Ki-67 index and galectin-3 as surrogate markers for predicting disease-free survival in luminal-type breast carcinoma treated with neoadjuvant chemotherapy
Article information
Abstract
Background
Residual cancer burden (RCB) is an important marker for patients with breast carcinoma treated with neoadjuvant chemotherapy (NACT). However, the association between RCB and prognosis is relatively less significant in residual luminal-type breast carcinoma (LTBC). Associations between clinicopathological variables, including RCB class, and immunohistochemical (IHC) markers with disease-free survival (DFS) were analyzed.
Methods
Expression of Ki-67, calreticulin, clusterin, galectin-3, mucin-1, and p27 was assessed in tissue microarray slides of 55 post-NACT resection specimens from LTBC patients treated with docetaxel and doxorubicin. Patients’ age, pre-NACT progesterone receptor status, lymphovascular invasion, histologic grade, ypT category, ypN category, RCB class, and IHC markers were analyzed for their association with DFS using Kaplan-Meier analysis and Cox proportional hazards models.
Results
Twenty of the 55 patients developed recurrence or distant metastasis. A Ki-67 index > 2.7% and high galectin-3 expression were associated with shorter DFS on Kaplan-Meier analysis (p < .001 and p = .018, respectively). Other clinicopathological and IHC markers were not significantly associated with DFS. In the multivariate Cox proportional hazards model, a Ki-67 index > 2.7% (hazard ratio, 7.23; p < .001) and high galectin-3 expression (hazard ratio, 4.51; p = .007) remained independent predictors of shorter DFS.
Conclusions
The Ki-67 index and high galectin-3 expression in post-NACT resection specimens may serve as surrogate markers for predicting recurrence in LTBC.
INTRODUCTION
Standardized pathological evaluation and reporting of breast carcinoma (BC) specimens after neoadjuvant chemotherapy (NACT) are recommended to predict prognosis and evaluate further therapeutic plans for clinicians and pathologists [1]. The residual cancer burden (RCB) index has been widely adopted as a common tool to evaluate the efficacy of NACT in BC by quantifying residual disease in both the primary tumor and regional lymph nodes, thereby predicting patient outcomes effectively [2-4]. However, its predictive power is relatively limited in luminal-type breast carcinoma (LTBC) compared with human epidermal growth factor receptor 2 (HER2)–positive BC or triple-negative breast carcinoma (TNBC), although the association between RCB class and survival showed a trend in a larger cohort study [4-6]. Moreover, our previous study showed that LTBC achieved a lower pathological complete response rate than HER2-positive BC or TNBC following NACT [7]. This underscores the necessity of predicting patient outcomes or determining further therapeutic strategies using residual carcinoma specimens.
In addition to conventional pathological examinations, including tumor (ypT) and nodal (ypN) categories, histologic grade, and RCB index, immunohistochemical (IHC) markers have been suggested as predictors of outcomes in patients who underwent NACT. The association between the Ki-67 index, one of the commonly used proliferation markers, and clinical outcomes has been evaluated in various ways in BC patients who received NACT [8-10]. In the setting of neoadjuvant endocrine and/or cyclin-dependent kinase (CDK) 4/6 inhibitor therapy, a decreased Ki-67 index from baseline and/or a low Ki-67 index in the resection specimen have been considered a marker of treatment response. Recent clinical trials have shown that complete cell cycle arrest (CCCA), determined as a Ki-67 index ≤ 2.7% (an approximate value derived from the base of the natural logarithm divided by 100), in the resection specimen after NACT is an independent factor associated with longer disease-free survival (DFS) in LTBC [11,12].
Studies based on complementary DNA analysis have also identified many genes as potential surrogate markers of resistance to NACT, with some being upregulated or overexpressed in resistant cell lines or tissues, whereas others exhibit the opposite pattern [13-15]. Among these genes, galectin-3 and clusterin act as apoptosis inhibitors, p27 and mucin-1 are associated with cell proliferation, and calreticulin is involved in immune surveillance, with some functions potentially overlapping across these genes [13-15]. These biomarkers have also been proposed as predictors of prognosis in BC patients who have undergone NACT in studies using IHC [16-19]. However, no IHC markers other than estrogen receptor (ER), progesterone receptor (PR), and HER2 have yet been formally incorporated into the guidelines [1]. Additionally, further assays, such as IHC and molecular studies, as well as conventional pathological examinations performed on needle biopsy specimens, are relatively limited compared with those performed on resection specimens; therefore, studies of resection specimens following NACT can provide important information to clinicians and/or pathologists.
