1 Division of Medical Oncology, Department of Medicine, 2 Division of Pathology and 3 Division of Epidemiology and Biostatistics, European Institute of Oncology, Milan, Italy
* Correspondence to: Dr L. Orlando, European Institute of Oncology, Via Ripamonti 435, 20141 Milan, Italy. Tel: +39-0-257-489460; Fax: +39-0-257-489457; Email: laura.orlando{at}ieo.it
![]() |
Abstract |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Patients and methods:: Evaluation was performed on node-negative tumor specimens from 40 patients who had either MC (12 patients), AMC (nine patients) or DC (19 patients), treated in a single institution. All had no hormonal receptor, Ki-67 30%, G3, expansive pattern of growth and peritumoral lymphocytic infiltration. In addition, p27, p21 and HER2/neu overexpression, p53, cyclin E and E-cadherin expression, presence of apoptotic cells, stromal tenascin (TN), and type of immune cell infiltration (CD3- and CD68-positive cells) were assessed.
Results:: No difference in expression of HER2/neu, p21, p27, p53, number of apoptotic cells and CD68-positive cells was detected. Lower levels of stromal TN expression were found in MC compared with DC (P=0.0007), but differences between MC and AMC were not significant (P=0.27). A higher proportion of intratumoral CD3-positive cells was seen in MC than in AMC (P=0.046). No differences were seen between MC and DC (P=0.73). With a median follow-up of 67 months, three patients with DC had relapsed in distant sites, while one patient with AMC had a second primary. Two patients with MC had reappearance of DC in the breast.
Conclusions:: The three distinct disease types, selected by having similar high proliferation, had similar expression of cell cycle regulators. The lower expression of TN and massive infiltration of T lymphocytes might both indicate a special interaction between tumor cells and microenvironment, important features for conferring improved prognosis through negligible invasive and metastatic potential to MC. In our series, however, patients with a previous MC are not free from the risk of developing a subsequent DC. Finally, defining AMC as a distinct entity from DC is not justified.
Key words: breast cancer, medullary phenotype, pathological features
![]() |
Introduction |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
MC is classified as such by pathologists because of its peculiar microscopic characteristics. If most, but not all of these are present, a tumor is identified as atypical medullary carcinoma (AMC). Lymphoplasmacytic reaction, microscopic circumscription, syncytial and expansive growth pattern, high nuclear grade, and mitotic rate are all well established diagnostic features [1, 4
]. Moreover, estrogen and progesterone receptor expression (biochemically determined in early series and immunohistochemically assessed in more recent studies) were reported as negative in virtually all cases [5
]. Good prognosis despite high cytological grade, absence of steroid hormone receptor expression and rapid growth rate, indicated the potential role of other, as yet undefined, biomarkers. Moreover, its poorly differentiated morphology represented a diagnostic challenge. The mutated p53 and overexpression of HER-2/neu were evaluated in studies, showing that the majority of MC had a positive nuclear expression of p53 [6
, 7
], while there was no immunostaining of HER-2/neu [8
].
The presence of some prominent lymphoplasmacytic cell infiltrates and their immunophenotype were hypothesized as an explanation for improved survival. These consist of T-cell predominance and abundant plasma cells, suggesting both cellular and humoral immune processes are involved [9, 10
]. The extent of apoptosis in medullary, atypical medullary and poorly differentiated carcinoma, the pattern of adhesion molecules involved, the role of cyclin and cyclin inhibitor and the activation of stroma were also analyzed [11
]. Given the particular prognosis of MC compared with other histological types of breast cancer, we retrospectively evaluated several biological and immunohistochemical features of typical medullary (MC), atypical medullary (AMC) and ductal (DC) breast carcinoma specimens selected for similar morphological and proliferation markers, as well as absence of expression of steroid hormone receptors, to explore differences among the three histopathological types and the characteristics of clinical course of disease associated with their presence.
![]() |
Materials and methods |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Formalin-fixed paraffin-embedded specimens from these 40 patients were used for immunohistochemical analyses. The samples were blinded for these analyses.
Serial sections from the same tissue blocks, selected for the presence of representative tumor tissue, were cut for histological verification and for assessment of steroid hormone receptors, grading, p21, p27, p53, HER-2/neu, Ki-67 labeling, cyclin E, E-cadherin, vascular invasion, immunotype analysis, terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick end labeling (TUNEL) evaluation for apoptosis and stromal tenascin (TN).
