Article Text

Download PDFPDF

Caveolin-1 expression as a prognostic marker in triple negative breast cancers of Asian women
  1. Joe Yeong1,2,
  2. Aye Aye Thike1,
  3. Murasaki Ikeda3,
  4. Jeffrey Chun Tatt Lim1,
  5. Bernett Lee2,
  6. Seigo Nakamura3,
  7. Jabed Iqbal1,
  8. Puay Hoon Tan1
  1. 1 Division of Pathology, Singapore General Hospital, Singapore
  2. 2 Singapore Immunology Network (SIgN), Agency of Science, Technology and Research (A*STAR), Singapore
  3. 3 Division of Breast Surgical Oncology, Showa University School of Medicine, Tokyo, Japan
  1. Correspondence to Dr Puay Hoon Tan, Division of Pathology, Singapore General Hospital, 20 College Road, Academia, Level 7, Diagnostics Tower, 169856 Singapore; tan.puay.hoon{at}


Background Triple-negative breast cancers (TNBCs) are defined by their lack of oestrogen receptor, progesterone receptor and epidermal growth factor receptor 2. Although heterogeneous, the majority are aggressive and treatment options are limited. Caveolin acts as tumour suppressor or promoter depending on the cancer type.

Aim In this study, we aimed to determine if the expression levels of the candidate biomarker caveolin-1 on stromal or tumour cells were associated with clinicopathological parameters and disease outcomes in TNBCs from an ethnically diverse cohort of Asian women.

Methods Tumour specimens from 699 women with TNBC were subjected to immunohistochemical analysis of the frequency and intensity of caveolin-1 expression in tumour and stromal cells. A subset of 141 tumour samples also underwent Nanostring measurement of CAV1 mRNA. Results were correlated with clinicopathological parameters and disease outcomes.

Results Expression of caveolin-1 in stromal cells was observed in 14.4% of TNBC cases. TNBCs of the basal-like phenotype (85% of samples) were significantly more likely to exhibit stromal cell caveolin-1 expression (p=0.028), as were those with a trabecular growth pattern (p=0.007). Lack of stromal caveolin-1 expression in both TNBCs and those with the basal-like phenotype was significantly associated with worse overall survival (p=0.009 and p=0.026, respectively): accordingly, increasing mRNA levels of CAV1 in TNBC samples predicted better overall survival. Caveolin-1 expression on TNBC tumour cells was not associated with clinical outcome.

Conclusion Stromal, but not tumoural, caveolin-1 expression is significantly associated with survival in Asian women with TNBC.

  • triple negative breast cancers
  • basal-like phenotype
  • caveolin-1
  • stroma

Statistics from

Request Permissions

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.


Breast cancer is the most common malignancy and leading cause of cancer death in females. An estimated 1.671 million new breast cancer cases and 522 000 deaths occurred worldwide in 2012 alone.1 Several molecular subtypes of breast cancer have been identified: triple-negative breast cancers (TNBCs) exhibit an absence of oestrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2) protein expression, and are generally associated with aggressive clinical behaviour, poor prognosis and a heterogeneous molecular profile.2–4 Oncological management options for this group of tumours are limited, and no reliable prognostic markers have been identified to aid clinical decision-making. There is an urgent need to understand the mechanisms that propel tumour cell growth of this group of neoplasms in order to develop specific new strategies for early diagnosis, prognosis and treatment.