In this study, we analyzed the association between survival and pathological markers in resection specimens, including histological and IHC results measured using H-score, in LTBC patients who were treated with NACT.
MATERIALS AND METHODS
Patients and specimens
Female patients who attended Seoul National University Hospital in Seoul, Korea, from October 2005 to November 2007 and received NACT for LTBC were included in this study. All patients had a primary tumor ≥ 5 cm and/or palpable ipsilateral axillary lymph nodes before NACT, which were assessed clinically or radiologically. All patients underwent needle biopsy of the primary tumor for pathological diagnosis before at least three cycles of NACT comprising docetaxel (75 mg/m2) and doxorubicin (50 mg/m2). IHC results of ER, PR, HER2, and Ki-67 expression in pre-NACT biopsy specimens were obtained, including HER2 fluorescence in situ hybridization in cases that showed equivocal HER2 expression in the pre-NACT specimens from each patient. Only cases of luminal type, defined as tumors with ≥10% ER-positive tumor cells and negative HER2 expression, were included in this study.
All patients underwent either mastectomy or breast-conserving quadrantectomy an average of 4 weeks after completion of NACT. All resection specimens were pathologically examined including ypT and ypN categories, lymphovascular invasion (LVI), and RCB index, based on whole-slide evaluation of the resected specimens. RCB class was determined according to the criteria suggested by Symmans et al. [2]. Ipsilateral axillary lymph nodes were also dissected and histologically assessed in every patient. According to pathological assessment, only cases with RCB class 2 or 3 were included in the study because RCB class 1 cases contained insufficient tumor cells for analysis on tissue microarray (TMA) slides. After resection, adjuvant chemotherapy and endocrine therapy were administered to every patient, and radiation therapy was applied to the breast in patients who underwent quadrantectomy. The mean follow-up period was 91.0 months (range, 3.2 to 150.5 months).
Immunohistochemistry and computer-based assessment
To construct TMAs, each representative tumor area was identified in the post-NACT resection specimens by examining hematoxylin and eosin slides, cut from formalin-fixed, paraffin-embedded blocks, and 3-mm-diameter cores were manually arranged into recipient TMA blocks. IHC using selected antibodies was performed on 4-μm sections of the TMA slides using the BenchMark XT (Roche Diagnostics, Basel, Switzerland) for calreticulin (1:200, polyclonal, GeneTex, Irvine, CA, USA) or the BOND-MAX (Leica Biosystems, Nussloch, Germany) automatic stainer for clusterin (1:100, 41D, Merck Millipore, Burlington, MA, USA), galectin-3 (1:200, 9C4, Leica Biosystems), Ki-67 (1:300, MIB-1, Agilent Technologies, Santa Clara, CA, USA), mucin-1 (1:200, MA695, Leica Biosystems), and p27 (1:3000, 57, BD Biosciences, Franklin Lakes, NJ, USA).
Every IHC-stained TMA slide was scanned by the Aperio GT450 Scanner (Leica Biosystems). Cell counting and categorization of intensity, such as weak, moderate, and strong positivity, for each TMA specimen were performed using QuPath software ver. 0.6.0 [20]. A minimum of 100 cells per IHC sample were counted and analyzed in the nucleus (Ki-67 and p27) and cytoplasm (calreticulin, clusterin, galectin-3, and mucin-1) for each IHC sample. For all markers except Ki-67, the H-score was derived from the product of staining intensity and the percentage of stained tumor cells (scale 0–300) [21]. Cutoff values for each marker were determined based on approximate H-score values near the median (e.g., 10, 50, 100, and 150) and cases were subsequently dichotomized into low and high expression groups. In contrast, the Ki-67 index was assessed using a different method: it was defined solely as the percentage of tumor cell nuclei showing positive staining, regardless of staining intensity.
