2025 Volume 58 Issue 5 Pages 173-182
Eribulin, a microtubule inhibitor, is effective as later-line therapy for metastatic breast cancer (MBC) and has been reported to remodel the tumor microenvironment and inhibit epithelial–mesenchymal transition (EMT). However, the association between pretreatment EMT status and eribulin efficacy remains unclear. We retrospectively analyzed 41 patients with MBC (excluding invasive lobular carcinoma) treated with eribulin between 2013 and 2020. Formalin-fixed, paraffin-embedded biopsy specimens were examined by immunohistochemistry (IHC) using anti–E-cadherin (24E10) and anti-vimentin (V9) antibodies. Complete membranous E-cadherin expression (3+) was defined as normal; reduced expression (2+, 1+, 0) as altered. Negative vimentin was considered normal; positive expression, altered. Co-localization of E-cadherin and vimentin was assessed by multi-immunofluorescent staining. Of the 41 patients, 24 responded to eribulin and 17 did not. Progression-free survival (PFS) and overall survival (OS) were significantly longer in responders than in nonresponders (p < 0.001 and p = 0.0044). Altered E-cadherin and/or vimentin expression was more frequently observed in responders (p = 0.013) and associated with longer progression-free survival (p = 0.048). These results suggest that eribulin efficacy may be predicted by altered E-cadherin and vimentin expression before treatment.
Breast cancer ranks first in incidence and second in mortality among malignant tumors in women, according to the latest global cancer statistics [1]. Although early-stage breast cancer is often curable, treatment of metastatic breast cancer (MBC) focuses on prolonging survival and alleviating symptoms [2]. Systemic therapies are the mainstay of treatment for MBC and are selected based on the immunohistochemical status of the estrogen receptor (ER), progesterone receptor (PgR), and human epidermal growth factor receptor 2 (HER2). Patients with HER2-positive or hormone receptor–positive MBC tend to have better median survival than those with triple-negative breast cancer (TNBC), which lacks ER, PgR, and HER2 expression [3]. However, treatment options for TNBC patients who have previously received multiple regimens remain limited. Likewise, no established predictive indicators exist for hormone receptor–positive MBC patients with resistance to hormone therapy.
Microtubule inhibitors have been widely used at various stages of breast cancer treatment, with paclitaxel and docetaxel being among the most commonly prescribed agents [4, 5]. Eribulin, another microtubule inhibitor, was developed within this therapeutic class. Eribulin is a synthetic analog of halichondrin B, a compound first isolated from Halichondria okadai off the coast of Japan in 1986 [6–8]. It binds near the GDP/GTP binding site of β-tubulin and specifically inhibits stathmin, a regulator of microtubule polymerization. Compared with paclitaxel and docetaxel, eribulin demonstrates antitumor activity at lower concentrations and is associated with fewer side effects, such as axonopathy [6, 9].
Clinical studies comparing eribulin with capecitabine in patients with MBC have shown no significant differences in median OS [10]. However, in the Eisai Metastatic Breast Cancer Study Assessing Physician’s Choice Versus E7389 (EMBRACE) trial, eribulin was compared with physician’s choice of treatment in patients with HER2-negative MBC who had received prior therapies. The results demonstrated a significantly longer OS in the eribulin group [11]. On the basis of these findings, eribulin monotherapy is now regarded as a viable option for patients with HER2-negative MBC, particularly those who have undergone intensive prior treatment. Further studies have highlighted eribulin’s distinct effects compared with other microtubule inhibitors. In addition to inhibiting microtubule polymerization, eribulin has been reported to remodel the tumor microenvironment through vascular normalization and to inhibit epithelial–mesenchymal transition (EMT) [12–14].