Further complicating the clinical landscape is the increasing appreciation of the role of ethnicity on cancer incidence, histopathology and prognosis.5–10 In the case of breast cancer, Asian women may be subjected to different risk factors owing to genetics or lifestyles. These same factors may also affect responses to treatment and disease outcomes. Of note, women of non-Caucasian backgrounds have been insufficiently studied, or studied only as a minor proportion of a Western population. Caveolin-1 has recently been identified as a promising biomarker that is directly involved in the pathogenesis of prostate, colorectal, tongue, pancreatic, gastric and other cancers.11–21 Caveolin-1 is the principal structural component of caveolae, which are specialised lipid rafts on the plasma membrane of terminally differentiated cell types that function as membrane-organising centres involved in signal transduction, cholesterol transport, mechano-sensing and clathrin-independent endocytosis.22 23

In the normal breast, caveolin-1 is expressed in a subset of epithelial and mesenchymal cells, including myoepithelial cells of the ductolobular system, endothelial cells, fibroblasts and adipocytes of the breast stroma.24–26 The human caveolin-1 gene maps to a fragile chromosomal region at 7a31.1/2 that is frequently deleted in human breast carcinomas.27 However, some studies have suggested that caveolin-1 is an oncogene.27–30 In breast cancer, loss of stromal caveolin-1 expression seems to be associated with poor prognosis,25 31 32 though no large studies have examined its prognostic significance in TNBC, or in ethnically diverse Asian populations.

Here we measured the intensity and frequency of caveolin-1 protein expression in both tumour and stromal cells in TNBC samples from a mixed-ethnicity Asian population in Singapore, correlating the results with clinicopathological parameters and disease outcomes.

Materials and methods

Patients and tumours

The study used archival specimens from a cohort of 699 patients with TNBC diagnosed at the Department of Anatomical Pathology, Singapore General Hospital between 1994 and 2010 (CIRB ref: 2011/433/F). Clinicopathological parameters, including age, ethnicity, tumour size, histological grade, histological subtype, associated ductal carcinoma in situ, lymphovascular invasion, axillary lymph node status, tumour borders and growth patterns, were reviewed.

Tissue microarray (TMA) construction

Histological slides were retrieved and reviewed. Two representative tumour areas of each formalin-fixed, paraffin-embedded (FFPE) tissue block were constructed into TMAs using a Beecher microarrayer with 1 mm punch, two cores per case.

Immunohistochemistry (IHC)

TMA sections (4 µm) were subjected to IHC using antibodies shown in table 1, according to published protocols.33 IHC for ER, PR and HER2 was conducted to reconfirm triple negativity; epidermal growth factor receptor (EGFR) and cytokeratins (CK14 and high molecular weight clone 34βE12) were used to detect basal-like phenotype.2 To define basal-like TNBC we applied a tri-panel of CK14, EGFR and 34βE12, with any positivity indicating a basal phenotype. This panel of markers achieved 100% specificity and 78% sensitivity as documented in our previous study.34 Presence of cytoplasmic decoration of CK14 and 34βE12, cytoplasmic membrane reactivity of EGFR in at least 1% of tumour cells, were regarded as positive immunohistochemical expression.35 Positive controls were included for each IHC run. Labelling percentage (%) and intensity (0: no labelling, 1+: weak, 2+: moderate, 3+: strong labelling) were scored by two observers (MI and AAT) independently and blinded to the clinicopathological information.36 An immunoreactive score (H-score) was calculated by the formula: (1 × % of weak labelling)+(2 × % moderate labelling)+(3 × % strong labelling). The higher immunoscore was selected from duplicate cores for data analysis. Positivity for ER, PR, CK14, EGFR, and 34βE12 was defined as the presence of ≥1% of tumour cells labelled.37 38 HER2-positive expression was defined as >10% of tumour cells exhibiting 3+ membrane labelling.39 H-scores were calculated for caveolin-1 expression and H-scores of ≥50 were regarded as positive for either membranous and/or cytoplasmic staining of stromal cells. For tumour cells we recorded membranous and cytoplasmic staining separately with the same H-scores as cut-off point (≥50) (table 2). In the stroma, we observed staining only on stromal cells but not immune cells. As scoring was conducted on TMA, the stroma adjacent to the tumour was evaluated.

