Identification of different subtypes of breast cancer

Published: Last Edited:

This essay has been submitted by a student. This is not an example of the work written by our professional essay writers.

Breast cancer may be classified based on gene expression profiling or immunohistochemical (IHC) characteristics. The aim of the study was to classify into molecular subtypes. Two hundred and seventeen (217) primary breast carcinomas cases were immunostained for ER, PR, HER2, CK5/6, EGFR, CK8/18, p53 and Ki-67 using tissue microarray technique. HER2 expression evaluated by IHC was confirmed by fluorescence in situ hybridization (FISH). A panel of five markers (ER, PR, HER2, CK5/6 and EGFR) were used to classify cases into luminal A, luminal B, HER2 overexpressing, basal-like and normal breast-like subtypes.The luminal A (53.4%) was the most common IHC-subtypes, followed in descending order by HER2+/ER- (14.7%), normal-like (13.4%), basal-like (7.4%) and luminal B (6.9%) subtypes. Luminal A was associated with smaller mean tumour size (p=0.032), mean number of lymph nodes positive (p=0.032), low histological grade (p=0.000) and p53 negativity (p=0.003). Luminal B, HER2 over expressing, normal breast-like, and basal-like subtypes were significantly associated with race of Malay origin (p=0.035), bigger mean tumour size and mean number of lymph node positive, high histological grade (p=0.000) and p53 positivity (p=0.003). The expression of CK8/18 was seen in more than 50 % of all subtypes (p=0.002). Majority of luminal B subtype was Ki67 positive. In conclusion, IHC can be useful to identify different subtypes and luminal A was the most common subtypes. This study demonstrated that all subtypes were only significantly associated with race, mean tumour size, mean number of lymph node positive, histological grade and all markers except EGFR and Ki67.

Key words: Immunohistochemistry; estrogen receptor; progesterone receptor; HER2; keratins.


Breast cancer is heterogeneous diseases that may respond to therapy and give prognosis differently despite similarities in histological types, grade and also stage. One factor that may influence breast cancers differences is the cell of origin. Normal human mammary gland epithelium characterized by ductal glandular or luminal, basal myoepithelial and mammary gland stem/progenitor cells. These mammary gland stem/progenitor cells expressed Ck5 besides capable of forming glandular epithelial/luminal and myoepithelial end cells/basal (Böcker et al. 2002).

From c-DNA microarray studies, gene expression profile divided breast carcinoma into two either expressing ER (ER +) form luminal subtype or less express ER (ER-) which can be divided into three subtypes; over expressing HER2, basal-like and normal-like (Sorlie et al. 2001; Nielsen et al. 2004). This ER+/luminal-like was further divided into at least 2 distinct subgroups; luminal subtype A tumours which demonstrated the highest expression of the ER α gene and others while some luminal subtype B tumours do express ERBB2 and also express gene associated with ERBB2 such as GRB7 (Sorlie et al. 2001; Cheang et al. 2009).

However, using c-DNA microarray techniques required high cost renders which unsuitable in clinical practice. Nevertheless, a c-DNA microarray technique is still the gold standard for classification of breast carcinoma. Validation gene expression (mRNA) can be done by using immunohistochemistry techniques that can detect overexpression of protein. Immunohistochemistry also demonstrated good and acceptable surrogate of the gene analysis (Nielsen et al. 2004; Bertucci et al. 2005; Carey et al. 2006). Therefore, other researchers used IHC techniques co-responding with TMA for classified breast carcinoma (Nielsen et al. 2004; Banerjee et al. 2006; Diallo-Danebrock et al. 2007; Cheang et al. 2008).

Various definitions can be used to classify breast carcinoma. However, definition of breast carcinoma subtypes by IHC techniques especially basal- like was not established yet. Some studies used three molecular markers of ER, PR and HER2 to classify breast carcinoma into four subtypes which basal-like and normal-like were include in one group also known as triple negative (Carey et al. 2007; Cheang et al. 2008). Basal-like and normal-like subtypes were found significantly related with triple negative tumour (Pintens et al. 2009). The use of three molecular markers was practical as they were commonly used for a routine laboratory testing and also important markers to determine treatment for breast carcinoma patients.