EMT is a biological process in which epithelial cells lose polarity and adhesive properties and acquire stromal features. This transition enables cells to develop migratory and invasive capabilities [15]. During EMT, breast cancer cells increasingly exhibit stromal characteristics, including a switch in intermediate filaments to vimentin [16], along with loss of intercellular adhesion due to reduced E-cadherin expression [17]. EMT has been implicated in the metastatic progression of multiple cancers, including breast cancer [18, 19]. Because the development of metastatic lesions worsens prognosis [20, 21], therapeutic strategies that account for EMT status may offer promise for improving outcomes. However, the relationship between EMT status in primary MBC lesions and the therapeutic efficacy of eribulin remains unclear. Therefore, this study aimed to investigate whether immunohistochemical factors related to EMT status before treatment can predict the efficacy of eribulin.
We reviewed 67 patients with MBC who received eribulin monotherapy, including but not limited to first-line treatment, at Nihon University Itabashi Hospital between January 2013 and December 2020. Eribulin at a dose of 1.4 mg/m2 was administered intravenously on days 1 and 8 of a 21-day cycle. If adverse effects occurred, dose reduction of up to 50% or administration every other week was permitted. Of these patients, 41 with available formalin-fixed, paraffin-embedded (FFPE) biopsy specimens obtained before anticancer treatment were included in this study. Invasive ductal carcinoma (IDC) and invasive lobular carcinoma (ILC) were not distinguished at the time of eribulin therapy. However, patients with ILC, which is usually negative for E-cadherin, were excluded. Patient characteristics are summarized in Table 1. Treatment response to eribulin was evaluated using the Response Evaluation Criteria in Solid Tumors. Patients were categorized as responders (complete response [CR], partial response [PR], and stable disease [SD] lasting ≥ 6 months) or nonresponders (SD lasting < 6 months or progressive disease [PD]). The study protocol was approved by the Institutional Review Board of Nihon University Itabashi Hospital (Approval No. 2407097) in accordance with the Declaration of Helsinki. Informed consent was obtained using an opt-out system; patients who declined participation were excluded.
Summary of patient characteristics
| Factors | Number (%) |
|---|---|
| Age at diagnosis, mean ± SD (range) | 57.7 ± 12.9 (26–81) |
| Histologya | |
| Invasive ductal carcinoma | 37 (90.2) |
| Others | 4 (9.8) |
| Clinical stage at diagnosis | |
| I | 4 (9.8) |
| II | 15 (36.6) |
| III | 8 (19.5) |
| IV | 14 (34.1) |
| Disease type | |
| Recurrence after surgery | 25 (61.0) |
| Locally advanced or de novo stage IV | 16 (39.0) |
| Metastatic sites at initiation of eribulin therapy (including duplicates) |
|
| Liver | 17 (41.5) |
| Lung (including pleura) | 19 (46.3) |
| Other visceral sites (e.g., ovary) | 1 (2.4) |
| Brain | 2 (4.9) |
| Bone | 16 (39.0) |
| Distant lymph nodes | 9 (22.0) |
| Local skin or regional lymph nodes | 17 (41.5) |
| Summary of metastatic sites | |
| Visceral | 31 (75.6) |
| Non-visceral only | 7 (17.1) |
| Local/regional only | 3 (7.3) |
| Subtype | |
| Luminal | 33 (80.5) |
| Triple-negative | 8 (19.5) |
| Chemotherapy line for eribulin | |
| 1st | 8 (19.5) |
| 2nd | 15 (36.6) |
| 3rd | 12 (29.3) |
| 4th or later | 6 (14.6) |
| Chemotherapies for metastases before eribulin (including duplicates)b |
|
| Anthracycline | 20 (48.8) |
| Taxanes | 20 (48.8) |
| Oral 5-FU derivatives | 7 (17.1) |
| Paclitaxel + bevacizumab | 10 (24.4) |
| Therapeutic effects of eribulin | |
| Complete response | 1 (2.4) |
| Partial response | 8 (19.5) |
| Stable disease ≥ 6 months | 15 (36.6) |
| Stable disease < 6 months | 7 (17.1) |
| Progressive disease | 10 (24.4) |
a Histology, others: Invasive mucinous carcinoma, invasive apocrine carcinoma, invasive medullary carcinoma, and invasive carcinoma with neuroendocrine differentiation.
b Chemotherapy regimens;
Anthracycline: Doxorubicin (A) 60 mg/m2 plus cyclophosphamide (C) 600 mg/m2, or epirubicin (E) 90 mg/m2 plus cyclophosphamide (C) 600 mg/m2, administered intravenously every three weeks.
