One of the most important current issues in breast cancer research is early identification of patients at high risk for relapse coupled with risk-adapted individualized therapy concepts. We have previously shown that the Y-box protein YB-1 is involved in regulating transcription of the P-glycoprotein gene mdr1 and that nuclear localization of YB-1 in human breast cancer is associated with increased P-glycoprotein expression.1 In the experimental setting, expression of P-glycoprotein confers cross-resistance to a variety of cytotoxic agents differing in structure and mechanism of action (e.g., anthracyclines, vinca alkaloids, epipodophyllotoxins and taxanes), resulting in a multidrug-resistant phenotype.2 However, the functional relevance of P-glycoprotein expression for clinical drug resistance in breast cancer is controversial, since evidence for an association between P-glycoprotein expression and survival is not supported by all investigations.3 At present, it is still unclear which functions can be directly attributed to P-glycoprotein, or whether P-glycoprotein expression is merely a surrogate marker for other genetic and biologic changes. Notably, YB-1 is not only a key regulator for P-glycoprotein expression1 but also functions in additional biologic processes that contribute to the development of a malignant phenotype. This includes cellular response to environmental stress4 as well as regulation of genes involved in cell proliferation5 and metastasis.6 However, the clinical relevance of YB-1 is yet unknown. Therefore, we investigated the impact of YB-1 expression on the clinical outcome in breast cancer patients (n = 83) and compared it with expression of the tumor-biologic factors HER2,7 urokinase-type plasminogen activator (uPA) and plasminogen activator inhibitor type 1 (PAI-1),8 which are already in use for clinical decision making.
Intrinsic or acquired resistance to chemotherapy is responsible for failure of current treatment regimens in breast cancer patients. The Y-box protein YB-1 regulates expression of the P-glycoprotein gene mdr1, which plays a major role in the development of a multidrug-resistant tumor phenotype. In human breast cancer, overexpression and nuclear localization of YB-1 is associated with upregulation of P-glycoprotein. In our pilot study, we analyzed the clinical relevance of YB-1 expression in breast cancer (n = 83) after a median follow-up of 61 months and compared it with tumor-biologic factors already used for clinical risk-group discrimination, i.e., HER2, urokinase-type plasminogen activator (uPA) and plasminogen activator inhibitor type 1 (PAI-1). High YB-1 expression in tumor tissue and surrounding benign breast epithelial cells was significantly associated with poor patient outcome. In patients who received postoperative chemotherapy, the 5-year relapse rate was 66% in patients with high YB-1 expression. In contrast, in patients with low YB-1 expressions, no relapse has been observed so far. YB-1 expression thus indicates clinical drug resistance in breast cancer. Moreover, YB-1 correlates with breast cancer aggressiveness: in patients not treated with postoperative chemotherapy, those with low YB-1 expression are still free of disease, whereas the 5-year relapse rate in those with high YB-1 was 30%. There was no significant correlation between YB-1 expression and either HER2 expression or uPA and PAI-1 levels. Risk-group assessment achieved by YB-1 differed significantly from that by HER2 or uPA/PAI-1. In conclusion, YB-1 demonstrated prognostic and predictive significance in breast cancer by identifying high-risk patients in both the presence and absence of postoperative chemotherapy, independent of tumor-biologic factors currently available for clinical decision making. © 2001 Wiley-Liss, Inc.
MATERIAL AND METHODS
The clinical relevance of YB-1 was assessed in 83 breast cancer patients treated at the Department of Obstetrics and Gynecology, Technical University of Munich, Germany. Before surgery, each patient signed informed consent permitting scientific investigation of the tumor tissue. Treatment decisions were solely based on consensus recommendations at the time, i.e., on established parameters such as nodal status, tumor size, histologic grade and steroid hormone receptor status. After surgery, 41 patients were treated by chemotherapy, most of whom (88%) were node-positive: 34 patients received anthracyclin-, 4 patients cyclophosphamide-, methotrexate-, 5-fluorouracil (CMF)-containing regimens and 3 patients other regimens. Five patients received tamoxifen alone as postoperative endocrine therapy and 13 patients both chemo- and endocrine therapy. Forty-two patients did not receive any postoperative chemotherapy; most of these (95%) were node-negative. Median patient age was 54 (range 31–82) years. Clinical and histomorphologic characteristics of the patients are summarized in Table I. Follow-up data were obtained at regular intervals.9 Median follow-up time of patients still alive at time of analysis was 61 (range 1–139) months. Within the follow-up period, 22 patients (27 %) experienced disease recurrence.