Tan (2008) found that majority triple negative tumour markers also expressed basal marker such as CK5/6, CK14 and CK17. This explains that triple negative is heterogeneous tumour and not a single entity (Tan et al. 2008). Several molecular markers such as basal CK5 / 6 and CK14 or molecular markers such as vimentin and p63 were also used to define basal-like subtype by expressing one or a combination of these markers (Banerjee et al. 2006; Kim et al. 2006; Livasy et al. 2006). It's parallel with gene expression studies of basal-like subtypes which showed high gene expression of CK5, CK17, vimentin, EGFR and c-kit (Sorlie et al. 2001; Nielsen et al. 2004; Diallo-Danebrock et al. 2007).

However, combination of these two entities which were triple negative tumour and also tumour expressed molecular markers especially CK5/6 and EGFR were more corresponding with gene expression profile using c-DNA microarray and yet better predicts for breast cancer survival (Cheang et al. 2008).

In 2006, breast cancer was the most common cancer among female and also the most important cancer among population regardless of sex in Peninsular Malaysia (Zainal et al. 2006). The aim of the study is to examine and identify basal subtypes of breast carcinomas in Malaysia population by immunohistochemistry using tissue microarray.

Material and Methods:


Two hundred and seventeen (217) patients confirmed with invasive breast carcinoma were derived from 2 hospitals, namely: Hospital Universiti Kebangsaan Malaysia and Hospital Putrajaya. There were 213 (98.2%) female and 4 male patients. Breast cancers were classified according to the World Health Organization (WHO 2003)(Tavassoli and Devilee 2003) while histologic grading was performed according to Modified Bloom-Richardson classification (Elston and Ellis 1991).

Tissue Microarray Construction

A tissue microarray of 217 breast carcinomas was constructed into four blocks from which a total of 31 tissue array blocks were built. These H & E stained sections were made from each original block to define representative tumour regions and to determine the spots that are suitable before encircling with a marker.

As guided by marked H&E stained slide, cores from the selected area of the donor blocks were punched with 0.6 mm diameter needle using manual MTABooster from Alphelys (Plaisir, France) and inserted into new paraffin block (recipient block) with 1.8 mm spacing between the cores at defined array coordinates. Two samples were taken from the centre and the other two were taken from the periphery of the tumour. When the TMA block was completed, the block was heated at 60°C for approximately 5 minutes which was able to melt the wax thus closing the small gap between the inserted cores.

Immunohistochemical staining:

The immunohistochemical staining was performed on tissue microarray and was done according to the manufacturer's recommended protocol with minor modifications and has been described at previous paper (Noranizah et al. 2010). The summarized of primary antibodies including their clone, dilution, and antigen retrieval were shown in Table 1.

The positive control slides were prepared each from different blocks of breast cancer (ER, PR, HER2 & CK8/18), adenocarcinoma colon (p53), tonsil (CK5/6 & Ki-67) and plasenta (EGFR). The negative control slides were prepared from the same tissue blocks but TBS buffer was used instead of the primary antibody.

Evaluation of immunohistochemistry

ER, PR, p53 and Ki67/ MIB-1 stained brown in the nucleus of the tumour cells. Oestrogen receptor, progesterone receptors, p53 and Ki67/ MIB-1 were evaluated as positive if ≥ 10 % of the cells stained nuclear positive [(Banerjee et al. 2006); (Kim et al. 2006; Diallo-Danebrock et al. 2007). Membranous staining was scored for HER2 as follows : 0, no staining or faint incomplete staining in <10% cells; 1+, faint incomplete staining in >10% cells; 2+, weak to moderate complete staining in >10% cells; 3+, strong complete staining in >30% cells (Wolff et al. 2007). Cases scoring 3+ were regarded as HER2 positive and the rest (0 and 1+) were HER2 negative. Cases scoring 2+ were regarded as equivocal and were subsequently assessed by Fluorescence in situ hybridization (FISH) (Cheang et al. 2008).

