High Levels of Expression of Human Stromal Cell–Derived Factor-1 Are Associated with Worse Prognosis in Patients with Stage II Pancreatic Ductal Adenocarcinoma

Pancreatic cancer is the fourth leading cause of cancer death in the United States, preceded only by lung, colon, and breast cancers (1). Pancreatic ductal adenocarcinoma (PDA) has the highest mortality rates among all major malignancies and virtually all patients with PDA die from it (2). Even for early stages of resectable PDAs, the tumor commonly recurs, with the most common sites of recurrence being the liver, lung, peritoneal cavity, and pancreatic surgery bed. Studies have shown that multiple genes are frequently altered in PDAs, including activation of the K-ras oncogene by point mutation and inactivation of the p16, SMAD4/DPC4, and p53 tumor suppressor genes (3-7). Although our understanding of the genetic changes associated with pancreatic carcinogenesis and progression is growing, the roles of specific genetic alterations involved in the aggressive growth and metastasis of PDA is unclear.

Stromal cell–derived factor-1 (SDF-1), also called chemokine (C-X-C motif) ligand 12 (CXCL12), is a chemotactic factor for T cells, monocytes, hematopoietic progenitor cells, dendritic cells, endothelial cells, and tumor cells (8-11). The receptor for this chemokine is CXCR4, which was previously called fusin (12). In addition to its chemotactic functions, SDF-1 has also been shown to regulate tumor cell proliferation, motility, metastasis, and survival through its binding and subsequent activation of CXCR4 (13-15). SDF-1 regulates angiogenesis by recruiting endothelial progenitor cells from the bone morrow through a CXCR4-dependent mechanism (16) and carcinogenesis and the neovascularization linked to tumor progression (17, 18). Recent studies have shown that SDF-1/CXCR4 overexpression is associated with pelvic lymph node metastasis in cervical adenocarcinoma (19), metastasis of colonic adenocarcinoma (20), and lymph node metastasis in esophageal squamous cell carcinoma (21).

Both SDF-1 and its receptor, CXCR4, have been shown to be expressed in human pancreatic intraepithelial neoplasia (PanIN) and PDAs, and mouse PanIN lesions developed in PDX-Cre/LSL-Kras^G12D mice (15, 22, 23). The expression of CXCR4 increases with the progression from low-grade to high-grade PanIN lesions and has been detected in 71.2% of PDA samples by immunohistochemistry (22). Treatment of mouse PanIN cells with SDF-1 leads to a dose-dependent increase in cell proliferation, which is in part due to the activation of mitogen-activated protein kinase signaling (22). SDF-1 not only enhances the migration and invasion of CXCR4-positive metastatic pancreatic cancer cells, but also promotes their survival and proliferation (15). Inhibition of SDF-1/CXCR4 signaling inhibits the progression of PDAs (15, 24). In addition, patients with CXCR4-positive PDAs have worse survival than those patients with CXCR4-negative PDAs (24). These data suggest that the autocrine loop of SDF-1/CXCR4 signaling plays an important role in PDA. However, the prognostic significance of SDF-1 in human PDA has not been systematically studied. In this study, we examined the expression of SDF-1 in 72 pancreaticoduodenectomy specimens of PDA and correlated the expression level of SDF-1 with survival and other clinicopathologic features in patients with PDA.

Formalin-fixed, paraffin-embedded archival tissue samples were retrieved from 72 patients with stage II PDA who had undergone pancreaticoduodenectomy at the University of Texas M. D. Anderson Cancer Center between 1990 and 2005 and had not received any preoperative chemotherapy and/or radiation therapy. Patients who received preoperative chemotherapy and/or radiation, or who died from postoperative complications, were excluded. Our study group consisted of 24 female patients and 48 male patients with a median age at time of surgery of 63.7 years (range, 39.8-79.9 years). This study was approved by the Institutional Review Boards of M. D. Anderson Cancer Center and Penn State Hershey Medical Center.

