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
The objective of this study was to clarify the prognostic and predictive value of immunoreactivity for the cyclin-dependent kinase inhibitor p27(Kip1) in patients with early-stage breast carcinoma and to investigate its relation with clinicopathologic features and other markers.
Immunoreactivity for p27 protein was analyzed on tumor slides from 461 patients who were enrolled in the International Breast Cancer Study Group (IBCSG) Trial V (median follow-up, 13 years), including 198 patients with lymph node negative disease and 263 patients with lymph node positive disease. Tumors with < 50% immunoreactive neoplastic cells were considered low expressors. Immunoreactivity for p27 was correlated with several clinicopathologic characteristics. Disease free survival (DFS) and overall survival were analyzed according to p27 immunoreactivity and treatment group.
In the lymph node negative population, decreased p27 immunoreactivity was associated with higher tumor grade (P = 0.001) and HER-2/neu overexpression (P = 0.04). In the lymph node positive population, low p27 expression was associated with higher tumor grade (P = 0.01), low expression of thymidylate synthase (P = 0.001), and higher Ki-67 expression (P = 0.007). DFS was not significantly different according to p27 status in either lymph node negative patients (10-year DFS: low p27 expression, 53% ± 5%; high p27 expression, 55% ± 5%) or in lymph node positive patients (10 year DFS: low p27 expression, 33% ± 4%; high p27 expression, 32% ± 4%). However, in the lymph node negative population, the benefit of one course of perioperative chemotherapy with cyclophosphamide, methotrexate, and 5-fluorouracil was confined exclusively to patients with tumors that showed reduced p27 immunoreactivity (P = 0.03; test for interaction).
For the vast majority of patients with early-stage breast carcinoma, it has been shown that several forms of adjuvant treatment provide a benefit in terms of disease free survival (DFS) and overall survival (OS). However, there remains considerable controversy over how to tailor the most beneficial treatment to each patient according to the individual characteristics of the patients and their tumors. Despite the variety of new tumor markers that have been investigated in the last 10–15 years, only a few provide clinically useful information in addition to lymph node status, hormone receptor status, and tumor grade, and there remains a substantial need for clinically useful tumor markers.1
Various kinases called cyclin dependent kinases (CDKs) are the key regulators of cell cycle progression and are under the positive control of cyclins and the negative control of CDK inhibitors. The p27 protein belongs to the family of the Kip inhibitors, which has a broad spectrum of inhibitory activity on different CDKs and acts as a tumor-suppressor protein.2 A large body of studies has shown that p27 activity is regulated almost exclusively posttranslationally at the level of protein degradation3 and that decreased p27 protein levels are common in several tumor types.4
Several retrospective studies have shown that reduced immunoreactivity of the p27 protein in patients with early-stage breast carcinoma was associated with unfavorable prognostic factors and a significant increased risk of disease recurrence or death.5–7 Down-regulated p27 was an independent, unfavorable prognostic factor in multivariate analysis, although it was correlated with high tumor grade in one study5 and with negative estrogen receptor (ER) status in two studies.5, 6 Tan et al. reported that low p27 expression was an independent prognostic marker in patients with small (T1a,b) lymph node negative tumors, suggesting the value of p27 in selecting patients with small, lymph node negative breast carcinoma who may benefit from adjuvant therapy. Compared with those early reports, other studies found that low p27 expression was correlated strongly with tumor grade and had no prognostic value in multivariate analyses.8–10 Unfortunately, those studies used different definitions for p27 status or different cut-off values when p27 scoring was based on the percentage of immunoreactive cells. Furthermore, the cohorts of patients investigated were not treated in randomized clinical trials. Consequently, the interaction between p27 status and the benefit from adjuvant treatment usually could not be addressed. We are aware of only a single report11 that addressed the question of the possible predictive value of p27 protein levels on the benefit of adjuvant chemotherapy. Those authors reported an improvement in DFS and OS for patients with lower p27 immunoreactivity but not for patients with high p27 immunoreactivity.
