High expression levels of p27 correlate with lymph node status in a subset of advanced invasive breast carcinomas

Relation to E-Cadherin alterations, proliferative activity, and ploidy of the tumors




The cyclin-dependent kinase inhibitor p27 plays a central role in cell cycle progression and is deregulated in breast carcinomas. Although its levels are inversely associated with tumor proliferation, overexpression of p27 has been reported in a subset of rapidly proliferating breast carcinoma cell lines.


p27 levels were determined by immunohistochemistry in a series of 52 sporadic invasive breast carcinomas consisting of 47 ductal, 2 lobular, and 3 mixed; most tumors were Grade 2 or 3 (46 of 52) and Tumor Node Metastasis (TNM) Stage II–IV (46 of 52). E-cadherin expression and its gene alterations at 16q22.1 were also studied, because in vitro evidence suggests a biologic association between p27 and E-cadherin-mediated growth suppression.


The mean p27 labeling index (LI; percentage of p27 positive tumor cells) was 33.3% ± 25.3% (range, 0.1–85%). High p27 levels (p27 LI, > 50%) were observed in 14 (26.9%) of 52 carcinomas and were significantly associated with metastatic disease in axillary lymph nodes (14 of 33 vs. 0 of 19; P = 0.0007 by Fisher exact test). In addition, p27 LI was higher in the group of lymph node positive vs. lymph node negative tumors (mean p27 LI, 40.9% vs. 20.1%; P = 0.008 by Mann-Whitney test). Reduced or absent E-cadherin expression was found in 27 of 45 (60%) informative cases. Allelic imbalance of the 16q22.1 locus was found in 14 (27.5%) of 51 cases by using the microsatellite markers D16S503, D16S752, and D16S512. p27 LI and E-cadherin alterations were not statistically related.


In summary, high p27 levels detected in a subset of advanced breast carcinomas correlate with lymph node metastasis, suggesting that other mechanisms may bypass the cell cycle inhibitory role of p27 and provide growth advantage in these tumors. Cancer 2002;94:2454–65. © 2002 American Cancer Society.

DOI 10.1002/cncr.10505

Breast carcinoma is the most frequent malignancy among women in Western countries.1 Many gene alterations possibly involved in breast carcinogenesis have been extensively investigated including regulators of cell cycle.2–4

Cyclins and cyclin-dependent kinase (CDK) complexes have a central role in controlling cell cycle progression.5 The activity of CDKs is regulated by the CDK inhibitors that are classified in two families: INK4 proteins including p16INK4a, p15INK4b, p18INK4c, and p14ARF and Cip/Kip family including p21, p27Kip1 (p27), and p57Kip2. All members of Cip/Kip family are capable of binding both to cyclin and CDK subunits.6 p27 is a universal CDK inhibitor that directly inhibits the enzymatic activity of cyclin-CDK complexes and arrests cells in G1-phase of cell cycle.7 The levels of p27 protein are increased in quiescent cells and rapidly decrease after stimulation with mitogens.8 Thus, p27 could be considered as a potent tumor suppressor gene. However, no homozygous deletions and only rare mutations of the p27 gene have been reported to date.9–11 Although transcriptional changes may occur, the cellular abundance of p27 is primarily regulated at posttranscriptional level by the ubiquitin-proteasome pathway and its degradation is faster in proliferating than quiescent cells.12

Previous studies have reported an inverse correlation between p27 and tumor cell proliferation in various tumor types including breast carcinomas (reviewed in Slingerland and Pagano).8 However, in vitro studies have shown that high expression levels of p27 are detected in a subset of rapidly proliferating breast carcinoma cell lines.13, 14 This p27 accumulation may result from posttranscriptional up-regulation or decreased degradation, but the exact mechanisms are still unclear.13

