastic carcinoma (11/1174) and others (including mucinous, invasive micropapillary, tubular and neuroendocrine carcinomas, 54/1174). Among those 1164 cases with complete IHC data for molecular classification, there were 531(45.6%) luminal A cases, 351(30.2%) luminal B cases, 114(9.8%) HER2-OE cases, 79(6.8%) BLBC cases and 89(7.6%) 5NP cases. Other clinicopathological features are summarised in table 1.
Correlation with clinicopathological features and biomarkers in all cases
CD147 was expressed 140 cases (11.9%), including 106 cases (75.7%) with low expression and 34 cases (24.4%) with high expression (figure 1). The overall expression was associated with higher tumour grade (p=0.027) and presence of EIC (p<0.001). There was no association with apocrine phenotype, FF, necrosis, LVI, pT stage, pN stage, TIL, patient’s age and tumour size (table 1). CD147 was differentially expressed in different molecular subtypes (p<0.001) and histological subtypes (p=0.041). For molecular subtypes, the highest expression rate was found in BLBC (34.2%), followed by non-basal (14.6%), luminal B (13.6%), luminal A (8.2%) and the least in HER2-OE (7.0%). For the histological subtypes, the highest expression rate was found in carcinoma with medullary features (26.0%) and the lowest in ILC (2.9%) (table 1). For the biomarkers, CD147 correlated positively with Ki67, EGFR, P63, c-kit, CK5/6, CK14, NGFR, vimentin and p-cadherin (p≤0.04), and negatively with ER (p=0.001) and PR (p=0.014). Additionally, it had marginal correlation with CD44 (p=0.057) and α–B crystalline (p=0.070) positively and HER2 (p=0.077) negatively. No correlation with E-cadherin, SOX2, ALDH and AR was observed (table 2).
Correlation with clinicopathological features and biomarkers in TNBC and non-TNBC
CD147 expression was observed in 40 cases (23.8%; 30 cases with low expression and 10 cases with high expression) among the 168 TNBCs. It was associated with high tumour grade (p=0.002), the presence of necrosis (p=0.02) and BLBC subtype (p=0.008). Further classified the TNBC cases into basal-like (TNBC, CK5/6+and/or EGFR+), quadruple negative-non-basal (QNBC; TNBC CK5/6−, EGFR−, AR−) and luminal androgenic (TNBC CK5/6−, EGFR−, AR+) TNBC. There was a significant difference expression of CD147 among the three TNBC subgroups. However, there was no significant difference observed between QNBC non-basal and luminal androgenic TNBC cases (p=0.095) (online supplementary table S2). There was a marginal correlation with the absence of EIC (p=0.070). No association was found with apocrine phenotype, FF, LVI, pT stage, pN stage, TIL, patients’ age and tumour size (table 3). For biomarkers, CD147 correlated positively with Ki67, EGFR, CK5/6, CK14, vimentin, CD44 and p-cadherin (p≤0.046) in TNBC. Interestingly, among these variables, CD44 (rs=0.322) and vimentin (rs=0.394), both cancer stem cell–related markers, showed the highest correlation coefficient if the data were analysed as continuous variables (online supplementary table S3). CD147 showed a marginal positive correlation with c-kit (p=0.062) and NGFR (p=0.066), and a marginal negative correlation with AR (p=0.078). No correlation with p63, α–B crystalline, E-cadherin, SOX2 and ALDH (table 3) was seen.
In non-TNBC, CD147 expression was observed in 99 out of 993 cases (10.0%), including 75 and 24 cases of low and high expression, respectively. CD147 did not show any significant association with the clinicopathological features, except a correlation with older patients’ age (p=0.042). For biomarkers, it was positively correlated with Ki67, EGFR, c-kit, CK5/6, NGFR and vimentin (p≤0.049), but not others (table 3).
High expression of CD147 associated with poor OS in chemotherapy treated TNBC
Survival data were available in 1159 cases with a mean survival time of 64.3months (ranged 1–210 months). Among them, there were 197 cases with breast cancer–related relapse/death. Treatment information was available in 681 cases. Among them, 390 cases received adjuvant chemotherapy and hormonal treatment, 208 cases with hormonal treatment only and 46 cases received chemotherapy only. There was no association between CD147 expression and patients’ OS and DFS in the overall cohort (figure 2A–B). Also, there was no association between CD147 expression and patients’ OS and DFS in TNBC and non-TNBC, respectively (figure 2C–F). Interestingly, when the cases were analysed according to treatment received, high CD147 showed a trend of worse survival in those patients treated with chemotherapy only (CD147 negative cases as reference: χ2=3.679, p=0.055) (figure 3). Further analysis on those patients receiving chemotherapy showed CD147 expression did not correlate with OS in luminal cases. In contrast, in patients with TNBC with chemotherapy, high CD147 expression was associated with poor OS (CD147 negative cases as reference: χ2=7.082, p=0.008) (figure 4). High CD147 expression also correlated with poor OS in HER2-OE cases. However, the number of cases with CD147 expression in HER2-OE cases was low (three cases with survival data).