The current study was planned to assess the prognostic and predictive value of p27 immunoreactivity in a large population of patients with early-stage breast carcinoma who were enrolled in a randomized clinical trial and to investigate the correlation of p27 status with established markers.
MATERIALS AND METHODS
The material used in this study was from a cohort of patients who were enrolled in the International Breast Cancer Study Group (IBCSG) Ludwig Trial V.12–14 In 1981, the IBCSG (formerly the Ludwig Breast Cancer Study Group) began a randomized clinical study to assess the effect of early commencement of adjuvant cyclophosphamide, methotrexate, and 5-fluorouracil (CMF) chemotherapy in patients with lymph node negative and lymph node positive breast carcinoma. A total of 2504 eligible patients were entered on the trial between 1981 and 1985. All patients had their breast tumors removed by either total mastectomy with axillary clearance or modified radical mastectomy. To be eligible, patients had to have disease of unilateral clinical stage T1, T2, or T3a; N0 or N1; and M0 according to the tumor-lymph node-metastasis classification system. Patients with lymph node negative breast carcinoma (n = 1275 women) were randomized to receive either a single cycle of perioperative chemotherapy (PeCT) or no adjuvant chemotherapy. Lymph node positive patients (n = 1229 women) were assigned to one of three treatments: PeCT, conventionally timed chemotherapy (ConCT), or both. The PeCT regimen was a combination of intravenous cyclophosphamide, methotrexate, and fluorouracil given on Days 1 and 8 and commencing within 36 hours of mastectomy. The ConCT regimen was cyclophosphamide, methotrexate, and fluorouracil (cyclophosphamide was given orally) plus low-dose, continuous prednisone started 25–36 days after mastectomy and continuing for 6 cycles, every 28 days. Postmenopausal patients with positive lymph node status who were assigned to receive ConCT also received concurrent tamoxifen for 6 months. The trial and the clinical results have been described in detail elsewhere.12–15 The PeCT + ConCT group (seven courses) and the ConCT group (six courses) were combined and labeled the prolonged treatment group in this analysis, because no difference was found between the two groups in the published trial report.12
In 1993, the IBCSG established a tissue bank of tumor blocks from a subset of Trial V patients. All breast tumors were assessed by conventional histopathologic assessment and were classified histologically according to the World Health Organization criteria. ER and progesterone receptor (PgR) status were measured by standard biochemical methods according to guidelines of the central laboratory, and levels ≥ 10 fmol/mg cytosol protein were classified as positive. These pathologic parameters have been merged with the clinical data base that was begun in 1981 and is updated annually for survival and disease status. The histologic material from the trial was used previously to investigate the relevance of HER-2/neu overexpression16 and of thymidylate synthase (TS) expression17 in patients with early-stage breast carcinoma, and these features were available in the data base. HER-2/neu expression was positive in 14 of 198 patients who were tested, because the antibody against HER-2/neu was titrated to stain only tumors that had gene amplification.16 The expression of p27 protein was analyzed by immunostaining available material from 461 patients: 198 patients with negative lymph nodes and 263 patients with positive lymph nodes. Low p27 immunoreactivity was defined as staining in < 50% of tumor cells, whereas high p27 immunoreactivity was defined as staining in ≥ 50% of neoplastic cells. This cut-off value was determined prior to data analysis and was selected because it had been used by other investigators.5, 7 All analyses were done separately according to lymph node status. The median follow-up was 13.3 years for lymph node negative patients and 13.6 years for lymph node positive patients as of the data base update in November 1999.
All histopathologic tumor parameters were recorded in the data base, and the current investigators were unaware of any of the clinical or histopathologic parameters when p27 was assessed, because all specimens were defined by a randomization number. Formalin fixed, paraffin embedded tissue sections were immunostained for p27 using a 1:200 dilution of the specific monoclonal antibody (clone 57; Transduction Laboratories, Lexington, KY) and a standard peroxidase labeled streptavidin-biotin staining procedure.18 Peroxidase activity was developed with diaminobenzidine as a chromogenic substrate.
