Predictive Value of p53, mdm-2, p21, and mib-1 for Chemotherapy Response in Advanced Breast Cancer¹ Free

Most breast cancer patients receive chemotherapy at some point of their disease. Only about one-half of patients with metastatic disease benefit from chemotherapy but suffer from severe adverse effects. Presently, no tumor biological factor is available for clinical use in prediction of chemotherapy response in breast cancer, contrary to estrogen receptor status which predicts response to hormonal treatment. It would be beneficial to find factors that could predict response to chemotherapy or more accurately to distinct chemotherapy regimens in advance to avoid useless treatment. In addition to metastatic breast cancer, such factors could also be used in the choice of neoadjuvant or adjuvant treatment.

During the last decade, it has become more obvious that regardless of distinct mechanisms of action, most anticancer agents, as well as radiation, ultimately kill cancer cells primarily by inducing apoptosis (1, 2, 3). Mutations in mediators of apoptosis can produce treatment-resistant tumors (4). p53 is a transcription factor that participates in cell cycle checkpoint processes and apoptosis (5). Under normal physiological conditions, p53 limits proliferation after damage to genomic DNA through two alternative mechanisms: by G₁ (or G₂) arrest; or by apoptosis. In response to DNA-damaging agents, the wild-type p53-activated fragment 1 (also known as p21) is an important downstream effector in the p53-specific growth arrest pathway (6, 7) but not in the p53-dependent apoptotic pathway (8). The mechanism of the arrest/death decision is unclear and may be quite complex. It has been suggested that apoptosis proceeds in the absence of G₁arrest or that repairable damage in wild-type p53 cells would result in a G₁ arrest allowing time for repair of the damaged DNA and irreparable DNA damage would result in apoptosis. Indeed, in response to anticancer therapeutics, cancer cells growing in vitro have been shown to either enter into a stable arrest or die, depending on the integrity of their cell cycle checkpoints (9). Moreover, in mice, tumors with intact checkpoint functions (p21^+/+) have been shown to regrow after treatment with γ-irradiation, whereas a significant fraction of checkpoint-deficient tumors(p21^−/−) were completely cured (10). Thus, expression of p21 in tumors might be of relevance concerning the possible therapeutic effects of agents causing DNA damage. By immunohistochemical detection, normal breast epithelium ducts are p21 negative or have weakly positive focal areas, whereas breast carcinomas are often p21 positive (11, 12, 13). Overexpression of p21 suppresses tumor growth in several experimental models but other alterations, such as point mutations, seem to be rare in human tumors (14).

The protein product of the mdm-2³plays a central role in the regulation of p53. mdm-2 binds to p53 concealing its transactivation domain (15), thereby inhibiting p53-dependent effects in cell cycle arrest and apoptosis (16). mdm-2 also promotes rapid proteosomal degradation of p53 (17).

The p53 gene is mutated in at least 50% of cancers (18). The mutations are generally single base point mutations in the DNA-binding central domain of p53 blocking its ability to bind to specific DNA sequences (19). The unfunctional protein is then observed at high concentrations in these cells. p53 protein accumulation usually indicates an abrogation of p53 function due to either abnormal protein structure (20) or sequestration by p53-binding proteins (21). In addition to p53 mutations, some tumors inactivate p53 by the amplification of the mdm-2gene. mdm-2 gene amplification occurs at a lower frequency(4–8%) in breast cancer than in nonepithelial tumors (22, 23). Alterations in mdm-2 and p53expressions may represent alternative pathways in tumorigenesis and combined mdm-2/p53 immunohistochemical phenotype may provide better prognostic and predictive information than evaluation of the expression of p53 alone.

In the treatment of breast cancer, it is unclear whether monitoring p53, mdm-2, or p21 expression provides additional information on how patients will respond to systemic therapy. The purpose of the present study was to evaluate the utility of these tumor biological factors as predictive indicators for chemotherapy response in advanced breast cancer.

Paraffin-embedded blocks of the primary tumor were available for 134 of 283 patients who took part in a randomized multicenter trial comparing docetaxel to sequential methotrexate and 5-fluorouracil in advanced breast cancer (24). To enter the trial, patients were required to have histologically proved primary breast cancer that had progressed during or after first line anthracycline treatment for advanced disease or relapsed within 12 months after discontinuation of adjuvant anthracycline therapy. The patients were required to be ≥18 and ≤70 years old with a performance score of ≤2 and have no more than one previous chemotherapy regimen for advanced disease (multiple endocrine treatments and radiotherapy were allowed). Only patients with measurable metastatic lesions or nonmeasurable but evaluable metastatic lesions were eligible. Response evaluation was done after every third course, at treatment discontinuation, and every 3 months during follow-up. Partial and complete responses were confirmed within 4–6 weeks after first identification of objective response. Response evaluation was performed according to WHO recommendations (25). At disease progression, the patients were recommended to cross over to the alternative treatment arm. The main patient and tumor characteristics of the 134 analyzed patients are seen in Table 1.