EGFR staining was considered positive if ≥ 10 % membrane staining of tumour cells was observed (Kim et al. 2006). A positive score for CK5/6 and CK8/18 is recorded if any cytoplasmic and/or membrane staining of any invasive malignant cells is present (Potemski et al. 2005)

Fluorescence in situ hybridization (FISH) technique

FISH was performed using PathVysis HER2 DNA Probe kit (Abbott Molecular, Canada) according to the manufacturer's instructions and with minor modification. Tissue whole section were cut into 3μm thick and adhered on silanized slide. The slides were baked overnight at 60°C, deparaffinized and pre-treatment with Skip-dewax (1:10, Dako, Glostrup, UK) for one hour. Then, slides washed in distilled water for 2 minutes each before incubated in Protease (Vysis Abbott Molecular, Canada) for 50 minutes at 37°C with agitation. Slides were washed twice in 2x sodium chloride/sodium citrate (SSC) at room temperature, 5 minutes each and then, allowed to air dry. Slides were incubated with DNA probe (LSI HER2/neu SpectrumOrange/ CEp17 SpectrumGreen, PathVysis, Abbott Molecular, Canada) overnight at 37°C after denatured for 15 minutes at 90°C. After hybridization, slide were washed for 2 minutes with agitation in 0.4 SSC/0.3%NP40 at 74°C and followed by 2SSC 0.1%NP40 at room temperature. Slides were allowed to air dry before counterstain with diamidino-2'-phenylindole (DAPI, PathVysis HER2 DNA Probe kit, Abbott Molecular, Canada and stored at 4°C in dark.

Evaluation of FISH

Slides were evaluated using fluorescence microscope. The number of LSI HER2/ neu in red signal and CEP17 in green signal per nucleus were recorded. Each case was counted 60 nuclei. Case was scored as amplified when mean ratio between LSI HER2/ neu and CEP17 was greater than 2.0 (Cheang et al. 2008)

Definition of breast cancer subtypes

The tumours were classified into molecular subtypes and according to immunohistochemical expression profiles of ER, HER2, EGFR, c-KIT and CK5/6. The cases were classified into molecular subtypes of luminal A (ER + and/or PR+, HER2-), luminal B (ER+ and/or PR+, HER2+), human epidermal growth factor receptor-2 (HER2) overexpressing (ER-, PR-, HER2+), basal-like (ER-, PR-, HER2-, CK5/6+ and/or EGFR+) and normal-like tumours (negative for all 5 markers) (Nielsen et al. 2004; Carey et al. 2006)).

The basal-like subtypes were examined on whole paraffin sections of breast carcinomas stained by hematoxylin & eosin for pathologic evaluation. All samples were examined by two pathologists (SA, RMZ)

Statistical analysis

All statistical analyses were performed using SPSS version 12. Analysis chi square Pearson was used to investigate associated between categorical variables. A p value of less than 0.05 was regarded as statistically significant.


Patients, tumour subtypes and tumour characteristics

The mean age at diagnosis of patients was 53.1 years, ranging from 27 to 87 years old. The patients were mostly above 50 years old (53.9%), postmenopausal (58.2%), Malay (59.0%), larger than 2 cm (76.1%), positive lymph nodes (53.9%), IDC (85.7%) and grade 2 (44.2%). Below 15.0% of patients had recurrent (10.6%) and died (13.4%) during 93 months of follow up.

Table 2 shows the tumour subtypes and demographic data breast cancer patients. There was a total of 217 cases, the luminal A (57.6%) was the most common IHC-subtypes, followed in descending order by normal-like (13.8%), HER2+/ER- (14.3%), basal-like (7.4%) and luminal B (6.9%) subtypes.

