Fax (713) 745–5768
Prognostic significance of phosphorylated P38 mitogen-activated protein kinase and HER-2 expression in lymph node-positive breast carcinoma
Version of Record online: 18 DEC 2003
Copyright © 2003 American Cancer Society
Volume 100, Issue 3, pages 499–506, 1 February 2004
How to Cite
Esteva, F. J., Sahin, A. A., Smith, T. L., Yang, Y., Pusztai, L., Nahta, R., Buchholz, T. A., Buzdar, A. U., Hortobagyi, G. N. and Bacus, S. S. (2004), Prognostic significance of phosphorylated P38 mitogen-activated protein kinase and HER-2 expression in lymph node-positive breast carcinoma. Cancer, 100: 499–506. doi: 10.1002/cncr.11940
- Issue online: 20 JAN 2004
- Version of Record online: 18 DEC 2003
- Manuscript Accepted: 4 NOV 2003
- Manuscript Revised: 30 OCT 2003
- Manuscript Received: 28 AUG 2003
- Nellie B. Connally Breast Cancer Research Fund
- tumor markers;
- p38 mitogen-activated protein kinases;
Chemotherapy-induced p38 mitogen-activated protein kinase (MAPK) phosphorylation reportedly leads to increased apoptosis in breast carcinoma cells. The goals of the current study were to assess the incidence of activated phosphorylated p38 MAPK (P-p38) expression in invasive breast carcinoma, correlate expression of P-p38 MAPK with HER-2, and estimate the prognostic value of this marker in patients with lymph node-positive breast carcinoma treated with adjuvant chemotherapy.
P-p38, HER-2, and Ki-67 were measured using immunohistochemistry (peroxidase method) in 96 patients with lymph node-positive breast carcinoma treated with adjuvant fluorouracil, doxorubicin, and cyclophosphamide chemotherapy. All markers were measured in the primary tumors, before the initiation of adjuvant chemotherapy. The median follow-up period was 11 years after initial cancer surgery. P-p38 MAPK expression was scored visually and quantified using an image analyzer.
The rate of P-p38 MAPK expression ranged from 19–24%, depending on the scoring system used. There was a trend toward shorter progression-free survival (PFS) for patients whose tumors expressed high levels of P-p38 MAPK, although the difference was not statistically significant (P = 0.39). PFS was shorter in patients whose tumors overexpressed P-p38 MAPK and had a high level of Ki-67 (P = 0.04). In HER-2–negative patients, P-p38 MAPK overexpression was associated with a shorter PFS (P = 0.05).
P38 MAPK phosphorylation occurred in 20% of primary breast carcinomas and may be associated with poor outcome in patients with lymph node-positive breast carcinoma. Further studies are needed to define the interaction between P-p38 MAPK and HER-2 expression in breast carcinoma. Cancer 2004. © 2003 American Cancer Society.
Overexpression of the erbB/HER family of growth factor receptors has been associated with poor prognosis in patients with breast carcinoma.1 Overexpression of the epidermal growth factor receptor 2 (HER-2) has been associated with an improved survival in association with doxorubicin-based adjuvant chemotherapy.2, 3 Novel therapies directed against the epidermal growth factor receptor and HER-2 either have been approved or are in clinical trials. Trastuzumab monoclonal antibody (MoAb) therapy is effective in patients with breast carcinoma with HER-2 overexpression.4, 5 However, only 20–40% of patients respond to this highly targeted therapy. Understanding the biologic relevance of signaling networks activated downstream of these receptors may provide novel prognostic markers and therapeutic targets.
