Abstract
The chemokine stromal cell-derived factor-1 (SDF-1) may function to attract CXCR4-expressing cancer cells to metastatic organs. We have previously demonstrated that low plasma SDF-1, a host-derived marker, increases distant metastatic risk in breast cancer. We therefore hypothesized that tumors overexpressing the SDF-1 receptor CXCR4 have an enhanced ability to metastasize in patients with low plasma SDF-1 levels. In this study, we determined the prognostic significance of activated CXCR4, or phosphorylated CXCR4 (p-CXCR4), and CXCR7, another receptor for SDF-1. Immunohistochemistry was performed on a tissue microarray built using 237 samples from the same cohort of patients for which we measured plasma SDF-1 levels. We found that the prognostic value of p-CXCR4 expression (hazard ratio or HR, 3.95; P = 0.004) was superior to total CXCR4 expression (HR, 3.20; P = 0.03). The rate of breast cancer-specific mortality was much higher in patients with both high p-CXCR4 expression and low plasma SDF-1 levels (HR, 5.96; P < 0.001) than either low plasma SDF-1 (HR, 3.59; P = 0.01) or high p-CXCR4 expression (HR, 3.83; P = 0.005) alone. The added prognostic value of low plasma SDF-1 was only effective in patients with high p-CXCR4 expression, and as such, provides clinical validation for modulation of the metastatic potential of tumor cells by an inherent host-derived metastatic risk factor.
Breast cancer is the second most common cancer in women and represents a major risk to women’s lives because of the life-threatening consequences of metastatic disease (SEER Cancer Statistics Review, http://seer.cancer.gov/csr/1975_2005/; accessed September 1, 2008).1 The process of metastasis has often been reported as a cascade of events, with emphasis placed on the tumor cell and its potential to proliferate, invade into the circulation, exit the bloodstream, and grow at the metastatic site.2 However, little is known about the manner in which the host can modulate tumor progression and the propensity of the tumor to metastasize. Indeed, the role of the host was recognized over a century ago in the “seed and soil” theory, whereby the presence of a “congenial” environment of the host metastatic organ influenced the colonization of tumor cells at specific distant organs.3 More recently, a chemokine-receptor model was proposed to help explain the manner in which the host influences the homing of cancer cells to specific target organs. Muller et al. proposed that chemokines, such as stromal cell-derived factor-1 (SDF)-1, are normally overexpressed by those target organs to which breast cancer metastasizes, such as lung, liver, and bone, and serve to attract breast cancer cells that express their receptors, such as CXCR4.4 Various animal studies have subsequently demonstrated the functional role of CXCR4 as the prime chemokine receptor involved in distant metastasis in breast and other types of cancers.5,6,7,8,9
Several studies, including our own, have since observed an association between CXCR4 expression and distant metastasis in primary breast cancer patients.10,11,12,13 Furthermore, we recently identified circulating levels of SDF-1 as a prognostic blood marker in a series of patients with primary breast cancers. Interestingly, circulating SDF-1 levels were found to be independent from tumor-derived SDF-1, and as such, plasma SDF-1 is the first candidate host-derived blood marker in breast cancer. We found that a low plasma SDF-1 level was predictive of distant metastasis, suggesting that low SDF-1 in the circulation may favor the extravasation of tumor cells from the circulation to the metastatic site.14 In accordance with Muller’s hypothesis, the differential concentration gradient of SDF-1, that is, low blood SDF-1 and high tissue SDF-1 at the metastatic site, may enhance the homing of CXCR4-expressing cancer cells. In this case, it would be expected that tumors with high CXCR4 expression would be especially sensitive to the SDF-1 gradient at metastatic target organs. To test this hypothesis, we determined the expression of CXCR4 in primary tumors using the same cohort of breast cancer patients in which we previously measured plasma SDF-1 levels. We determined if patients with an innate susceptibility for metastasis, associated with low levels of plasma SDF-1, demonstrated a greater risk of metastasis when their tumors expressed higher levels of CXCR4. In addition to tumor expression of CXCR4, we also measured the levels of the phosphorylated CXCR4 receptor as a means of quantifying CXCR4 activity and compared its expression in the primary tumor and metastatic lymph nodes. We also measured tumor expression of the two factors that may activate CXCR4: its ligand SDF-1, and another chemokine receptor, CXCR7, which may activate CXCR4 via heterodimerization.15 In this way, we provide a more detailed picture of metastatic risk associated with the activity of CXCR4 in the primary breast tumor and relate it with the risk of metastasis associated with low plasma SDF-1 levels.
Materials and Methods
Patients
We used the same cohort of patients as described previously.14 Three hundred five patients with primary breast cancers of stages I, II, and III were recruited from 2000 to 2003 with a median follow-up of 3.3 years, with informed consent, as per the Research Ethics Committee of the Centre Hospitalier de l’Université de Montréal. Thirty-seven patients were excluded due to unavailability of tissue blocks. Thirty-one patients were further excluded due to absence of the prognostic tumor lesion, leaving 237 patients for correlation with clinicopathological characteristics and survival analysis. Tissue cores from the microarray were damaged for up to three other patients, resulting in a minimum of 234 patients. Due to incomplete data available regarding HER2 status from the pathology reports, HER2 was re-stained using our tissue microarray. Nine percent of the patients were HER2 positive by immunohistochemistry (either 2+ or 3+), and 18% were estrogen receptor (ER) negative/progesterone receptor (PR) negative/HER2 negative (ER−/PR−/HER2−), also known as triple negative. Corresponding plasma samples were available for 212 patients.