Several clinical studies have correlated absent or low p27 expression with poor prognosis in various malignancies including breast carcinomas.8, 15–20 However, several published breast carcinoma series included only subgroups of cases such as younger patients or tumors of small size.17, 18 Other investigators failed to reproduce or only partially confirmed the previously reported results.9, 21, 22 Furthermore, the statistical correlation between p27 expression and lymph node status was controversial in the aforementioned studies. Of note, low levels of p27 were detected more frequently in lymph node negative breast carcinomas compared with the lymph node positive ones in a recent published series.21

Evidence from in vitro studies suggests a biologic association between p27 and E-cadherin proteins. More specifically, E-cadherin, in addition to its inhibitory role in tumor invasion, also may suppress tumor cell proliferation, possibly through cell cycle-regulating molecules including p27.23E-cadherin gene mapped on 16q22.1 locus encodes the E-cadherin protein, a calcium-dependent cellular adhesion molecule crucial for epithelial organization and adhesion.24–26 Thus, E-cadherin originally was thought to be an important suppressor of epithelial tumor invasiveness and metastasis.27, 28 In various tumors including breast carcinomas, E-cadherin expression is decreased or even absent because of truncating gene mutations, loss of the wild-type allele by loss of heterozygosity (LOH), and other mechanisms.24, 28–42 It has been shown that significantly elevated levels of p27 are associated with E-cadherin-dependent growth suppression, and, in addition, that p27 is up-regulated by E-cadherin in breast, colon, and lung carcinoma cell lines.23 Previous studies also have implicated p27 in regulation of contact inhibition.43–47 However, the possible association between p27 overexpression and E-cadherin status has not been yet investigated in vivo.

In the current study, we hypothesized that a subset of breast tumors may overexpress p27, because the effect of p27 on cell cycle arrest may be bypassed by other mechanisms involved in cell cycle machinery. We therefore investigated the expression levels of p27 in a series of invasive breast carcinomas and correlated our findings with clinicopathologic parameters including axillary lymph node status as well as tumor proliferative activity and ploidy. On the basis of previous in vitro studies, we also decided to explore possible associations between p27 levels and E-cadherin status in vivo.


Patients and Tissue Samples

Fifty-two cases of invasive breast carcinoma were included in this study. All patients were female with a mean age 63.6 years (range, 35–90 years) and were surgically treated in the Second Department of Surgery, University of Athens School of Medicine, Greece, from November 1995 to April 1997. None of the patients had a strong family history of breast carcinoma. Selection of specimens was based on the availability of fresh frozen tumor tissue necessary for molecular analysis and corresponding paraffin blocks for immunohistochemical studies. Most of the cases included in this study group represented advanced breast tumors: 46 (88.4%) had Grade 2 or 3, 46 (88.4%) had Stage II–IV, and all but 1 tumor were larger than 1 cm in maximum dimension (Table 1). The material comprised 47 ductal, 2 lobular, and 3 mixed invasive carcinomas. The clinicopathologic characteristics of the patients are shown in Table 1. The cutoff for positive estrogen and progesterone receptors was 10 fmol/mg.

Table 1. Summary of Clinicopathologic Features, LOH Analysis, E-Cadherin Expression, P27 Status, Proliferative Activity, and Ploidy Status
SampleAge (yrs)HistologyTumor size (cm)GradeLN metastasisStageER positivePR positiveImmunohistochemistryMolecular analysis
p27 (LI) (%)E-cadherinMIB-1 (P1) (%)(LOH of 16q22.1 [E-cadherin])
  1. LOH: loss of heterozygosity; LN: lymph node; ER: estrogen receptor; PR: progesterone receptor; LI: labeling index; IDC: invasive ductal carcinoma; N: no; ND: not done; ILC: invasive lobular carcinoma; Y: yes; H: heterozygous.