A well characterized intracellular signal transduction pathway in breast carcinoma progression involves activation of mitogen-activated protein kinases (MAPK).6–9 Growth factor receptor phosphorylation and dimerization activate Ras, which, in turn, activates MEK kinase (MEKK) and Raf. MEKK and Raf independently phosphorylate and activate MEK, which, in turn, activates MAPK.10, 11 The MAPK family includes the extracellular signal-regulated protein kinases (ERK1, ERK2), Jun NH2-terminal protein kinase (JNK), and p38 MAPK.12–16 Activation of these cascades results in induction of gene expression, leading to increased proliferation, invasion, and metastasis.8, 9 The various MAPK family members play a complex role in the determination of cell growth and cell death. The cellular decision is believed to involve a balance among competing MAPK pathways.17, 18
Phosphorylation of p38 MAPK has been reported in response to ultraviolet irradiation, biologic inducers (e.g., growth factors and cytokines), and chemicals.19 Bacus et al.20 reported that paclitaxel chemotherapy activates p38 MAPK in MCF7 cells before apoptosis. SB203580, a specific inhibitor of p38 MAPK activity, significantly decreased apoptosis, leaving the surviving cells arrested in the G2/M-phase of the cell cycle. Tamoxifen therapy resulted in p38 MAPK activation in BT-474 cells.21, 22 In a previous study, Esteva et al.23 showed that phosphorylated p38 MAPK (P-p38) levels were detected in 17% of primary breast carcinomas. The goals of the current study were to further define the incidence of activated p38 MAPK expression in breast tumor specimens and to estimate the prognostic value of this marker in patients with lymph node-positive breast carcinoma treated with adjuvant chemotherapy.
MATERIALS AND METHODS
Patients and Histology
Specimens were obtained from 96 lymph node-positive patients with breast carcinoma who participated in a prospective clinical trial of adjuvant systemic therapy at The University of Texas M. D. Anderson Cancer Center. In this clinical trial, 789 patients were registered and followed for a median of 10 years. They were treated in two groups. In Group 1, patients < 50 years or > 50 years but with either negative or unknown estrogen receptor (ER) status were randomized to receive six cycles of 5-fluorouracil, doxorubicin, and cyclophosphamide (FAC) alone or followed by four cycles of methotrexate and vinblastine (MV). In Group 2, patients ≥ 50 years with ER-positive disease were randomized to receive either tamoxifen or combination chemotherapy (FAC and MV) for 10 cycles.24 The 96 patients selected for immunohistochemistry study were the only patients treated with FAC from whom tissue specimens were available. All patients underwent mastectomy or breast-conserving surgery between December 1986 and January 1994. None of the patients had evidence of distant metastases at the time of surgery. The median age of the patients was 46 years (range, 28–63 years). Steroid hormone receptor status was determined by immunohistochemistry. After surgery, all patients received FAC chemotherapy as part of a randomized clinical trial (Protocol ID86-012). In this clinical trial, premenopausal patients were randomized to receive FAC alone or FAC and MV in combination. Postmenopausal, ER-positive patients were randomized to receive FAC versus tamoxifen. We selected patients treated with FAC. None of the 96 patients selected for this biomarker study received tamoxifen, regardless of ER/progesterone receptor (PR) status.
The primary tumor specimens were fixed in 10% neutrally buffered formalin and embedded in paraffin. Pathologic confirmation of breast carcinoma was obtained using hematoxylin-eosin stain on slides prepared from each block. Five-micrometer sections were prepared from each block. Immunohistochemical staining was performed using the peroxidase method. Briefly, sections were deparaffinized and hydrated in the usual manner. Sections were antigen retrieved using citrate buffer (pH 6) in a microwave. The sections were quenched by 3% hydrogen peroxide/methanol and then sections were blocked by 10% goat serum/0.1% Triton-X (Sigma Aldrich, St. Louis, MO). After cooling for 60 minutes, the sections were incubated with primary antibodies overnight at 4 °C in a moist chamber. The primary antibody used was a MoAb that detects human p38 MAPK only when activated by dual phosphorylation at Thr180 and Tyr182 (clone 28B10; New England Biolabs, Beverly, MA), diluted 1:100. The sections were incubated with biotinylated goat anti-mouse immunoglobulin (1:200 dilution) for 15 minutes, followed by horseradish peroxidase-conjugated streptavidin complex for 10 minutes (1:100 dilution). Immunoreactivity was visualized using diaminobenzidine tetrachloride (Dako Corporation, Carpinteria, CA). The sections were washed well with Tris-buffered saline between each step. After exposure to antibodies, sections were counterstained with ethyl green (Sigma Chemical Company, St. Louis, MO). For the negative controls, the primary antibody was omitted but all incubation steps were identical. No staining was observed in control slides. Previously identified strongly staining tumor tissue sections were used as positive controls. Immunohistochemistry for HER-2 and Ki-67 was performed using the peroxidase method, as previously described.25–27 The primary antibody directed against the HER-2 protein was the MoAb Ab8 (Neomarkers, Lab Vision Corporation, Fremont, CA). Ki-67 was determined by staining for KiS5 (Dako Corporation) binding to Ki-67. KiS5 is an antibody similar to MIB1 in its ability to recognize the Ki-67 protein.
