High Ki-67 expression is a marker of poor survival in apocrine breast carcinoma

Introduction

Apocrine carcinoma is rare type of breast cancer, which represents 0.5 to 4% of all breast cancers [1, 2, 3] with evident apocrine differentiation in more than 90% of cancer cells [2]. Microscopically, apocrine tumour cells are distinguishable as large cells with abundant acidophilic, granular cytoplasm and nuclei with prominent nucleoli [2, 3]. It is reported that apocrine carcinomas are usually androgen receptor (AR)-positive and oestrogen/progesterone (ER/PgR)-negative tumours [5]. However, this phenotype (AR+/ER–/PgR–), which is very often observed in intraductal apocrine carcinoma is more variable in invasive carcinomas [3, 4]. Farmer et al. on the basis of microarray data distinguished molecular apocrine group, defined as AR+/ER– [6]. Based on morphological features, another term was introduced: pure apocrine carcinoma [7]; however, some authors define this category as tumours characterised by apocrine differentiation with AR+/ER–/PgR– immunohistochemical (IHC) profile [8, 9]. These two definitions are overlapping because some authors use immunochemistry phenotype AR+/ER–/PgR– to define molecular apocrine breast carcinoma [10, 11]. The clinicopathological and biological characteristics of apocrine carcinoma are intensively studied; however, comparing data might be a problem because some authors investigate differently characterized groups: apocrine carcinoma/pure apocrine carcinoma (based on only morphological feature) [4, 8, 12, 13, 14, 15, 16, 17, 18] or with additional AR+/ER–/PgR– IHC profile [8, 9, 19], molecular apocrine [10, 20, 21, 22, 23, 24], or the subgroup of apocrine carcinomas in triple-negative breast carcinoma subtype [25]. The expression of a number of biological markers have been investigated, mainly: AR, ER, and PgR [4, 8, 9, 11, 13, 16, 17, 18, 19, 20, 24, 25] as well as epidermal growth factor receptor 1 and/or 2 (EGFR, HER2) [8, 9, 10, 11, 13, 15, 16, 17, 18, 19, 20, 22, 23, 25], proliferation rate based on assessment of Ki-67 expression [8, 10, 11, 16, 18, 19, 20, 25], gross cystic disease fluid protein-15 (GCDFP-15) [8, 13, 18, 21, 23], p53 [10, 11, 13, 16, 18, 19], CK5/6 [18, 20, 22] and bcl2 [4, 16]. In intraductal apocrine carcinoma and in apocrine invasive carcinoma gains of 1p, 1q, and 2q and losses of 1p, 17q, 22q, 16q, and 12q were detected [26]. Some authors noted the same prognosis for invasive apocrine carcinoma and invasive ductal carcinoma [17] and others did not observe differences between molecular apocrine and others subtypes of breast carcinoma [20]. However another study stated that molecular apocrine breast carcinoma (MABC) subgroup developed distant metastasis earlier than nonMABC subgroup, and patients with MABC had poorer prognosis [11]. Therefore the search for markers useful for treatment selection is still important. Herein we analysed expression of ER, PgR, HER2, AR, GCDFP-15, EGFR, CK5/6, CK8/18 and Ki-67 in 57 patients with apocrine breast carcinoma defined using morphological classification. Relationships between the parameters listed above and their influence on patients’ survival were investigated.

Material and methods

Patients

The studied group consist of 57 patients with apocrine breast carcinoma, who underwent treatment between 1987 and 2010 at Maria Sk³odowska-Curie Memorial Institute – Oncology Center, Cracow Branch. Those patients accounted for 0.5% of all breast carcinoma patients treated during that period. Apocrine carcinoma was defined according to the new WHO classification of breast tumours revised and updated in 2012. The age of patients ranged from 34 to 83 years with a mean value of 58.1 ±12.0 (SD) years. All patients underwent radical surgery. Radiotherapy, chemotherapy and hormonotherapy were applied after surgery, according to recommendations valid at that time. Clinical and histological characteristics of the patients are presented in Table I. The study was approved by the Ethical Committee at the Regional Medical Chamber in Krakow (decision from 17th April 2014). This study was conducted on archived tissues, with no direct patient contact, no modification of diagnostic or treatment procedures, and full confidentiality of patients’ personal data. Formal consent is not required for this type of study.