|Variable||Patients without postoperative chemotherapy||Patients with postoperative chemotherapy|
|Lymph node metastasis|
|Steroid hormone receptor status§|
Determination of tumor biologic factors
Immunostaining of formalin-fixed, paraffin-embedded breast tissue sections for YB-1 was performed as described earlier.1 As a reference, YB-1 expression was examined in normal breast tissue from patients undergoing elective breast reductive surgery (n = 12). The percentage of YB-1-positive cells was divided into 4 groups: 1, <10%; 2, < 50%; 3, <80%; and 4, ≥80%. A scale from 0 (no detectable immunoreactivity) to 3 (strong immunoreactivity) was assigned to staining intensity. An immunoreactive score (0–12) was then calculated by multiplying percentage score of YB-1-positive cells times staining intensity score.10 HER2 immunostaining was performed using polyclonal antibody A0485 (Dako, Denmark), and membrane staining was scored from 0 to 3+, as recommended for clinical routine. Since 1987, in all breast cancer patients treated at our institution, uPA and PAI-1 antigens have been prospectively measured by ELISA (uPA, Imubind #894; PAI-1, Imubind #821; both from American Diagnostica, Greenwich, CT).11
Correlations between continuous variables were analyzed using the Spearman rank test. Associations between continuous and/or categorical variables were analyzed using the Wilcoxon test, the Mann-Whitney U-test, the χ2 test or the McNemar test, as appropriate. Continuous variables were coded as binary variables to identify low- and high-risk patients.11 For uPA and PAI-1, the previously optimized and reevaluated cutoffs of 3 ng uPA/mg protein and 14 ng PAI-1/mg protein were used.9 The clinically used combination of both factors, uPA/PAI-1, is defined as both factors low vs. either or both factors high. An immunoreactive score of 3+ was considered HER2 overexpression. For YB-1, optimized cutoff values were determined using log-rank statistics. For patients with postoperative chemotherapy, the cutoff values were immunoreactivity score 4 for YB-1 expression in tumor tissue and score 0 for YB-1 in peritumoral breast epithelial cells. For the group of low-risk patients without postoperative chemotherapy, the cutoffs were score 1 for YB-1 expression in tumor tissue and score 4 for YB-1 in peritumoral breast epithelial cells.
The combination of YB-1 expression in tumor tissue and YB-1 expression in peritumoral breast epithelial cells was dichotomized as low in tumor and peritumoral cells vs. high in tumor and/or adjacent normal tissue. Disease-free survival (DFS) rates were calculated according to Kaplan and Meier, and survival curves were compared using log-rank statistics. To account for the small number of events, p-values were determined from exact distributions of the log-rank statistics. All tests were performed at a significance level of α = 0.05.
YB-1 is expressed in tumor tissue and adjacent normal tissue of breast cancer patients
We determined YB-1 expression levels by immunohistochemistry in primary tumor tissue of 83 breast cancer patients. The cohorts consisted of 41 high-risk patients treated by postoperative chemotherapy and 42 patients with low-risk breast cancer without postoperative chemotherapy. As a reference, we examined YB-1 expression in normal breast epithelial cells of women who underwent elective breast reductive surgery (n = 12).