Chi square analysis showed a significant relationship between subtypes with races (0.035), mean size of tumor (0.009), mean number of positive lymph nodes (p=0.032) and histological grade (p < 0000) [See Table 2]. There were no significant relationship between subtypes with age (p = 0.189), menopausal status (p = 0.669), tumor size (p = 0069), lymph nodes (p = 0.079), recurrent (p=0.242), tumor stage (p = 0.242) and patients status (p=0.749)

However, luminal A had the oldest mean age (55.1 years) and the highest percentage of patients (60.8%) above 50 years old as compared with other subtypes. Luminal A (62.4%), HER2+ (56.0%) and basal-like (53.8%) subtypes were seen mainly in the postmenopausal women. In this study, the large racial distribution was Malay (59.0%) followed by Chinese (31.8%) and others (9.2%) including Indians. The scenario also showed in all subtypes except for luminal B which had similar percentage (6.7%) for both Chinese and others respectively. However, slightly different was showed for luminal A subtype in Malays and Chinese patients comprising 62 (49.6%) and 51 (40.8%) respectively. In other subtypes, more than 60% were Malays and below 30% were Chinese patients.

The size of the tumor ranged between 0.5 to 24.0 cm (mean = 4.2 ± 3.1 cm) and the largest mean size (6.2 ±5.1; range 1.4 - 24.0) was seen in HER2+ subtype (p=0.009). All the subtypes were larger than 2 cm (76.1%) and the tumour sizes ranged from 2.1 to 5cm. The mean lymph node was significant largest in luminal B subtypes 5.4±6.5 (0-17) and smallest in luminal A subtype 2.6±4.7 (0-21) (p=0.032). Majority of luminal A subtype was lymph node negative contrary with other subtypes albeit no significant were found.

The most common histological types were infiltrating ductal carcinomas (IDC), not otherwise specified (NOS) comprising 186 (85.7%) cases while the rest in descending orders were 11 (5.1%) infiltrating lobular carcinomas, 6 (2.8%) infiltrating mucinous carcinoma, 5 (2.3%) mixed ductal & lobular carcinomas, 4 (1.8%) infiltrating cribriform carcinoma, 3 (1.4%) metaplastic carcinomas and 2 (0.9%) invasive papillary carcinomas. Luminal A, luminal B, HER2+, basal-like and normal-like subtypes consisted of 83.2%, 86.7%, 100%, 81.3% and 83.3% of IDC, NOS of the cases respectively. The other special types were mainly seen in luminal A subtype.

Majority (44.2%; 96/217) of the patients were in Grade 2, and significantly followed by Grade 3 (29.5%; 64/217) and Grade 1 (26.3%; 57/217). The luminal A and luminal B subtypes were 46.4% (58/105) and 60.0% (9/15) respectively in Grade 2 while HER2+ (48.4%; 15/31), normal-like (50.0%; 15/30) and basal-like subtype (56.3%; 8/16) were mainly in grade 3 (p=0.000). Among the subtypes, recurrence was mainly seen in luminal B subtype (4/15; 26.7%). Stage I comprised of 7.4% (16/217) of all cases and was seen in all subtypes except HER2+ and basal-like but was not significant statistically (p=0.212).

Immunohistochemical results

(ER, PR, HER2, CK5/6, EGFR, c-KIT, CK8/18, p53, Ki67 immunostaining )

Table 3 shown immunohistochemistry results in different subtypes of invasive breast carcinomas whereas Figure 1. shown representative of immunostaining of molecular markers in breast carcinoma by tissue microarray. This table shows positive frequency of immunohistochemical expression of ER, PR, HER2, CK5/6, EGFR, CK8/18, p53 and Ki67 was 55.8%, 41.9%, 21.2%, 17.8%, 1.5%, 80.5%, 55.3% and 41.8% respectively. In this study, all markers except EGFR and Ki67 were significantly associated with subtypes.