477IDC2.22NIIYY< 0.5+9LOH
948IDC3.81NIIYN< 0.5++NDLOH
1042IDC2.83NIINY< 0.511H
1435IDC23YIINN< 0.5+42LOH
3588IDC7.53YIVNDND< 0.5++NDH
4264IDC32YIINDND< 0.5++12H

Two samples of each tumor were taken; 1 was snap-frozen in liquid nitrogen and stored at −70 ° C; the other was fixed in buffered formalin and embedded in paraffin (FFPE). In addition adjacent normal tissue was included from each case examined. The patients had not undergone any chemo- or radiotherapy before surgical resection, thus avoiding up- or down-regulation of cell cycle proteins because of DNA damage.48 Consecutive 5-micrometer-thick paraffin embedded sections were cut from each tumor specimen and processed for immunohistochemical analysis as described below. Representative hematoxylin and eosin-stained sections from each lesion were histologically examined by two pathologists (V.G.G. and G.Z.R.) to confirm that all tumor samples contained 60% or more tumor cells. Microscopic grading and surgical staging of the carcinomas was based on the Nottingham modification of the Bloom-Richardson and TNM systems, respectively.



For immunohistochemical analysis, the following monoclonal antibodies were used: 1) p27: clone SX53K8, dilution 1:200 (Dako, Copenhagen, Denmark); 2) E-cadherin: clone HECD-1, dilution 1:400 (Takara Biomedicals, Shuzo, Japan), that recognizes an extracellular epitope on human E-cadherin protein;31 and 3) Ki-67: MIB-1, dilution 1:100 (Immunotech, Westbrook, ME)


In brief, the immunohistochemical technique was the following: 5-μm-thick sections of FFPE tissues were mounted on poly-L-lysine-coated slides, dewaxed, rehydrated, and incubated for 30 minutes with 0.3% hydrogen peroxide to quench the endogenous peroxidase activity. For the unmasking of the E-cadherin, p27, and MIB-1 antigens, the sections were incubated in 10 mM sodium citrate (pH 6.0) and heated in a microwave oven at 600-W power for 3 cycles of 5 minutes each. After endogenous protein blocking, the sections were incubated with the monoclonal antibodies at the aforementioned dilutions at 4 °C overnight. Biotin-conjugated secondary antibody was added at a 1:200 dilution for 30 minutes at room temperature. In the next 30-minute incubation step, the streptavidin-hyperoxidase complex (1:100 stock; Dako) was applied. 3,3′-Diaminobenzidine tetrahydro-chloride (DAB) was used as chromogen and hematoxylin as counterstain.


To distinguish false-positive (background staining) from positive cells, we performed an additional control assay in which each immunostaining-assay step was sequentially eliminated. Tissue sections stained with immunoglobulin G isotype (Dako) were used as negative controls in all immunostainings. Epithelial cells of the adjacent normal ducts of the breast were used as internal positive controls for the expression of p27 and E-cadherin proteins.



Evaluation of all immunostained slides was performed independently by two pathologists (V.G.G., G.Z.R.) by counting at least 1000 tumor cells in 10 representative high-power fields. Interobserver variability was minimal (P < 0.01).

Tumor cells were considered as p27 positive when nuclear staining was observed irrespective of the intensity. The percentage of p27 positive cells was designated as p27 labeling index (LI). Therefore, we considered p27 LI as a continuous variable. However, based on the distribution of data (histogram) and for the purpose of more detailed statistical analysis, several cutoffs were used for p27 expression including 1%, 10%, 20%, 50%, and 60%. Labeling index less than or equal to 10% was considered as “low p27 expression,” whereas p27 LI greater than 50% was considered as “high p27 expression” as has been suggested by others.16, 17, 21 The same areas of the breast were selected in serial tissue sections for examining p27 and Ki-67 immunoreactivity.


For the evaluation of E-cadherin expression, we used previously published criteria.31, 49 The intensity of immunohistochemical staining in tumor cells was estimated semiquantitatively by comparison with normal pre-existing epithelial cells in the same section as an internal control. If the staining intensity of tumor cells was the same as that of normal epithelial cells, E-cadherin was considered as “preserved expression” (++). In cases in which the intensity of the staining was weaker than in cells of the normal ducts but still present, E-cadherin expression was evaluated as “reduced” (+). Absence of E-cadherin immunoreactivity in tumor cells was considered as complete loss of expression or “absent” (−).