Immunohistochemistry revealed that YB-1 was not expressed in any of the normal breast tissues (Fig. 1a). In the breast cancer specimens, however, YB-1 expression in tumor cells was detectable in 63 of 83 cases (76%), indicating that upregulation of YB-1 expression takes place during breast cancer development. In tumor cells, 2 distinct YB-1 expression patterns were observed: cytoplasmic and nuclear localization (Fig. 1b,c). In 11 cases (13%), distinct nuclear YB-1 expression was noted in tumor cells. In most of these cases (10/11), nuclear expression coincided with strong cytoplasmic expression. Moreover, in a fraction of breast cancers, YB-1 was also expressed in normal tissue adjacent to breast cancer tissue (Fig. 1d). YB-1 immunoreactivity in normal tissue adjacent to the tumor could be determined in only 55 of the 83 breast cancer specimens because the remaining 28 specimens consisted solely of tumor cells. YB-1 scores in tumor and peritumoral cells were strongly correlated (r = 0.84; p < 0.001), with significantly higher scores in tumor (median 6) than in adjacent normal cells (median 2; p < 0.001). In 35/55 cases, YB-1 scores were high in both tumor and peritumoral cells; in 7/55 cases, YB-1 was high in tumor cells but low in peritumoral cells; and in 2/55 cases, only the peritumoral cells expressed low levels of YB-1. In 11/55 cases, tumor and peritumoral cells did not detectably express YB-1 protein.
YB-1 expression and established or novel tumor biologic factors
No significant association was found between YB-1 scores in breast cancer tissue or peritumoral breast epithelium and tumor size, histologic grade or lymph node status. Between YB-1 expression in tumor tissue and steroid hormone receptor status, a negative association was observed, with a median YB-1 score of 8 in hormone receptor-negative and 4 in hormone receptor-positive tumors (p = 0.007). Looking separately at the 2 steroid hormone receptors, the correlation of YB-1 with progesterone receptor status reached borderline significance (p = 0.085), whereas that with estrogen receptor status was not significant. No significant correlation was found between YB-1 expression in tumor or adjacent normal cells and HER2 scores or tumor tissue levels of PAI-1. Levels of uPA in tumor tissue had a borderline significant correlation with YB-1 expression in tumor cells (r = 0.21; p = 0.063) and adjacent normal breast epithelium (r = 0.26; p = 0.074). Risk-group classification acccording to YB-1 differed significantly from that obtained by HER2 (p < 0.001) or by the combination of uPA/PAI-1 (p = 0.006). Moreover, nuclear YB-1 staining coincided in only 2/10 cases with HER2 overexpression and in only 1/8 cases with high uPA/PAI-1.
YB-1 expression indicates resistance to postoperative chemotherapy
In patients who received postoperative chemotherapy (n = 41), YB-1 expression in tumor cells was low in 21 cases and high in 20 cases. The 5-year relapse rate for patients with a low YB-1 score was 39%, compared with 68% in patients with a high YB-1 score (Fig. 2a). This difference reached borderline statistical significance (p = 0.071). Most relapses (6/7) in patients with a high YB-1 score occurred within the first 1.5 years after surgery, indicating resistance to the administered chemotherapy. Risk-group discrimination by YB-1 in patients with postoperative chemotherapy became even more evident (p = 0.037) when YB-1 expression in tumor cells and peritumoral epithelial cells was taken into account. Figure 2b shows that all women with a low YB-1 score in both breast carcinoma and peritumoral epithelial cells survived free of disease (n = 10). In contrast, those patients with a high YB-1 score in breast cancer and adjacent normal cells had a 5-year relapse rate of 66% (p = 0.037). This group included 5 patients with a high YB-1 score in peritumoral normal tissue and a low YB-1 score in breast cancer tissue.
In this group of patients with postoperative chemotherapy, HER2 overexpression was found in 8/32 cases (25%). High uPA/PAI-1 levels were found in 23/35 patients (66%). Neither HER2 (p = 0.723) nor uPA/PAI-1 (p = 0.449) had any significant impact on disease-free survival.