All tissues had been fixed in 4% buffered formalin, processed, and embedded in paraffin according to the normal schedule used in the laboratory. From each block, 5-μm-thick sections were cut on coated slides and dried overnight at 37°C. The sections were deparaffinized in xylene, rehydrated through graded concentrations of ethanol to distilled water, and then boiled in citrate buffer (pH 6.0) in a microwave oven for 20 min. Immunohistochemical stainings were performed by using a commercial Elite ABC Kit (Vectastain; Vector Laboratories,Burlingame, CA) directed against mouse IgG. Blocking serum was applied for 15 min followed by overnight incubation with the diluted monoclonal primary antibody p53 (1:300; clone DO7; Dako, Glostrup, Denmark), mdm2(1:100; clone IF2; Oncogene Research Products, Cambridge, MA), p21(1:200; WAF1 protein, clone 4D10; Novocastra Laboratories, Ltd.,Newcastle upon Tyne, United Kingdom) or Ki-67 (1:500; clone MIB-1;Immunotech, Marseille, France). The sections were then incubated with the biotinylated second antibody and the peroxidase-labeled ABC for 30 min each. All dilutions were made in PBS (pH 7.2), and all incubations were performed in humid chambers at room temperature. Between each step in the staining procedures (except before incubation with the primary antibody), the slides were rinsed three times in PBS. Sections stained with p21 and mdm-2 were further stained with biotinyl tyramide and streptavidin conjugated to horseradish peroxidase, using these two reagents from a commercial amplification kit (Catalyzed Signal Amplification; Dako). Both reagents were diluted 1:8 in 0.1% Triton X(Sigma, St. Louis, MO) in PBS, and the sections were incubated for 15 min at room temperature in humid chambers with each, rinsing in 0.1%Triton X in PBS between and after the amplification steps.

Bound peroxidase was visualized in all slides with a 3-amino-9-ethylcarbazole solution (Sigma; 0.2 mg/ml in 0.05 m acetate buffer containing 0.03% perhydrol, pH 5.0) at room temperature for 15 min. Finally, the sections were lightly counterstained in Mayer’s hematoxylin and mounted in Aquamount mountant (BDH Ltd., Poole, United Kingdom).

Known positive sections for p53, mdm-2, p21 and Ki-67 were included in every staining batch as positive controls. Breast cancer cases stained with antivimentin (clone V9; Dako) served as negative controls.

Cells were considered positive for p53, mdm-2, p53, and mib-1 only when distinct nuclear staining was identified. The percentage of immunoreactive nuclei was evaluated by scanning the whole section at medium magnification and by counting at least 500 cells in the most densely stained tumor areas at high magnification (×400). All of the stained sections were independently scored by at least two of three investigators (P. H, A. R-S., J. S.) who were blinded to the clinical data. For depiction of the material (Table 1) and phenotype analysis (Table 6), a 10% cutpoint for low and high expression for p53, mdm-2, and p21 and a 25% cutpoint for mib-1 were chosen.

Spearman correlation coefficients were calculated for the investigated tumor biological factors. For the statistical analysis of response prediction we divided clinical response into two categories: response(complete response + partial response); and nonresponse (stable disease and progression). In both the univariate and multivariate analyses,differences in treatment response according to p53, mdm-2, p21, and mib-1 were tested with a logistic regression model with the tested tumor biological factors regarded as continuous variables. The univariate and multivariate analysis of time to progression and overall survival were done by the Cox logistic regression model. All of the parameters tested in the univariate analysis were included also in the multivariate analysis. All of the analyses were performed in the whole immunohistochemical study population (n = 134) as well as in the docetaxel and MF arms separately. In the analyses of response and time to progression in the whole patient population, in addition to the investigated tumor biological factors, given treatment was also included because in the whole randomized trial population(n = 283) the treatment arm had significant impact on treatment outcome. Three patients in the docetaxel group had a nonevaluable response and were excluded from the analyses of response.