Ki-67 (MIB-1).

Nuclear staining of MIB-1 was considered positive irrespective of the intensity. Proliferation index (PI) was designated the percentage of MIB-1 positive tumor cells. Examination of the immunostained slides was performed as described above for p27.

Microdissection and DNA Extraction


To extract DNA, we processed contiguous 5-μm sections. The first section was stained with hematoxylin and eosin to visualize the extent of the tumor cells within each sample. The boundaries of the cancerous tissue were delineated microscopically, and excess normal tissues were removed with sterile surgical blades.

DNA extraction

A portion of the neoplastic material was digested in 500 μL lysis solution of: 50 mM Tris-HCl (pH 8.0), 150 mM NaCl, 5 mM ethylenediamine tetraacetic acid, containing 1% sodium dodecyl sulfate and proteinase K (Boehringer, Athens, Greece) at a final concentration of 100 μg/mL. Lysis was conducted at 55 °C for 72 hours. Additional proteinase K (50 μg/mL) was added on each day of lysis. DNA was extracted using the phenol/chloroform/isoamylalcohol method. The quality and quantity of DNA was assessed by agarose gel electrophoresis and spectrophotometry at 260 and 280 nm.

Allelic Imbalance Analysis of Chromosomal Region 16q22.1

The primers for highly polymorphic human microsatellite repeats D16S503, D16S752, D16S512 were purchased from Research Genetics (Huntsville, AL).


Reactions were cycled in a Perkin-Elmer Cetus thermal cycler. The reaction mixture consisted of 10× polymerase chain reaction (PCR) buffer, dNTPs (1.25 mM), 25 mM MgCl2, and 5 U/mL Taq DNA polymerase (Promega, Madison, WI). The PCR conditions for these markers was as follows: initial denaturation for 5 minutes at 95 °C, 30 cycles of 55 seconds at 94 °C, and annealing time ranged from 50 seconds to 1 min and 10 sec at 55 °C. Extension step was performed at 72 °C for 60 seconds, followed by a final extension step of 7 minutes at 72 °C. Polymerase chain reaction products were loaded on 8% acrylamide gel. Visualization was conducted by silver staining. Most markers chosen had a heterozygosity of greater than or equal to 70%.


Loss of heterozygosity was determined by a combination of naked-eye assessment and scanning densitometry. When the product ratio values were less than or equal to 0.65 or greater than or equal to 1.54, the samples were scored immediately as LOH, whereas when their values ranged from 0.77 to 1.23 they were scored as negative. Samples producing values between 1.23 and 1.54 or 0.65 and 0.77 were subjected to a second assay and scored as LOH only if another positive value was obtained.

Nuclear DNA Ploidy Analysis


The samples were stained using the thionin-Feulgen procedure.50 Briefly, 5-μm paraffin sections were dewaxed, rehydrated, and then subjected to acidic hydrolysis in 5 N HCl at room temperature for 1 hour. The samples were stained in thionin-Schiff reagent (0.5% thionin, 0.5% sodium bisulfide, 0.1 N HCl) for 1.5 hours at room temperature. The specimens then were washed 3 times (30 seconds, 5 minutes, and 10 minutes) in freshly prepared sulfide rinsing solution (0.5% sodium bisulfide, 0.05 N HCl) and rinsed under tap water for 10 minutes. An additional rinsing step in acidic alcohol (70% alcohol, 0.1 N HCl) was used to increase the staining contrast. Finally, the sections were dehydrated and mounted in a xylene-based material.


