MCP-1 Promoter Polymorphism at −2518 Is Associated with Metastasis of Nasopharyngeal Carcinoma after Treatment

Nasopharyngeal carcinoma (NPC) is a unique tumor with a remarkably geographic and ethnic distribution. NPC is more prominent in southeastern Asia, including southern China, Hong Kong, and Taiwan, where the annual incidence rate is ∼25-fold higher than that in Western world (1). NPC is closely associated with EBV infection (2), but EBV infection is ubiquitous in human, and the incidence of NPC remains high among Chinese people who have migrated to North America. Moreover, the survival of NPC patients varied among different ethnic groups. Together, these reports suggested an underlying genetic component to NPC risk, progression, and outcomes (3–5). Consistent with this notion, host genetic factors are likely to influence cancer susceptibility and outcomes (6). In the context of NPC, HLA haplotype (7, 8) and polymorphisms in DNA repair enzymes such as XRCC1 and hOGG1 (9) have been associated with NPC susceptibility, but no genetic predictor has been reported.

Monocyte chemoattractant protein-1 [MCP-1 or tumor-derived chemotactic factor (TDCF)], belongs to CC chemokine family, a member of the small inducible gene, and is encoded by the CCL2 gene, which was mapped to chromosome 17q11. Induction of MCP-1 by inflammatory cytokine interleukin-1 (IL-1)can be modulated through a functional single nucleotide polymorphism (SNP−2518G/A) in MCP-1 distal regulatory region (10), and cells with AG or GG genotype produce ∼2-fold more MCP1 compared with cells with AA genotype. MCP-1 is known as a potent chemoattractant for monocytes and T lymphocytes (11); it is thought to be involved in several diseases characterized by intense macrophage infiltration, including atherosclerosis (12) and cancers. The functional SNP−2518 has been reported to be associated with increased susceptibility or severity of diseases such as asthma (13), atherosclerosis (14), and rheumatoid arthritis (15). Previous studies have shown that MCP-1 is overexpressed in many tumors such as glioma (16), ovarian (17), breast (18), and esophagus cancers (19). Furthermore, expression of MCP-1 in tumor cells correlates with macrophage infiltration and tumor vascularity (19, 20). Notably, the patient's serum with high MCP-1 level is correlated with better prognosis (21–23), but the molecular mechanism remains to be investigated. To date, no studies have been carried out to examine the role of this SNP in tumor progression.

In NPC, the overexpression of MCP-1 has been detected in the infiltrated macrophages, leading to intensive leukocyte infiltration (24). However, it is unclear whether (a) MCP-1 is overexpressed in NPC tumor cells; (b) MCP-1 expression level can be modulated by the functional SNP; and (c) MCP-1 SNP−2518 genotype is associated with NPC patients' prognosis.

In this study, we hypothesized that MCP-1 SNP−2518 may be associated with the clinical outcome of NPC patients under current therapeutic protocols. We retrospectively analyzed MCP-1 SNP−2518 genotypes and their association with clinical outcomes in 411 NPC patients receiving complete radiotherapy. This is the first report suggesting a host genetic polymorphism that may serve as a prognostic predictor for NPC.

Patients, clinical staging protocol, oncological treatment, and clinical outcome assessment. This retrospective cohort comprises 411 NPC patients who had been admitted to Chang Gung Memorial Hospital (CGMH), Lin-Kou, from March 1994 to November 2004. The tumor-node-metastasis (TNM) stage was defined according to the 2002 cancer staging system revised by the American Joint Committee on Cancer (AJCC; ref. 25), and histologic typing was done according to the WHO classification criteria (26). This study was reviewed and approved by the institutional review board and ethics committee of CGMH. Informed consent was obtained from all patients.

All enrolled patients had been treated with definitive radiotherapy (cumulative dose of external beam radiotherapy ≧64.8 Gy). Among them, 109 patients received additional chemotherapy in the Department of Radiation Oncology at CGMH. Patients who were diagnosed with distant metastatic disease at presentation (M₁ stage) and/or who had undergone previous treatment at another institute were excluded from the present study. For all enrolled patients, pathology records were retrieved from pathologic databases and medical records and reviewed for confirmation of the NPC diagnosis. Information on stage, treatment, follow-up, and limited information on family history were collected from hospital tumor registries and medical files.