Statistical analysis
Statistical analysis and visualization were performed in RStudio using R ver. 4.5.2 software (R Foundation for Statistical Computing, Vienna, Austria). DFS was defined as the time from surgery to the first locoregional relapse or distant metastasis. Kaplan-Meier analysis and univariate and multivariate Cox proportional hazards models were assessed using ‘survminer’ ver. 0.5.1, ‘survival’ ver. 3.8-3, ‘broom’ ver. 1.0.12, ‘forestmodel’ ver. 0.6.2, ‘dplyr’ ver. 1.1.4, and ‘ggplot2’ ver. 4.0.0. Each p-value of <.05 was considered statistically significant.
RESULTS
Clinicopathological characteristics
A total of 55 cases were included in this study. The average age of the patients was 45.7 ± 8.87 years. According to the IHC results of pre-NACT biopsy specimens, 41 cases (74.5%) were PR-positive, determined by positive staining of ≥1%, and 11 (20.0%) and 24 cases (43.6%) showed a Ki-67 index ≥15% and <15%, respectively. According to the pathological examinations of resection specimens, 45 cases (81.8%) had lower ypT categories (1 or 2), whereas 10 cases (18.2%) had higher ypT categories (3 or 4). Ten cases (18.2%) were classified as ypN0, and axillary lymph node metastases were identified in the remaining cases. LVI was observed in 42 cases (76.4%). Based on the pathological assessment, 26 cases (47.3%) and 29 cases (52.7%) were classified as RCB classes 2 and 3, respectively (Table 1).
Clinicopathological data of the patients
In the analysis of TMA slides, the average Ki-67 index was 3.00%. The mean H-scores of the other IHC markers were 44.29 for calreticulin, 33.22 for clusterin, 56.84 for galectin-3, 131.99 for mucin-1, and 126.25 for p27 (Fig. 1). The mean counts of tumor cells per core for each marker were as follows: 3,868.60 for Ki-67, 1,549.98 for calreticulin, 876.18 for clusterin, 1,661.69 for galectin-3, 1,498.49 for mucin-1, and 1,822.47 for p27 (Table 2, Fig. 2, Supplementary Fig. S1). Based on these results, the cutoff values for the IHC markers other than the Ki-67 index were determined as follows: clusterin, 10; calreticulin, 50; galectin-3, 50; p27, 100; and mucin-1, 150 (all H-scores).
QuPath-based analysis of immunohistochemical staining for Ki-67 and galectin-3. (A) Ki-67–positive and –negative cells are outlined with red and blue lines, respectively. (B) Weakly, moderately, and strongly galectin-3 positive cells are outlined with yellow, orange, and red lines, respectively.
Immunohistochemical results and numbers of tumor cells assessed for each marker in each tissue microarray core
Immunohistochemical expression levels of Ki-67 index and galectin-3 in resection specimens after neoadjuvant chemotherapy. (A) Ki-67 index (%) for each case, displayed in ascending order. (B) Galectin-3 expression evaluated using the H-score, displayed in ascending order. Red bars denote patients who developed recurrence or distant metastasis during follow-up, while blue bars indicate patients who remained event-free. The black dashed line indicates the cutoff value used to dichotomize the lower and higher expression groups.
Analysis of clinicopathological factors associated with disease-free survival
During a maximum follow-up period of 150.5 months, 20 of the 55 patients (36.3%) experienced recurrence or distant metastasis during follow-up. Since only two deaths (3.63%) occurred due to recurrence and/or distant metastasis, an analysis of the association between clinicopathological markers and overall survival was precluded. Kaplan-Meier analysis revealed that a Ki-67 index > 2.7% (p < .001) and high expression of galectin-3 (p = .018) were significantly associated with shorter DFS, while higher ypT categories (p = .227) showed a non-significant tendency towards shorter DFS (Fig. 3). All other clinicopathological factors, including age < 45 years (p = .343), histologic grade 3 (p = .333), pre-NACT PR (p = .448), RCB class (p = .778), ypN category (p = .534), LVI (p = .420), calreticulin (p = .851), clusterin (p = .461), mucin-1 (p = .806), and p27 (p = .583), did not reach statistical significance (Supplementary Fig. S2).