YB-1 identifies a subgroup of women at a high risk for relapse in breast cancer patients without postoperative chemotherapy
In the group of predominantly node-negative patients who did not receive any postoperative chemotherapy (n = 42), the YB-1-immunoreactive score in breast cancer tissue was high in 32 patients and low in 10 patients. So far, none of the 10 patients with a low YB-1 score in the tumor has relapsed. In contrast, 30% of the patients with a high YB-1 score in the tumor relapsed within the first 5 years after surgery (p = 0.011; Fig. 3). Again, combination of YB-1 expression in breast cancer and adjacent epithelial cells was also correlated with clinical outcome: none of the 8 patients with low YB-1 expression in both tumor tissue and adjacent normal epithelial cells experienced disease recurrence during follow-up. In contrast, a 5-year relapse rate of 30% was observed among 32 patients who had high YB-1 immunoreactivity scores (p = 0.0563).
In this group of low-risk breast cancer patients, high uPA/PAI-1 tumor tissue levels were found in 20/39 patients (51%). Patients with high uPA/PAI-1 levels had a 5-year relapse rate of 36% compared with only 6% in patients with low levels (p = 0.0045). HER2 overexpression was found in 6/39 patients (15%); it did not show any significant impact on disease-free survival (p = 0.273).
We have analyzed the clinical relevance of YB-1 expression in breast cancer by evaluating patients treated with postoperative (primarily anthracyclin-containing) chemotherapy and low-risk patients who did not receive any postoperative chemotherapy. In both groups, high YB-1 expression in breast cancer tissue is linked with an unfavorable clinical course of the disease, indicating that YB-1 plays a role in clinical drug resistance as well as tumor aggressiveness. Moreover, not only high YB-1 expression in tumor cells but also high YB-1 expression in peritumoral epithelial cells is associated with poor clinical outcome. This indicates that YB-1 expression in the adjacent epithelial cells reflects a distinct, obviously more aggressive tumor phenotype.
How peritumoral YB-1 expression is brought about is currently unknown, but it is likely that a paracrine effect between breast cancer and adjacent normal tissue is responsible for this phenomenon. Expression of tumorigenesis-associated factors in normal tissue adjacent to breast cancer has been described for major matrix metalloproteases, the mdr1 gene and fibroblast growth factors.12–14 In tumor cells, not only cytoplasmic but also distinct nuclear YB-1 staining is present. Due to the small number of cases with distinct nuclear YB-1 staining, separate analysis of its clinical impact was not possible. However, nuclear YB-1 expression coincides with high cytoplasmic expression in most cases, indicating that high cytoplasmic YB-1 expression may precede its nuclear transport. We previously showed that nuclear localization of YB-1 is associated with expression of P-glycoprotein.1 In the present study, we demonstrate for the first time that cytoplasmic YB-1 expression is of clinical relevance. This observation could be explained by the fact that Y-box proteins regulate both transcription and translation, i.e., they exert a variety of functions in the nucleus as well as in the cytoplasm.15
The observed impact of YB-1 on the clinical outcome in breast cancer patients is in line with experimental data that indicate an involvement of YB-1 in biologic processes determining drug resistance and tumor aggressiveness. Y-box factors represent a highly conserved class of proteins implicated in the cellular response to environmental stress, e.g., cold-shock, nutritional deprivation, hypoxia, or genotoxic stress.4 YB-1 binds to damaged DNA, and its downregulation by antisense constructs increases sensitivity to DNA-damaging agents.16 In addition, YB-1 regulates transcription of the matrix metalloproteinase 2 (MMP-2)/gelatinase A gene, which encodes a protease facilitating cell migration.6 High MMP-2 expression is correlated with tumor aggressiveness and decreased survival in breast cancer patients.17 YB-1 regulates the mdr1 gene1 encoding P-glycoprotein, a marker not only for chemoresistance2 but also for tumor aggressiveness.18 Moreover, a role of Y-box proteins in regulation of cell proliferation has been discussed, and high expression of Y-box proteins occurs under conditions of cell proliferation, e.g., in certain embryonal tissues, in fetal liver, in the regenerating liver after tissue damage5 and in the proliferating compartment of colorectal mucosa.19 YB-1 is induced in T lymphocytes by interleukin-2 (IL-2) stimulation, where it is involved in stabilizing IL-2 mRNA;20; likewise, YB-1 is induced in quiescent fibroblasts by serum stimulation.21 Several cell cycle-regulated genes contain Y-box sequences in their promoter or enhancer regions.5 Recently, several groups have reported high levels of YB-1 expression in human osteosarcomas,22 colorectal carcinomas,19 ovarian serous adenocarcinomas,23 malignant melanomas,24 and nasopharyngeal carcinomas,25 indicating a broader role for YB-1 in tumor development.