To construct the tissue microarrays used in this study, formalin-fixed, paraffin-embedded archival tissue blocks and their matching H&E-stained slides were retrieved, reviewed, and screened for representative tumor regions and nonneoplastic pancreatic parenchyma by a gastrointestinal pathologist (H. W.). For each patient, two cores of tumor were sampled from representative areas using a 1.0-mm punch. The tissue microarrays were constructed with a tissue microarrayer (Beecher Instruments) as described previously (25).

Immunohistochemical staining for SDF-1 was done on 4-μm unstained recut sections from the tissue microarray blocks using a mouse anti-human SDF-1/CXCL12 monoclonal antibody (R&D Systems). To retrieve the antigenicity, the tissue sections were treated at 100°C in a steamer containing 1 mmol EDTA (pH 8.0) for 20 minutes. The sections were then immersed in methanol containing 0.3% hydrogen peroxidase for 10 minutes to block the endogenous peroxidase activity. Sections were incubated for 30 minutes at room temperature with primary anti-SDF-1 at a 1:300 dilution. DAKO Envision⁺ Polymer Detect was used according to the manufacturer's recommendations (DAKO). Diaminobenzidine was used as a chromogen, and hematoxylin was used for counterstaining. The staining results were evaluated independently by two pathologists (J.J.L. and S.Z.) to determine the average percentage of positive tumor cells in two cores. Using the 25th percentile value of the average percentage of positive tumor cells as a cutoff, we categorized SDF-1 expression as high expression of SDF-1 (SDF-1-H; cytoplasmic staining of ≥10% of the tumor cells) and low expression of SDF-1 (SDF-1-L; cytoplasmic staining of <10% of the tumor cells or no cytoplasmic staining).

Fisher's exact tests were used to compare categorical data. Overall and recurrence-free survival curves were constructed using the Kaplan-Meier method, and the log-rank test was used to evaluate the statistical significance of differences. The patients' follow-up information through September of 2008 was extracted from the prospectively maintained institutional pancreatic cancer database managed in the Department of Surgical Oncology at M. D. Anderson Cancer Center and, if necessary, updated by review of the U.S. Social Security Index. The recurrence information was updated every time a patient came to the clinic/institution for a follow-up visit. Recurrence-free survival was calculated as the time from the date of surgery to the date of first recurrence after surgery (in patients with recurrence) or to the date of last follow-up (in patients without recurrence). Overall survival was calculated as the time from the date of diagnosis to the date of death or the date of last follow-up (if death did not occur).

The prognostic significance of clinical and pathologic characteristics was determined using univariate Cox regression analysis. Cox proportional hazards models were fitted for multivariate analysis. After interactions between variables were examined, a backward stepwise procedure was used to derive the best-fitting model. Statistical analysis was done using Statistical Package for Social Sciences software (for Windows 12.0, SPSS Inc.). We used a two-sided significance level of 0.05 for all statistical analyses.

Twenty-five of 72 (35%) PDAs showed high levels of SDF-1 expression (Fig. 1A). SDF-1 showed cytoplasmic and membranous staining in PDA cells (Figs. 1A and 2) The acinar cells and islet cells were negative for SDF-1 expression. Clinicopathologic characteristics of the study population are summarized in Table 1. According to WHO classification standards, 10 cases (14%) were well differentiated, 44 (61%) were moderately differentiated, and 18 (25%) were poorly differentiated adenocarcinoma. Of the 72 cases, 57 (79%) had R0 resection (defined as all surgical margins negative microscopically), 15 (21%) patients had R1 resection, and no patients had a grossly positive margin of resection. Fifty-eight (81%) and 54 (75%) patients underwent postoperative chemotherapy and radiotherapy due to recurrence or metastasis, respectively. Recurrence/distant metastasis were present in 66% (31 of 47) of the patients with SDF-1-L tumors compared with 80% (20 of 25) of patients with PDAs that were SDF-1-H (P = 0.28). There were no significant differences in tumor size, the frequencies of lymph node metastasis, or other clinicopathologic features between the SDF-1-L and SDF-1-H groups (Table 1).