The results were evaluated independently by two pathologists, taking into account the percentage of neoplastic cells that showed nuclear staining of the same or greater intensity compared with the staining in normal cell counterparts or in stromal lymphocytes, which served as internal positive controls for immunostaining. At least 2000 neoplastic cells were evaluated in different randomly chosen high-power fields (magnification, × 400) from the tissue sections.
Fisher exact tests were performed to assess correlations between p27 expression (low or high) and patient characteristics.19 Patient characteristics of interest included pathologic tumor size, ER status, PgR status, tumor grade, histology, vessel invasion, HER-2/neu overexpression, TS level, treatment group, menopausal status, and Ki-67 expression. Wilcoxon rank-sum tests for comparing the medians of two independent samples were computed to detect a difference in age by p27 expression category.20 There was no attempt to correct for multiple comparisons.
DFS was defined as the time from randomization to recurrence, metastasis, appearance of a second primary tumor, or death from any cause, whichever occurred first. OS was defined as the time from randomization to death from any cause.
Hazards ratios, 95% confidence intervals (95%CIs) for the hazards ratios, and associated P values were calculated based on Cox proportional hazards regression models with p27 immunoreactivity as a single covariate as well as treatment as a single covariate.21 The models were created for each lymph node status group and used both DFS and OS as endpoints. Kaplan–Meier plots were created to graphically show differences in DFS and OS according to p27 immunoreactivity by lymph node status as well as by treatment group.22
In all samples, the vast majority of epithelial cells from normal or hyperplastic ducts and lobules entrapped within the tumor or surrounding it consistently displayed intense nuclear staining, as did some stromal cells and most infiltrating lymphocytes. Overall, 201 tumors (44%) exhibited p27 nuclear immunoreactivity in < 50% of neoplastic cells. The prevalence of p27 down-regulation was similar in both lymph node negative patients (43%) and lymph node positive patients (44%).
Clinicopathologic Correlates of p27 Immunoreactivity
In patients with lymph node negative tumors, no significant associations were detected between p27 immunoreactivity and patient age, patient menopausal status, pathologic tumor size or type, peritumoral vascular invasion, Ki-67 labeling index, or treatment; whereas there was a statistically significant association of p27 down-regulation with high tumor grade (P = 0.001) and HER-2/neu overexpression (P = 0.04) (Table 1). In addition, patients who had tumors with negative ER status or negative PgR status and patients with low TS expression more frequently had low p27 levels. The associations of p27 status with ER, PgR, and TS status did not reach the level of statistical significance. With regard to histologic features, invasive lobular histology was associated less commonly with low p27 levels. Within the group of patients with lymph node positive disease, low p27 immunoreactivity was associated significantly with high tumor grade (P = 0.01), low TS immunostaining (P = 0.001), and high Ki-67 labeling index (defined as ≥ 8% cells immunostained for Ki-67) (P = 0.007) (Table 2).
Table 1. Characteristics of Lymph Node Negative Patients with p27 Expression Data Available
Given the importance of the correlation between low p27 levels and HER-2/neu overexpression, we specifically analyzed the degree of decreased p27 immunoreactivity in relation to HER-2/neu overexpression. It was very interesting to note that, among the 21 patients who had lymph node negative tumors with positive HER-2/neu overexpression, 8 patients (38%) had very low levels of p27 immunoreactivity (< 20% positive cells); whereas only 18% of patients who had lymph node negative tumors that did not overexpress HER2/neu had such low levels of p27 immunoreactivity (P = 0.04), indicating that the correlation between HER-2/neu overexpression and down-regulation of p27 is particularly robust. In contrast, among the patients with lymph node negative, histologic Grade 3 tumors, which also are correlated significantly with low p27 immunoreactivity, only a small proportion (23% of all patients with lymph node negative, Grade 3 tumors) had very low p27 immunoreactivity (< 20% positive cells).