The measuring procedure was performed using the Optipath (Meyer Instruments, Houston, TX) TV-based image analysis system equipped with a microscope and a video charge-coupled device camera. As internal reference control, lymphocytes or granulocytes were used. In each analysis, approximately 100 control and 350 tumor cells were measured. To distinguish nondiploid cells from diploid ones, an upper limit of 2.5c was set for diploid values. Because the fraction above 2.5c might include proliferating diploid cells, we also calculated the fraction of tumor cells with DNA values above the 5c level, which exceeds those of proliferating diploid cells. Cases with greater than 5% of cells with DNA content above 5c were considered as aneuploid as previously described.50

Statistical Analysis

The statistical associations of p27 LI (continuous variable) with clinicopathologic parameters, E-cadherin expression, LOH of 16q22.1, and ploidy were assessed by nonparametric Mann-Whitney U and Kruskal-Wallis tests. Spearman rank correlation coefficient was used to evaluate the strength of correlation between p27 LI and PI. Chi-square and Fisher exact tests also were applied for statistical comparisons between various groups. All the analysis was implemented with the SPCC PC+, version 9.0, statistical package (Chicago, IL). The statistical difference was considered significant when the P value was less than 0.05.


p27 Expression

Any p27 immunoreactive tumor cells were detected in all 52 breast carcinomas examined (Fig. 1; Table 1). p27 protein also was expressed in the nucleus of coexisting normal epithelial, myoepithelial, and stromal cells, which served as internal positive controls. The percentage of p27 positive tumor cells (p27 LI) ranged extensively from 0.1% (very rare positive cells) to 85% with a mean (± standard deviation) of 33.3% ± 25.3% and a median 34.1%. p27 LI greater than 50% was observed in 14 (26.9%) of 52 carcinomas.

Figure 1.

p27 expression in invasive breast carcinomas. (a) Moderate levels (40%) of p27 labeling index (LI) in a Grade 3 invasive breast carcinoma (Case 18) with metastatic disease in the axillary lymph nodes. (b) High p27 LI (78%) in a Grade 3 invasive breast carcinoma (Case 40) with metastatic disease in the axillary lymph nodes (original magnification ×400, DAB, hematoxylin counterstain). (c) Histogram that shows the distribution of tumors according to p27 labeling index in the current breast carcinoma series.

Statistical analysis revealed a significant association between p27 LI and the presence of metastatic invasion of the axillary lymph nodes (mean p27 LI, 40.9% for lymph node positive vs. 20.1% for lymph node negative tumors; P = 0.008 by Mann-Whitney test). The distribution of p27 LI in lymph node positive and lymph node negative tumors is shown in Figure 2. Using various cutoffs, p27 expression was associated with lymph node status as shown in Table 2. None (0%) of 19 lymph node negative cases expressed high levels of p27 compared with 14 (42%) of 33 lymph node positive tumors (Table 2; P = 0.0007 by Fisher exact test). In more detail, all 14 carcinomas expressing high levels of p27 were associated with lymph node metastatic disease, and 11 of these 14 cases were Grade 2 or 3 tumors (Table 1). If the subgroup of high p27 expressing tumors was not included in the statistical analysis, no significant correlation between p27 LI and lymph node disease was found (P = 0.8, Mann-Whitney test). Table 3 summarizes the associations of p27 expression with clinicopathologic parameters of the patients by using a 10% and a 50% cutoff for low and high p27 levels, respectively. The p27 LI did not differ in cases with E-cadherin alterations compared with those with preserved E-cadherin expression (Table 4). Similarly, no significant association was found between p27 LI and other clinicopathologic parameters including estrogen and progesterone receptor status. p27 LI and PI were found to be inversely but not significantly correlated (Spearman rank correlation coefficient, −0.2; P = 0.2).

Figure 2.

Box plots of p27 labeling index in lymph node positive and lymph node negative cases of sporadic invasive breast carcinoma.

Table 2. Association of p27 Expression with Metastatic Disease to Axillary Lymph Nodes
Cutoff for p27 expression (%)No. (%) of tumors with p27 LI higher than cutoff (%)No. (%) of tumors with LN metastasis and p27 LI higher than cutoff (%)P valuea
  • LI: labeling index; LN: lymph node.