The ER was expressed significantly 55.8% (120/217) (Fig. 1B) and was ER negative in 44.2% (95/217) of all cases. The ER was mainly expressed in 90.3% (112/125) in luminal A and 53.3% (8/15) in luminal B of all cases (p=0.000). ER positive was not seen in HER2+, basal-like and normal-like subtypes but was 100.0% ER negative in these subtypes.(p=0.000). PR negative was seen in 58.1% (126/217) of all cases and present in 100.0% of HER2+, basal-like and normal-like subtypes (p=0.000). PR was expressed in 64.8% (81/125) in luminal A (Fig. 1C) and 73.3% (11/15) in luminal B subtypes of all cases

HER2+ was expressed in 21.2% (46/217) of all cases (Fig. 1D) and expressed in Luminal B (100%; 15/15), HER2+ (100.0%; 32/32) but not in luminal A (100.0%; 125/125), basal-like (100.0%; 16/16) and normal-like subtypes (100.0%; 29/29) (p=0.000). CK5/6 was positive in 17.8% (38/217) of all the cases (Fig. 1E) and seen in 100.0% (16/16), 18.7% (6/31), 12.2% (15/125), 7.1% (1/15) of Basal-like, HER2+, Luminal A and Luminal B subtypes respectively (p=0.000).

EGFR (Fig. 1F) was rarely seen in all subtypes and was not statistically significant (p=0.426). CK8/18 positive was 80.5% (171/217) of all cases (Fig. 1G) and was seen in all subtypes (p=0.009). P53 was expressed in 55.3% (120/217) of all cases (Fig. 1H) and was seen in 87.5% (14/16) and in descending order 80.0% (12/115), 56.2% (18/32), 51.7% (15/29), 48.8% (61/125) of Basal-like, Luminal B, HER2+, Normal-like and Luminal A respectively (p=0.013). 43.0% (89/217) expressed Ki67 (Fig. 1I) and Luminal B showed high expression of 66.7% in descending order followed by Basal-like (57.1%; 8/16), HER2+ (41.9%; 13/32), Luminal A (40.0%; 48/125) and Normal-like 37.0%; 17/29) subtypes (p=0.250).

Basal-like subtype:

The mean age of basal-like tumours was 48.3 years and the youngest was 27 years old. The majority affected was the Malays (p=0.035) and 11 patients (68.8%) were postmenopausal (p=0.669). It was significantly associated with bigger size [mean tumour size=4.7 cm; p=0.009] and more lymph node positive [mean number of lymph node positive=4.7; p=0.032] compared to luminal A [mean size=3.6cm; mean number of lymph node positive=2.6]. Majority (81.3%; 13/16) of basal-like tumours were infiltrating ductal carcinomas of no special type, 2 (12.5%) were metaplastic carcinoma and one (6.3%) was mixed ductal and lobular carcinoma. Nine (9) of the cases (56.3%; 9/16) were grade 3, 6 cases (37.5%) were grade 2 and 1 case (6.3%) was grade 1 and it was significant (p<0.0000). Only two (2/16; 12.5%) recurrence was seen in this subtype and 7 patients were in Stage II (7/16; 43.8%) and alive (14/16; 87.5%). Positive expression of basal CK 5/6 (100%; 16/16) and was statistically significant (p=0.000) and only one case expressed EGFR (7.7%, 1/16) and 9 (56.3%; 9/16) cases in Ki67 but was not significant (p=0.438; p=0.405; p=0.117 respectively). CK8/18 (68.7%; 11/16) and p53 (93.8%; 15/16) were expressed in this subtype and were statistically significant (p=0.002; p=0.003 respectively). Majority of basal-like subtypes showed necrosis [10/16(68.75)] (Fig. 2A), margin infiltrative [13/16 (81.3%)] and scanty to moderate inflammation [8/16 (50.0%)] at tumours. But only a few cases demonstrated solid pattern [5/16 (31.25%)], margin/pushing [3/16 (18.75)] (Fig. 2B), central acellular scar [1/16 (6.3%)] (fig. 2C) and focal squamous differentiation [2/16 (12.5%)] (fig. 2D).