Methods

The study was performed using archival formalin-fixed paraffin-embedded (FFPE) tissues. Tumour specimens were reexamined by a pathologist (J.W.) to confirm histological evaluation and to assess tumour grade. Apocrine carcinoma was defined according to morphological features listed in the 2012 WHO classification. The presence of emboli and extent of lymphocytic infiltration (TILs) were assessed on hematoxylin-eosin-stained (HE) slides. The magnitude of infiltration was assessed using International TILs Working Group recommendation [27] with the authors’ modification: tumours were stratified into two groups: low (0-10% of stromal TILs) and high (> 10% of stromal TILs).

Immunohistochemistry staining

Expression of investigated proteins was assessed immunohistochemically. The primary antibodies used were as follows: AR (SP107, 1:100, Cell Marque), CK5/6 (D5/16 B4, 1:50, DakoCytomation, Agilent Technologies, Inc), CK5 (XM26, 1:80, Thermo Fisher Scientific), CK8/18 (NCL-5D3, 1:200, Leica Biosystems), EGFR (H11, 1:200, DakoCytomation, Agilent Technologies, Inc), ER (SP1, 1:250, Thermo Fisher Scientific), PR (SP2, 1:100, Thermo Fisher Scientific), Ki-67 (MIB, 1:100, DakoCytomation, Agilent Technologies, Inc), GCDFP (NCL-GCDFP15, 1:20, Leica Biosystems), and HER2 (HercepTest, DakoCytomation, Agilent Technologies, Inc). The sections from formalin-fixed paraffin-embedded tissues (4 µm) were mounted on SuperFrost Plus (Menzel-Gläser, Germany) slides. Proteinase K was used to unmask epitope in the case of EGFR and CK8/18 proteins, for AR, CK5/6 and CK5 incubation with TRS pH6 (50 min, 96°C, DakoCytomation) and for the rest heating in citrate buffer pH6 in microwave oven (2 × 600 W) was applied. After incubation with primary antibody, for protein visualization, detection system BrighVision Immunologic or Ultra-vision (Thermo Fisher Scientific) and DAB (Vector Laboratories, Inc., Burlingame, USA) were used. Hematoxylin was applied for nuclear counterstaining. Due to insufficient amount of tissue in paraffin blocks or small fragments of tumour tissue, which has precluded obtaining reliable results, whole panel of IHC staining was not available for some cases.

IHC evaluation

Tumours were considered as ER or PR (Fig. 1A, B) immunopositive if nuclear staining was observed in more than 1% of tumour cells. Overexpression of HER2 (Fig. 1C) was assessed using HercepTest (Dako Denmark A/S, Glostrup Denmark) and evaluated according to recommended standards [28]. In case of: AR, GCDFP-15, EGFR, CK 5/6 and CK 8/18 expression (Fig. 1D-H), tumours were classified as positive if staining was detected in more than 10% of tumours cell. In most cases strong expression of above mentioned proteins was observed in > 50% cells. Finally, Ki-67 labeling index (Ki-67 LI) was calculated as the percentage of tumour cells with Ki-67 nuclear staining (Fig. 1I).

Statistical analysis

For continuous variables descriptive statistics were used to determine mean values and standard deviation of means (SD). The Mann-Whitney test was applied for evaluation of differences between two groups. Relationships between categorical variables were analysed using Pearson χ2 test, or in the case of 2 × 2 tables two-sided Fisher exact test for independence was used in order to avoid problems with small samples. Overall survival (OS) and disease-free survival (DFS) were estimated using the Kaplan-Meier method and the log-rank test was used to evaluate the statistical significance of the differences observed between two groups. A Cox proportional hazard model was applied to test independence of variables. In all statistical procedures, p-value below 0.05 were considered significant. The statistical analysis was carried out using STATISTICA v.13 software (TIBCO Software Inc., Palo Alto, CA, USA).

Results

Relationships between molecular and clinical parameters

Emboli were observed in 19 tumours (34.5%) and high lymphocyte infiltration in 15 tumours (28.8%). Almost all cases were CK8/18, GCDEP-15 and AR positive (96.2%, 94.4%, and 90.7%, respectively). Expression of EGFR, HER2 and CK5/6 was noted in approximately half of the tumours (50.0%, 45.5%, and 48.1%, respectively). About one third of tumours were ER or PgR positive (29.1%). Statistical analysis revealed that the presence of smaller tumours (pT1) was related to less frequent lymph node involvement (p = 0.001, pN0 vs. pN+) and more frequent presence of G2 tumours (p = 0.012, G2 vs. G3). Absence of emboli was more frequent in tumours without lymph node metastasis (p = 0.011, pN0 vs. pN+). Less prominent lymphocyte infiltration was observed in the group of tumours with pN0 + pN1 (p = 0.003). The ER/PgR-negative tumours were more often characterised by CK5/6 and EGFR positivity (p = 0.014, and p = 0.006, respectively; Table II). Presence of androgen receptors was related to HER-2 and GCDFP-15 expression (p = 0.061, and p = 0.021, respectively; Table II). Additionally, tumours with expression of CK5/6 were more likely to be EGFR positive (p < 0.001; Table II). Expression of CK5/6 was also related to higher Ki-67 LI (p = 0.017, Table II).