  • a

    Fisher exact test.

144/52 (84.6)29/44 (65.9)0.44
1039/52 (75.0)27/39 (69.2)0.18
2034/52 (65.4)26/34 (76.5)0.01
5014/52 (26.9)14/14 (100)0.0007
607/52 (13.5)7/7 (100)0.03
Table 3. Associations of Low and High p27 Expression with Clinicopathologic Features and Ploidy in Sporadic Invasive Breast Carcinomas
CharacteristicLow p27 expression (p27 LI < 10%)High p27 expression (p27 LI >50%)
n%P valuen%P value
  • LI: labeling index; IDC: invasive ductal carcinoma; ILC: invasive lobular carcinoma; ER: estrogen receptor; PR: progesterone receptor.

  • a

    Chi-square test.

  • b

    Fisher's exact test.

Histologic type0.38a0.35a
ER status0.70b0.46b
PR status0.70b0.73b
Table 4. p27 Labeling Index and Proliferative Activity in Relation to E-Cadherin Status in Sporadic Invasive Breast Carcinomas
Characteristicp27 labeling indexProliferation index
MeanSDP valueaMeanSDP valuea
  • SD: standard deviation; LOH: loss of heterozygosity.

  • a

    Mann-Whitney U test.

E-cadherin expression0.990.5
 Reduced or absent32.723.522.418.6
LOH of 16q22.10.920.9
 Not detected33.925.119.916.4

E-Cadherin Protein Expression

Of the 45 informative cases, 4 (8.9%) were negative, 23 (51.1%) were weakly positive (“reduced” expression, +), and 18 (40%) were strongly positive (“preserved” expression, ++) for E-cadherin protein (Table 1, Fig. 3). In both invasive lobular carcinoma of the current study group, E-cadherin was totally absent (Table 1). In three carcinomas of mixed type, invasive ductal carcinoma areas showed preserved or reduced expression of E-cadherin whereas in invasive lobular carcinoma areas the tumor cells were principally negative.

Figure 3.

E-cadherin expression in invasive breast carcinomas. (a) A case of invasive lobular carcinoma with complete loss of E-cadherin expression. Entrapped normal ducts express E-cadherin. (b) A case of invasive ductal carcinoma demonstrating preserved E-cadherin membranous immunoreactivity in an invasive ductal carcinoma case. (original magnification ×400, DAB chromogen, hematoxylin counterstain).

Reduced or absent E-cadherin expression was slightly more frequent in aneuploid compared with diploid tumors (P = 0.1, Fisher exact test). No significant correlation was observed between E-cadherin expression and p27 LI, PI (Table 4) or clinicopathologic parameters of tumors.

Allelic Imbalance (AI) Analysis

A total of 51 microdissected tissue specimens of invasive breast carcinoma were assayed for AI with the use of a panel of microsatellite markers for 16q22.1 locus (D16S503, D16S512, and D16S752). Fourteen (27.5%) of 51 cases had AI of the 16q22.1 as detected by any marker used (Table 1; Fig. 4). Our findings for each marker are illustrated in Figure 5. Absent or reduced E-cadherin expression was not statistically related to the LOH of the 16q22.1. (P > 0.1, Fisher exact test). AI of the D16S752 locus, which better characterizes the localization of E-cadherin gene, was more frequently detected in patients with advanced stage (P = 0.04, chi-square test). AI of 16q22.1 was not significantly associated with p27 LI and PI as presented in Table 4, as well as other clinicopathologic parameters and tumor ploidy.

Figure 4.

Allelic imbalance (AI) of 16q22.1 locus. (a) Cases 5, 13, 14, and 32 show AI of D16S752 (N: DNA from adjacent normal tissue; T: tumor DNA). (b) Cases 5 and 6 exhibit AI of D16S503 (N: DNA from adjacent normal tissue; T: tumor DNA).