Western Blot Analysis
Human umbilical vein endothelial cells (Cambrex BioScience, Walkersville, MD) were serum starved for 3 hours before being stimulated with recombinant human SDF-1 (R&D Systems, Minneapolis, MN). Cell lysates were prepared using lysis buffer consisting of 1% Triton X-100, 25 mmol/L Tris (pH 7.5), 150 mmol/L NaCl, and 5 mmol/L EDTA was supplemented with protease inhibitor cocktail set III (Calbiochem, Gibbstown, NJ) and 1 mmol/L sodium orthovanadate. Cell lysates (30 μg protein) were solubilized in NuPAGE lithium dodecyl sulfate sample buffer, incubated at 37°C for 30 minutes, and run through 10% NuPAGE Bis-Tris gels (Invitrogen, Carlsbad, CA). After transfer, Immobilon-P membranes (Millipore, Billerica, MA) were incubated overnight with antibody against p-CXCR4 (courtesy of Dr. Joshua Rubin, Washington University, St. Louis, MO, 1:1000).16 Relative protein expression levels were estimated by membrane rehybridization with anti-mouse CD184 (2B11, 1:250, BD PharMingen, San Jose, CA). Antibody detection was performed using enhanced chemiluminescence (ECL; Amersham Pharmacia Biotech, Piscataway, NJ).
Tissue Microarray Construction
Formalin-fixed, paraffin-embedded tissue blocks were collected from the Department of Pathology from the Centre Hospitalier de l’Université de Montréal. All blocks were re-sectioned and stained for hematoxylin phloxine saffron stain before marking of histological lesions. We constructed a tissue microarray as described previously17 using a Manual Tissue Arrayer I (Beecher, Sun Prairie, WI). In total, 1619 cores were punched and distributed into four recipient blocks. Lesions were placed in either duplicate or triplicate cores adjacent to one another. Six-micrometer sections were cut using the tape transfer system (Instrumedics, St. Louis, MO).
Immunohistochemistry
Immunohistochemistry was performed via the labeled streptavidin biotin method for phosphorylated CXCR4 (p-CXCR4), SDF-1, CXCR7, and Ki-67 as previously described.10 Primary antibodies and concentrations used were: p-CXCR4, (courtesy of Dr. Joshua Rubin) at a 1:250 dilution; SDF-1 (MAB350, clone 79018, R&D Systems) at 10 μg/ml; CXCR7 (MAB4227, clone 358426, R&D Systems) at 10 μg/ml; and Ki-67 at a 1:50 dilution (M7240, clone MIB-1, Dako, Denmark). All primary antibodies were incubated overnight at 4°C. A biotin-labeled secondary antibody was used, either goat anti-mouse at 2.4 μg/ml or 9 μg/ml for Ki-67 (catalog no. 115-065-003, Jackson ImmunoResearch Laboratories, West Grove, PA) or goat anti-rabbit at 2.75 μg/ml (catalog no. 111-065-003, Jackson ImmunoResearch Laboratories). Biotin detection was performed with peroxidase-conjugated streptavidin (catalog no. 016-030-084, Jackson ImmunoResearch Laboratories) at 0.2 μg/ml for p-CXCR4, 0.1 μg/ml for SDF-1, 0.08 μg/ml for CXCR7, and 0.25 μg/ml for Ki-67. CXCR4 expression was detected using a biotin-labeled CXCR4 antagonist, TN14003, synthesized in the Saragovi laboratory, following reported methods.6,18 The staining intensity (0, 1, 2, 3) and percentage of positively stained cells (0 to 100%) were scored in a blinded manner.
Statistical Analysis
Analysis for each biomarker was performed using the product score, whereby the product of the staining intensity and percentage of positive cells of the cytoplasm was used to create a continuous score from 0 to 300. The product score of each biomarker was analyzed as both a continuous variable for correlations between biomarkers and with clinicopathological characteristics, and as a categorical variable, for survival and comparative analysis between primary tumor and lymph nodes. The product score was divided into low, medium, and high expression categories using outcome-derived cut points from X-tile (version 3.6.1, Robert Camp, Yale University, New Haven, CT19). For survival analysis, high expression of the biomarker was compared with low expression, whereas the medium and high categories were combined and termed as “high” for all other categorical variable analysis. Correlations between biomarkers and clinicopathological characteristics were performed using Spearman’s rank correlation and with χ2 or Fisher’s exact test for categorical variable analysis, as previously described.14 Clinicopathological correlations examined include age, tumor size, lymph nodes, stage, tumor grade, ER, PR, HER2 status, triple negative disease. Survival analysis was performed for breast cancer-specific survival and distant disease-free survival as described previously.14 Since this is the first survival analysis of p-CXCR4 in cancer patients, an a priori sample size could not be determined. Survival analysis was first performed on X-tile from which cut-points were obtained with subsequent cross-validation. Subsequently, a Cox proportional hazards regression model20 was used for univariate (n = 237) and multivariate (n = 196) analysis. Covariates included in multivariate analysis were: age, tumor size, lymph node status, tumor grade, ER, PR, and neoadjuvant or adjuvant hormonal therapy or chemotherapy. Correlation analysis between tissue biomarkers and plasma SDF-1 levels was performed using Spearman’s rank correlation. Survival analysis for the combination of tissue biomarker and plasma SDF-1 levels was performed as described above for univariate (n = 212) and multivariate (n = 177) analysis. No statistical significance was identified for patients who were excluded due to unavailability of blood or missing information for multivariate analysis for each endpoint. All reported P values are two-sided. All statistical analysis was performed using STATA version 9.2 (College Station, TX).