Taxanes: Docetaxel 60–75 mg/m2 intravenously every three weeks; paclitaxel 80 mg/m2 intravenously weekly; or nab-paclitaxel 260 mg/m2 intravenously every three weeks.
Oral 5-FU derivatives: S-1, 80–120 mg/day orally, twice daily for 28 days followed by a 14-day rest period; or capecitabine, 1800–3000 mg/day orally, twice daily for 21 days followed by a 7-day rest period. Paclitaxel + bevacizumab: Paclitaxel 80 mg/m2 intravenously on days 1, 8, and 15, combined with bevacizumab 10 mg/kg on days 1 and 15 of a 28-day cycle.
SD, standard deviation.
The expression of E-cadherin (an epithelial marker) and vimentin (a mesenchymal marker) was evaluated by immunohistochemistry. Formalin-fixed, paraffin-embedded tissue sections were cut at a thickness of 4 μm and mounted on silane-coated glass slides. Following deparaffinization, antigen retrieval was performed by boiling the sections in EDTA buffer (pH 9.0). After washing with phosphate-buffered saline (PBS), endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 5 min at 25°C. The samples were incubated with rabbit monoclonal anti–E-cadherin antibody (1:1,000; clone 24E10, #3195, Cell Signaling Technology, Massachusetts, USA) and mouse monoclonal anti-vimentin antibody (1:1,000; clone V9, M0725, Agilent Technologies, California, USA) for 30 min at 25°C. After washing with PBS, the sections were treated with a polymer secondary antibody (Simple Stain Max PO Multi; Nichirei Bioscience, Tokyo, Japan) for 30 min at 25°C. Visualization was achieved with 3,3'-diaminobenzidine for 10 min at room temperature, followed by counterstaining with hematoxylin, dehydration, and mounting.
Normal breast tissue served as a positive internal control for E-cadherin (Supplementary Fig. S1), and stromal cells in each sample served as a positive internal control for vimentin. For negative controls, sections were processed without the primary antibody. In addition, mouse IgG (Vector Laboratories, California, USA) was used as an isotype control at the same concentration as the primary anti-vimentin antibody to confirm staining specificity. The control samples treated with mouse IgG showed no staining. Therefore, positive vimentin expression in the tumor cytoplasm was confirmed to be specific and not due to background signal or nonspecific antibody binding (Supplementary Fig. S2).
After confirming staining specificity, immunohistochemical expression was evaluated using an intensity scoring system: 0 (negative), 1+ (weak), 2+ (moderate), and 3+ (strong) (Fig. 1A–H). Diffuse strong E-cadherin expression (3+, Fig. 1D) was defined as normal, whereas negative to moderate expression (0–2+, Fig. 1A–C) was defined as altered. For vimentin, negative expression (0, Fig. 1E) was defined as normal, whereas weak to strong expression (1+ to 3+, Fig 1F–H) was defined as altered.
Immunohistochemical evaluation of E-cadherin and vimentin expression. E-cadherin expression at the tumor cell membrane was scored as (A) negative (0), (B) weak (1+), (C) moderate (2+), or (D) strong (3+). Vimentin expression in the tumor cytoplasm was scored as (E) negative (0), (F) weak (1+), (G) moderate (2+), or (H) strong (3+). Bars = 50 μm.