In our study, the luminal A (57.6%) was the most common IHC-subtypes, followed in descending order by HER2+/ER- (14.3%), normal-like (13.8%), basal-like (7.4%) and luminal B (6.9%) subtypes. Other studies found that luminal A cases was from 27% up to 73.4% whereas other subtypes were between 5.2% to 32% (Carey et al. 2007; Diallo-Danebrock et al. 2007; Tamimi et al. 2008; Spitale et al. 2009).

There were no significant associated between breast carcinoma subtypes with age, menopausal status, tumour size, lymph nodes status and stage except race and histological grade found in this study (Rouzier et al. 2005; Kim et al. 2006; Carey et al. 2007; Yang et al. 2007; Nofech-Mozes et al. 2009; Spitale et al. 2009). Despite that, several studies found significant relationship between subtypes with age, menopausal status, size tumour the status of lymph nodes, histological grade and stage (Banerjee et al. 2006; Carey et al. 2006; Carey et al. 2007; Nofech-Mozes et al. 2009). In this study, luminal A cases were usually older in age and occur in women postmenopausal (Carey et al. 2006; Nofech-Mozes et al. 2009). In addition, luminal A subtype also was less frequent among Malay patients compared with other subtypes.

A population-based study from the South of Switzerland showed that luminal A subtype had the highest proportion of patients aged 70 or over (35.5%) and HER2/neu was more frequent in postmenopausal women (Spitale et al. 2009). While, the Carolina Breast Cancer Study showed that the luminal A subtype was less frequent among premenopausal African American women (36%) compared with post menopausal African American (59%) or non- African American (54%) women (Carey et al. 2006). Compared with luminal A subtype, patients with basal-like tumours were 2.1 times more likely to be African American (Carey et al. 2006).

Luminal A had better prognosis than other subtypes (Nielsen et al. 2004; Potemski et al. 2005; Spitale et al. 2009). In fact, basal-like and overexpression HER2 subtypes had a shorter DFS (disease-specific death rate) and OS (overall survivor) compared with luminal A (Potemski et al. 2005; Kim et al. 2006; Banerjee et al. 2006). From this study, basal-like and overexpression HER2 compared luminal subtypes more likely had large tumour size in line with other study (Spitale et al. 2009). Luminal A also had high percentage of tumour size less than 2 cm (Yang et al. 2007)).

Our study found that mean number of lymph nodes positive and percentages of lymph nodes positive were high in overexpression HER2 and / or luminal B subtypes (Carey et al. 2006; Kim et al. 2006; Yang et al. 2007; Tamimi et al. 2008). Consistent with other studies, low percentage of positive lymph nodes was seen at the basal-like subtype in this studies (Carey et al. 2006; Yang et al. 2007; Cheang et al. 2008; Spitale et al. 2009). Foulkes (2004) suggested that the favoured route of metastatic spread may differ for basal and non-basal breast cancers, because CK5/6-positive cancers are less likely than CK5/6-negative cancers to be node-positive when large in size, but are nevertheless associated with a worse outcome.

Prior to immunohistochemistry and expression-profiling, breast cancer defined mainly by histology and grade (Nofech-Mozes et al. 2009). The most common histological type of breast carcinoma is invasive ductal carcinoma (Li et al. 2005). Thus, most subtypes were also invasive ductal carcinoma which all cases of overexpression HER2 subtype represent only this histological type whereas other subtypes represent with other histological types. Patients' diagnosis with IDC and mixed lobular and ductal not only had severe prognosis, but likely recurrence compared with medulary, papillary or tubular types (Li et al. 2005). This study demonstrated that most metaplastic carcinoma represented in basal-like subtype similar with other studies (Kim et al. 2006).