Results
Correlation of p-CXCR4 with CXCR4, SDF-1, and CXCR7
To gain a more complete understanding of the role of the CXCR4/SDF-1 receptor/ligand axis in breast cancer, we measured the expression of total CXCR4 receptor together with phosphorylated-CXCR4, the activated form of the receptor, its ligand, SDF-1, as well as the CXCR7 receptor. To verify the specificity of the p-CXCR4 antibody, we treated human umbilical vein endothelial cells that are known to express CXCR4 endogenously, with recombinant human SDF-1 for 15 minutes. Expression of p-CXCR4 was induced on SDF-1 stimulation of human umbilical vein endothelial cells (Figure 1). To detect the expression of total CXCR4, we synthesized a biotinylated anti-CXCR4 peptide, biotinylated-TN14003, as this peptide was previously reported to show greater specificity in immunohistochemistry in comparison with a commercially available antibody.6 Immunohistochemical analysis of p-CXCR4 and CXCR4 from the tissue microarray revealed cytoplasmic and nuclear expression for both biomarkers (Figure 2, A–D). Cytoplasmic p-CXCR4 was expressed at moderate to high levels in 47% of breast tumors (see Supplemental Figure S1 at http://ajp.amjpathol.org). Expression of p-CXCR4 correlated positively with tumor progression (rho = 0.42, P < 0.0001): 9% of normal lesions, 54% of ductal carcinoma in situ, 47% of tumors, and 54.8% of lymph nodes demonstrated high expression of p-CXCR4, showing that stage 0, I, II and III breast cancer had much higher levels of p-CXCR4 expression than normal breast tissues (see Supplemental Figure S2 at http://ajp.amjpathol.org). Levels of cytoplasmic tumor p-CXCR4 expression correlated strongly with both cytoplasmic CXCR4 (rho = 0.58, P < 0.0001) and nuclear CXCR4 (rho = 0.54, P < 0.0001) expression. These results are in concordance with our previous results for CXCR4.10 For the sake of simplicity, from here on, we will only refer to the cytoplasmic expression of both markers. To further understand the significance of p-CXCR4, we examined the expression of SDF-1 and CXCR7. SDF-1 expression was found mainly in the cytoplasm of tumor cells (Figure 3, A and B). Interestingly, expression of SDF-1 correlated positively with p-CXCR4 (rho = 0.19, P = 0.004), but not with CXCR4 (rho = 0.05, P = 0.41), suggesting that autocrine stimulation of CXCR4 may contribute to CXCR4 phosphorylation in breast tumors. CXCR7 was predominantly expressed in the cytoplasm, and less so in the nucleus, and thus we refer only to cytoplasmic expression (Figure 3, C and D). A strong positive correlation was found between CXCR7 and SDF-1 (rho = 0.32, P < 0.0001), CXCR4 (rho = 0.45, P < 0.0001) and, above all, p-CXCR4 (rho = 0.49, P < 0.0001) expression. Therefore it is plausible that the phosphorylation of CXCR4 may be induced by SDF-1 and/or by co-expression of CXCR4 and CXCR7 in breast cancers (Table 1).
Figure 1.
Detection of CXCR4 phosphorylation in primary human endothelial cells. Human umbilical vein endothelial cells were incubated in medium alone or treated with SDF-1 (100 ng/ml) for 15 minutes. p-CXCR4 expression was evaluated by immunoblotting with an antibody to p-CXCR4 and reblotting with an anti-mouse CXCR4 (CD184) antibody. Results are representative of three independent experiments.
Figure 2.
Immunohistochemical analysis of p-CXCR4 and CXCR4 using 20× objective lens magnification. A: High expression of p-CXCR4. B: Low expression of p-CXCR4. C: High expression of CXCR4. D: Low expression of CXCR4.
Figure 3.
Immunohistochemical analysis of SDF-1 and CXCR7 using 20× objective lens magnification. A: High expression of SDF-1. B: Low expression of SDF-1. C: High expression of CXCR7. D: Low expression of CXCR7.
Table 1.
Correlation Analysis between Biomarkers
| Variable | Rho | P value |
|---|---|---|
| CXCR4 and p-CXCR4 | 0.58 | <0.0001 |
| SDF-1 and CXCR4 | 0.05 | 0.41 |
| SDF-1 and p-CXCR4 | 0.19 | 0.004 |
| CXCR7 and SDF-1 | 0.32 | <0.0001 |
| CXCR7 and CXCR4 | 0.45 | <0.0001 |
| CXCR7 and p-CXCR4 | 0.49 | <0.0001 |
P-CXCR4 Has a Better Prognostic Value than CXCR4
To determine the prognostic significance of CXCR4 and p-CXCR4 expression, survival analysis was performed using the product score. Categories of low, medium, and high expression were obtained using cut points derived from the X-tile software. For CXCR4, the population was divided into low (41%, with product score or PS, ranging from 0 to 53.3), medium (30%; PS, 55.6–130), and high (29%; PS, 131–300). For p-CXCR4, the population was divided into low (53%; PS, 0–150), medium (24%; PS, 152–203), and high (24%; PS, 209–300). High p-CXCR4 expression demonstrated a greater prognostic value than high CXCR4 expression for breast cancer-specific survival and distant disease-free survival in univariate analysis. Patients with high p-CXCR4 expression demonstrated a fourfold higher rate of death (hazard ratio or HR, 3.95; 95% confidence interval or CI, 1.55–10.03; P = 0.004) due to breast cancer-related causes, which is greater than that for patients with high CXCR4 expression (HR, 3.20; 95% CI, 1.09–9.37; P = 0.03). Furthermore, for the risk of distant disease-free survival, high p-CXCR4 expression exhibited greater significance (HR, 2.38; 95% CI, 1.13–5.00; P = 0.02) than high total CXCR4 expression (HR, 1.90; 95% CI, 0.85–4.25; P = 0.12). To determine whether p-CXCR4 or CXCR4 is an independent marker for survival, multivariate analysis was performed for both endpoints, and no statistical significance was found for either p-CXCR4 or CXCR4 (data not shown). Although this may be due to the small size of our patient cohort, the superiority of p-CXCR4 over CXCR4 in breast cancer-specific survival and distant disease-free survival suggests that p-CXCR4 expression may be a more sensitive marker than CXCR4 expression for metastatic risk.