The co-expression of E-cadherin and vimentin was assessed by multi-immunofluorescent staining. FFPE tissue sections (4 μm thick) were prepared and mounted on silane-coated glass slides. Following deparaffinization, antigen retrieval was performed by boiling the sections in EDTA buffer (pH 9.0). Endogenous peroxidase activity was blocked with 3% hydrogen peroxide for 5 min at room temperature. The sections were incubated with primary antibodies: rabbit monoclonal anti–E-cadherin antibody (1:10,000; clone 24E10, #3195, Cell Signaling Technology, Massachusetts, USA) and mouse monoclonal anti-vimentin antibody (1:10,000; clone V9, M0725, Agilent Technologies, California, USA) for 30 min at 25°C. After washing with PBS, the sections were incubated with secondary antibodies: Alexa Fluor 488–labeled anti-mouse (1:500; ab150117, Abcam, Cambridge, UK) and Alexa Fluor 594–labeled anti-rabbit (1:500; A11072, Thermo Fisher Scientific, Massachusetts, USA) for 30 min at 25°C. The slides were mounted with Diamond Antifade Mountant containing DAPI (4',6-diamidino-2-phenylindole; Thermo Fisher Scientific). Fluorescence images were acquired using an Axio Imager Z2 Upright Microscope (Zeiss Microscopy GmbH, Jena, Germany) and analyzed with ZEN 2 Pro software (Zeiss Microscopy GmbH).
Statistical analysisContinuous variables were summarized as medians (range), and categorical variables as frequencies (%). Group comparisons of continuous variables were performed using unpaired or Welch’s t tests, depending on variance equality as determined by the F test. Categorical variables were analyzed using the χ2 test. Progression-free survival (PFS) was defined as the time from initiation of eribulin to the first documented tumor progression or death from any cause, whichever occurred first. Overall survival was defined as the time from initiation of eribulin to death from any cause. PFS and OS curves were estimated using the Kaplan–Meier method and compared with a two-sided log-rank test. Statistical significance was set at p < 0.05. All statistical analyses were performed using JMP software (JMP Statistical Discovery LLC, Cary, NC, USA).
The effects of eribulin therapy in patients with MBC were analyzed. The median PFS from initiation of eribulin therapy was 7.5 months (95% confidence interval [CI], 4.9–10.3) across all 41 patients. In the responder group, the median PFS was significantly longer than in the nonresponder group (11.8 months [95% CI, 8.9–14.7] vs 3.2 months [95% CI, 1.6–4.8]; p < 0.001; Fig. 2A). The median OS from initiation of eribulin therapy was 23.3 months (95% CI, 16.6–29.9) across all patients. Median OS was significantly longer in the responder group than in the nonresponder group (29.4 months [95% CI, 14.7–44.1] vs 14.2 months [95% CI, 4.7–23.6]; p = 0.0044; Fig. 2B).
Survival analysis from the start of eribulin therapy in the eribulin responder and the nonresponder groups. (A) Progression-free survival and (B) overall survival. The blue line represents the responder group, and the red line represents the nonresponder group. *p < 0.05, **p < 0.001 (log-rank test).
Table 2 summarizes the correlation between eribulin efficacy and clinicopathological characteristics. No significant differences in age, stage at diagnosis, tumor subtype, metastatic site, or eribulin treatment line were observed between the two groups.