The patient materials used were derived from archival tumor specimens available from patients who had been enrolled previously in a randomized two-arm multicenter study where docetaxel was compared with MF in advanced breast cancer after anthracycline failure (24). The results of that study indicated that the response rate was higher (46% versus 22%; P < 0.001) and time to progression was longer (6.3 versus 3.0; P < 0.001) in patients who received docetaxel. From this original study with 283 patients, 134 specimens of primary tumors were available for immunostaining with p53,mdm-2, p21, and mib-1. Of these 134 cases, 72 were in the docetaxel arm and 62 were in the MF arm (Table 1). The response rate to docetaxel and MF in this subgroup of 134 patients was 50 and 24%, respectively.

Of the 134 investigated tumors, 48% showed high expression of p53,77% of mdm-2, 49% of p21, and 78% of mib-1 (Table 1). The correlation between the investigated tumor factors is shown in Table 2. Estrogen receptor positivity correlated with low MIB(P < 0.001) and negative p53 (P <0.001). Low MIB correlated with negative p53 (P =0.001). mdm-2 and p21 correlated positively to each other(P < 0.001). Estrogen receptor positivity(P = 0.009), low MIB (P = 0.02), and high mdm-2 correlated (P = 0.04) with a longer disease-free interval after primary operation of breast cancer.

p53, mdm-2, p21, and mib-1 were not significantly associated with response to chemotherapy in the whole patient population or in the docetaxel group (Tables 3 and 4). Time to progression (data not shown) and overall survival (data not shown) were also similar in the whole patient population and in the docetaxel group irrespective of p53, mdm-2, p21, or mib-1 expression of the primary tumor. However, in the MF group, a low mib-1 (<25%) and a high mdm-2 (≥10) predicted a better response to treatment(P = 0.0014 and P = 0.046,respectively; Table 5; Fig. 1, A and B). These findings remained significant also in the multivariate analysis (P < 0.01 for mib-1 and P = 0.01 for mdm-2; data not shown). Patients with primary tumors expressing either a low mib-1 or a high mdm-2 seemed to have a longer time to progression both in the univariate(P = 0.06 for both; data not shown) and in the multivariate analyses (P = 0.03 and P =0.04, respectively; data not shown). Overall survival was similar also in the MF group irrespective of mib-1 or mdm-2 status of the primary tumor.

Phenotype analysis of combined information of p53, mdm-2, and p21 expression also revealed some difference in response to treatment in the two groups with distinct chemotherapy regimens (Table 6). Tumors with no expression of both mdm-2 and p21 staining,irrespective of p53 status, had a high response rate to docetaxel but no response to MF (Fig. 1, C and D). In other phenotypes, the response rates were more equally distributed.

Interestingly, different tumor biological factors were associated with response to chemotherapy in the two groups of patients receiving distinct chemotherapy regimens with different modes of action. In the MF group, a low mib-1 and a high mdm-2 predicted a better treatment outcome as measured by a higher response rate and a longer time to progression. Methotrexate and 5-fluorouracil are both antimetabolites and are believed to inhibit cell proliferation mainly in the S phase of the cell cycle. Thus, contrary to our results, a higher response rate might have been expected in the highly proliferating subgroup of tumors. Previously, a high proliferation activity of the primary tumor has generally predicted a better response to neoadjuvant chemotherapy,whereas in metastatic disease the results are conflicting (26, 27, 28, 29). One explanation for this might be the inaccuracy in measuring the proliferation index in the primary tumor and correlating it to treatment outcome after various therapies for metastatic disease. Another explanation for our findings might be that,on the whole, our patient material represents aggressive tumors, as,indeed, the median mib-1 staining (38.5%) was very high compared with typical breast cancer specimens of all stages (10–15%) (11, 23). Hence, the tumors with a high index of mib-1 staining in our material probably represent more aggressive tumors associated with other unfavorable characteristics that may have an impact on treatment outcome.

To the best of our knowledge, the present study is the first one to assess the clinical utility of mdm-2 as a predictive indicator for chemotherapy response. The better response to MF in tumors with a high mdm-2 staining might mean that the expression of mdm-2 reflects the functional wild-type p53 status more accurately than p21 or p53 expression measured by immunohistochemistry although this remains to be confirmed in larger materials. A high mdm-2 expression may also be an indicator of mdm-2 gene amplification and overexpression,although this is more unlikely due to the reports on low frequency of mdm-2 amplification in breast cancer. As for p21, it has previously been shown that p21 can be induced also by p53-independent pathways (30, 31, 32, 33, 34). Consequently, it is probably not a good indicator of functional p53. As for measuring functional p53,immunohistochemistry can detect nearly all missense mutations of p53 in the core domain, but its sensitivity is lower for mutations that produce protein truncation or for mutations that lie outside the core domain as well as null mutations (35). Moreover, positive immunohistochemical staining for p53 may be present in the absence of mutations for several reasons. Namely, detectable levels may exist as a result of normal cell cycle fluctuation, response to DNA damage,stabilization caused by interaction with other cellular proteins such as mdm-2, or failure of feedback loops or degradation pathways.