P-CXCR4 Enhances Prognostic Value of Plasma SDF-1 level
To better understand the metastatic risk of high p-CXCR4 expression in the context of host-derived risk, we examined the prognostic significance of p-CXCR4 in combination with blood SDF-1 levels. We previously measured SDF-1 blood levels from the same cohort of breast cancer patients and found that plasma SDF-1 is a host-derived marker predictive of distant metastasis.14 To confirm once again that circulating SDF-1 levels are independent of the tumor, we compared the tumor expression of SDF-1 with plasma SDF-1 levels and found no correlation between the two variables (rho = −0.08, P = 0.23). We now investigated the prognosis of patients who expressed both high levels of p-CXCR4 and low plasma SDF-1 levels. Using the median value of plasma SDF-1, as in our previous study, the cohort was again divided into two groups, high and low SDF-1. Patients with both a low plasma SDF-1 level and high p-CXCR4 expression (n = 29, or 14% of the entire cohort) (Table 2) demonstrated a significant correlation with the development of distant metastasis (rho = 0.25, P = 0.0003), stronger than that of plasma SDF-1 alone (rho = −0.17, P = 0.01). As there were 22 fewer patients for whom both plasma and tissue samples were available (n = 212), the prognostic value for each variable was recalculated for breast cancer-specific survival and revealed similar values: low plasma SDF-1 (HR, 3.59; 95% CI, 1.33–9.74; P = 0.01) and high p-CXCR4 (HR, 3.83; 95% CI, 1.49–9.90; P = 0.005) (Figure 4, A and B). Patients with the combination of both low plasma SDF-1 and high p-CXCR4 showed a very poor prognosis (HR, 5.96; 95% CI, 2.57–13.81; P < 0.001) (Figure 4C), which remained significant after multivariate analysis (adjusted HR, 3.78; 95% CI, 1.31–10.94; P = 0.01). After adjustment for Ki-67 labeling index (n = 142), a marker for cellular proliferation, the combination remained significant for breast cancer-specific survival (HR 3.70; 95% CI, 1.02–11.48; P = 0.005).
Table 2.
Clinicopathological Properties of Patients Who Expressed Both Low Plasma SDF-1 and High Tumor p-CXCR4
| Variable | Number of patients | % |
|---|---|---|
| Age (yr) | ||
| ≤50 | 12 | (41.4) |
| >50 | 17 | (58.6) |
| Tumor size (cm) | ||
| T1 (≤2) | 10 | (34.5) |
| T2 (2–5) | 14 | (48.3) |
| T3 (>5) | 4 | (13.8) |
| T4 | 1 | (3.4) |
| Nodal status | ||
| N0 | 13 | (44.8) |
| N1 | 7 | (24.1) |
| N2 | 4 | (13.8) |
| N3 | 2 | (6.9) |
| Lymphadenectomy not performed | 3 | (10.3) |
| Stage | ||
| I | 7 | (24.1) |
| II | 12 | (41.4) |
| III | 7 | (24.1) |
| Unavailable | 3 | (10.3) |
| Tumor grade | ||
| 1 | 0 | (0) |
| 2 | 9 | (31.0) |
| 3 | 15 | (51.7) |
| Unavailable | 5 | (17.2) |
| ER Status | ||
| Negative | 17 | (58.6) |
| Positive | 12 | (41.4) |
| PR status | ||
| Negative | 20 | (69.1) |
| Positive | 9 | (31.0) |
| Triple negative | ||
| Present | 13 | (44.8) |
| Absent | 13 | (44.8) |
| Unavailable | 3 | (10.3) |
| Luminal A/B | ||
| Present | 11 | (37.9) |
| Absent | 15 | (51.7) |
| Unavailable | 3 | (10.3) |
| HER2 status | ||
| Positive | 3 | (10.3) |
| Negative | 23 | (79.3) |
| Unavailable | 3 | (10.3) |
Figure 4.
Kaplan-Meier survival curves for breast cancer-specific survival for plasma SDF-1 (A), tumor p-CXCR4 (B), and combination of low plasma SDF-1 and high p-CXCR4 (C).