Analysis of Survival
No correlation was found between DFS or OS and p27 status in patients with lymph node negative breast carcinoma (Fig. 1; Tables 3, 4). The same was true for patients with lymph node positive breast carcinoma (Fig. 2; Tables 3, 4). Although, for patients with lymph node negative tumors, there was an apparent shift between the survival curves for high and low p27 expression in the first 2 years of follow-up, there was no statistically significant difference in DFS at any of the time points.
Table 3. Disease Free Survival According to p27 Status: Overall and by Treatment Group
The hazard ratio (p27 low:p27 high), 95% confidence intervals, and P values were calculated based on Cox proportional-hazards regression models with p27 expression used as a single covariate.
Lymph node negative patients
77 ± 5
64 ± 5
81 ± 4
73 ± 4
81 ± 5
69 ± 6
80 ± 5
70 ± 5
68 ± 9
51 ± 10
84 ± 6
79 ± 7
Lymph node positive patients
63 ± 4
49 ± 5
69 ± 4
43 ± 4
70 ± 6
59 ± 6
73 ± 4
43 ± 5
54 ± 7
35 ± 7
62 ± 7
42 ± 7
Association of p27 Immunoreactivity with Outcome of Adjuvant Chemotherapy
DFS and OS patterns of patients with lymph node negative and lymph node positive breast carcinoma were analyzed to determine whether p27 immunoreactivity predicted for outcome of CMF chemotherapy. Patients with lymph node positive breast carcinoma demonstrated a benefit from prolonged CMF chemotherapy compared with one cycle of perioperative chemotherapy, irrespective of their p27 status. There was no difference in the magnitude of the benefit in patients with tumors that showed high or low p27 immunoreactivity (Tables 3, 4; Fig. 3A,B). In contrast, patients with lymph node negative tumors and decreased p27 immunoreactivity had a greater benefit from one course of perioperative chemotherapy compared with patients who had higher p27 levels (10-year DFS for patients with low p27: 58% ± 7% for perioperative chemotherapy; 43% ± 9% without perioperative chemotherapy; hazard ratio, 0.58; 95%CI, 0.33–1.03; P = 0.06). Conversely, patients with lymph node negative tumors and high p27 immunoreactivity showed an increasing risk in the perioperative chemotherapy group, although this effect was not significant (Tables 3, 4; Fig. 4A,B). Based on a Cox proportional hazards model, it was found that the interaction between p27 status and treatment in patients with lymph node negative disease was significant (P = 0.03).