In the docetaxel group, we found no association between any of the investigated tumor biological factors and the response to treatment. This is not surprising because docetaxel as a microtubulin stabilizer exerts its cytotoxic effect in the G₂M cell cycle phase and thus may not depend so much on factors regulating mainly the G₁S cell cycle checkpoint. Indeed, for some years clinicians have been aware that taxanes do not have cross-resistance with most other chemotherapy regimens, especially with anthracycline-based regimens (36). This phenomenon may be explained by taxanes mediating their functions at different point of the cell cycle. However, in the phenotype analysis of the present study, tumors with a p21⁻/mdm2⁻ phenotype were highly responsive to docetaxel but completely resistant to MF irrespective of their p53 status. To date, only one other study has assessed the clinical utility of p21 in predicting response to chemotherapy in advanced breast cancer (37). Neither p21 nor any p21/p53 phenotype as assessed by immunohistochemistry was associated with clinical response to neoadjuvant treatment with weekly Adriamycin in that study. One explanation to our findings could be that the combined information of negative p21 and mdm-2 expression would indicate the loss of p53 function better than immunohistochemical p53 expression itself due to the shortcomings of this assay as discussed above. If this is the case, then functional p53 might after all be a significant factor in determining the response to taxanes. Indeed,recently some investigators have suggested that either p53 or p21 status might interfere with sensitivity to taxanes in vitroand that cases with nonfunctional p53 and/or negative p21 might be more susceptible to treatment with drugs that interfere with the G₂ M checkpoint (38, 39, 40).

Many groups have tried to find predictive factors for chemotherapy in breast cancer patients, but the results are contradictory. Moreover,the results are somewhat different in the neoadjuvant and metastatic settings, perhaps indicating that metastatic tumors are different from primary tumors (41). The two largest recent preoperative trials showed that neither p53 nor c-erbB-2 detected by immunohistochemistry had an impact on treatment outcome, whereas high proliferation activity was associated with chemosensitivity to 5-fluorouracil-Adriamycin-cyclophosphamide (42) or to epiadriamycin-vincristine-methotrexate and mitomycin C-thiotepa-vindesine (29) regimens. However, in some other studies, no association between response to preoperative anthracycline (43) or mitoxantrone-methotrexate-mitomycin C (44) regimen and proliferation was found. In addition, a worse response rate to preoperative 5-fluorouracil-Adriamycin-cyclophosphamide or 5-fluorouracil-thiotepa-cyclophosphamide treatment (45) or 5-fluorouracilfolinic acid-vinorelbine treatment (46)was found in patients with tumors expressing wild-type p53.Currently, in metastatic breast cancer, there are few data on p53 and response to chemotherapy. p53 expression was not associated with chemotherapy response in previous studies or in the present study (47, 48, 49). In one study, p53mutations as assessed by the constant denaturant gel electrophoresis or by direct sequencing of cDNA could predict resistance but not sensitivity to weekly doxorubicin therapy (50). On the contrary, p53 status as assessed by immunohistochemistry was not predictive.

The conflict in the results of different studies may be due to differences in the type of chemotherapy, which would be in line with our results. If this is true, predictive factors could be used to assist in tailoring individual chemotherapy regimens for tumors with distinct biological phenotypes. Of the investigated tumor biological factors in the present study, we could not identify any single factor that would predict response to docetaxel. However, an immunohistochemical staining of mib-1 or mdm-2 may assist in selecting patients for MF chemotherapy.

In conclusion, our results suggest that different tumor biological factors may predict response to different chemotherapy regimens with distinct mechanisms of action. Due to the multiplicity of possible defects in tumor cells resulting in various phenotypes and heterogeneity, it is unlikely that any single factor will be found that would predict response to any chemotherapy. Instead, we believe that a panel of factors measuring the function of different cell cycle checkpoints or apoptotic pathways should be investigated in detail.

We thank the departments of pathology of the participating institutions for providing us with tumor specimens and Helena Huotarinen for technical assistance.