A similar enhancing effect was also apparent in distant disease-free survival, whereby patients with the combination showed an almost fourfold greater rate of distant recurrence (HR, 3.75; 95% CI, 1.82–7.76, P < 0.001), greater than either biomarker alone: low plasma SDF-1 (HR, 2.15; P = 0.04) and high p-CXCR4 (HR, 2.31; P = 0.03). The combination was also significant after multivariate analysis for distant disease-free survival (adjusted HR, 2.80; 95% CI, 1.14–6.83; P = 0.02). On the other hand, patients with both low levels of p-CXCR4 expression and low plasma SDF-1 levels did not exhibit a significantly poorer prognosis for breast cancer-specific survival (HR, 0.69; 95% CI, 0.23–2.04; P = 0.50) or distant disease-free survival (HR, 0.78; 95% CI, 0.36–1.69, P = 0.52) than the remainder of the entire cohort. Therefore, the poor prognostic value that we observed in patients with low plasma SDF-1 levels is enhanced in patients with tumors showing high expression of p-CXCR4, and not low p-CXCR4. No interaction between plasma SDF-1 and p-CXCR4 expression was observed in univariate or multivariate analysis for both endpoints (data not shown). Therefore, the prognostic value of high tumor p-CXCR4 and low plasma SDF-1 levels are independent from one another, reflecting the independent source of each marker. Thus, we have identified a specific cohort of primary breast cancers that express high levels of p-CXCR4, suggesting a propensity for significant CXCR4 activity, whose later extravasation into metastatic target sites may be especially promoted in the presence of low plasma SDF-1 levels. We also examined the prognostic value of tumor expression of total CXCR4 and plasma SDF-1 levels. We found that patients with high CXCR4 tumor expression and low plasma SDF-1 demonstrated a significantly worse prognosis due to breast cancer-related causes (HR, 3.45; 95% CI, 1.49–7.99; P = 0.004), compared with patients with both low plasma SDF-1 and low CXCR4 expression (HR, 0.96; 95% CI, 0.33–2.83; P = 0.95). Therefore, in patients with low plasma SDF-1 levels, tumor metastasis appears to be promoted particularly in cancer cells that express high levels of p-CXCR4 or CXCR4.
Elevated Expression of CXCR4 in Lymph Nodes
If tumor cells expressing high CXCR4 or p-CXCR4 are more likely to metastasize, we would expect to find more of these cells in the first site of metastasis, regional lymph nodes. Lymph nodes were available for 34 patients with their matched primary tumor also present on the tissue microarray. Although the frequency of elevated p-CXCR4 expression was the same in primary tumor and lymph nodes, 88% of patients demonstrated high total CXCR4 expression in the lymph nodes, in comparison with 64% in the primary tumor, which was statistically significant via McNemar’s test (P = 0.02). Paired t-test analysis demonstrated a significant difference in the product score of CXCR4 in lymph nodes compared with primary tumor (P < 0.0001), which was not the case for p-CXCR4. 64% of patients demonstrated a higher product score of CXCR4 in the lymph nodes versus primary tumor. Therefore, these results suggest that tumor cells with higher expression of the CXCR4 receptor are more likely to undergo regional metastasis. Given the lack of difference in p-CXCR4 expression levels between the matched primary tumor and lymph nodes, it may be that the microenvironment of the lymph nodes does not particularly select for activation of the CXCR4 receptor.
Clinical Implications for Tumor Expression of p-CXCR4, CXCR4, and CXCR7
CXCR4 and p-CXCR4 expression were correlated with clinicopathological characteristics using Spearman’s rank correlation analysis (Table 3). Categorical variable analysis of low and high expression of each biomarker is provided in Supplemental Table S1 at http://ajp.amjpathol.org. We found that levels of both CXCR4 and p-CXCR4 expression inversely correlate with ER and PR positivity, and positively correlate with tumor grade and ER-/PR-/HER2- (triple negative) status. Triple negative tumors were almost twice as likely to have medium or high expression of p-CXCR4 as all other tumors (77% versus 41%, P < 0.001).
Table 3.
Clinicopathological Correlations of Tumor CXCR4, P-CXCR4, and CXCR7 Expression
| Variable | CXCR4
|
P-CXCR4
|
CXCR7
|
|||
|---|---|---|---|---|---|---|
| Rho | P value | Rho | P value | Rho | P value | |
| Age | −0.02 | 0.73 | −0.05 | 0.41 | 0.08 | 0.21 |
| Tumor size | 0.11 | 0.11 | 0.13 | 0.05 | 0.02 | 0.76 |
| Lymph nodes | −0.02 | 0.76 | 0.01 | 0.84 | 0.005 | 0.94 |
| Stage | −0.002 | 0.98 | 0.08 | 0.25 | 0.03 | 0.71 |
| Grade | 0.35 | <0.0001 | 0.37 | <0.0001 | 0.14 | 0.04 |
| ER positivity | −0.27 | <0.0001 | −0.32 | <0.0001 | −0.18 | 0.007 |
| PR positivity | −0.26 | <0.0001 | −0.31 | <0.0001 | −0.20 | 0.002 |
| HER2+ | −0.06 | 0.39 | 0.09 | 0.23 | 0.004 | 0.95 |
| ER−/PR−/HER2− | 0.37 | <0.0001 | 0.33 | <0.0001 | 0.14 | 0.05 |
Interaction analysis from the multivariate Cox model of p-CXCR4 revealed an interaction between p-CXCR4 expression and ER status (HR, 0.13; 95% CI, 0.02–1.12; P = 0.06) such that ER-positive patients with high p-CXCR4 expression (15% of the total) had a 6.5-fold worse prognosis than all other ER-positive patients (HR, 6.49; 95% CI, 1.08–38.9; P = 0.04). This remained essentially unchanged after multivariate analysis (HR, 6.40; 95% CI, 0.95–42.9; P = 0.06). No such correlation with survival was found with high phosphorylated CXCR4 expression in ER-negative patients. Thus, despite the otherwise good prognosis of all ER-positive patients, high p-CXCR4 expression has the power to identify a subset of these patients with poor prognosis. Furthermore, subgroup analysis revealed that the prognostic value of the combination was the greatest among patients with Luminal A or B subtype (ER+/PR+/HER2-, or ER+/PR+/HER2+) (n = 133; HR 8.08, 95% CI, 2.28–28.7; P = 0.001) and the least in the triple negative group of breast cancers (n = 33, HR 2.24, 95% CI, 0.60–8.37; P = 0.23). Thus the presence of the combination marker appears to have a great effect in ER+ breast cancers, while it may not contribute as much to prognostic information in triple negative breast cancers (although numbers in these subgroup analyses are limited).