Table 1

Details of antibodies used for immunohistochemistry (IHC) of triple-negative breast cancer sections

Table 2

Caveolin-1 immunohistochemical expression in different cellular localisations (n=699)

RNA extraction, NanoString gene expression analysis

One hundred and forty-one TNBCs were subjected to caveolin-1 (CAV1) gene expression analysis. Unlabelled FFPE standard sections (10 µm) were subjected to RNA extraction using the RNeasy FFPE kit (Qiagen, Hilden, Germany) on a QIAcube automated sample preparation system (Qiagen, Hilden, Germany) and was quantified by an Agilent 2100 bioanalyzer system (Agilent, Santa Clara, California, USA). Functional RNA (100 ng; >300 nucleotides) was assayed on an nCounter Custom CodeSet (NanoString Technologies, Seattle, Washington, USA). NanoString counts were normalised using the positive control probes and the housekeeping genes shown in online supplementary table 1.

Supplementary file 1


Follow-up and statistical analysis

Follow-up data were obtained from medical records of the entire cohort. Disease-free and overall survival (DFS; OS) were defined as time from diagnosis to recurrence, or death/date of last follow-up, respectively. Statistical analysis was performed using SPSS for Windows, version 18. The relationship between clinicopathological parameters and immunohistochemical expression of caveolin-1 was tested using χ² and Fisher’s exact tests. Survival outcomes were estimated by Kaplan–Meier analysis and compared between groups with log-rank statistics. Multivariate Cox regression analysis was carried out to evaluate the effect of various cellular localisations of caveolin-1 protein and CAV1 mRNA expression levels on survival. Multivariate analysis was done with adjustment of patient age, tumour size, histological grade and lymph node status. A p value <0.05 defined statistical significance.


Stromal caveolin-1 protein expression achieved H-scores ≥50, and therefore defined as positive, in 101 (14.4%) cases, while cytoplasmic and membranous expression of caveolin-1 in tumour cells was present in 20 (2.9%) and 21 (3.0%) cases, respectively (table 2). Representative images of labelled sections on the TMA are shown in figure 1.

Figure 1

A) Caveolin-1 shows expression in myoepithelial cells of the ductolobular system, endothelial cells and fibroblasts of the normal breast. (B) Triple-negative breast cancer (TNBC) with positive caveolin-1 stromal expression mainly on fibroblasts. (C) TNBC with membranous expression of caveolin-1 on tumour cells. (D) TNBC with cytoplasmic expression of caveolin-1 on tumour cells.

In this study, 593 (85%) tumours had the basal-like phenotype, assessed by immunohistochemical staining for CK14, EGFR and 34βE12. The clinicopathological characteristics of this cohort have been previously reported.2 Positive stromal caveolin-1 expression did not vary with differences in age, ethnicity, tumour size, association with ductal carcinoma in situ, grade, histological subtype, tumour border or lymph node stage (table 3). In contrast, significant associations were found between stromal caveolin-1 positivity and the trabecular growth pattern (p=0.007) (table 3), as well as with the basal-like TNBC phenotype (p=0.028) (table 3). Positivity of caveolin-1 in tumour cells showed no significant association with all clinicopathological parameters or with the basal-like phenotype.

Table 3

Correlation between clinicopathological characteristics and caveolin-1 stromal expression

Follow-up of this cohort ranged from 1 to 213 months (mean 101, median 97), with recurrence and death occurring in 29% and 24% of these women, respectively. We found no evidence that DFS was significantly improved for women with stromal caveolin-1-positive TNBC (p=0.468).

However, Kaplan-Meier analysis showed that patients with tumours expressing caveolin-1 in the stromal compartment exhibited significantly longer OS in both TNBCs and basal-like TNBCs (p=0.009 and p=0.026, respectively) (figure 2). This was further confirmed by multivariate Cox regression analysis (table 4, see online supplementary table 3); the presence of stromal caveolin-1 expression predicted better outcome in all TNBCs as well as in the basal-like subtype (HR=0.484, 95% CI 0.261 to 0.897, p=0.021 and HR=0.535, 95% CI 0.287 to 0.999, p=0.049, respectively). Caveolin-1 expression on tumour cells was not associated with prognosis, regardless of whether caveolin-1 was expressed on the cytoplasmic membrane or within the cytoplasm (TNBC p=0.855, p=0.770; basal-like TNBC p=0.636, p=0.688).