Assessment of Immunohistochemistry
The interpretation of the immunostaining was done without knowledge of the patients' outcome. The sections were scanned at low and high-power magnifications covering all fields. At least three areas with the highest degree of positive cells were selected. Typically, 400–500 tumor cells in each field were counted irrespective of immunoreactive status. P-p38 MAPK expression was evaluated in tumor cells. Because protein expression is reflected not only by the quantity of labeled cells but also by their staining intensity, we used an intensity-adjusted scoring system to evaluate immunostaining indices. With respect to the invasive cell count, < 10% positive cells were scored as 1, 10–50% were scored as 2, and > 50% were scored as 3. The staining intensity was graded 1–3, corresponding to weak, moderately strong, and intense staining, respectively. Absence of staining was recorded as zero. By multiplying these two factors, an immunoreactive score ranging from 0 to 9 was obtained. The ductal carcinoma in situ (DCIS) component was scored as either positive or negative for P-p38 MAPK. Image analysis was performed as described.23, 28 Briefly, quantitation was performed on the CAS 200 image analyzer (Becton Dickinson Corporation, San Jose, CA) at Quantitative Diagnostics Laboratory (QDL, Westment, IL). This is a microscope-based image analyzer equipped with a two-color imaging channel system. One color channel finds the total area of the tissue specimen. The other channel is specifically matched to have maximum absorption for the chromogen used to recognize the specific amount of antigen. Quantitative results are reported in arbitrary units of optical density corresponding to the staining, which is indicative of the amount of antigen on the tissues (QDL score). As the amount of the specific antigen is increased, so would the immunohistochemistry staining indicating increased levels of specific antigens. HER-2 and Ki-67 were scored as previously described.25, 26
The current study was a retrospective exploratory study to determine the incidence of P-p38 MAPK expression using immunohistochemistry in breast tumor specimens and to evaluate the prognostic role of this protein. The incidence of expression of P-p38 MAPK in breast carcinoma was determined with 95% confidence intervals (95% CI). With a sample size of 96 invasive tumor specimens, we expected to be able to estimate the true incidence with 95% CI of ± 0.08. The chi-square test was used to investigate the independence between two categorical variables. The simple kappa coefficient was used to measure agreement between the QDL score and the visual scoring methods, corrected for chance agreement (< 0.40, poor agreement; 0.40–0.75, fair level of agreement; > 0.75, strong agreement). Levels of P-p38 MAPK expression between groups were compared by Wilcoxon rank sum tests.
We estimated the prognostic value of P-p38 MAPK in this group of patients. The sample size was determined by availability of tissue specimens for patients who participated in a clinical trial of adjuvant chemotherapy (Protocol ID86-012). We identified 100 patients with axillary lymph node involvement who received FAC chemotherapy in the adjuvant setting. Four tumor specimens were not evaluable because of an insufficient tumor specimen. P-p38 MAPK, HER-2, and Ki-67 expression levels were assessed in the remaining 96 evaluable patients. Progression-free survival (PFS) was estimated using the Kaplan–Meier product-limit method. The two-sided log-rank test was used to test the association between variables and PFS. Multivariate analysis was performed using the Cox proportional hazards regression model to determine the association between treatment and PFS after adjusting for the role of other factors. This model was also used to calculate residuals for investigation of the association between P-p38 MAPK levels and PFS. All P values presented are two-sided values. Statistical analyses were performed using SAS 8.0 (SAS, Cary, NC) and Splus 2000 software (Insightful Corporation, Seattle, WA).