BLCs have common morphologic features including marked cellular pleomorphism, high nuclear-cytoplasmic ratio, vesicular chromatin, prominent nucleoli, lack of tubule formation, high mitotic index, frequent apoptotic cells, scant stromal content, pushing invasion borders, central geographic or comedo-type necrosis, atypical medullary features, and stromal lymphocytic infiltrate (Jones et al. 2001; Livasy et al. 2006)) which also seen in this study. They are also characterised by the presence of metaplastic elements such as spindle cells and squamous cell metaplasia, presence of a central scar, glomeruloid microvascular proliferation and a stromal lymphocytic response (Tsuda et al. 2000; Foulkes et al. 2004)).

The majority of good histological grade (1 and 2) in this study as in other studies was luminal A and B whereas poor histological grade (3) was overexpression HER2 and basal-like subtypes (Rouzier et al. 2005; Carey et al. 2006)) (Yang et al. 2007) (Tamimi et al. 2008; Nofech-Mozes et al. 2009; Spitale et al. 2009). This was consistent with the progressive model of breast cancer, suggesting hormone receptor-negative and frequently positive for either HER2 or basal markers tumour progress towards histological grade 3 (Simpson et al. 2005).

Tumour stages were classified according to greatness of tumour size, involvement of lymph nodes and metastasis to other site. This study showed that no stage I was seen in basal-like and overexpression HER2 subtypes. Nevertheless, basal-like and overexpression HER2 were more found at stage III and IV similar to other studies (Spitale et al. 2009). Basal-like subtype most likely recurrent compared to other subtypes (Banerjee et al. 2006) (Carey et al. 2007).

Based on the pattern of protein expression, luminal subtypes classified into two luminal A and luminal B in which both expresses this luminal molecular markers such as ER, PR, BCL2, CK8 and CK8/18 (Matos et al. 2005; Diallo-Danebrock et al. 2007; Honrado et al. 2007) and low expression of basal molecular markers (CK5, CK5 / 6 and CK17) as well as other molecular markers (EGFR and vimentin) (Diallo-Danebrock et al. 2007; Honrado et al. 2007). BCL2 showed significant relationship with ER (Yang et al. 2007)).

Compared with luminal A, luminal B were less expressed ER, PR and BCL2 (Matos et al. 2005; Diallo-Danebrock et al. 2007; Honrado et al. 2007). This study showed that the percentage of BCL2, PR and ER were low compared with luminal A. luminal B subtype also expresses genes associated with HER2 such as ERBB2 and GRB7 (Cheang et al. 2009). Therefore, HER2 had being used to distinguish luminal A and luminal B. However, HER2 positive only covers 30% of luminal B consequently showed that using HER2 alone was not sufficiently sensitive to identify luminal B (Cheang et al. 2009). Luminal B expressed more Ki67 compared to luminal A although there was no associated between subtypes and Ki67 (Lerma et al. 2007). Cheang (2009) found that Ki67 together with ER, PR dan HER2 can be used to distinguish luminal A and luminal B.

Most protein profile for overexpression HER2 and basal-like subtypes were similar except for HER2 was expressed by overexpression HER2 subtype whereas basal molecular markers such as CK5, CK17, c-Kit, EGFR and vimentin were expressed by basal-like subtype (Diallo-Danebrock et al. 2007; Honrado et al. 2007). In addition, overexpression HER2 and basal-like subtypes were less expressed (negative) ER and PR (Diallo-Danebrock et al. 2007; Honrado et al. 2007) along with cDNA- mikcoarray studies (Sorlie et al. 2001; Nielsen et al. 2004).

Many studies found that basal molecular markers not only can be seen in basal-like subtypes (Matos et al. 2005; Kim et al. 2006; Livasy et al. 2006; Pintens et al. 2009) but also in other subtypes except normal-like. In this study, overexpression HER2 had express CK5 / 6 higher than the luminal subtype similar to other study (Pintens et al. 2009). Banerjee (2006)(Banerjee et al. 2006) found that basal-like expresses ER, PR and HER2 in 18.4%, 20.4% and 8.2% of cases, respectively. Basal cytokeratin expression showed up as two microscopically distinguishable subtypes, i.e., a uniformly positive type (''basal'') and a partially positive type (''basoluminal'') often displaying a checkerboard-type intratumoural heterogeneity (Laakso et al. 2006).