The patients whose tumors were SDF-1-H had a worse prognosis than those patients whose tumors were SDF-1-L. The median overall and recurrence-free survival for patients with SDF-1-H PDAs were 26.1 and 11.1 months, respectively, compared with 44.3 and 22.3 months for patients with SDF-1-L tumors (log-rank test, P = 0.047 and P = 0.021; Fig. 3). In univariate analysis, high SDF-1 expression was associated with poor overall and recurrence-free survival (P = 0.05 and P = 0.02, respectively; Table 2). In multivariate analysis, high SDF-1 expression was significantly correlated with poor overall and recurrence-free survival in patients with stage II PDAs (P = 0.02 and P = 0.02, respectively) independent of tumor size, differentiation, and lymph node status (Table 3).

In this study, we examined the SDF-1 expression in 72 stage II PDAs by immunohistochemistry and found that 35% (25 of 72) of PDAs expressed high levels of SDF-1. High level of SDF-1 expression was significantly associated with poor overall and recurrence-free survival and was a prognostic factor for patients with stage II PDAs independent of tumor size, differentiation, and lymph node metastasis.

There is growing evidence that SDF-1/CXCR4 signaling is important for tumor proliferation, survival, vascularization, and metastasis (17). In 2001, Muller et al. (26) showed that the SDF-1/CXCR4 pathway mediates human breast cancer metastasis. Several studies followed, providing evidence that blocking of CXCR4 signaling in vivo using a specific antibody (26), selective synthetic polypeptide (27), or small interfering RNA (28) resulted in significant inhibition of breast cancer metastasis to regional lymph nodes and the lung. Similar results have also been reported for colonic adenocarcinoma and non–small cell lung cancer (29). Inhibition of SDF-1/CXCR4 signaling significantly reduces the frequency of metastasis of colonic adenocarcinoma and non–small cell lung cancer (29). CXCR4 was reported to be expressed in 71.2% of human PDA samples (22). Patients with CXCR4-positive PDAs were found to have worse survival than patients with CXCR4-negative PDAs (24). More recently, CD133-positive pancreatic cancer stem cells were studied by Hermann et al. (30). These investigators identified a subpopulation of migrating CD133(+)/CXCR4(+) tumor cells at the invasive front of pancreatic cancer specimens. Using an in vivo xenograft mouse model and PDA cell lines, they further showed that the invasiveness and migration of CD133-positive cells was primarily mediated through the SDF-1/CXCR4 system (30). More recently, studies have shown that SDF-1 also binds to another receptor called CXCR7 (18, 31-33). CXCR7 expression has been reported in a variety of tumor cell lines and human cancer samples, including malignant tumors from breast, brain, lung, kidney, esophagus, and skin (31, 32, 34-36). High levels of CXCR7 expression correlate with metastasis and recurrence in non–small cell lung cancer (37). In this study, we showed that SDF-1 was overexpressed in 35% of stage II PDAs. Overexpression of SDF-1 was associated with poor prognosis in patients with stage II PDAs. Therefore, targeting SDF-1/CXCR4 or SDF-1/CXCR7 pathways may offer new potential therapeutic strategies for pancreatic cancers. However, we did not see significant correlation between SDF-1 expression and lymph node metastasis in pancreaticoduodenectomy specimens or postoperative recurrence/metastasis in our study group.

Our results provide further evidence of the relevance of SDF-1 expression to the behavior of PDA, and suggest that this pathway may represent a potential target for therapeutic intervention, as well as prediction of prognosis. Additional, larger studies are needed to define the importance of SDF-1 as a prognostic marker in PDA, and to ascertain the mechanisms of its actions in this aggressive malignancy.

No potential conflicts of interest were disclosed.