Association between eribulin therapeutic responses and clinicopathological factors
| Factors | Eribulin therapeutic responsesa | p-value | |
|---|---|---|---|
| Responders (n = 24) | Nonresponders (n = 17) | ||
| Age at diagnosis (mean ± SD) | 58.8 ± 13.0 | 56.2 ± 13.1 | 0.54 |
| Histologyb | |||
| Invasive ductal carcinoma | 21 (87.5) | 16 (94.1) | NA |
| Others | 3 (12.5) | 1 (5.9) | |
| Clinical stage at diagnosis | |||
| I | 4 (16.7) | 0 (0.0) | NA |
| II | 6 (25.0) | 9 (52.9) | |
| III | 6 (25.0) | 2 (11.8) | |
| IV | 8 (33.3) | 6 (35.3) | |
| Disease typec | |||
| Recurrence | 14 (58.3) | 11 (64.7) | 0.68 |
| De novo | 10 (41.7) | 6 (35.3) | |
| Metastatic sited | |||
| Visceral | 17 (69.2) | 14 (82.4) | 0.40 |
| Non-visceral | 7 (30.8) | 3 (17.6) | |
| Subtype | |||
| Luminal | 19 (79.2) | 14 (82.4) | 0.80 |
| Triple-negative | 5 (20.8) | 3 (17.6) | |
| Chemotherapy lines for eribulin | |||
| 1st | 4 (16.7) | 4 (23.5) | NA |
| 2nd | 11 (45.8) | 4 (23.5) | |
| 3rd | 7 (29.2) | 5 (29.5) | |
| 4th or later | 2 (8.3) | 4 (23.5) | |
| Chemotherapy before eribulin (including duplicates) | |||
| Anthracycline | 11 (45.8) | 9 (52.9) | NA |
| Taxanes | 12 (50.0) | 8 (47.1) | |
| Oral 5-FU derivatives | 4 (16.7) | 3 (17.6) | |
| Paclitaxel + bevacizumab | 4 (16.7) | 6 (35.3) | |
a Patients with complete response, partial response, and stable disease lasting ≥ 6 months were categorized as responders, whereas patients with stable disease lasting < 6 months or progressive disease were categorized as non-responders.
b Histology, others: invasive mucinous carcinoma, invasive apocrine carcinoma, invasive medullary carcinoma, and invasive carcinoma with neuroendocrine differentiation.
c Disease type: recurrence indicates relapse after primary surgery; de novo indicates locally advanced or metastatic disease at diagnosis.
d Metastatic site: visceral includes liver, lung (including pleura), brain, and other visceral organs; non-visceral includes lymph nodes, bone, and skin.
SD, standard deviation; NA, not analyzed due to cells with an expected value < 5.
The expression status of E-cadherin and vimentin was classified as normal or altered. Concurrent normal expression of both markers was observed in 15 patients (36.6%). No patients showed altered vimentin expression with concurrent normal E-cadherin expression. Among patients with altered E-cadherin expression, 15 (57.7%) also exhibited altered vimentin expression, whereas 11 (42.3%) exhibited normal vimentin expression. Both the “normal/normal” and “altered/altered” expression patterns were significantly more frequent (p < 0.001; Table 3). Overall, altered expression of E-cadherin and/or vimentin was detected in 26 patients (63.4%), and all 15 patients (36.6%) with altered vimentin expression also had altered E-cadherin expression.
Combined immunohistochemical expression status of E-cadherin and vimentin
| Immunohistochemical expression statusa | E-cadherin | p-value | ||
|---|---|---|---|---|
| Normal (n) | Altered (n) | |||
| Vimentin | Normal (n) | 15 | 11 | <0.001** |
| Altered (n) | 0 | 15 | ||
a Immunohistochemical scoring: normal E-cadherin expression was defined as 3+; altered E-cadherin expression was defined as 0–2+. Normal vimentin expression was defined as 0; altered vimentin expression was defined as 1+–3+.
**p < 0.001
Neither altered E-cadherin expression nor altered vimentin expression was significantly correlated with clinicopathological factors, including stage at diagnosis, recurrence, metastatic site, or tumor subtype (Supplementary Table S1). Similarly, no significant correlation was observed between combined E-cadherin/vimentin expression status and clinicopathological factors (Table 4).