In this study, we examined the clinicopathologic correlates and the clinical implications of p27 immunoreactivity as a prognostic marker of disease recurrence and survival and as a predictor of outcome to chemotherapy in patients with early-stage breast carcinoma. We showed that slightly less than 50% of tumors had decreased immunoreactivity for p27, regardless of tumor size or lymph node status. The prevalence of reduced p27 immunoreactivity is consistent with that reported in previous studies, ranging from 31% to 69% of patients.5–10 Down-regulation of p27 likely is an early event in breast carcinoma progression, because we detected it with the same prevalence in small lymph node negative tumors with limited invasion and in larger, lymph node positive tumors. Overall, our findings suggest that decreased levels of p27 are more common in tumors with an aggressive phenotype, such as those with ER negative status and PgR negative status, high histologic grade, HER-2/neu overexpression, high Ki-67 labeling indexes, and low TS expression. These observations also are consistent with the recent report of a correlation between p27 down-regulation and BRCA1 germ-line mutations.23
To our knowledge, this is one of the few studies in which p27 immunoreactivity and HER-2/neu overexpression have been investigated in the same tumors from a large cohort of patients with early-stage breast carcinoma.6, 7 In prior published reports, data on the relation between HER-2/neu and p27 immunoreactivity were not shown in detail; HER-2/neu overexpression and low p27 immunoreactivity, however, were considered independent prognostic factors for reduced survival in a multivariate analysis reported by Porter et al.6 Low p27 immunoreactivity also was an independent prognostic factor for reduced survival in patients with small tumors (T1a,b) in a multivariate analysis by Tan et al.,7 who took into account several biologic parameters, including HER-2/neu overexpression. In the current study, we have documented that lymph node negative tumors with HER-2/neu overexpression are more likely to show reduced p27 immunoreactivity. Similar findings were reported recently in a cohort of 51 patients with breast carcinoma in which 41% of tumors overexpressed HER-2/neu and 92% of tumors had decreased p27 immunoreactivity; all HER-2/neu overexpressing tumors in that cohort exhibited down-regulation of p27 and increasing HER-2/neu activity correlated with decreasing p27 immunostaining.24 The data described above are supported by the current study, in which a very stringent definition of HER-2/neu overexpression was applied, in that the immunostaining experiments were titrated to detect only tumors with definite gene amplification (more than three copies of the HER-2/neu gene).16 The observation of a correlation between p27 and HER-2/neu is of particular relevance, because it confirms laboratory data showing that one of the downstream effects of HER-2/neu overexpression is p27 down-regulation.25 Furthermore, preclinical work also showed that inhibition of the epidermal growth factor (EGF)/ErbB1 receptor led to up-regulation of p27 in a model of breast carcinoma tumorigenesis,26 and analyses of tissue samples from patients who were treated with inhibitors of the EGF receptors has revealed up-regulation of p27. Taken together, these data suggest that p27 is a common downstream target of several signaling pathways, including not only HER-2/neu but also other receptors of the EGFR/Her family.
In the current study, p27 immunoreactivity did not show any prognostic value in terms of disease recurrence or survival, issues that have been debated in the recent literature.5–10 The current study is the first to evaluate patients who were enrolled in a randomized clinical trial, a population that has received long and careful follow-up. This has the important advantage of reducing the risk of a selection bias; furthermore, although adjuvant treatment, in theory, can obscure the power of a prognostic marker in an untreated population, half of the population under investigation (patients with lymph node negative tumors) either received no adjuvant systemic treatment or received a single course of perioperative chemotherapy, making it possible to evaluate a prognostic effect in the absence of adjuvant treatment. IBCSG Trial V enrolled a large population of premenopausal and postmenopausal patients with lymph node negative tumors and lymph node positive tumors. Because only 461 patients were available for this investigation, the number of patients with defined features, such as lymph node negative status, was insufficient to detect prognostic effects of small magnitude. However, the size of the population investigated was larger than in most previously reported studies on p27 and was adequate to detect large prognostic effects. The current cohort was a subset of the patient cohort reported previously by Gusterson et al.,16 in which it was found that HER-2/neu overexpression was significant prognostically for patients with lymph node positive tumors, but not for patients with lymph node negative tumors. Our finding of a correlation between low p27 expression and HER-2/neu overexpression in patients with lymph node negative tumors and the lack of prognostic value of decreased p27 immunoreactivity are consistent with previous findings.