Immunohistochemistry
Immunohistochemical investigations were performed on routinely processed, formalin-fixed, paraffin-embedded tissues. Consecutive 4-µm tissue sections were cut from blocks selected for the presence of representative tumor tissue and mounted on poly-L-lysine-coated slides. Prior to incubation with specific antibodies, the sections were treated with 0.1 mM citrate buffer (pH 6) under microwave irradiation for 10 min at 750 W for antigen retrieval, and with 3% hydrogen peroxide in distilled water for 5 min at room temperature to quench endogenous peroxidase activity. Non-specific binding was blocked with normal horse serum. Immunohistochemistry was performed using an automated immunostainer (TechMate 500; Dako, Glostrup, Denmark) and a peroxidase-based detection system in kit form (ChemMate; Dako) according to the manufacturer's instruction.
Primary specific monoclonal antibodies to the following antigens were used at the specified dilutions: estrogen receptors (clone 1D5; Dako; 1:50); progesterone receptors (clone 1A6; Novocastra, Newcastle-upon-Tyne, UK; 1:20); Ki-67 (clone MIB-1; Immunotech, Marseille, France; 1:100); HER-2 (clone CB11; Biogenex; 1:80); p27 (clone AB1; Transduction Laboratories, Lexington, KY, USA; 1:600); p21 (clone 57; Oncogene Science, Cambridge, MA, USA; 1:100); and p53 (clone D0-7; Dako; 1:1000). For detection of CD3- and CD68-positive cells, immunohistochemical reactions were performed according to avidinbiotin peroxidase complex (ABC) method after antigen retrieval pretreatment in a microwave oven. For the immunodetection of TN, deparaffined sections were treated with pepsin (0.1% in 0.01 M HCl) for 50 min at 37 °C. Antibodies were diluted to 1:40 and to 1:25 in PBS/bovine serum albumin/sodium azide and sections were incubated for 90 min at 37 °C. STRAviGen Multilink (Biogenex) and StreptABComplex-AP (Dako) were used as secondary antibody. Immunoreaction was developed with New Fuchsin chromogen substrate (Dako) or with 20% 3.3-diaminobenzidine in distillated water and 0.2% H2O2. TUNEL assay for apoptotic cell detection was applied on paraffin-embedded tissues using In Situ Cell Death Detection, AP kit (Boehringer Mnnheim GmbH, Mannheim, Germany). Rehydratated tissue sections were microwave-pretreated in 10 µM citrate buffer, pH 6. After washing in PBS, the specimens were incubated in buffer containing nucleotide mixture with fluorescein-labeled deoxy-UTP and TdT for 1 h at 37 °C. After washing, the slides were incubated with blocking solution (10% non-immune goat serum) for 30 min at room temperature. To localize cells containing DNA strand breaks, section were washed and incubated with sheep antifluorescein antibody, Fab fragment, conjugated with alkaline phospatase at 37 °C for 30 min. TUNEL-positive color development was obtained by incubating the sections with BCIP/NBT substrate (Zymed Laboratories). Slides treated in a similar way but without adding TdT served as negative controls. Positive controls consisted of DNAseI-treated slides. The apoptotic index was defined as the ratio of positive cells or fragments to all tumor cells and expressed as percentage.
For cyclin E, standard indirect immunoperoxidase procedures were used for immunohistochemistry (ABC-elite; Vector Laboratories, Burlingame, CA, USA). A monoclonal antibody to human cyclin E was employed for detection of cyclin E expression (clone 13A3; Diagnostric Biosystem, Pleasanton, CA, USA), with pressure cooker pretreatment for 5 min and a dilution of 1:120.
For E-cadherin expression, after deparaffination, slides were pretreated in a pressure cooker in citrate buffer for 2 min. E-cadherin antibody (HECD-1; R&D Systems, Abingdon, UK) was used at a dilution of 1:150 and incubated overnight at 4 °C. This was followed by the use of a streptavidin biotin kit (Biotek Dako, Trappes, France).
Scoring
The cut-offs of all biological markers were established before the analyses. Immunoreactivity for estrogen and progesterone receptors was scored according to the percentage of neoplastic cells. We chose a percentage 1% to identify hormone receptor-positive tumors.
p21 and p27 immunoreactivity was scored in four groups according to the percentage of neoplastic cells showing definite nuclear staining. Cases were categorized as negative when <1% of cells were immunostained, low when 125%, moderate when 2650% and intense when >50% were immunostained. Within tumor cells, the staining reaction for p27 was often localized both in the nucleus and in the cytoplasm, but the nuclear staining was predominant and, in most cases, exclusive.