We evaluated the expression of P-p38 MAPK, HER-2, and Ki-67 in 96 patients with primary breast carcinomas known to have axillary lymph node involvement. Patients were followed for a median time of 10 years after initial cancer surgery. A total of 44 patients experienced disease recurrence and 33 patients have died. PFS was determined from the date of cancer surgery. The estimated 5-year and 10-year PFS proportions were 0.65 and 0.55, respectively. Overall, premenopausal patients and patients with larger tumors had a significantly worse PFS compared with postmenopausal patients and patients with smaller tumors. Patients with ER-negative tumors had a longer PFS than ER-positive patients. When ER and/or PR were considered, there was no difference in PFS between ER/PR-positive and ER/PR-negative patients (P < 0.1).
P-p38 MAPK, HER-2, and Ki-67 expression levels were measured using immunohistochemistry. Representative stained sections are shown in Figure 1. The P-p38 MAPK slides were scored quantitatively using computerized image analysis (QDL score) and visually (H score). Of 96 tumor specimens, 18 (19%) had an H score > 5 and 23 (24%) had a QDL score > 20. The agreement between the image analysis and visual scoring methods was moderate (κ = 0.55). There were no major differences in PFS according to the level of expression of P-p38 MAPK using either scoring method (log-rank test, P = 0.39). The P-p38 MAPK QDL score was grouped into four levels based on quartiles of the distribution. The group with a P-p38 MAPK QDL score in the fourth quartile (> 20) had worse PFS than other groups. A similar result was found using the H score. Accordingly, P-p38 MAPK was combined into two levels using the QDL score (≤ 20, > 20) or the H score (≤ 5, > 5). The PFS curves for the two levels are illustrated in Figure 2. The suitability of the cutoff point at 20 was confirmed by a plot of regression residuals, which suggested different levels of risk for patients with tumors having P-p38 MAPK expression below and above this point.
As an initial step in considering jointly the correlation between P-p38 MAPK and PFS, correlations between P-p38 MAPK expression and other prognostic factors were assessed. First, the correlation between the original P-p38 MAPK QDL score and ER status was investigated. There were 44 and 41 patients in the ER-negative and ER-positive groups, respectively. The median P-p38 MAPK QDL scores for ER-negative and ER-positive tumors were 10 and 12, respectively. The P value using the Wilcoxon test was 0.51, indicating that the distributions of the P-p38 MAPK QDL score were not statistically significantly different (similar results were obtained for PR status). Tumor size was grouped into two levels (< 2 cm, ≥ 2 cm). There was no strong evidence that P-p38 MAPK was statistically significantly associated with tumor size or menopause status, although a trend was observed between tumor size and P-p38 MAPK expression. Expression of P-p38 MAPK in invasive tissue specimens was associated with P-p38 MAPK expression in adjacent DCIS tissue specimens (P < 0.01).
The Cox proportional hazards model was used to assess the correlation between P-p38 MAPK measured by QDL and PFS, while adjusting for effects of other prognostic factors. Histologic grade and age were omitted from further consideration based on their respective close correlation with ER status and menopause status. Factors considered in the Cox proportional hazards model were P-p38 MAPK expression (≤ 20 vs. > 20), menopause status, tumor size, and ER status. Parameter estimates and hazard ratios are summarized in Table 1. Results suggest that menopause status and tumor size had statistically significant correlations with PFS, after adjusting for other effects in the model. Postmenopausal status and larger tumor size had a tendency to be correlated with a shorter PFS. The effect of ER status approached the 0.05 significance level after adjusting for other factors.
|Variables||No.||Parameter estimate||P value||Hazard ratio|
|P-p38 MAPK QDL|
|Tumor size (cm)|
We assessed correlations between P-p38 MAPK expression and PFS based on HER-2 and Ki-67 status. In patients with HER-2–negative breast carcinoma, high P-p38 MAPK expression was associated with a shorter PFS (P = 0.05; Fig. 3). P-p38 MAPK expression had no prognostic value in the subgroup of patients with HER-2–overexpressing tumors. Patients whose tumor specimens had a high Ki-67 index had a significantly worse PFS if the P-p38 MAPK level was elevated (P = 0.04; Fig. 4).