E-cadherin and vimentin expression status and clinicopathological characteristics
| Factors | Normal E-cadherin and vimentin expression (n = 15)a | Altered E-cadherin and/or vimentin expression (n = 26)b | p-value |
|---|---|---|---|
| Age at diagnosis (mean ± SD) | 58.0 ± 12.7 | 57.6 ± 13.2 | 0.92 |
| Histologyc | |||
| Invasive ductal carcinoma | 14 (93.3) | 23 (88.5) | NA |
| Others | 1 (6.7) | 3 (11.5) | |
| Clinical stage at diagnosis | |||
| I | 0 (0.0) | 4 (15.4) | NA |
| II | 5 (33.3) | 10 (38.5) | |
| III | 3 (20.0) | 5 (19.2) | |
| IV | 7 (46.7) | 7 (26.9) | |
| Disease typed | |||
| Recurrence | 8 (53.3) | 17 (65.4) | 0.45 |
| De novo | 7 (46.7) | 9 (34.6) | |
| Metastatic sitee | |||
| Visceral | 12 (80.0) | 19 (73.1) | 0.62 |
| Non-visceral | 3 (20.0) | 7 (26.9) | |
| Subtype | |||
| Luminal | 11 (73.3) | 22 (84.6) | 0.38 |
| Triple-negative | 4 (26.7) | 4 (15.4) | |
| Chemotherapy line of eribulin | |||
| 1st | 5 (33.3) | 3 (11.5) | NA |
| 2nd | 2 (13.4) | 13 (50.0) | |
| 3rd | 5 (33.3) | 7 (26.9) | |
| 4th or later | 3 (20.0) | 3 (11.5) | |
| Chemotherapy before eribulin (including duplicates) |
|||
| Anthracycline | 9 (60.0) | 11 (42.3) | NA |
| Taxanes | 9 (60.0) | 11 (42.3) | |
| Oral 5-FU derivatives | 1 (6.6) | 6 (23.1) | |
| Paclitaxel + bevacizumab | 2 (13.3) | 8 (30.1) | |
Data are presented as number (%), unless otherwise specified.
a Normal expression: E-cadherin 3+ and vimentin 0.
b Altered expression: E-cadherin 0–2+ and/or vimentin 1+–3+.
c Histology, others: invasive mucinous carcinoma, invasive apocrine carcinoma, invasive medullary carcinoma, and invasive carcinoma with neuroendocrine differentiation.
d Recurrence indicates relapse after primary operation; de novo indicates locally advanced or metastatic cases at diagnosis.
e Visceral: metastases to liver, lung, brain, or other organs. Non-visceral: metastases to lymph nodes, bones, or skin.
EMT, epithelial–mesenchymal transition; SD, standard deviation; NA, not analyzed because of expected cell counts < 5.
We investigated the correlation between E-cadherin and vimentin expression status and therapeutic response to eribulin, as shown in Table 5. Based on evaluation of E-cadherin expression alone, patients with altered E-cadherin expression were significantly more frequent in the responder group (19 of 26, 73.1%), whereas patients with normal E-cadherin expression were significantly more frequent in the nonresponder group (10 of 15, 66.7%; p = 0.01). Based on combined evaluation of E-cadherin and vimentin expression, patients with altered E-cadherin and/or altered vimentin expression—including all patients with altered E-cadherin—were significantly more frequent in the responder group (19 of 26, 73.1%), whereas patients with concurrent normal expression of both markers were significantly more frequent in the nonresponder group (10 of 15, 66.7%; p = 0.01). No correlation was observed between vimentin expression alone and therapeutic response to eribulin.
Correlation between E-cadherin and vimentin expression status and eribulin therapeutic response
| Immunohistochemical expression status | Number | Eribulin therapeutic responsesa | p-value | |
|---|---|---|---|---|
| Responders (n = 24) |
Nonresponders (n = 17) |
|||
| E-cadherin expression | ||||
| Normal | 15 | 5 (33.3) | 10 (66.7) | 0.01* |
| Altered | 26 | 19 (73.1) | 7 (26.9) | |
| Vimentin expression | ||||
| Normal | 26 | 13 (50.0) | 13 (50.0) | 0.14 |
| Altered | 15 | 11 (73.3) | 4 (26.7) | |
| Combined E-cadherin and vimentin expression | ||||
| Normal E-cadherin and normal vimentin expression | 15 | 5 (33.3) | 10 (66.7) | 0.01* |
| Altered E-cadherin and/or altered vimentin expression | 26 | 19 (73.1) | 7 (26.9) | |
Data are presented as number (%). *p < 0.05 was considered statistically significant.
ᵃ Responders included patients with complete response, partial response, or stable disease lasting > 6 months; non-responders included patients with stable disease lasting < 6 months or progressive disease.