Figure 5.

Schematic representation of16q22.1 allelotype in sporadic invasive breast carcinomas. (red boxes) Loss of heterozygosity; (gray boxes) heterozygous; (blue boxes) noninformative cases; (green boxes) not done.

Proliferative Activity of the Tumors

The PI as assessed by MIB-1 immunostaining ranged from 5.4% to 83.1% (median, 15.8) with a mean 19.8% ± 14.9% (Fig. 6). Proliferation index was higher in tumors with advanced grade and stage (P = 0.05 and P = 0.1, respectively, Kruskall-Wallis test). Progesterone receptor negative tumors demonstrated slightly higher PI (P = 0.08, Mann-Whitney test). Proliferation index was not significantly associated with E-cadherin alterations (Table 4).

Figure 6.

Immunohistochemical detection of the tumor cell proliferation marker Ki-67 using the monoclonal antibody MIB-1 in a case of ductal breast invasive carcinoma (Case 8; proliferation index, 26%)

Ploidy Status

Eleven (37.9%) of 29 tumors examined were scored as aneuploid whereas the remaining 18 cases showed diploid DNA content. No statistical associations were found between ploidy and p27 expression (Table 3), clinicopathologic features, E-cadherin, or PI of the tumors.


In the current study, we investigated the hypothesis that p27 may be overexpressed in a subset of high-grade breast carcinomas because previous in vitro studies have shown high p27 levels in rapidly proliferating breast carcinoma cell lines.13, 14 We therefore assessed the expression levels of p27 in a series of invasive breast carcinomas, mostly consisted of tumors with large size, high grade, and advanced stage. We report the distribution of tumors according to p27 LI (Fig. 1c) with the median p27 LI being 34.1%, ranging extensively from almost 0% to 85%. These results led us to choose several cutoffs to distinguish high versus low p27 expression for the purpose of additional statistical analysis.

Using a 50% cutoff for high versus low p27 expression,16, 17, 21 we found that 14 tumors (27%) expressed high levels of p27. Of note, a strong statistical correlation was found between high p27 levels and the presence of axillary lymph node metastasis, and this association remained statistically significant using several cutoffs for high p27 expression (Fig. 2; Table 2). More detailed analysis revealed that all 14 carcinomas with a p27 LI higher than 50% were associated with lymph node disease, and most of them were Grade 2 and 3 tumors (Tables 1 and 3). Notably, if the subset of tumors expressing high levels of p27 is not included in the statistical analysis, no significant correlation between p27 LI and lymph node status is observed.

As far as we are aware, this is the first study that correlates overexpression of p27 with lymph node status in a series of predominantly advanced breast carcinomas. These results seem paradoxical because previous studies have shown that reduced levels of p27 significantly correlate with lower patient survival,16, 17, 19, 20 and because lymph node status is the strongest prognosticator in breast carcinoma. However, in some of the aforementioned studies, the authors failed to show any significant association between p27 levels and lymph node status.16, 17 Furthermore, in the largest breast carcinoma series published recently, Barbareschi et al.21 reported that p27 expression predicted clinical outcome only in the group of lymph node negative patients but not in the group of lymph node positive patients. More importantly, in the same study, p27 levels were significantly higher in the lymph node positive compared with lymph node negative cases,21 a finding that is in agreement with our results.