According to Laakso (2006)(Laakso et al. 2006), both basal and basoluminal subtypes were hormone receptor negative and of high grade, but differed with respect to the Ki-67 labeling index, vimentin and c-kit which were more frequently expressed in basal than in basoluminal tumours. In contrast, the amplification of HER2 was found almost exclusively in the basoluminal subgroup. Therefore, 19.4% of overexpression HER2 subtype which expressed CK5/6 but not ER and PR possibly was basoluminal subtype.

Normal-like or multiple marker negative (MMN) subtype gave negative result on the basal markers CK5 and CK7 as well as other molecular markers such as EGFR, c-kit, vimentin and Ki67 pengekspresan low (Diallo-Danebrock et al. 2007; Honrado et al. 2007). In addition, majority of normal-like subtype express CK8/18 without expresses CK5/6 and most likely these cells were derived from the luminal glands cell.

This study showed a significant relationship among all subtypes most probably basal-like with p53 (Bertucci et al. 2005). In prognosis tumour, TP53 gene mutations often show severe and less responsive to treatment of either chemotherapy or radiotherapy (Geisler et al. 2001). Basal-like and overexpression HER2 subtypes showed more mutation in p53 compared to luminal subtype (Sorlie et al. 2001; Carey et al. 2006). Overexpression of p53 also occurs in mutate gene either BRCA1 or BRCA2 (Greenblatt et al. 2001).

High BCL2 level was seen at luminal subtype while less on other subtypes. Freneaux (2000) found that less BCL2 found in BRCA1 mutations tumours. Overexpression CK8/18 can be viewed at all subtypes as a result it not suitable for specific luminal marker and less useful for this classification (Kim et al. 2006). Majority normal-like and basal-like subtypes do express luminal keratins (CK8/18), albeit at lower levels than those found in other subtypes(Livasy et al. 2006).

CK8/18 existence without basal markers in breast cancer was indicator showing tumour origin from luminal glands cell (Böcker et al. 2002)). Mammary gland lactation has the distinction complete TLDU and expresses CK8/18 not CK5 / 6 (Böcker et al. 2002). Hypothesis from Foulkes (2003)(Foulkes et al. 2003) stated that BRCA1wild type can act as stem cell regulator and to promote differentiation of gland epithelium in normal breast tissue. Therefore, tumours of BRCA1 mutation expresses less CK8/18 (Laakso et al. 2005).

Russo et. al (2001) suggests that breast cell in women who has BRCA1 mutation were failed to differentiate to acinus breast cell consequently increase the risk of developing breast cancer (Foulkes et al. 2003). Basal-like subtype showed profile of BRCA1 mutation which approximately low expresses ER, PR and HER2, while positive for p53 and basal markers such as CK5 / 6 or CK14 (Foulkes et al. 2003; Bertucci et al. 2005; Laakso et al. 2005). In addition, CK5 / 6 positive cases showed a tendency for more than five times (5x) to detect mutations in the BRCA1 (Foulkes et al. 2004). In this way, researchers can also perform IHC staining for BRCA1 mutation screening and detection. In contrast, most BRCA2 mutations occur in men and luminal subtype with high histological grade (Bane et al. 2007).

Conclusion:Overall, luminal A was the most common subtypes and likely in older, postmenopausal women, slightly smaller in size, negative lymph node, good histological grade and low expression of p53 and Ki67. Contrary, basal-like subtype most likely Malay, poor histological grade and high expression of p53 and Ki67. On the other hand, all subtypes showed high expression of CK8/18 thus this marker was not suitable to distinguish subtypes. However, only race, histological grade, CK8/18 and p53 along with ER, PR, HER2 and CK5/6 were associated with subtypes. This study found that majority of luminal B subtype was Ki67 positive even though no associated were found between them. Therefore, Ki67 may be used together with HER2 to distinguish luminal A and luminal B.