Invasive micropapillary carcinomas (IMPC) of the breast account for less than 2% of all breast cancers and have been recently described as luminal B carcinomas. CD24, CD44, ALDH1 and EZH2 are commonly used as stem-cell markers that display differential expression as a function of stage and molecular type, but their pattern of expression according to this rare histological type remains poorly defined and unknown for EZH2. We assessed expression of these markers in a series of 28 micropapillary breast carcinomas and compared the results with those obtained in a series of luminal A (27 cases) and B (34 cases) invasive carcinomas not otherwise specified (IC-NST). CD24 and CD44 were expressed in most cases. However, CD24 was expressed at the inverted apical membrane in 85% of invasive micropapillary carcinoma and at the apical pole of gland-forming cells in 45% of luminal A (p-val = 6.8 × 10-4) and 13% of luminal B cases (p-val = 1.1 × 10-7). ALDH1 was expressed in the stroma in most tumors, but in only 25%, 11% and 15% in epithelial cells of IMPC, luminal A and B IC-NST, respectively. Nuclear expression of EZH2 was not observed in luminal A tumors, and was detected in 35% (12/34) of luminal B carcinomas (p-val = 6.1 × 10-3) and only 4% (1/28) of invasive micropapillary carcinomas. This series shows that invasive micropapillary carcinomas harbor a CD24-positive inverted apical pole associated with weak EZH2 expression, phenotypical characteristics that distinguish this entity from other luminal carcinomas.
Breast carcinoma comprises many different entities characterized by specific molecular alterations. Tumor gene expression analyses have identified major molecular subtypes: luminal A, luminal B, HER2+, basal-like and normal breast-like groups [1,2] and, more recently, the apocrine and claudin-low groups [3,4].
Breast cancer-initiating cells have been defined as cells with properties responsible for tumor initiation, potentially driving tumor growth and metastasis, although these statements are still a subject of debate [5,6]. Xenotransplant assays in non-obese diabetic severe combined immunodeficiency (NOD/SCID) mice [
[
EZH2 belongs to the Polycomb group of proteins, which are involved in chromatin-modifying complexes, and stem cell self-renewal, a property of BrCICs, and are deregulated in cancer [9,10]. Although rarely chosen as a stem cell marker, EZH2 expression is poorly known in specific histological subtypes and is therefore assessed.
The identification in clinical practice of these BrCICs should help to understand chemo-radiotherapy resistance as stem cells and BrCICs have been shown to be more resistant to treatment than more differentiated cells [
Few studies have yet addressed the specific pattern of this combination of four markers expression in histological rare sub-types [
Though, we investigated the expression of CD44, CD24, ALDH1 and EZH2 in a series of 28 IMPC from a single institution. As IMPC have recently been proposed to belong to the group of luminal carcinomas [
Most of the IMPC harbored a specific pattern of CD24 expression different from that of the other luminal tumors analyzed and demonstrated a rare nuclear expression of EZH2.
We retrospectively selected 89 cases of invasive breast cancer—28 IMPC, 27 luminal A and 34 luminal B invasive carcinomas of no special type (IC-NST)—on the basis of the availability of clinical data, paraffin blocks and clinical follow-up information from our tumor bank. Initial treatment was surgery in all selected cases. These cases were reviewed by two experienced breast pathologists (AVS and PF), according to the World Health Organization (WHO) classification criteria. IMPC cases were recognized on the basis of inside-out MUC-1 staining at the inverted apical pole. IC-NST were defined as luminal A if grade I or II and ER+/PR+/ERBB2−, and as luminal B if ER+ and grade III or ERBB2 (3+) [
Tissue microarrays (TMA) were constructed with three representative cores (1 mm in diameter) of each tumor and one core of normal surrounding breast tissue for each case. Four µm thick sections were obtained from formalin-fixed, paraffin-embedded tissues for TMAs. These sections were cut, dried, deparaffinised and rehydrated according to standard procedures. All sections were subjected to heat-induced antigen retrieval in citrate buffer (pH 6.1). Estrogen receptor (ER, clone 6F11, 1:200, Novocastra), progesterone receptor (PR, clone 1A6, 1:200, Novocastra), ERBB2 (clone CB11, 1:1,000, Novocastra), CD44 (Thermo Scientific, clone 156/3C11, 1:100), CD24 (Thermo Scientific, clone SN3b, 1:100), ALDH1 (BD Transduction Laboratories, clone 44/ALDH, 1:200) and EZH2 (Novocastra, clone 6A10, 1:100) expressions were then evaluated. Internal and external positive and negative controls were included for each antibody. Staining was detected with the Vectastain Elite ABC peroxidase mouse IgG kit (Vector Burlingame, CA), with diaminobenzidine (Dako A/S, Glostrup, Denmark) as chromogen. Cases were considered positive for ER and PR when 10% of cells were positive for these markers [
For CD24 and CD44 the most frequently used cut-off of 10% of positive cells was chosen [14,25,26]. For ALDH1, the reported cut-offs in literature ranged from one positive cell to 10% of positive cells [14,15,27]. One positive cell cut-off was chosen. For EZH2, we determined our interpretation method according to Kleer et al. [
The homogeneity between groups of different tumor types was evaluated by the Pearson Chi-square test. Statistical analyses of survival were carried out with MedCalc® software. Cumulative overall survival was calculated by the Kaplan-Meier method, and the log-rank test was used to analyse differences in survival times. A pvalue ≤ 0.05 was considered significant.