The current study is the first to address the influence of p27 immunoreactivity on the outcome of adjuvant treatment. We found that low p27 immunoreactivity predicted a benefit from a single course of perioperative chemotherapy in patients with negative lymph node status, whereas it had no influence on the magnitude of the benefit of prolonged treatment with CMF-based chemotherapy in patients with positive lymph node status. It is possible that the effect observed in the lymph node negative patients may have been related to the timing, and not to the type, of chemotherapy (CMF). We previously showed that perioperative CMF was more effective for patients who had ER negative tumors compared with patients who had ER positive tumors, more effective for patients who had histologic Grade 2 and 3 tumors compared with patients who had histologic Grade 1 tumors, and more effective for patients in whom no axillary micrometastases were found after serial sectioning.27, 28 The current results, therefore, are consistent with these previous observations. There is no suggestion that low p27 levels may be predictive of a particular benefit from CMF-based chemotherapy, given the lack of an interaction between p27 status and the magnitude of effect of prolonged CMF treatment in patients with lymph node positive disease. However, given its inverse correlation with HER-2/neu overexpression, low p27 immunoreactivity may be predictive of a benefit from anthracycline-based adjuvant chemotherapy, which has been suggested for the treatment of patients with tumors that overexpress HER-2/neu. An analysis of patients who are treated on clinical trials that randomize patients to CMF versus anthracycline-containing chemotherapy may be of interest in this regard.
The results of the current study found that decreased p27 immunoreactivity was not an independent prognostic factor in patients with early-stage breast carcinoma and was correlated with known prognostic factors, such as high histologic grade. In addition, low p27 levels were associated with HER-2/neu overexpression in patients with negative lymph node status and with higher Ki-67 labeling indices in patients with positive lymph node status. We also found that p27 immunoreactivity was not predictive of a benefit from CMF-based chemotherapy per se but may be predictive of an important benefit from starting (perioperative) adjuvant chemotherapy very early. In addition, this is the first study in a large population of patients who were treated on a randomized clinical trial that has addressed the correlation between HER-2/neu overexpression and p27 immunoreactivity, and our findings substantiate in vivo a biologic model in which p27 down-regulation is an important event in cells that overexpress HER-2/neu. These findings suggest that p27 may become a useful adjunct to HER-2/neu status for predicting response to drugs that act on EGF/Her pathways, such as trastuzumab and receptor tyrosin-kinase inhibitors, which target other members of the EGFR/Her family. Further studies clearly are warranted to investigate the usefulness of p27 in the selection of adjuvant treatments for patients with early-stage breast carcinoma.
The authors gratefully acknowledge the initial support provided by the Ludwig Institute for Cancer Research, the Cancer League of Ticino, and the Swiss Cancer League. They further acknowledge the continuing support for central coordination, data management, and statistics provided by the Swedish Cancer League, the Cancer Council Australia, the Australian-New Zealand Breast Cancer Trials Group, the Frontier Science and Technology Research Foundation, the Swiss Group for Clinical Cancer Research, and the United States National Cancer Institute.
The authors thank the patients, physicians, nurses, and data managers who participate in the International Breast Cancer Study Group trials. The skillful technical assistance of Miss Alessandra Cavallon in the immunocytochemical experiments also is recognized.
The authors acknowledge the following participants in this study of International Breast Cancer Study Group (IBCSG) Trial V: IBCSG Central Pathology Review (B. Davis, W. Hartmann, R. Bettelheim, A. and M. Neville); IBCSG Data Management (M. Isley, R. Hinkle, and L. Blacher); The Institute of Oncology, Ljubljana, Slovenia (J. Lamovec, J. Jancar, J. Nowak, D. Erzen, M. Naglas, M. Sencar, J. Cervak, O. Cerar, B. Stabuc, S. Sebek, and T. Cufer); Madrid Breast Cancer Group, Madrid, Spain (C. Mendiola, F. Cruz-Caro, M. L. Larrodera, F. Calero, A. Suarez, F. Pastrana, S. Cruchaga, C. Guzman, and B. Rodriguez); Spedali Civili and Fondazione Beretta, Brescia, Italy (L. Morassi, G. Marini, P. Marcpicati, U. Sartori, A. Barni, D. DiLorenzo, A. Albertini, G. Marione, and M. Zorzi); and Sir Charles Gairdner Hospital, Nedlands, West Australia, Australia (P. M. Reynolds, H. J. Sheiner, S. Levitt, D. Kermode, R. Hahnel, and G. Van Hazel).