CD147 plays various roles in cancer biology and is proposed to be a cancer-associated biomarker for pathological diagnosis, prognostic evaluation and targeted therapy in a broad range of cancer types.1 The role of CD147 in breast cancer invasion, proliferation and treatment resistance has been demonstrated in several model systems.6 9–12 In clinical studies, CD147 was overexpressed in breast cancer tissue and increased during progression from atypical ductal hyperplasia to invasive breast cancers.15 It was also found to be associated with poor outcome in breast cancer.1 16 However, these studies had several limitations. Some involved small cohorts.1 13 A population bias was apparent in other reports as the cohorts showed relatively high proportion of hormonal receptor negative cases (>30%hour negative).15 16 Poor OS for CD147-positive TNBC has been demonstrated.14 However, no similar trend could be observed by others in BLBC.20 Moreover, the prognostic effects of CD147 on non-TNBC cases were not known. In this study, we examined the relationship of CD147 with patients’ survival in a large cohort of breast cancer grouped into different subtypes. In line with previous data, the prognostic impact of CD147 in TNBC was observed. Further, we demonstrated that the effect could be related to chemotherapy treatment. Moreover, a relationship of CD147 with basal features in breast cancers was also found.
To understand the role of CD147 in clinical breast cancers, its expression was correlated with various clinicopathological features and expression of a comprehensive panel of biomarkers. Similar to previous studies, CD147 was associated with higher histological grade and negative hormonal receptor expression.15 16 Additionally, we demonstrated a positive correlation of CD147 with the presence of EIC and different histological subtypes (with the highest in IBC with medullary features). Regarding biomarkers, it was associated with Ki67 and basal markers. The data were consistent with the high prevalence of CD147 in BLBC subtype. Of note, in the TNBC subset analysis, CD147 continued to show a strong association with basal markers including EGFR, basal CK, vimentin and p-cadherin. As these markers had been used for BLBC identification,21–23 this strong association may suggest CD147 as a possible candidate BLBC marker.
Most previous studies examined the mechanisms of CD147 inducing tumour cell infiltration and metastasis via promoting MMP expression.24–26 Some other focused on its action via MCT interaction.27 Therefore, in most studies of clinical breast cancers, the association with CD147 expression was limited to these two markers.13 14 20 Its relationship with other markers has rarely been explored. Novel involvement of CD147 in cancer stem cell (CSC) pathway has been recently demonstrated.9 28 29 In CSC, CD147 was found to be significantly upregulated.28 29 In addition, CD147 appears to play a role in regulation of the several CSC signalling pathways including Wnt/β-catenin signalling30 and STAT3/HIF-1α signalling.31 Moreover, a close interaction and co-localisation of CD147 with CD44, a major CSC marker, in breast cancer10 as well as other cancers32 has been observed. In accordance with these studies, the current findings supported a possible relationship of CD147 with CSC in clinical breast cancers. CD44 and vimentin were the markers that showed the strongest correlation with CD147 in TNBC. As mentioned, CD44 was one of the major CSC markers in breast cancer while vimentin was upregulated during EMT, a property exhibited in CSC. Of note, high CD147 expression was linked to poor OS in chemotherapy treated patients with TNBC. A similar observation was also found in HER2-OE cases but not luminal cases. However, the number of cases with CD147 expression in the HER2-OE groups was small. Further investigations with larger studies are required. As CSC had been implicated in treatment failure and cancer progression and high CD147 expression was linked to TNBC chemoresistance, CD147 could be involved in CSC-mediated treatment resistance, thus a potential therapeutic target. In fact, a combined treatment with CD147 targeted therapy was found to increase sensitivity of cancer cells to chemotherapeutic drug and can result in more effective inhibition of tumour proliferation and recurrence.33 34 As the treatment strategy for TNBC is limited, biologically driven combinatorial therapies should be considered. Thus, CD147 as a potential therapeutic target in TNBC warrant further investigation.
In summary, this study highlighted the correlation of CD147 with TNBC, particularly BLBC, as evidenced by the correlation of CD147 with other basal markers, high tumour grade and histological subtypes within TNBC. Further, high CD147 expression indicated poor OS in chemotherapy treated patients with TNBC; thus, CD147 could be a potential prognostic marker and treatment target in this subset of patients.
Take home messages
CD147 expression was associated with clinicopathological features and biomarkers related to triple-negative breast cancer (TNBC), particularly basal-like breast cancer (BLBC).
High CD147 expression indicated poor overall survival in chemotherapy treated patients with TNBC.
CD147 is a putative marker in identifying TNBC or BLBC, and may be useful as a prognosis indicator for patients with TNBC or BLBC post chemotherapy.
Handling editor Cheok Soon Lee.
Contributors ML collected the cases, reviewed the slides and wrote the manuscript; JYST analysed data and wrote the manuscript; ML, JH and HH collected and arranged clinicopathological data of the cases; Y-BN, S-KC and S-YC reviewed the slides; GMKT conceived the idea of the study, reviewed the cases, provided guidance and critically revised the paper. All authors read and approved the final submitted version.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests None declared.
Patient consent Not required.
Ethics approval Joint Chinese University of Hong Kong–New Territories East Cluster clinical research ethics committee.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.