Material and methods

PATIENTS

Our study comprised 307 women with primary infiltrating breast carcinomas (T1–4, N0–2). The patients were treated between 1983 and 1989 at our department. All women had no evidence of metastases (M0) at the time of diagnosis. The patients underwent mastectomy (225 patients) or tumorectomy with postoperative irradiation (linear accelerator: 50 Gy and 10 Gy boost to tumour bed) of the breast (82 patients). Two hundred and ninety five patients underwent axillary lymphonodectomy with removal of more than nine (≥ 10) nodes to rule out nodal metastatic disease. No adjuvant systemic treatment was administered to node negative patients (n = 134). In the node positive patient subgroup, premenopausal or postmenopausal women without steroid hormone receptors received six cycles of a chemotherapy regimen consisting of a combination of either cyclophosphamide, methotrexate, and 5-fluorouracil (600/40/600 mg/m²) or epirubicin and cyclophosphamide (60/600 mg/m²). Postmenopausal women with positive steroid hormone receptors were treated with 20–30 mg tamoxifen for up to five years.

Physical, laboratory, and apparative checks (mammography, thorax x ray, and abdominal ultrasonography) on patients were carried out regularly as a part of an organised follow up programme. Follow up ranged from 24 to 114 months (median, 82). Histological classification and grading were based on the World Health Organisation criteria (1981) and the suggestions of Bloom and Richardson, respectively. Tumour stage followed the TNM system of UICC. Oestrogen receptor (ER) and progesterone receptor (PR) status were determined by immunohistochemistry (ER: ERICA-System, Abbott, Wiesbaden, Germany; PR: monoclonal antibody mPR1, Dianova, Hamburg, Germany). Steroid hormone receptor status was considered to be positive when ER and/or PR were positive, and negative when both ER and PR were negative. The proliferation marker proliferating cell nuclear antigen (PCNA; NA03; Dianova), the tumour suppressor p53 (MAb 1801; Dianova), and the Her2/neu protein (OPA 01/1; Medac, Hamburg, Germany) were determined immunohistochemically in adjacent sections from the same tumour tissue block. Technical procedures and the immunoreactive scoring system have been described previously.^18–20

IMMUNOHISTOCHEMISTRY

Immunohistochemical analyses of p21^waf were performed on routinely processed blocks of formalin fixed, paraffin wax embedded surgical specimens of the primary carcinomas. The 3–4 μm sections of carcinoma tissues were mounted on 3-aminopropyltriethoxysilane (APES) covered glass slides. After drying, paraffin wax was removed with xylene (30 minutes), the sections were rehydrated, and the tissues digested with 0.1% trypsin (15 minutes). A modified three step avidin–biotin complex method to detect the p21^waf protein was used. All incubations with antibodies were performed in a moist chamber. The primary monoclonal antibody (clone 4D10, mouse IgG1 subtype; Novocastra, Newcastle upon Tyne, UK) was incubated at 4°C for 24 hours. Second and third antibodies were incubated at room temperature for 30 minutes. At intervals of different incubations, slides were washed with phosphate buffered saline. Antigen–antibody complexes were visualised using 3,3′-diaminobenzidine tetrahydrochloride (10 minutes). Cell nuclei were counterstained with haematoxylin (two minutes). A highly positive breast cancer and a negative control (non-specific immunoglobulins) were used in each run.

Specific staining was evaluated independently by two investigators semiquantitatively, yielding an immunoreactive score (IRS) ranging from 0 to 9. The IRS was calculated by multiplying the number of positive nuclear staining tumour cells (0, none; 1, < 10%; 2, 10–49%; 3, ≥ 50% positive tumour cells) by the staining intensity (1, weak; 2, moderate; 3, strong). Tumours were considered p21^waf positive with IRS ≥ 2. When the scores of the two investigators differed, consensus was reached after examination using a teaching microscope. To demonstrate the reproducibility of p21^waf detection, consecutive sections from 40 carcinomas were stained at one week intervals.

STATISTICS

We used SPSS 9.0 for Windows (Statistical Package for the Social Sciences; Munich, Germany) for statistical analysis. The χ² test was used for univariate comparison of data, whereas follow up data were analysed using the log rank test. Multivariate analyses were based on the Cox proportional hazards model and calculated relative risks.

Results

The p21^waf protein was detected in the nuclei of tumour cells. Intracytoplasmic reactions were very rare and much weaker than nuclear staining. The staining pattern in tumours was heterogeneous, revealing a mixture of positive and negative cells. Among the specimens with specific p21^waf staining, most of the tumours expressed p21^waf only in up to 20% of the cells. Non-diseased lobular or ductal epithelia found in the tumour periphery did not express detectable amounts of p21^waf (fig 1). The staining of different consecutive series showed similar results. Thus, the immunohistochemical detection of p21^waf expression was considered reproducible and reliable.