For Ki-67 labeling, the absolute percentage of the neoplastic cells showing nuclear immunoreactivity was recorded. Thirty per cent was used as cut-off point to select DC controls.
For HER-2/neu the score was assessed as follows. No staining at all, or membrane staining in <10% of the observed tumor cells was considered negative (0). A faint/barely perceptible membrane staining in >10% of tumor cells or staining of part of their membrane was scored as negative (1+). A weak to moderate staining of the entire membrane in >10% of the tumor cells was considered weakly positive (2+). A moderate to strong staining of the entire membrane in >10% of the tumor cells was scored as strongly positive (3+).
Depending on the percentage of nuclei exhibiting positive staining, tumors were categorized for p53 as follows: negative/patchy (0% to 10% of positive nuclei) or positive (>10%).
Immunoreaction for TN was evaluated in the stroma in a distance of no more than 0.5 mm from epithelial elements of the tumor. TN positivity was given as percentage of stromal positive cells on total stromal cells. The absence of TN immunoreactivity was scored as negative. A staining of <20% of the stromal cells was considered weakly positive. A staining of stromal cells between 20% and 70% of the stroma was scored as moderately positive. A staining of >70% was scored as strongly positive. Moreover, the presence (+) or absence () of TN immunoreactivity was also scored for epithelial cells. Depending on the percentage of nuclei exhibiting positive staining, tumors were categorized for cyclin E as <10% or >10% of positive nuclei. E-cadherin immunoreactivity was analyzed as a continuous variable, according to the percentage of neoplastic cells showing membrane staining. For intratumoral CD3-positive cells, cases were categorized as negative when 1% of intratumoral lymphocytes were immunostained for CD3, low when 220%, moderate when 2150% and intense when >50% were immunostained. For peritumoral CD3-positive cells, we identified negative tumors, weakly positive tumors (+), moderately positive tumors (++) and strongly positive tumors (+++), based on a qualitative assessment of the presence of peritumoral CD3-positive infiltration. CD68 cells and TUNEL were analyzed as continuous variables, according to the percentage of positive cells. Scoring of immunostaining results was independently performed by two pathologists, who had no knowledge of the tumor clinico-pathological data and patient survival. In each case, at least 2000 tumor cells were counted from 10 randomly selected areas, ensuring that the whole section was scanned. When the scoring discrepancy between the examiners was >10%, a consensus interpretation was reached after re-examination of the slides with a double-headed microscope.
Statistical analysis
We investigated whether the distribution of the biological parameters were similar among MC, AMC and DC using the Fisher's exact test for categorical parameters and the KruskalWallis rank sum test for the continuous parameters. The same tests were used to assess possible differences in the distribution of biomarkers by age and tumor characteristics. All tests were two-sided and considered significant in case of P value < 0.05.
![]() |
Results |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
|
|
|
![]() |
Discussion |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
Theoretically, a specific molecular finding of MC could facilitate the differential diagnosis with AMC and highly anaplastic DC. Moreover, the presence or absence of a particular aberration in key molecules involved in cell growth and progression able to explain the controlled MC expansion may be of clinical and therapeutic importance. We have not found significant differences in levels of biomarkers otherwise associated with poor prognosis in many histological types of breast cancer. Level of p53, found to be increased in MC in other series, was found mutated in 64% of MC, without statistically significant differences (P=0.49) from AMC (67%) and DC (42%). HER-2 status, often associated with high histological grade, larger tumor size, degree of axillary lymph node involvement, low or absent sex steroid hormone receptor expression, high proliferative activity and poor patient survival, has been evaluated in previous immunohistochemical study, showing that the majority of MC have absence of expression of HER-2 [8]. In our series, most (11 tumors) of MC and most (eight tumors) of AMC showed absence of HER-2, as did 17 of DC. p21 and p27 belong to the family of cyclin-dependent kinase (cdk) inhibitors and are well known prognostic factors in a broad array of cancers [12
, 13
]. As for p53, in our series we did not find any difference in level of p21 expression between the three groups, with similar distribution of values. Down-regulation of p27 is considered an early event in breast cancer progression. We showed that p27 expression distribution was similar in the three groups. This is not surprising, taking into account that loss of p27 has recently been found in breast cancer with an aggressive phenotype defined by hormonal receptor negativity, high histologic grade, c-erbB2 overexpression and high Ki-67 [14
], all features of MC. In the present study, the role of cyclin E was also investigated. Previous studies about the prognostic importance of cyclin E in breast cancer have yielded controversial results [15
, 16
]. Cyclin E is believed to control G1/S-phase progression. It is associated with cdk2 and activates its kinase activity shortly before entry of cells in S-phase. Cyclin E was reported to be overexpressed in a significant fraction of breast cancer and it is correlated with high grade and stage [17
]. Recently, amplification of cyclin E gene was observed in medullary breast cancer, in association to protein overexpression [16
]. In contrast with these data, we observed similar levels of cyclin E in the three groups analyzed. All these findings suggest that the good prognosis of MC compared with the other similar phenotypes could be unrelated to alterated expression of molecules involved in cell-cycle control usually related to aggressive phenotype. Otherwise, it is possible that additional alteration of such biomarkers might be implicated, as suggested in the recent work by Keyomarsi et al. [18
], in which the role of a truncated form of cyclin E was shown to be of relevant prognostic significance.