A set of 28 IMPC, 27 luminal A and 34 luminal B ICNST was analyzed. The median age of the patients, not statistically different among the three groups (
The expression of the CD44, CD24, ALDH1 and EZH2 markers was analyzed, as these markers have been shown to be associated with either more differentiated luminal epithelial (CD24) or stem cell-like (CD44, ALDH1, EZH2) characteristics (
and co-workers [
The expression of these markers was further investigated in tumor cells (
ling was positive in 96% of IMPC (27/28 patients), 74% of luminal A (20/27 patients) and 88% of luminal B IC-NST cases (30/34 patients). Two different patterns of CD24 staining were observed, with labelling either only localized at the apical membrane or at the cytoplasmic membrane (circumferential staining). In most IMPC cases, CD24 staining was observed at the apical membrane (
CD44 staining was detected in most tumors of all three types [79% of IMPC (22/28 patients) (
The coexistence of cells expressing CD44 and cells expressing CD24 within the same tumor was then analyzed. CD44+ and CD24+ cells were observed in most cases of the three subtypes studied (75% of IMPC, 67% of luminal A tumors and 62% of luminal B tumors). However, the luminal A subgroup comprised the highest percentage of cases with CD44+/CD24− cells: 26%, versus only 4% for IMPC cases. No significant difference was observed between luminal B tumors and IMPC in terms of the proportion of cases with CD44+/CD24− cells (9% versus 4%; p-val = ns).
ALDH1 has been recognized as a putative marker of BrCICs. ALDH1 expression was therefore assessed in this series of cases. ALDH1 was expressed in only 25% of IMPC and 11% of luminal A and 15% of luminal B tumor cells per case. The proportion of ALDH1+ cells was 90% in only one IMPC case, which was negative for both ER and PR. All but three IMPC cases presented ALDH1+ stromal cells. ALDH1 was clearly expressed in the stroma in most cases of all three tumor types, but more frequently in IMPC (25/28, 89%) and luminal B (27/34, 79%) than in luminal A IC-NST (14/27, 52%) (
EZH2 has been identified as a marker of BrCICs and of breast carcinomas with poor prognosis. EZH2 expression was therefore evaluated in this series of IMPC and luminal A and B tumor controls. One of the 28 IMPC (4%) displayed nuclear EZH2 expression, whereas no EZH2 expression was detected in any of the luminal A tumors (100% negative). Twelve of the 34 cases of luminal B tumors displayed nuclear EZH2 expression (35% cases), with 1% to 25% of cells positive for this marker in each case. EZH2 was more frequently expressed in luminal B tumors than in the other tumor types (luminal B versus IMPC, p-val = 6.1 ´ 10−3; luminal B versus luminal A, p-val = 6.1 ´ 10−3) (Figures 2(j)-(l)). Notably, the EZH2 positive IMPC case was grade 2 whereas all grade 3 IMPC cases were EZH2 negative.
The definition of IMPC patients outcome remains controversial. Clinical data were available for all patients. We tried to assess the outcome of the three groups (IMPC, luminal A and B IC-NST) knowing the putative weaknesses of this evaluation (the retrospective nature of the study, the different periods of patient’s clinical management, the different clinico-pathological characteristics among the groups). In addition, within the IMPC group, tumors were further classified as “luminal A IMPC” or “luminal B IMPC”. This sub-classification allowed us to compare the specific outcome of these two groups of IMPC together and to that of luminal A and B IC-NST respectively.