Kaplan-Meier of disease-free survival curves according to age (A), residual cancer burden class (B), ypT category (C), ypN category (D), histologic grade (E), lymphovascular invasion (F), Ki-67 index (G), and galectin-3 expression (H).
In the univariate Cox proportional hazards model, a Ki-67 index > 2.7% (hazard ratio [HR], 5.15; p < .001) and high galectin-3 expression (HR, 2.90; p = .024) were significantly associated with shorter DFS. Higher ypT categories showed a non-significant tendency towards shorter DFS (p = .235). In the multivariate analysis, after adjusting for age, histologic grade, ypT and ypN categories, LVI, RCB class, Ki-67 index, and galectin-3, both a Ki-67 index > 2.7% (HR, 7.23; p < .001) and high galectin-3 expression (HR, 4.51; p = .007) remained independently associated with shorter DFS (Table 3, Fig. 4).
Results of univariate and multivariate Cox proportional hazards models
Forest plots demonstrate the association between clinicopathologic factors, including Ki-67 index and galectin-3 expression, and disease-free survival. (A) Results of the univariate Cox proportional hazards model. (B) Results of the multivariate Cox proportional hazards model including selected covariates. NACT, neoadjuvant chemotherapy; PR, progesterone receptor; LVI, lymphovascular invasion; RCB, residual cancer burden.
DISCUSSION
The association between the Ki-67 index and clinical outcomes has been evaluated using various approaches in BC patients receiving NACT, including assessments of post-NACT resection specimens [8]. In this meta-analysis, the Ki-67 index measured after NACT predicted DFS more accurately than the pre-NACT Ki-67 index, suggesting that elevated proliferative activity in residual carcinoma after NACT reflects aggressive tumor biology and that more intensive adjuvant treatment should be considered to prevent recurrence or distant metastasis.
A key distinction of our study is that a cutoff value of 2.7%, referred to as the CCCA cutoff, is similarly effective in patients treated with taxane- and adriamycin-based NACT and in those receiving neoadjuvant endocrine therapy and/or CDK4/6 inhibitors. In contrast, previous studies using conventional chemotherapy have typically applied higher cutoff values (≥14%), originally suggested to discriminate luminal B subtype from the low-grade, slow-growing luminal A subtype [8,22]. Furthermore, these studies applied a uniform cutoff value regardless of BC subtype. Our previous study revealed that LTBC cases showed a higher rate of residual disease, whereas the Ki-67 index decreased significantly in the resection specimens following NACT [7]. Therefore, we propose that cutoff values for the Ki-67 index in post-NACT specimens should be established in a subtype-specific manner. Our study demonstrated that a post-NACT Ki-67 index > 2.7% serves as a significant independent predictor of shorter DFS in patients with LTBC, consistent with findings from studies using neoadjuvant endocrine therapy and/or CDK4/6 inhibitors [11,12]. For patients exceeding this threshold, more aggressive therapeutic strategies should be strongly considered to mitigate the increased risk of recurrence.
Galectin-3 is a member of the lectin family that contains a carbohydrate-binding domain with affinity for β-galactosides, and its overexpression has been implicated in tumor progression, metastasis, and immune evasion [18]. Several studies demonstrated that galectin-3 expression was significantly higher in TNBC than in other types, and higher in less responsive groups (stable and progressive diseases) than in more responsive groups (complete and partial responses). They also showed that galectin-3 overexpression was associated with chemotherapy resistance in TNBC cell lines and that its knockdown increased sensitivity to chemotherapy [23,24]. Raiter et al. [24] suggested that suppression of galectin-3 may represent a novel therapeutic strategy, because galectin-3 derived from TNBC cells induced the recruitment of regulatory T-cells into the tumor microenvironment, disrupted T-cell mitochondrial function, and subsequently led to an increase in exhausted T cells and loss of anti-tumor immune responses. Our study revealed that high galectin-3 expression was associated with shorter DFS even in LTBC. However, it remains unclear whether this reflects NACT-induced upregulation or intrinsic overexpression in tumors with worse outcomes, as galectin-3 expression was not assessed in pre-NACT biopsy specimens, although LTBC has generally been characterized by low galectin-3 expression [18]. Further studies, including the assessment of galectin-3 expression in pre-NACT biopsy specimens, the examination of the association between galectin-3 expression and the antitumor immune response in LTBC, and in vitro analyses involving the induction or inhibition of galectin-3 expression in hormone receptor-positive cell lines and the evaluation of subsequent changes in hormone receptor expression, are needed to clarify the role of galectin-3 in chemotherapy resistance.