An alternative hypothesis to that of proliferative molecules involvement was to investigate the role of microenvironment on determining and controlling the MC growth pattern. TN is a large glycoprotein of the extracellular matrix expressed transiently during fetal development, inflammation, wound healing and neoplasia. TN is suspected to play a significant role in cancer invasion. It is highly expressed by tumor cells themselves and stromal cells such as fibroblast and endothelial cells. Previous studies on breast cancer have demonstrated that higher TN expression is associated with tumor progression and unfavorable outcome of patients [19]. We hypothesized that the loss of TN expression could be correlated with the less invasive behavior of MC compared with AMC and DC. As for ductal carcinoma in situ, in which TN expression may be related to early invasion capacity [20
], the lack of TN expression in MC could contribute to explain the weak invasive potential of such subtype of breast cancer. In our series of 40 tumors, a decreased intratumoral expression of TN was observed in MC compared with DC. Although no significant differences were detected between MC and AMC, a greater proportion of MC showed a complete absence of TN expression than AMC (45% versus 22%; P=0.27), albeit this difference did not reach statistical significance, probably owing to the small number of cases. It is reasonable that different isoforms of TN might be implicated, as recently shown by Adams et al. [21
]. A recent work suggested a reciprocal regulation of different matrix components; in particular TN seems to up-regulate metalloproteinase-9 and transforming growth factor-ß [22
]. We performed analysis of E-cadherin, another protein involved in cellcell interaction and adhesion. In humans, loss of E-cadherin function is associated with increased proliferation, invasion and metastasis in a variety of tumors. In breast cancer, inactivation and decreased expression of this molecule are frequently observed in invasive lobular breast carcinomas, in high-grade tumors and in tumors associated with syncytial pattern [23
]. In our series, no difference was observed between the three groups analyzed, with high level of expression in all histological types, in contrast to previous studies, in which loss of E-cadherin has been advocated as useful parameter to identify MC [24
]. Several other hypotheses have been put forward to explain the biological basis for the favorable prognosis of patients with MC, including the prominence of an effective host immune response [25
, 26
] and enhanced tumor cells apoptosis [27
]. Previous series have shown that the MC immune infiltrate predominantly consists of T, B and plasma cells. In this study, we determined the immunological profile of tumor infiltrating cells, by analyzing CD3- and CD68-positive cells. We found that the number of intratumoral CD3-positive cells were significantly increased in MC compared with the AMC group (P=0.046), while no differences between MC and DC were detected. In contrast, in a previous work by Tamiolakis et al. [28
], the infiltration of CD8- and CD20-positive cells was considered a feature of both MC and AMC, and not of DC. It is reasonable to hypothesize that our finding was owing to our choice to select DC based also on the presence of peritumoral lymphocytic infiltration. While the immune infiltration in MC is invoked as an explanation for the favorable clinical outcome, its presence in cases of DC does not necessarily imply an effective immune response against the tumor. Moreover, it is reasonable that humoral immunity can also have a role in MC, as suggested by the presence of abundant plasma cells found in some studies [26
]. It is also possible that differential analysis of CD8/CD4 subpopulations could find differences between the three groups under investigation. The absence of differences in CD68 distribution between the three groups supports the hypothesis that immune response in MC is a consequence of specific stimuli (perhaps tumor associated antigens) and not of inflammatory response. Owing to the fact that activated cytotoxic lymphocytes may trigger apoptosis, the degree of apoptosis was also evaluated in our series by TUNEL assay. Previous data about the prognostic impact of apoptosis in breast cancer have led to controversial results and in some studies apoptosis data did not provide independent prognostic value in any, node-positive or node-negative, breast cancer patients [29
]. The presence of programmed cell death has been advocated by some authors to explain the controlled tumor growth in MC; however, in our series a similar degree of apoptosis was detected among the three groups of tumors.