Survival analysis

In studied group mean time of follow-up was 99.4 months (range 6-285 months, median 109 months). During this time 22 patients died and 15 deaths were cancer related. Locoregional recurrence developed in 8 patients and distant metastases were observed in 13 cases. The probability of 10-years overall survival in the whole group was 66.2% and disease-free survival 62.7%. For Ki-67 LI, a value of 20% was established as cut-off point using the minimal p-value method from the log-rank test. In univariate analysis, longer OS was observed in patients with tumours characterised by Ki-67 LI < 20% (p = 0.036; Table III, Fig. 2A) and an favourable trend was noted for patients with smaller tumours (p = 0.053; Table III), without lymph node metastases (p = 0.074; Table III) and without EGFR expression (p = 0.060; Table III). In the case of DSF only Ki-67 LI < 20% was a statistically significant favourable factor (p = 0.009, Table III; Fig. 2B), and a trend was not observed for any variables. Cox multivariate analysis confirmed that Ki-67 LI < 20% is independent positive prognostic factors influencing overall and disease-free survival of patients with apocrine carcinoma (Table IV).

Discussion

The analysed group comprised 57 apocrine carcinomas treated between 1987 and 2010 at Maria Sklodowska-Curie Memorial Institute – Oncology Center, Cracow Branch. Apocrine carcinoma was defined according to 2012 WHO morphological criteria. This group accounted for 0.5% of all breast cancer patients treated during that time period, which is within the reported range [1, 2, 3]. As we mentioned before comparing biological features between different studies can be problematic. Firstly, because apocrine carcinoma is such a rarely occurring subtype of breast carcinoma, the analysed groups were rather small. Therefore, differences concerning a few cases can strongly influence the final results. Secondly, the authors selected groups using various definitions: apocrine carcinoma/pure apocrine carcinoma (based only on morphological feature) or with additional AR+/ER–/PgR– IHC profile or molecular apocrine which can also be differently defined. In order to have better insight and better understanding of this diversity and problems concerning comparing of studies, we have prepared Table V, which contains information about classifications used, the number of patients, and the investigated parameters. Our analysed group of 57 apocrine carcinomas was defined according to morphological features and is one of the largest studied groups (Table V). Studies concerning molecular apocrine carcinoma usually are conducted on larger groups. But even within studies concerning apocrine carcinoma defined by morphological features, authors used different scales and cut off points for expression of proteins. In the case of HER2 it was probably caused by changes in the ASCO/CAP recommendations. In our group ER/PgR and HER2 positivity was observed in 29.1% and 45.5% respectively, which is within observed range (Table V). Apocrine carcinoma is assumed to be ER/PgR-negative and AR-positive. Some authors combining morphological features with this IHC phenotype created pure apocrine carcinoma subtype, while others use IHC phenotype to distinguish molecular apocrine. In our group 90.1% of the tumours were AR-positive; however, one third were ER/PgR positive. This is consistent with reports noting that the phenotype (AR+/ER–/PgR–) is characteristic for benign apocrine proliferation, very often observed in intraductal apocrine carcinomas, but less frequent in invasive carcinomas [3, 4]. Expression of EGFR, GCDFP-15 was similar to the results reported by other authors. However, we observed expression of CK5/6 more frequently than other authors investigating similarly defined apocrine groups. Similar results to ours concerning CK8 expression were reported by Guo et al. [22]. The relationship between ER/PgR lack of expression and EGFR expression was reported not only in apocrine carcinomas [10, 18] but also in other breast carcinoma subtypes [29]. Because GCDFP-15 expression is reported to be a frequent feature of apocrine carcinoma [4, 13, 14] as well as AR expression, it is not surprising that the relationship between the presence of these two proteins was observed not only in our study but also by others authors [21]. Lehman-Che et al. suggested that molecular apocrine subtype could be better defined using IHC phenotype ER– and HER2+ or GCDFP15+ [23]. One the most important factors from a clinical point of view is patients` survival. Table VI represents a review of studies concerning the survival of patients with apocrine carcinoma. Usually the studies investigated the differences between apocrine/molecular apocrine carcinomas and other breast carcinoma groups (defined by the authors; Table VI) [7, 8, 10, 11, 12, 17, 20, 23, 24, 25, 30, 31]. In our study the probability of 10-years overall survival was 66.2% and disease-free survival was 62.7%. This is in agreement with observation of other authors, who reported 6-years OS for apocrine carcinoma of 72% against 52% for invasive ductal carcinoma [7], or 5-year disease-free survival and OS 66% and 77%, respectively for molecular apocrine [23]. The estimated 10-year OS in the group studied by di’Amore was 52% [31]. However, higher values of  OS, at around 80%, for apocrine and invasive ductal carcinoma [17] or non-apocrine carcinoma [30] were noted in Japanese women after 10 years observation. Comparison between our result concerning survival of apocrine carcinoma and other subtypes of invasive ductal cancer is difficult due to wide range of estimated OS and DFS for those phenotypes [20, 24, 32, 33]. It is also worth mentioning that those subtypes could be differently selected, and different treatment. protocols could influence probability of survival [33]. Only a few studies have investigated survival according to biological factors in apocrine carcinomas groups [10, 20, 21, 22, 23, 30]. The problem concerning analysis of survival is very complex, especially in the case of apocrine carcinomas classified according to morphological features. Because this type of cancer is so rare, groups consist of cases collected over a long time (Table VI). It is obvious that during that time, treatment regimens might change considerably and new drugs might be implemented, which could influence patient survival. Nevertheless, in our study longer OS and DFS were observed in patients with tumours characterised by low Ki-67 LI (< 20%), and a favourable trend was observed for OS in cases with smaller tumours, without lymph node metastases and without EGFR expression. Similar results acquired Liu et al., who reported better DFS and OS for patients with lower Ki-67 LI (< 20%) tumours, however in that study a molecular apocrine subgroup was investigated [10]. Guo et al. did not find a relationship between OS and Ki-67, but those authors investigated molecular apocrine carcinoma and used lower cut-off point (> 14%) [22]. Ki-67 is a well-known biological factor, which, along with ER, PgR, and HER2, can be useful in the assessment of molecular breast cancer subtypes. St. Gallen’s recommendation from 2013 stated that cut-off point of Ki-67 ≥ 20% could be used to define luminal B, however, a lack of methodological standarization was also mentioned [34]. Two recent meta-analyses concerning Ki-67 in breast cancer, confirmed that high Ki-67 LI is unfavourable prognostic factor [35, 36]. Both studies also addressed the issue of cut-off point. Liu et al. demonstrated that the most appropriate Ki-67 cut-off point for predicting chemosensitivity, may be a value close to ≥ 19% in hormone receptor-positive patients [35]. A second study which included more than 64 000 patients stated that although the Ki-67 threshold with the greatest prognostic significance is still unknown, a cut-off > 25% is related to higher probability of death compared with lower expression rates [36]. Because many studies have demonstrated, that EGFR expression might be feature of more invasive breast carcinoma phenotype, it is not surprising that also in our studied apocrine carcinoma group the expression of EGFR tended to be related with poorer OS. Similarly, Liu et al. in a molecular apocrine tumours group noted relation between shorter DFS and EGFR expression [10]; however, other study did not confirm connection between OS and EGFR [22]. We noted that HER2 expression was not related to OS or DFS. Conversely, some authors observed that HER2-positivity was associated with shorter survival [10, 20, 22], however all those studies were conducted in molecular apocrine groups defined by IHC phenotype (Tables V and VI). In a molecular apocrine group defined by qRT-PCR signature, Lehmann-Che et al. did not report this relation [23]. Similarly to others authors, we also did not noted relation between GCDFP-15, CK5/6, AR, ER, and PgR expression and survival [21, 22, 23, 30].