In our study, YB-1 expression did not correlate significantly with either HER2 expression or tumor tissue levels of uPA/PAI-1, i.e., with novel tumor biologic factors currently used in clinical decision making. Nuclear YB-1 expression did not coincide with HER2 overexpression or elevated levels of uPA/PAI-1. Risk-group assessment obtained by YB-1 differed significantly from that according to HER2 or uPA/PAI-1.
Whereas data on the impact of HER2 overexpression on patient prognosis remain controversial,7 there is increasing evidence that HER2 overexpression is predictive of response to chemotherapy, particularly for sensitivity to anthracyclin-containing regimens.26 Our patients with HER2-overexpressing tumors who received chemotherapy had a similar DFS to those without HER2 overexpression, indicating a survival benefit from the predominantly anthracyclin-containing chemotherapy in the high-risk patients. In contrast, patients with high YB-1 expression, compared with those with low YB-1 expression, had a significant DFS disadvantage after chemotherapy, suggesting resistance to the postoperative chemotherapy in YB-1-overexpressing tumors.
There is abundant evidence from the literature that the tumor invasion factors uPA and PAI-1 permit clinically relevant risk-group discrimination in breast cancer, particularly in low-risk, i.e., node-negative, patients.8 A prospective randomized multicenter therapy trial in node-negative breast cancer has recently validated the prognostic impact of the combination uPA/PAI-1 and indicated that patients with high uPA/PAI-1 levels do benefit from adjuvant chemotherapy.27 Thus, our findings that uPA/PAI-1 significantly predict patient outcome in low-risk breast cancer patients, but not in patients with postoperative chemotherapy, are in accord with this published evidence. In our study, immunohistochemical YB-1 determination renders similar prognostic information as uPA/PAI-1 in low-risk patients but—unlike uPA/PAI-1 or HER2—it allows identification of patients who do not respond to postoperative, predominantly anthracyclin-containing chemotherapy.
In the clinical setting, adequate risk-group assessment and prediction of therapy response are prerequisites for individualized therapy concepts in breast cancer. A number of established as well as novel tumor biologic prognostic factors such as nodal status, tumor size, histologic grade and uPA/PAI-1 already allow relevant risk-group discrimination. Unfortunately, useful predictive markers are still scarce, in particular for chemosensitivity.28 So far, anthracyclin-containing regimens are considered the single most effective chemotherapeutic agents next to taxanes.29 Thus, the results of our pilot study implicating YB-1 as a new tumor biologic marker involved in determining chemotherapy resistance are of immediate clinical relevance. Of course, these preliminary findings warrant independent validation by larger clinical studies. Also, considering that YB-1 expression in both tumor cells and adjacent normal breast epithelium had an impact on patient outcome, different determination techniques (e.g., ELISA) may be more suitable for clinical routine. Moreover, future research needs to reveal the mechanisms of YB-1-mediated drug resistance at a molecular level. Nevertheless, the results presented here do indicate that YB-1 represents a promising target molecule for novel therapeutic strategies aimed at overcoming multidrug resistance and tumor aggressiveness.
We thank Ms. S. Metzner for technical assistance and the Berliner Krebsgesellschaft and the state of Bavaria for funding our work.