Our study failed to show that RCB class was associated with DFS in these patients. However, this result should not be interpreted as indicating that the assessment of RCB index after NACT is not useful in LTBC patients, because our study included only cases of residual LTBC classified as RCB-2 or RCB-3. Notably, patients with RCB-1 were excluded due to insufficient tumor cells for TMA construction, which may have introduced selection bias and limited the representation of the whole RCB spectrum. Although pathological complete response is a rare event in patients with LTBC who receive NACT, it has been generally reported that patients with RCB-0 have longer DFS and overall survival than those with RCB-3 [4-6]. In a recent meta-analysis, DFS was similar between cases with RCB-0 and RCB-1 in LTBC [25]. Therefore, the exclusion of RCB-1 cases, which may have prognostic outcomes closer to RCB-0, could have attenuated potential differences in DFS across RCB classes in this study. These findings suggest that the prognostic significance of RCB may differ among breast cancer subtypes with respect to its detailed prognostic implications, and its utility in LTBC requires further investigation in larger cohorts.
This study has several limitations. First, the relatively small sample size from a single institution limits statistical power and external validity, which may in part reflect the restricted inclusion criteria, including the selection of cases requiring long-term follow-up and the inclusion of only a subset of LTBC cases with residual disease categories (RCB-2 and RCB-3). In addition, the wide confidence intervals of the HRs for Ki-67 index and galectin-3 suggest that the estimates may be imprecise and unstable, thereby limiting the robustness of the observed independent prognostic effects. Second, the potential confounding effects of pre-NACT clinicopathological variables and post-NACT treatment modalities, including pre-NACT clinical T and N categories, adjuvant chemotherapy, endocrine therapy, and radiation therapy, could not be completely excluded, and these factors may have influenced patient outcomes. Finally, the use of TMA specimens instead of whole tumor sections for image analysis represents another limitation, given the possibility of intratumoral heterogeneity in the expression of IHC markers, especially following NACT. This limitation is particularly relevant in the post-NACT setting, where treatment-related changes may accentuate spatial heterogeneity and lead to discordant expression patterns across different tumor areas. Therefore, the results of this study should be interpreted with caution, and further studies using whole sections of resection specimens are warranted. Despite these limitations, our results may provide meaningful insights into the potential prognostic value of the Ki-67 index and/or galectin-3 expression in this specific clinical context. Further larger-scale, multi-institutional studies are required to validate these findings.
Supplementary Information
The Data Supplement is available with this article at https://doi.org/10.4132/jptm.2026.06.22.
Notes
Ethics Statement
All procedures performed in this study were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. This study was approved by the Institutional Review Board of Seoul National University Hospital (IRB No. 2006-086-1132) and informed consent was waived for this study.
Availability of Data and Material
The datasets generated or analyzed during the study are available from the corresponding author on reasonable request.
Code Availability
Not applicable.
Author Contributions
Conceptualization: IAP. Data curation: HCL. Formal analysis: HCL. Investigation: HCL. Methodology: HCL, IAP. Project administration: IAP. Resources: WH, SAI. Supervision: IAP. Visualization: HCL. Writing—original draft: HCL. Writing—review & editing: WH, SAI, IAP. Approval of final manuscript: all authors.
Conflicts of Interest
The authors declare that they have no potential conflicts of interest.
Funding Statement
No funding to declare.