It is not possible to draw conclusions about the impact of these biomarkers on patient outcome owing to the short follow-up and the small number of patients investigated, and mainly to differences in the adjuvant treatment performed in the three groups; however, we described event rates that help to differentiate MC and AMC cancers from non-endocrine-responsive DC. The latter were described as a subpopulation with poor prognosis, if left untreated (Zambetti et al. [30] described a 42% 5-year disease-free survival). In our series we observed three distant recurrences in the DC subgroup and a second tumor (rectal carcinoma) in AMC. The relatively favorable outcome of non-medullary cancer in our series could be explained by the extended use of systemic adjuvant chemotherapy. A further observation is noteworthy: two patients with MC had local ductal infiltrating reappearance (possibly second tumors) and subsequent distant metastases. We do not know whether systemic therapy (performed in five out of 12 patients) might have a role in reducing the risk of relapse for any type of subsequent breast disease in patients with previous diagnosis of MC, because the two patients who relapsed received chemotherapy.
In conclusion, despite a similar uncontrolled proliferation potential in MC as well as in AMC and DC, it is reasonable to hypothesize a strong impact of microenvironment (in particular TN and T-cell immune response triggered by some unknown specific antigens) in limiting tumor expansion in MC. Uncertainty exists also about the role of some matrix molecules (TN, adhesion molecules) in enhancing or starting lymphocyte recruitment in the MC phenotype. Further efforts are also needed to completely characterize AMC, which is not universally considered a distinct entity.
In the future, to further define the particular behavior of MC, the complex balance between tumor proliferation potential and stromal influence on invasiveness, and probably also on cellular growth rate and death, might be stressed.
Received for publication October 22, 2004. Revision received February 9, 2005. Accepted for publication February 10, 2005.
![]() |
References |
---|
![]() ![]() ![]() ![]() ![]() ![]() ![]() |
---|
2. Richardson WW. Medullary carcinoma of the breast. A distinctive tumor type with a relatively good prognosis following radical mastectomy. Br J Cancer 1956; 10: 415423.[ISI][Medline]
3. Jensen ML, Kiaer H, Andersen J et al. Prognostic comparison of three classifications for medullary carcinoma of the breast. Histopathology 1997; 30: 523532.[CrossRef][ISI][Medline]
4. Ridolfi R, Rosen P, Port A et al. Medullary carcinoma of the breast. A clinicopathologic study with 10 year follow-up. Cancer 1977; 40: 13651385.[ISI][Medline]
5. Reiner A, Reiner G, Spona J et al. Histopathologic characterization of human breast cancer in correlation with estrogen receptor status. A comparison of immunocytochemical and biochemical analysis. Cancer 1988; 61: 11491154.[ISI][Medline]
6. de Cremoux P, Salomon AV, Liva S et al. p53 mutation as a genetic trait of typical medullary breast carcinoma. J Natl Cancer Inst 1999; 91: 641643.
7. Dendale R, Vincent-Salomon A, Mouret-Fourme E et al. Medullary breast carcinoma: prognostic implications of p53 expression. Int J Biol Markers 2003; 18: 99105.[ISI][Medline]
8. Piccart MJ, Di Leo A, Hamilton A. HER2. A predictive factor ready to use in the daily management of breast cancer patients? Eur J Cancer 2000; 36: 17551761.[CrossRef][ISI][Medline]
9. Hansen MH, Nielsen H, Ditzel HJ. The tumor-infiltrating B cell response in medullary breast cancer is oligoclonal and directed against the autoantigen actin exposed on the surface of apoptotic cancer cells. Proc Natl Acad Sci USA 2001; 98: 1265912664.
10. Coronella JA, Telleman P, Kingsbury GA et al. Evidence for an antigen-driven humoral immune response in medullary ductal breast cancer. Cancer Res 2001; 61: 78997899.