Next, immunohistochemical expression scores of E-cadherin and vimentin were compared between responders and nonresponders (Fig. 3A). Normal E-cadherin expression (score 3+) was more prevalent in the nonresponder group, whereas no trend was observed for vimentin expression scores. An inverse correlation was observed between E-cadherin and vimentin expression scores (R2 = 0.192, p = 0.0042; Fig. 3B).
Distribution of E-cadherin and vimentin expression scores. (A) Expression scores in the eribulin responder group (gray bars) and the nonresponder group (white bars). (B) Box sizes and numbers indicate case counts. The dotted line indicates the inverse correlation between E-cadherin and vimentin expression scores (regression line: y = –0.539x + 1.773; R2 = 0.192; p = 0.0042).
The co-localization of E-cadherin and vimentin was further examined using multi-immunofluorescent staining (Fig. 4). In cases with normal expression of both markers, only membranous E-cadherin expression was observed in tumor cells (Fig. 4A). In cases with altered E-cadherin and normal vimentin expression, weakened membranous E-cadherin expression was observed (Fig. 4B). In cases with altered expression of both markers, weakened membranous E-cadherin expression was accompanied by cytoplasmic vimentin expression (Fig. 4C). These findings suggest that alterations in E-cadherin and vimentin occur within the same tumor cells, rather than reflecting a mixture of cells with different expression patterns.
Multi-immunofluorescence staining of E-cadherin and vimentin. (A) Normal expression (E-cadherin 3+, vimentin 0). (B) Altered E-cadherin expression (2+) with normal vimentin expression (0). (C) Altered E-cadherin expression (1+) with altered vimentin expression (3+).
Progression-free survival and OS from initiation of eribulin therapy were compared according to E-cadherin and vimentin expression status (Fig. 5). Patients with altered E-cadherin and/or vimentin expression had significantly longer median PFS than those with normal expression of both markers (7.8 months [95% CI, 2.8–12.8] vs 5.7 months [95% CI, 4.1–7.2]; p = 0.048; Fig. 5A, left). Among patients with altered E-cadherin expression, no significant association was observed between vimentin expression status and PFS (Fig. 5A, center). Similarly, no correlation was observed between vimentin expression status and PFS (Fig. 5A, right), or between E-cadherin/vimentin expression status and OS (Fig. 5B).
Survival analysis from the start of eribulin therapy stratified by E-cadherin and vimentin expression status. (A) Progression-free survival and (B) overall survival. Left: combined E-cadherin and vimentin expression status (blue line: altered E-cadherin and/or altered vimentin; red line: both normal). Center: vimentin expression status among patients with altered E-cadherin expression (blue line: altered vimentin; red line: normal vimentin). Right: vimentin expression status (blue line: altered; red line: normal). *p < 0.05 (log-rank test).
Eribulin is a microtubule inhibitor that has demonstrated efficacy in later lines of therapy. Patients who respond to eribulin therapy have been reported to maintain good performance status without an increase in the number of metastatic lesions [22]. In our cohort, the eribulin responder group similarly exhibited longer PFS and OS. However, reliable stratification factors for selecting patients who would benefit most from eribulin therapy have not yet been established. Eribulin has also been reported to remodel the tumor microenvironment and inhibit EMT [12–14]. Therefore, we investigated the correlation between eribulin response in MBC and the immunohistochemical expression status of E-cadherin and vimentin, both of which are well known to change during EMT. Specifically, EMT is characterized by decreased E-cadherin expression and increased vimentin expression.