Due to the functional significance of CXCR7 previously reported in breast cancer tumorigenesis and metastasis,21,22 we analyzed the clinical relevance of CXCR7. High expression of CXCR7 was associated with poorer outcome in breast cancer-specific survival (HR, 3.63; 95% CI, 1.35–9.76; P = 0.01), and distant disease-free survival (HR, 2.21; 95% CI, 1.00–4.87; P = 0.05), both of which were not significant after multivariate analysis (data not shown).
Discussion
To date, much of the cancer literature has interpreted the metastatic process to be largely dependent on the aggressive potential of the tumor and its ability to invade surrounding tissues and metastasize. However, in addition to the tumor, recent evidence has introduced the significance of the host and its role in predicting metastatic propensity. For example, a genetic influence on metastatic progression has been observed: a Swedish study reported that mothers and daughters of patients with breast cancer of poor outcome who developed breast cancer themselves demonstrated poor prognosis like their first-degree relatives.23 We previously identified the first host-derived blood marker predictive of distant metastasis in breast cancer, the SDF-1 chemokine,14 and found that low levels of plasma SDF-1 were predictive of distant metastasis, suggesting that the concentration gradient of SDF-1 between metastatic site and plasma may play a critical role in promoting the extravasation of cancer cells. Since we and others have previously shown an association between overexpression of CXCR4 in primary breast tumors and metastatic risk,10,11,12 we investigated whether the metastatic potential of CXCR4 overexpression could be further augmented in the context of low blood SDF-1 levels. Indeed, we found that patients who showed both low blood SDF-1 levels and high tumor CXCR4 expression demonstrated a significantly worse prognosis in comparison with patients with low plasma SDF-1 levels whose tumors did not express high levels of CXCR4. These results suggest that a low plasma SDF-1 level may favor the extravasation of tumor cells expressing high CXCR4. This hypothesis is further corroborated by the higher levels of CXCR4 expression we observed in lymph nodes compared with matched primary tumors, although the mechanism of lymphatic dissemination may be different from hematogenous spread. Enrichment for CXCR4-expressing tumor cells at the metastatic site has been reported previously.24,25 The tumor-derived risk of metastasis (CXCR4) was thus enhanced with an intrinsic host-derived risk (SDF-1). As a result, we present here, for the first time, evidence for a biologically plausible scenario, providing insight into a dysfunctional relationship between the tumor and its host, which impacts the capacity of breast cancers to form metastasis.
We also found that overexpression of CXCR4 was frequently associated with activation of CXCR4 via phosphorylation. Consistent with a previous report in brain tumors,16 we found that expression of p-CXCR4, and not total CXCR4, highly correlated with SDF-1 expression, suggestive of the presence of autocrine stimulation of CXCR4 in primary breast tumors. Furthermore, p-CXCR4 expression also correlated strongly with the expression of CXCR7, a recently discovered receptor for SDF-1, implying that heterodimerization of CXCR7 with CXCR415,21 may also contribute to activation of CXCR4 patients with primary breast cancers. We then found that high expression of p-CXCR4 is predictive of a fourfold higher rate of breast cancer-specific mortality, and that the prognostic value of high p-CXCR4 is superior to that of high CXCR4 expression for breast cancer-specific survival and distant disease-free survival. Moreover, patients with both high p-CXCR4 levels and low blood SDF-1 levels had a nearly sixfold higher rate of mortality due to breast cancer-related causes, which was more significant than either p-CXCR4 expression or plasma SDF-1 levels alone, and remained significant after multivariate analysis. Although our immunohistochemical analysis deals with the primary tumor and not the distant metastatic site, the presence of activated CXCR4 receptor in this setting may imply a particular dependence of the tumor cell on its CXCR4 receptor, facilitating the selection of CXCR4 expressing cells during the metastatic process.
Finally, several therapeutic agents have been designed to target the SDF-1/CXCR4 ligand/receptor axis,26,27,28 one of which is presently being tested in a clinical trial.29 In preclinical models, such treatments have been shown to be effective not only in decreasing metastasis from breast cancer, but also in inhibiting primary tumor growth in breast cancer.5,6 As most breast cancers express at least moderate to high levels of CXCR4,10,11,30 there is a risk that these therapeutic agents may not be adequately targeted, perhaps impeding their clinical development. Since we found that most (77%) patients with triple negative disease express high levels of p-CXCR4, it is possible that these patients who do not benefit from hormonal or anti-HER2 therapy may potentially benefit from agents targeting CXCR4 activity. Most interestingly, we also identified a subset of ER+ patients with high p-CXCR4 expression/low plasma SDF-1 that demonstrated an eightfold higher risk of mortality, who may also potentially benefit from anti-CXCR4 therapy. In these patients with a good prognosis, the measurement of p-CXCR4 tumor expression and plasma SDF-1 may contribute most to provide novel prognostic and potentially predictive information. Although our findings will require follow-up and validation with independent clinical material, elevated p-CXCR4 expression together with low plasma SDF-1 levels may provide a new paradigm for breast cancer biomarkers, highlighting the interaction between corresponding host and tumor molecular factors.
Supplementary Material
Acknowledgments
We thank Martin Demers, Lucien Tremblay, and Eleanor Garofalo from the Archives of the Department of Pathology from the Centre Hospitalier de l’Université de Montréal and McGill University, respectively, for their technical support. We also thank Dr. Joshua Rubin and his laboratory for graciously providing us with the anti-p-CXCR4 antibody.