Conclusions

In the studied group of apocrine carcinomas, defined by morphological features, the ER/PgR-negative tumours were more often characterised by CK5/6 and EGFR positivity. Additionally, the presence of androgen receptor was related to HER-2 and GCDFP-15 expression. Data suggest that in breast apocrine carcinomas, like in other breast cancers, high Ki-67 proliferation index is one of the most important factors related to shorter OS and DFS survival. Expression of other investigated proteins (ER, PgR, HER2, AR, GCDFP-15, EGFR, CK 5/6, CK 8/18) was not statistically significant for OS and DFS It is worth mentioning that there is a problem in the comparison of studies concerning apocrine carcinomas, due to the different types of classification applied by authors. In the literature, the following terms are used presently: apocrine carcinoma, pure apocrine carcinoma, and molecular apocrine carcinoma, which are defined differently by authors and could overlap. Sometimes the same “term” could mean differently defined groups or different “terms” are applied to describe the same group. It seems that there is an urgent need for unification of classification of different apocrine carcinoma subgroups. Clinical practice points:
1. Our data suggest that in breast apocrine carcinomas one of the most important factor related to shorter survival, like in other breast cancers, is high Ki-67 proliferation index.
2. There is a problem concerning the clear definition of apocrine carcinoma, pure apocrine carcinoma, molecular apocrine carcinoma, because in different studies the same “name” could mean differently selected groups or different “names” describe the same group. It seems that there is urgent need for unification of these classifications.
The authors declare no conflict of interest.