Table 4

Association of caveolin-1 protein expression and clinicopathological parameters with overall survival in triple-negative breast cancers (TNBCs) and basal-like TNBCs

Figure 2

A) Women with stromal caveolin-1-positive triple-negative breast cancer (TNBC) had significantly better overall survival (p=0.009) (n=699). (B) Women with stromal caveolin-1-positive basal-like TNBC had significantly better overall survival (p=0.026) (n=593).

To understand the quantitative nature of the relationship between caveolin expression in TNBCs and survival outcomes, we used a NanoString assay40 41 to measure expression of caveolin-1 at the transcription level (CAV1) in 141 TNBC samples and then compared the transcript abundance with length of survival. We found that every incremental 1 unit of NanoString count was associated with better OS (HR=0.84, 95% CI 0.74 to 0.84) (table 5). As expected, the mRNA level of CAV1 correlated with positivity with stromal, tumour cytoplasmic and tumour membranous caveolin-1 protein expression by IHC (see online supplementary table 2).

Table 5

Multivariate analysis of the association between the mRNA levels of caveolin-1 (CAV1) and overall survival in patients with triple-negative breast cancer


Effective clinical management of TNBC is hampered by the lack of robust and reliable prognostic biomarkers; this situation is exacerbated in populations for which few data are available, including Asian women. Here we confirmed and extended previous studies in populations of predominantly Caucasian women living in the West, and concluded that stromal levels of caveolin-1 protein and CAV1 mRNA are potential prognostic markers in TNBC capable of predicting Asian women’s clinical outcome.

As shown in table 6, several studies, including a meta-analysis, have suggested that low stromal caveolin-1 is correlated with poorer prognosis in patients with breast cancer.25 31 32 42–46 Of these, only two have included a sizeable cohort of TNBC samples, and both were conducted in small ethnically homogeneous cohorts of women (Korean and Caucasian). Using a 1% positive cut-off point, Koo et al observed that 91.7% of 144 Korean TNBC cases expressed stromal caveolin-1, while only 5.6% of cases were positive based on tumour cell expression of caveolin.25 In another small study, Witkiewicz et al found that 28.9% of their 83 Caucasian TNBC samples exhibited high-intensity stromal caveolin-1 expression, and that low/absent stromal caveolin-1, but not tumour caveolin-1, predicted poorer DFS and OS, and was significantly associated with lymph node metastasis.31 In our cohort, 14.4% of TNBC samples were defined as positive based on stromal expression, using the H-score system, which takes both staining percentage and intensity into consideration. This approach, alongside published studies, allows a broad consensus to emerge, suggesting that stromal caveolin-1 expression is strongly and independently associated with OS in TNBC, irrespective of the method used to measure expression, patient ethnicity and geographical region. However, the association between caveolin-1 expression and DFS and lymph node status reported by others,25 31 and that between caveolin-1 with tumour growth pattern in our study, appear to vary depending on experimental approach and/or ethnicity/geographical region. In addition, with the Nanostring assay,40 41 which is a technology that generates more accurate and easily reproducible results, our study also found that CAV1 is correlated with the caveolin-1 protein and that the loss of either mRNA or protein level is associated with worse OS. Thus measurement of mRNA expression of CAV1 in TNBC samples may prove an easily scalable approach to applying these findings as part of a prognostic panel to refine clinical decision-making in TNBC.