In the present study, immunohistochemical analysis revealed altered expression of E-cadherin and vimentin in MBC cases, which was correlated with eribulin therapeutic response. Patients with altered E-cadherin and/or altered vimentin expression were more frequent in the responder group, and these patients also had longer PFS. Furthermore, E-cadherin and vimentin expression scores demonstrated an inverse correlation, with both alterations occurring within the same tumor cells. These findings suggest that eribulin monotherapy may be particularly effective in patients whose tumors are in an EMT state, as reflected by altered expression of these markers. Although current clinical indications for eribulin therapy do not distinguish between IDC and ILC, in this study, ILC cases—characteristically negative for E-cadherin expression—were excluded. Loss of E-cadherin in ILC is a histological hallmark and is not considered an EMT-related change; consistent with this, eribulin therapy was not effective in these patients in our observational analysis (data not shown). In the future, larger studies will be required to validate these findings and to incorporate additional markers capable of distinguishing between IDC- and ILC-related EMT status.
EMT is associated with cancer cell invasion and metastasis in solid tumors [18, 19, 23]. Cancer cells with high vimentin expression and reduced E-cadherin expression may exhibit an advanced EMT state. Positive vimentin expression has been linked to adverse prognostic factors such as higher histological grade and increased Ki-67 index in breast cancer [24]. Moreover, vimentin expression has been shown to increase during cancer metastasis [16], whereas loss of E-cadherin expression in breast ductal carcinoma has been associated with an increased risk of death [25]. When combined with other factors, vimentin expression may worsen disease-free survival and is considered a poor prognostic indicator [26, 27]. Additionally, elevated vimentin expression has been reported in metaplastic carcinoma of the breast, a subtype that is often difficult to treat [28]. Thus, higher vimentin expression together with lower E-cadherin expression in breast cancer may predict poor prognosis. In the present study, although patients with altered E-cadherin expression, regardless of vimentin status, exhibited longer PFS after initiation of eribulin therapy, patients with positive vimentin expression tended to have shorter OS. Previous gene expression studies using droplet digital polymerase chain reaction have shown that SNAIL1 and TWIST1—genes associated with EMT—are also associated with increased eribulin efficacy [29]. These findings are consistent with our results, which suggest that EMT status may predict treatment response. In contrast, paclitaxel, another microtubule inhibitor widely used in breast cancer, has been reported to potentially induce EMT based on the expression levels of N-cadherin, vimentin, and E-cadherin in breast cancer cells [30]. Our study further demonstrated that eribulin therapeutic efficacy was not affected by differences in prior anticancer treatments.
These findings indicate that the cellular EMT status before treatment correlates with the therapeutic response to eribulin in MBC. EMT status has been reported to change after a single chemotherapy regimen [13]. Because MBC patients typically undergo multiple interventions during their treatment course, further studies are needed to clarify the impact of EMT status on OS at treatment completion. Moreover, EMT status could not be evaluated both before and after eribulin therapy, which represents a limitation of this study. Nevertheless, eribulin therapy demonstrated efficacy even in highly malignant cases characterized by EMT. Therefore, for patients with HER2-negative breast cancer—a subgroup with limited effective treatment options—the efficacy of eribulin may be predicted by examining the pre-treatment expression status of E-cadherin and vimentin. In particular, E-cadherin expression in invasive ductal carcinoma prior to any treatment may serve as a predictive marker of eribulin response.
Histological examination plays a crucial role in diagnosis and surgical procedures, making it relatively straightforward to obtain tumor tissue samples before therapeutic intervention. Recently, FFPE tissue samples have been increasingly employed as companion diagnostics for various drugs. In this context, our findings using FFPE samples obtained before therapeutic intervention may provide valuable clinical insights. Direct evaluation of the effects of eribulin on EMT suppression and therapeutic efficacy remains challenging. However, the ability to predict eribulin efficacy using samples that can be easily collected before treatment may aid in selecting optimal therapies for patients with HER2-negative MBC.
In conclusion, altered expression of E-cadherin and/or vimentin was correlated with a better therapeutic response to eribulin and longer PFS in MBC. Although cancer patients with EMT may generally have worse prognoses, immunohistochemical evaluation of E-cadherin and vimentin before initiation of therapy may be useful for identifying patients who are more likely to benefit from eribulin, regardless of treatment line.
The authors declare no conflicts of interest.
The authors thank Mrs. Emiko Nishikiori for her clerical support.