References

1. Celis JE, Gromova I, Gromov P, et al. Molecular pathology of breast apocrine carcinomas: a protein expression signature specific for benign apocrine metaplasia. FEBS Lett 2006; 580: 2935-2944.
2. Makki J. Diversity of Breast Carcinoma: Histological Subtypes and Clinical Relevance. Clin Med Insights Pathol 2015; 8: 23-31.
3. O’Malley FP, Bane A. An update on apocrine lesions of the breast. Histopathology 2008; 52: 3-10.
4. Honma N, Takubo K, Akiyama F, et al. Expression of GCDFP-15 and AR decreases in larger or node-positive apocrine carcinomas of the breast. Histopathology 2005; 47: 195-201.
5. Masood S, Rosa M. The challenge of apocrine proliferations of the breast: a morphologic approach. Pathol Res Pract 2009; 205: 155-164.
6. Farmer P, Bonnefoi H, Becette V, et al. Identification of molecular apocrine breast tumours by microarray analysis. Oncogene 2005; 24: 4660-4671.
7. Japaze H, Emina J, Diaz C, et al. “Pure” invasive apocrine carcinoma of the breast: a new clinicopathological entity? Breast 2005; 14: 3-10.
8. Dellapasqua S, Maisonneuve P, Viale G, et al. Immunohistochemically defined subtypes and outcome of apocrine breast cancer. Clin Breast Cancer 2013; 13: 95-102.
9. Vranic S, Tawfik O, Palazzo J, et al. EGFR and HER-2/neu expression in invasive apocrine carcinoma of the breast. Mod Pathol 2010; 23: 644-653.
10. Liu X, Feng C, Liu J, et al. The importance of EGFR as a biomarker in molecular apocrine breast cancer. Hum Pathol 2018; 77: 1-10.
11. Liu X, Yang Y, Feng X, et al. Early versus late distant metastasis and adjuvant chemotherapy alone versus both radiotherapy and chemotherapy in molecular apocrine breast cancer. Oncotarget 2016; 7: 48905-48917.
12. Dreyer G, Vandorpe T, Smeets A, et al. Triple negative breast cancer: clinical characteristics in the different histological subtypes. Breast 2013; 22: 761-766.
13. Kaya H, Bozkurt SU, Erbarut I, et al. Apocrine carcinomas of the breast in Turkish women: hormone receptors, c-erbB-2 and p53 immunoexpression. Pathol Res Pract 2008; 204: 367-371.
14. Kasashima S, Kawashima A, Ozaki S, et al. Expression of 5α-reductase in apocrine carcinoma of the breast and its correlation with clinicopathological aggressiveness. Histopathology 2012; 60: E51-E57.
15. Mills AM, Gottlieb CE, Wendroth SM, et al. Pure Apocrine Carcinomas Represent a Clinicopathologically Distinct Androgen Receptor-Positive Subset of Triple-Negative Breast Cancers. Am J Surg Pathol 2016; 40: 1109-1116.
16. Matsuo K, Fukutomi T, Hasegawa T, et al. Histological and immunohistochemical analysis of apocrine breast carcinoma. Breast Cancer 2002; 9: 43-49.
17. Tanaka K, Imoto S, Wada N, et al. Invasive apocrine carcinoma of the breast: clinicopathologic features of 57 patients. Breast J 2008; 14: 164-168.
18. Vranic S, Marchiò C, Castellano I, et al. Immunohistochemical and molecular profiling of histologically defined apocrine carcinomas of the breast. Hum Pathol 2015; 46: 1350-1359.
19. Tsutsumi Y. Apocrine carcinoma as triple-negative breast cancer: novel definition of apocrine-type carcinoma as estrogen/progesterone receptor-negative and androgen receptor-positive invasive ductal carcinoma. Jpn J Clin Oncol 2012; 42: 375-386.
20. Cha YJ, Jung W, Koo JS. The Clinicopathologic Features of Molecular Apocrine Breast Cancer. Korean J Pathol 2012; 46: 169-176.