Table 6

Association of stromal caveolin-1 expression and survival rate in breast cancers

Caveolin-1 has attracted much attention in recent years for its potential role in tumour progression. Expression of stromal caveolin-1 protein has been associated with rapid progression to invasive ductal carcinoma from ductal carcinoma in situ,25 31 with similar findings in inflammatory and metaplastic breast cancers.29 30 However, some studies strongly suggest that caveolin-1 is a tumour suppressor, and in particular, that stromal expression of caveolin-1 positively correlates with better prognosis.25 31 32 42–46 In summary, the picture remains unclear, and so one might speculate that caveolin-1 can play variable roles dependent on a range of tumour-intrinsic and -extrinsic factors. Taken together, caveolin-1 appears to be a promising prognostic biomarker, but its use must be underpinned by consistent methods of evaluation, and limited to tumours and populations in which robust data exist to support its role.

Caveolin-1 plays important homoeostatic roles in a range of tissue and cell types, and therefore it is perhaps unsurprising that it should also be involved in cancer. Recent years have seen an increasing appreciation of the role of the tumour microenvironment in regulating neoplastic growth and progression, and our data highlighting the significance of stromal caveolin-1 expression, over and above tumour cell expression of caveolin-1, support this view. Stromal activation is a critical step for tumour formation, and the emergence of the cancer-associated fibroblast (CAF) phenotype, which can be defined by the level of caveolin-1 expression,47 48 has been proposed to lead to tumour progression through promoting neoangiogenesis, tumour cell proliferation and tumour cell invasion.49–51 Recent data have now linked the loss of normal caveolin-1 expression with induction of the CAF phenotype52; increased growth factor secretion by mammary fibroblasts53; AKT phosphorylation and transforming growth factor (TGF)-β1 expression and subsequently acceleration of the development of aggressive cancer.54 The secretion and expression of TGF-β by CAF have been linked to cancer progression in multiple studies.55 56 Using the same cohort, we had previously reported that women with TNBC expressing cytoplasmic TGF-β showed poorer metastasis-free survival but which was without statistical significance.57 Recent studies have shown that CAV1 is involved in tumour metabolic alterations—a pro-survival strategy adopted by cancer cells. Based on evidence from literature, Nwosu et al have proposed that CAV1 modulates cancer metabolism by (a) enabling the expression of membrane-localised metabolic enzymes, (b) coordinating a switch between metabolic pathways (c) relaying signals for transcriptional activation or suppression of metabolic regulators, or inhibiting antimetabolic stimuli.58

In summary, our study demonstrated that low or absent CAV1 mRNA level and caveolin-1 protein expression on stromal cells in TNBC are significantly associated with worse clinical outcome in a large ethnically diverse cohort of Asian women in Singapore. These data may facilitate the development of caveolin-1 expression into a reliable prognostic marker in TNBC to enable optimised clinical management of this hard-to-treat cancer.

(2220 words)

Take home messages

  • Loss of caveolin-1 in the stromal compartment is an independent prognostic indicator, associated with significantly worse overall survival in triple-negative breast cancer (TNBC). However, caveolin-1 protein expression in the tumour compartment is not associated with clinical outcome.

  • Transcript abundance of caveolin-1 (CAV1) is a continuous measure that is linked with better overall survival in these patients.

  • To our knowledge, this is the largest study of the role of caveolin-1 in TNBC on a purely Asian population.


We thank Dr Huihua Li from the Department of Clinical Research, Singapore General Hospital for her advice on statistical analysis.



  • JY and AAT contributed equally.

  • Handling editor Tahir S Pillay.

  • Contributors The study was designed and directed by PHT and coordinated by JY. JY, AAT, JCTL and MI acquired the data. The analysis was done by AAT and BL; SN and JI provided advice and guidance from breast pathology and clinical perspectives. JY, JCTL and AAT drafted the manuscript, which was commented on and revised by all authors.

  • Funding This article was funded by the A*STAR Biomedical Research Council, National Medical Research Council Stratified Medicine Programme Office (SMPO201302) awarded to Dr PH Tan. Dr J Iqbal is a recipient of the Transition Award from the Singapore National Medical Research Council (NMRC/TA/0041/2015).

  • Competing interests None declared.

  • Ethics approval SingHealth centralised institutional review board.

  • Provenance and peer review Not commissioned; externally peer reviewed.