At 10 years, IMPC patients had overall survival rates similar to those for patients with luminal A IDC tumors (
overall survival rates than patients with luminal B IDC tumors (
We then compared the outcome according to the molecular subgroups defined as described in material and methods. Luminal A and B IC-NST showed statistically significant different outcome (p-val = 1.2 ´ 10−2 (DFS); p-val = 3.7 ´ 10−2 (OS)). Luminal A and B IMPC experienced the same overall survival and disease-free survival (p-val = ns). Luminal A IMPC and luminal A IC-NST also experienced the same outcome (OS and DFS, p-val = ns). Identically, luminal B IMPC and luminal B IC-NST had also same overall and disease-free survival (Supplementary
The various molecular subtypes of breast carcinomas and ductal carcinoma in situ (DCIS) have been shown to present patterns of so called “stem-cell related” marker expression different from those of IC-NST. Basal-like tumors have more CD44+/CD24− cells than luminal (A and B) and ERBB2 tumors [
We assessed the cellular pattern of expression of these markers (CD44, CD24, ALDH1 and EZH2) in IMPC and compared this pattern to that observed in luminal A and B IC-NST.
IMPC is a rare and unusual histological type, characterized by a very high rate of vascular and axillary lymph node invasion (about 60% to 70% of cases) and features of local and regional aggressiveness. IMPC have recently been shown to share phenotypic and genomic characteristics with luminal B carcinomas [
This study demonstrated that 1) IMPC differ from luminal B carcinomas by presenting a high level of membranous inverted apical CD24 expression; 2) EZH2 was much less frequently expressed in IMPC than in luminal B carcinomas; 3) IMPC patients had a favourable overall survival (85% at 10 years).
Different patterns of CD44 and CD24 expression have been reported for different tumor stages and molecular subtypes [
CD44 expression is associated with basal-like stem cells [
A shift in CD24 staining from the apical membrane to a membranous/cytoplasmic distribution during progression from DCIS to IC-NST has been reported [
No staining for EZH2 was observed in the luminal A group and EZH2 expression levels were very low in IMPC (4% positive). The only IMPC case displaying EZH2 expression overexpressed ERBB2. In contrast, the frequency of EZH2 expression was significantly higher in luminal B tumors, which are known to have a poor prognosis.
IMPC tumors are associated with a high frequency of vascular and axillary lymph node invasion. Interestingly, in this series, eventhough the large majority of the IMPC cases was luminal, a small majority (54% of the cases) were luminal B (ER+ and grade III or HER2 3+) but demonstrated an outcome identical to that of luminal A IMPC. These observations suggest that despite the high rates of axillary lymph node metastasis and vascular invasion, this outcome could be related either to the small tumors size being mostly T1 and T2 in this series or to IMPC histological type per se and its biological properties without excluding also the possibility of a high sensitivity to treatments. The ERBB2 overexpression pattern of IMPC has been reported to be unusual in that it is confined to three sides of the cells, excluding the apical inverted pole [
Paradoxically, CD44+/CD24− cells have been reported to be associated with greater invasiveness in an in vitro model [
The percentage of ALDH1-positive tumors in the present study was lower than reported in other studies (7% to 30% ALDH1+ tumors) [8,15]. However, ALDH1 staining within the stroma was observed in most cases in this study, even in the absence of epithelial staining. This observation challenges the use of ALDH1 as an epithelial stem cell-related marker, at least in this rare histological subtype. In luminal A and B IC-NST, ALDH1 staining patterns were consistent with the low frequency of cells positive for stem cell-related markers on immunohistochemistry. In previous studies based on cell sorting approaches, basal-like carcinomas were identified as the molecular group enriched in tumor-initiating cells/stem cells [
In conclusion, IMPC should be considered to be different from other luminal carcinomas, as CD24 is expressed at the inverted apical cellular pole associated with small numbers of EHZ2+ and ALDH1+ epithelial cells.
This work was supported by grants from INSERM and the Institut Curie (Programme Incitatif et Coopératif “Polarité et mitose”). Dr. Anne Vincent-Salomon was supported by an “Interface INSERM” grant.