21. Darb-Esfahani S, von Minckwitz G, Denkert C, et al. Gross cystic disease fluid protein 15 (GCDFP-15) expression in breast cancer subtypes. BMC Cancer 2014; 14: 546.
22. Guo W, Wang W, Zhu Y, et al. HER2 status in molecular apocrine breast cancer: associations with clinical, pathological, and molecular features. Int J Clin Exp Pathol 2015; 8: 8008-8017.
23. Lehmann-Che J, Hamy AS, Porcher R, et al. Molecular apocrine breast cancers are aggressive estrogen receptor negative tumors overexpressing either HER2 or GCDFP15. Breast Cancer Res 2013; 15: R37.
24. Lakis S, Kotoula V, Eleftheraki AG, et al. The androgen receptor as a surrogate marker for molecular apocrine breast cancer subtyping. Breast 2014; 23: 234-243.
25. Montagna E, Maisonneuve P, Rotmensz N, et al. Heterogeneity of triple-negative breast cancer: histologic subtyping to inform the outcome. Clin Breast Cancer 2013; 13: 31-39.
26. Masuda S. Breast cancer pathology: the impact of molecular taxonomy on morphological taxonomy. Pathol Int 2012; 62: 295-302.
27. Salgado R, Denkert C, Demaria S, et al. The evaluation of tumor-infiltrating lymphocytes (TILs) in breast cancer: recommendations by an International TILs Working Group 2014. Ann Oncol 2015; 26: 259-271.
28. Wolff AC, Hammond ME, Hicks DG, et al. Recommendations for human epidermal growth factor receptor 2 testing in breast cancer: American Society of Clinical Oncology/College of American Pathologists clinical practice guideline update. Arch Pathol Lab Med 2014; 138: 241-256.
29. Adamczyk A, Grela-Wojewoda A, Domaga³a-Haduch M, et al. Proteins Involved in HER2 Signalling Pathway, Their Relations and Influence on Metastasis-Free Survival in HER2-Positive Breast Cancer Patients Treated with Trastuzumab in Adjuvant Setting. J Cancer 2017; 8: 131-139.
30. Takeuchi H, Tsuji K, Ueo H, et al. Clinicopathological feature and long-term prognosis of apocrine carcinoma of the breast in Japanese women. Breast Cancer Res Treat 2004; 88: 49-54.
31. d’Amore ES, Terrier-Lacombe MJ, Travagli JP, et al. Invasive apocrine carcinoma of the breast: a long term follow-up study of 34 cases. Breast Cancer Res Treat 1988; 12: 37-44.
32. Adamczyk A, Niemiec JA, Ambicka A, et al. CD44/CD24 as potential prognostic markers in node-positive invasive ductal breast cancer patients treated with adjuvant chemotherapy. J Mol Histol 2014; 45: 35-45.
33. Niemiec JA, Adamczyk A, Ambicka A, et al. Prognostic role of lymphatic vessel density and lymphovascular invasion in chemotherapy-naive and chemotherapy-treated patients with invasive breast cancer. Am J Transl Res 2017; 9: 1435-1447.
34. Untch M, Gerber B, Harbeck N, et al. 13th st. Gallen international breast cancer conference 2013: primary therapy of early breast cancer evidence, controversies, consensus – opinion of a german team of experts (Zurich 2013). Breast Care (Basel) 2013; 8: 221-229.
35. Liu Y, Yin W, Yan T, et al. The clinical significance of Ki-67 as a marker of prognostic value and chemosensitivity prediction in hormone-receptor-positive breast cancer: a meta-analysis of the published literature. Curr Med Res Opin 2013; 29: 1453-1461.
36. Petrelli F, Viale G, Cabiddu M, et al. Prognostic value of different cut-off levels of Ki-67 in breast cancer: a systematic review and meta-analysis of 64,196 patients. Breast Cancer Res Treat 2015; 153: 477-491.