Footnotes
Address reprint requests to Mark Basik, M.D.C.M., F.R.C.S.(C), Department of Oncology, Lady Davis Institute, Sir Mortimer B. Davis Jewish General Hospital, McGill University, 3755 Cote Ste. Catherine, Montréal, Québec H3T1E2, E-mail: markbasik@gmail.com.
Supported by Canadian Breast Cancer Research Alliance grant 14598 (to M.B.) and Fonds de la recherche en santé du Québec Reseau de Recherche sur le Cancer for the tumor bank.
Supplemental material for this article can be found on http://ajp.amjpathol.org.
References
- Ries LAG, Melbert D, Krapcho M, Stinchcomb DG, Howlader N, Horner MJ, Mariotto A, Miller BA, Feuer EJ, Altekruse SF, Lewis DR, Clegg L, Eisner MP, Reichman M, Edwards BK, editors. Bethesda, MD: National Cancer Institute,; SEER Cancer Statistics Review, Surveillance Research Program, :pp 1975–2000. [Google Scholar]
- Weinberg RA. The biology of cancer. New York: Garland Science,; 2007:pp 589–595. [Google Scholar]
- Paget S. The distribution of secondary growths in cancer of the breast. Lancet. 1889;1:571–573. [PubMed] [Google Scholar]
- Muller A, Homey B, Soto H, Ge N, Catron D, Buchanan ME, McClanahan T, Murphy E, Yuan W, Wagner SN, Barrera JL, Mohar A, Verastegui E, Zlotnik A. Involvement of chemokine receptors in breast cancer metastasis. Nature. 2001;410:50–56. doi: 10.1038/35065016. [DOI] [PubMed] [Google Scholar]
- Smith MC, Luker KE, Garbow JR, Prior JL, Jackson E, Piwnica-Worms D, Luker GD. CXCR4 regulates growth of both primary and metastatic breast cancer. Cancer Res. 2004;64:8604–8612. doi: 10.1158/0008-5472.CAN-04-1844. [DOI] [PubMed] [Google Scholar]
- Liang Z, Wu T, Lou H, Yu X, Taichman RS, Lau SK, Nie S, Umbreit J, Shim H. Inhibition of breast cancer metastasis by selective synthetic polypeptide against CXCR4. Cancer Res. 2004;64:4302–4308. doi: 10.1158/0008-5472.CAN-03-3958. [DOI] [PubMed] [Google Scholar]
- Darash-Yahana M, Pikarsky E, Abramovitch R, Zeira E, Pal B, Karplus R, Beider K, Avniel S, Kasem S, Galun E, Peled A. Role of high expression levels of CXCR4 in tumor growth, vascularization, and metastasis. FASEB J. 2004;18:1240–1242. doi: 10.1096/fj.03-0935fje. [DOI] [PubMed] [Google Scholar]
- Phillips RJ, Burdick MD, Lutz M, Belperio JA, Keane MP, Strieter RM. The stromal derived factor-1/CXCL12-CXC chemokine receptor 4 biological axis in non-small cell lung cancer metastases. Am J Respir Crit Care Med. 2003;167:1676–1686. doi: 10.1164/rccm.200301-071OC. [DOI] [PubMed] [Google Scholar]
- Balkwill F. The significance of cancer cell expression of the chemokine receptor CXCR4. Semin Cancer Biol. 2004;14:171–179. doi: 10.1016/j.semcancer.2003.10.003. [DOI] [PubMed] [Google Scholar]
- Salvucci O, Bouchard A, Baccarelli A, Deschenes J, Sauter G, Simon R, Bianchi R, Basik M. The role of CXCR4 receptor expression in breast cancer: a large tissue microarray study. Breast Cancer Res Treat. 2006;97:275–283. doi: 10.1007/s10549-005-9121-8. [DOI] [PubMed] [Google Scholar]
- Blot E, Laberge-Le Couteulx S, Jamali H, Cornic M, Guillemet C, Duval C, Hellot MF, Pille JY, Picquenot JM, Veyret C. CXCR4 membrane expression in node-negative breast cancer. Breast J. 2008;14:268–274. doi: 10.1111/j.1524-4741.2008.00573.x. [DOI] [PubMed] [Google Scholar]
- Holm NT, Byrnes K, Li BD, Turnage RH, Abreo F, Mathis JM, Chu QD. Elevated levels of chemokine receptor CXCR4 in HER-2 negative breast cancer specimens predict recurrence. J Surg Res. 2007;141:53–59. doi: 10.1016/j.jss.2007.03.015. [DOI] [PubMed] [Google Scholar]
- Li YM, Pan Y, Wei Y, Cheng X, Zhou BP, Tan M, Zhou X, Xia W, Hortobagyi GN, Yu D, Hung MC. Upregulation of CXCR4 is essential for HER2-mediated tumor metastasis. Cancer Cell. 2004;6:459–469. doi: 10.1016/j.ccr.2004.09.027. [DOI] [PubMed] [Google Scholar]
- Hassan S, Baccarelli A, Salvucci O, Basik M. Plasma stromal cell-derived factor-1: host derived marker predictive of distant metastasis in breast cancer. Clin Cancer Res. 2008;14:446–454. doi: 10.1158/1078-0432.CCR-07-1189. [DOI] [PubMed] [Google Scholar]