Ninety nine of 307 tumours (32.2%) were considered p21^waf positive. This observation was independent of the nodal status (N0, 32.3%; N+, 32.8%).

Table 1 summarises the clinical, morphological, and biological data of the study group. The study group showed the expected distribution of established parameters. There was no significant correlation between p21^waf and the parameters studied (p > 0.05), except for PCNA expression (p = 0.029). One of the major subdivisions of the patients concerned their nodal status. Therefore, the group was further divided into patients with node negative and node positive tumour tissues (tables 2 and 3). There was no significant correlation between p21^waf and the other parameters in the node positive subgroup (p > 0.05), although p21^waf expression correlated with PCNA expression (p = 0.017) and larger tumour size (p = 0.001) in the node negative subgroup.

Using the log rank test, tumour size, tumour grade, and the expression of steroid hormone receptors, PCNA, Her2/neu, and p53 correlated significantly with clinical outcome (table 4). In node negative patients, all markers failed to indicate recurrence. Disease free survival was calculated according to Kaplan and Meier (fig 2). There was a tendency to a favourable outcome for node positive patients with detectable p21^waf expression, but this was not significant. Similarly, node negative patients seemed to benefit moderately from p21^waf expression, although again the results were not significant (p > 0.05). Data for overall survival revealed similar results (not shown).

Multivariate analysis for disease free survival, as indicated by Cox regression analysis, found nodal status to be the strongest parameter (relative risk (RR), 1.74; p = 0.00001), followed by receptor status (RR, 0.59; p = 0.0085), tumour size (RR, 1.42; p = 0.02), and Her2/neu expression (RR, 1.56; p = 0.033) in the entire study group. In the subgroup of node positive patients, tumour size (RR, 1.83; p = 0.0004), steroid hormone receptor status (RR, 0.43; p = 0.0002), p53 expression (RR, 1.8; p = 0.009), and Her2/neu expression (RR, 1.74; p = 0.019) were independent parameters. Node negative patients showed no independent marker (p > 0.05).

Discussion

In selected breast cancer tissues, p21^waf expression rates range from 32% to 57%.^13,21,22 In our study group, 32.2% of tumours had p21^waf expression. With respect to the scoring system used we excluded spotted, weakly stained tumour cells from the p21^waf positive group. Thus, p21^waf detection in our study confirmed earlier results.

The role of p21^waf expression in breast cancer is controversial: whereas some studies reveal an inverse correlation between p21^waf expression and the apoptotic marker bcl-2,^23,24 the tumour suppressor p53,^22,25 and to histological grading,^22,26 other reports demonstrate p53 independent p21^waf expression^13,27 and association with high histological grading,^13,28 positive nodal status,²⁸ and large tumour size.²⁸ It is postulated that p21^waf expression is a prognostic marker for relapse free survival and improved overall survival.^22,25,28 In combination with nodal status, p21^waf expression is thought to have predictive value. On the contrary, in more recent studies p21^waf expression was not a prognostic factor²⁹ and did not correlate with clinical outcome.^13,30 Significant correlations were restricted to a lobular subtype.¹³ Elledge and Allred summarised numerous clinical studies with respect to p21^waf and concluded that p21^waf expression plays a subordinate role in breast cancer prognosis.³¹ Our study confirms these latter results. We show here that p21^waf expression is an independent factor in breast carcinoma progression, but lacks a clear predictive and prognostic relevance. Only trends could be seen with regard to clinical outcome.

Interestingly, in node negative women, p21^waf expression correlated significantly with tumour size and PCNA expression (table 2). The subgroup of tumours > 5 cm showed an opposite distribution of p21^waf positive and p21^waf negative cells. Although more patients need to be investigated, this suggests that p21^waf might have a protective effect on nodal involvement during tumour development. This is confirmed by earlier results that show a correlation between p21^waf expression and negative nodal status,²⁶ and vice versa.²² Further studies should be designed to prove the hypothesis of p21^waf mediated node protection.

One of the main functions of p21^waf uncoupled from cell cycle regulation is its inhibitory effect on the proliferation of tumour cells.¹⁸ p21^waf expression strongly correlates with Ki67 expression.³⁰ Furthermore, p21^waf can affect DNA replication via physically binding to PCNA.³² p21^waf disrupts the PCNA–Fen1 complex, thereby prohibiting DNA replication.³³ This inhibitory role is concentrated on the operation of PCNA in DNA replication but not in DNA repair.³⁴ In our study, p21^waf expression correlated with increased PCNA expression in node negative patients. To our knowledge, this is the first clinical study to deal with both parameters. We assume that increased PCNA expression induces p21^waf expression in primary node negative breast cancer. Consistent with the reports discussed above, this might be a protective event that results in decreased nodal involvement. However, these effects are abolished if tumour cells are capable of inducing nodal spread.

In conclusion, the immunohistochemical detection of p21^waf expression is of no use when making decisions about the treatment of breast cancer, although it is useful for understanding the biology of breast carcinogenesis.