Our findings are not unexpected because previous in vitro studies have detected high expression levels of p27 in rapidly proliferating breast carcinoma cells.13, 14 More specifically, Fredersdorf et al.13 demonstrated that several breast carcinoma cell lines with high growth rate tolerated p27 overexpression, which was more likely controlled at the posttranscriptional level.13 In the latter study, high expression levels of p27 were significantly associated with overexpression of cyclin D1 in a series of breast tumors.13 In fact, cyclin D1 is overexpressed in greater than 50% of mammary tumors51 and seems to have a causative role in breast carcinogenesis.52 Moreover, it has been shown recently that cyclin D1 may be the target of c-erb-B2- and Ras-mediated oncogenic pathways.53

Of note, recent molecular studies suggest that p27 and other CDK inhibitors of Cip/Kip family, in contrast with their inhibitory function for cyclin E- and cyclin A-dependent CDKs, also may act as positive regulators of cyclin D-dependent kinases (reviewed in Sherr and Roberts54). Indeed, it has been shown that cyclin D3-CDK6-p27 and cyclin D1/2-CDK4/6-p27 complexes are enzymatically active.55, 56 Furthermore, on the basis of experiments with transgenic animal models, Cheng et al. found that p27 may promote and stabilize cyclin D1/2-CDK4 assembly.57

It is also tempting to speculate that other mechanisms involving the balance between cyclins/CDKs and CDK inhibitors may result in cell cycle progression from G1- to S-phase. For instance, in G1-arrested rat fibroblasts, cyclin D3-CDK6 complex retains kinase activity mainly because of its ability to evade inhibition by p27.58 In a recent study, transgenic expression of CDK4 caused hyperproliferative phenotype in mice and was independent of D-type cyclin expression.59 In the latter study, p27 was overexpressed and sequestered by the increased levels of CDK4 possibly resulting in increased CDK2 activity and subsequent cell proliferation.59

Another possibility is that high expression levels of p27 might represent abnormal accumulation of p27 protein, which is probably inactive when joined to cyclin D-CDK4-6 complexes. Sanchez-Beato et al.60 demonstrated overexpression and colocalization of p27 and cyclin D3 proteins in a subgroup of aggressive non-Hodgkin lymphomas and assumed that high cyclin D3 levels could lead to stabilization of p27 protein that is probably inactive. Of note, high p27 expression was associated with worse prognosis in the latter study,60 and similar associations with clinical outcome have been reported in other low-grade lymphoid malignancies, as well.61

We further correlated the levels of p27 with E-cadherin expression to better clarify the biologic relation between p27 and E-cadherin in vivo. We report absent or reduced expression of E-cadherin in 60% of tumors, which is in agreement with most published data.29, 30, 42, 49, 62–65 E-cadherin was negative in both invasive lobular carcinomas, thus confirming previous observations.29–31 We also analyzed the allelic imbalance of 16q22.1, the locus of E-cadherin gene, and we report allelic losses in 28% of tumors (Table 1; Fig. 4). Previous studies reported comparable or higher percentages of LOH of 16q22.1 using various microsatellite markers.66–68 Reduced or absent E-cadherin expression was not significantly associated with LOH of 16q22.1 in the current study group, thus suggesting that various underlying mechanisms may affect E-cadherin expression in mammary tumors. These mechanisms may be complex and include E-cadherin gene mutations resulting in a truncated protein found in invasive and in situ lobular breast carcinomas,31, 32, 34, 49, 69 homozygous deletions, mutations or CpG methylation of the promoter region and transcriptional repression.70–73

Reduced or absent E-cadherin was not significantly related to lymph node status in our series. The clinical significance of E-cadherin expression remains controversial, although it is generally accepted that reduction of E-cadherin expression is associated with invasive growth or axillary lymph node metastasis,28, 30, 74, 75 However, such a correlation has been explicitly denied in other studies.27, 76 p27 LI and proliferation rate were not statistically related to E-cadherin expression or alterations of 16q22.1 in this study. Nevertheless, it remains unclear whether possible biologic association between p27 and E-cadherin-mediated growth suppression observed previously in cell lines23 plays any role in breast oncogenesis, in vivo.

In summary, we have showed that in a subset of high-grade and advanced invasive breast carcinomas, p27 may be expressed at high levels and significantly associated with axillary lymph node metastasis suggesting that other mechanisms override the effect of p27 on the cell cycle arrest. It seems unlikely that p27 expression in breast carcinoma is associated with E-cadherin alterations, in vivo.