- Sierro F, Biben C, Martinez-Munoz L, Mellado M, Ransohoff RM, Li M, Woehl B, Leung H, Groom J, Batten M, Harvey RP, Martinez AC, Mackay CR, Mackay F. Disrupted cardiac development but normal hematopoiesis in mice deficient in the second CXCL12/SDF-1 receptor, CXCR7. Proc Natl Acad Sci USA. 2007;104:14759–14764. doi: 10.1073/pnas.0702229104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Woerner BM, Warrington NM, Kung AL, Perry A, Rubin JB. Widespread CXCR4 activation in astrocytomas revealed by phospho-CXCR4-specific antibodies. Cancer Res. 2005;65:11392–11399. doi: 10.1158/0008-5472.CAN-05-0847. [DOI] [PubMed] [Google Scholar]
- Kononen J, Bubendorf L, Kallioniemi A, Barlund M, Schraml P, Leighton S, Torhorst J, Mihatsch MJ, Sauter G, Kallioniemi OP. Tissue microarrays for high-throughput molecular profiling of tumor specimens. Nat Med. 1998;4:844–847. doi: 10.1038/nm0798-844. [DOI] [PubMed] [Google Scholar]
- Tamamura H, Omagari A, Hiramatsu K, Gotoh K, Kanamoto T, Xu Y, Kodama E, Matsuoka M, Hattori T, Yamamoto N, Nakashima H, Otaka A, Fujii N. Development of specific CXCR4 inhibitors possessing high selectivity indexes as well as complete stability in serum based on an anti-HIV peptide T140. Bioorg Med Chem Lett. 2001;11:1897–1902. doi: 10.1016/s0960-894x(01)00323-7. [DOI] [PubMed] [Google Scholar]
- Camp RL, Dolled-Filhart M, Rimm DL. X-tile: a new bio-informatics tool for biomarker assessment and outcome-based cut-point optimization. Clin Cancer Res. 2004;10:7252–7259. doi: 10.1158/1078-0432.CCR-04-0713. [DOI] [PubMed] [Google Scholar]
- Cox DR. Regression models and Life-Tables. Journal of Royal Statistical Society Series B. 1972;34:187–220. [Google Scholar]
- Burns JM, Summers BC, Wang Y, Melikian A, Berahovich R, Miao Z, Penfold ME, Sunshine MJ, Littman DR, Kuo CJ, Wei K, McMaster BE, Wright K, Howard MC, Schall TJ. A novel chemokine receptor for SDF-1 and I-TAC involved in cell survival, cell adhesion, and tumor development. J Exp Med. 2006;203:2201–2213. doi: 10.1084/jem.20052144. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Miao Z, Luker KE, Summers BC, Berahovich R, Bhojani MS, Rehemtulla A, Kleer CG, Essner JJ, Nasevicius A, Luker GD, Howard MC, Schall TJ. CXCR7 (RDC1) promotes breast and lung tumor growth in vivo and is expressed on tumor-associated vasculature. Proc Natl Acad Sci USA. 2007;104:15735–15740. doi: 10.1073/pnas.0610444104. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Hartman M, Lindstrom L, Dickman PW, Adami HO, Hall P, Czene K. Is breast cancer prognosis inherited? Breast Cancer Res. 2007;9:1–8. doi: 10.1186/bcr1737. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Kim J, Takeuchi H, Lam ST, Turner RR, Wang HJ, Kuo C, Foshag L, Bilchik AJ, Hoon DS. Chemokine receptor CXCR4 expression in colorectal cancer patients increases the risk for recurrence and for poor survival. J Clin Oncol. 2005;23:2744–2753. doi: 10.1200/JCO.2005.07.078. [DOI] [PubMed] [Google Scholar]
- Oda Y, Yamamoto H, Tamiya S, Matsuda S, Tanaka K, Yokoyama R, Iwamoto Y, Tsuneyoshi M. CXCR4 and VEGF expression in the primary site and the metastatic site of human osteosarcoma: analysis within a group of patients, all of whom developed lung metastasis. Mod Pathol. 2006;19:738–745. doi: 10.1038/modpathol.3800587. [DOI] [PubMed] [Google Scholar]
- Zhan W, Liang Z, Zhu A, Kurtkaya S, Shim H, Snyder JP, Liotta DC. Discovery of small molecule CXCR4 antagonists. J Med Chem. 2007;50:5655–5664. doi: 10.1021/jm070679i. [DOI] [PubMed] [Google Scholar]
- Kim SY, Lee CH, Midura BV, Yeung C, Mendoza A, Hong SH, Ren L, Wong D, Korz W, Merzouk A, Salari H, Zhang H, Hwang ST, Khanna C, Helman LJ. Inhibition of the CXCR4/CXCL12 chemokine pathway reduces the development of murine pulmonary metastases. Clin Exp Metastasis. 2008;25:201–211. doi: 10.1007/s10585-007-9133-3. [DOI] [PMC free article] [PubMed] [Google Scholar]
- De Clercq E. The bicyclam AMD3100 story. Nat Rev Drug Discov. 2003;2:581–587. doi: 10.1038/nrd1134. [DOI] [PubMed] [Google Scholar]
- Wong D, Korz W. Translating an Antagonist of Chemokine Receptor CXCR4: from bench to bedside. Clin Cancer Res. 2008;14:7975–7980. doi: 10.1158/1078-0432.CCR-07-4846. [DOI] [PubMed] [Google Scholar]
- Cabioglu N, Sahin A, Doucet M, Yavuz E, Igci A, E OY, Aktas E, Bilgic S, Kiran B, Deniz G, Price JE. Chemokine receptor CXCR4 expression in breast cancer as a potential predictive marker of isolated tumor cells in bone marrow. Clin Exp Metastasis. 2005;22:39–46. doi: 10.1007/s10585-005-3222-y. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.