11. Yakirevic E, Maroun L, Cohen O et al. Apoptosis, proliferation, and Fas (APO-1, CD95)/Fas ligand expression in medullary carcinoma of the breast. J Pathol 2000; 192: 166173.[CrossRef][ISI][Medline]
12. Elledge RM, Allred DC. Prognostic and predictive value of p53 and p21 in breast cancer. Breast Cancer Res Treat 1998; 52: 7998.[CrossRef][ISI][Medline]
13. Gillett CE, Smith P, Peters G et al. Cyclin-dependent kinase inhibitor p27kip1 expression and interaction with other cell cycle-associated proteins in mammary carcinoma. J Pathol 1999; 187: 200206.[CrossRef][ISI][Medline]
14. Spataro VJ, Litman H, Viale G et al. Decreased immunoreactivity for p27 protein in patients with early-stage breast carcinoma is correlated with HER-2/neu overexpression and with benefit from one course of perioperative chemotherapy in patients with negative lymph node status: results from International Breast Cancer Study Group Trial V. Cancer 2003; 97: 15911600.[CrossRef][ISI][Medline]
15. Donnellan R, Kleinschmidt I, Chetty R. Cyclin E immunoexpression in breast ductal carcinoma: pathologic correlations and prognostic implications. Hum Pathol 2001; 32: 8994.[CrossRef][ISI][Medline]
16. Schraml P, Bucher C, Bissing H et al. Cyclin E overexpression and amplification in human tumors. J Pathol 2003; 200: 375382.[CrossRef][ISI][Medline]
17. Nielsen NH, Arnelov C, Emdin SO et al. Cyclin E overexpression, a negative prognostic factor in breast cancer with strong correlation to oestrogen receptor status. Br J Cancer 1996; 74: 874880.[ISI][Medline]
18. Keyomarsi K, Tucker SL, Buchholz TA et al. Cyclin E and survival in patients with breast cancer. N Engl J Med 2003; 347: 15661575.[ISI]
19. Ishihara A, Yoshida T, Tamaki et al. Tenascin expression in cancer cells and stroma in human breast cancer and its prognostic significance. Clin Cancer Res 1995; 1: 10351041.[Abstract]
20. Jahkola T, Toivonen T, Nordling S et al. Expression of tenascin-C in intraductal carcinoma of human breast: relationship to invasion. Eur J Cancer 1998; 34: 16871692.[CrossRef][ISI][Medline]
21. Adams M, Jones JL, Walker RA et al. Changes in Tenascin-C isoforms expression in invasive and preinvasive breast disease. Cancer Res 2002; 62: 32893297.
22. Kalembey I, Inada H, Nishiura R et al. Tenascin-C upregulates metalloproteinase-9 in breast cancer cells: direct and synergistic effects with transforming growth factor beta 1. Int J Cancer 2003; 105: 5360.[CrossRef][ISI][Medline]
23. Charpin C, Bonnier P, Garcia S et al. E-cadherin and beta-catenin expression in breast medullary carcinoma. Int J Oncol 1999; 15: 285292.[ISI][Medline]
24. Jacquemier J, Eisinger F, Nogues C et al. Histological type and syncythial growth pattern affect E-cadherin expression in a multifactorial analysis of a combined panel of sporadic and BRCA1 breast cancer. Int J Cancer 1999; 83: 4549.[CrossRef][ISI][Medline]
25. Bacus SS, Zelnick CR, Chin DM et al. Medullary carcinoma is associated with expression of intercellular adhesion molecule-1. Implication to its morphology and its clinical behavior. Am J Pathol 1994; 145: 13371348.[Abstract]
26. Gaffey MJ, Frierson HF Jr, Mills SE et al. Medullary carcinoma of the breast. Identification of lymphocytes subpopulations and their significance. Mod Pathol 1993; 6: 721728.[ISI][Medline]
27. Yakirevich E, Maroun L, Cohen O et al. Apoptosis, proliferation, and Fas (APO-I, CD95)/Fas ligand expression in medullary carcinoma of the breast. J Pathol 2000; 192: 166173.[CrossRef][ISI][Medline]
28. Tamiolakis D, Simopoulos C, Cheva A et al. Immunophenotypic profile of tumor infiltrating lymphocytes in medullary carcinoma of the breast. Eur J Gynecol Oncol 2002; 23: 433436.[ISI][Medline]
29. Liu S, Edgerton SM, Moore DH et al. Measures of cell turnover (proliferation and apoptosis) and their association with survival in breast cancer. Clin Cancer Res 2001; 7: 17161723.
30. Zambetti M, Bonadonna G, Valagussa P et al. Adjuvant CMF for node-negative and estrogen receptor-negative breast cancer patients. J Natl Cancer Inst Monogr 1992; 11: 7783.[Medline]