Mitogenic signaling via the Ras-MAPK pathway was observed in a proportion of breast carcinomas and may contribute to tumor progression.29 Overexpression and/or activation of the MAPK family members, including ERK1 and ERK2, JNK-1, and p38, has been associated with aggressive breast tumor phenotypes including increased invasiveness, hormone receptor negativity, and advanced stage.29–31 Patients whose tumor specimens had activated ERK1 and ERK2 exhibited a poor response to endocrine therapy.31 Importantly, significantly reduced disease recurrence-free survival rates occurred among patients whose primary breast tumor specimens possessed heightened MAPK activity.32
To our knowledge, little is known regarding the role of P-p38 MAPK expression in human tumors. In the current study, P-p38 MAPK was overexpressed in 19–24% of patients, depending on the scoring method. This rate of expression is consistent with a previous report, in which P-p38 MAPK was overexpressed in 17% of patients.23 Our results revealed a correlation between active P-p38 MAPK signaling in tumor specimens and signaling in adjacent DCIS tissue specimens, suggesting that activated p38 MAPK may be an early event in the development of invasive tumors. Alternatively, excess production of growth factors in surrounding tumor stroma may activate the MAPK pathway, driving tumor cell proliferation. A previous study demonstrated increased MAPK activity in breast carcinoma tissue specimens but normal signaling in fibrocystic or normal mammary tissue specimens.29
The results of the current study suggest that P-p38 MAPK expression may be associated with reduced PFS, but is not an independent prognostic factor in lymph node-positive patients with breast carcinoma treated with adjuvant FAC chemotherapy. All available tumor samples for the selected patient group were included in the study and follow-up was sufficient for observation of disease recurrence in 44 patients. This sample size would have provided 90% power to detect a hazard ratio of approximately 3 associated with P-p38 MAPK expression and PFS outcomes (depending on the proportion of patients demonstrating a P-p38 MAPK abnormality), at a significance level of P = 0.05. Patients selected for the current biomarker study had participated in a clinical trial of adjuvant systemic therapy. Patients with ER and/or PR-negative tumors had improved survival. However, it should be noted that none of the patients received tamoxifen in the adjuvant setting. It is noteworthy that P-p38 MAPK levels were not related to ER status, in contrast with previous reports demonstrating an association between hormone receptor negativity and active MAPK signaling.31 The role of P-p38 MAPK expression as a predictive factor of response to hormone therapy is currently under investigation.33
In vitro data indicate that activation of p38 MAPK by phosphorylation plays an important role in radiation and drug-induced cell death.7, 33–38 We could not measure P-p38 MAPK levels before and after chemotherapy because treatment was administered after the primary tumor had been removed. Investigating P-p38 MAPK levels in response to neoadjuvant chemotherapy would be of interest, to determine whether p38 MAPK activation in response to chemotherapy is associated with clinical outcome.
HER-2 overexpression has been associated with a poor prognosis in lymph node-positive patients with breast carcinoma.1 In the current study, P-p38 MAPK expression was an adverse prognostic factor in patients with HER-2–negative tumor specimens. Ki-67 is a marker of proliferation that has been associated with poor prognosis in patients with breast carcinoma.26 We found that in patients whose tumor specimens had a high Ki-67 index, PFS was significantly shorter if p38 MAPK was phosphorylated. Although these findings are hypothesis generating and need to be confirmed in larger, prospective trials, it is possible that activation of p38 MAPK may help predict response to chemotherapy in a subset of patients with breast carcinoma. Further studies will be needed to define the role of P-p38 MAPK as a prognostic and predictive factor in breast carcinoma.
The authors thank Debbie Frye for assistance with data management.
- 31Phosphorylation of ERK1/2 mitogen-activated protein kinase is associated with poor response to anti-hormonal therapy and decreased patient survival in clinical breast cancer. Int J Cancer. 2001; 95: 247–254., , , .
- 33Onset of endocrine resistance in breast cancer is associated with increased active p38 MAPK [abstract]. Breast Cancer Res Treat. 2001; 69: 254a., , , .