Patients were followed-up at 2- to 3-month intervals during the first 3 years after therapy and at 6-month intervals thereafter. The minimal follow-up period was 28 months. The primary end point was disease-specific survival (DSS), which was calculated from the date of diagnosis to the date of death or the last follow-up. Progression-free survival (PFS), distant metastasis-free survival (DMFS), and local recurrence-free survival (LRFS) were also assessed. The time to local recurrence or distant metastasis was calculated using the date on which local recurrence or distant metastasis status was detected as the end point. Patients who died without occurrence of local recurrence or distant metastasis were censored in the analyses of LRFS and DMFS.

DNA extraction and genotyping. Blood samples were collected at the time of enrollment, and genomic DNA was obtained using a DNA isolation kit (Qiagen). The target MCP-1 SNP−2518 (CCL2 gene −2518G/A, rs1024611) was detected using a commercially available TaqMan genotyping assay kit and a GeneAmp PCR System 9700 (both from Applied Biosystems Inc.) according to the manufacturer's instructions. Two template-free controls and six DNA samples of known genotype were included in each plate as negative and positive controls, respectively. The Sequence Detection Software provided by ABI was used for genotyping analysis.

Immunohistochemical staining analysis. Immunohistochemical analyses were done using an automatic immunohistochemistry (IHC) staining device according to the manufacturer's procedures (Bond, Vision Biosystems). Anti–MCP-1 antibody (R&D Systems) and anti-CD68 antibody (DAKO, Dakopatts) were used. The detailed procedure of immunohistochemical staining was described in Supplementary Data. The MCP-1 intensity was scored as negative or 0; mild or 1; moderate or 2; or intense or 3 at 100× magnification according on the scoring method described in Supplementary Data. The amount of CD68⁺ macrophages infiltrated was measured using ImageScope (Aperio Technologies) software at 200× magnification. A total of 3 to 10 fields per slide were selected, counted, and averaged.

In situ hybridization.In situ hybridization (ISH) for detecting EBV-encoded RNA transcripts (EBER) was done using the EBV Probe ISH kit (Novocastra) according to the manufacturer's instructions.

MCP-1 reporter construct and luciferase activity assay. The distal regulatory region, which contains either G or A at the MCP-1 SNP−2518 was amplified from NPC patient genomic DNA by PCR. IL-1β–induced promoter activity was examined in 293T cells as described in Supplementary Data.

Statistical analysis. Patients were separated into subgroups based on their MCP-1 SNP−2518 (CCL2 gene −2518G/A, rs1024611) genotypes (AA, AG, or GG). Box plots were used to show the MCP-1 IHC score in subjects with different genotype and number of macrophages infiltrated in NPC biopsies with different MCP-1 expression levels. The boundary of the box closet to the zero indicates the 25th percentile; and the boundary farthest from zero indicates the 75th percentile; a line in the box marks the median. Whiskers above and below the box indicate the 90th and 10th percentiles. ANOVA was used to evaluate the MCP-1 IHC score by genotype and amount of infiltrated macrophages by MCP-1 expression levels. The age at diagnosis, distribution of gender, overall stage, N stage, T stage, and use of chemotherapy were compared between patient subgroups using the χ² test. Kaplan-Meier plots of DSS, PFS, DMFS, and LRFS were established, and statistical significance was measured by the log-rank test. The Cox proportional regression model was used to evaluate the effect of MCP-1 SNP−2518 genotype and other potential prognostic factors on survival tests. All statistical tests were two sided, and a P value of ≦0.05 was considered statistically significant. All statistical analyses were done using the SPSS software version 13.0 (SPSS Inc.).

In vivo correlation of mCP-1 expression in tumor cells with genotypes. To test whether MCP-1 expression can be detected in NPC tumors, 37 NPC biopsies were examined by IHC. The distribution of NPC tumor cells was assayed by the presence of EBV noncoding transcripts (EBERs) as previously described (27). Among 37 NPC biopsies, all of them had EBV infection, and 30 samples (81%) had MCP-1 overexpression. MCP-1 was observed in both tumor and stromal cells but mainly in the cytoplasm of tumor cells (Fig. 1A). However, little or comparatively low expression of MCP-1 was detected in the adjacent cells. To correlate the MCP-1 expression level with the MCP-1 SNP−2518 genotype, genomic DNA collected from the same patients was genotyped by TaqMan genotyping kit. The expression levels of MCP-1 in biopsies was scored and compared among patients with three genotypes. As shown in Fig. 1A and B and Supplementary Table S1, the biopsies from the GG-genotype patients had the highest MCP-1 expression level, whereas AA genotype had the lowest (P < 0.001). Together, these results suggested that MCP-1 is overexpressed by NPC tumor cells, and the expression level is associated with the MCP-1 SNP−2518 genotype.

Previous reports suggested that the extent of macrophage accumulation in tumors is correlated with the MCP-1 expression level (19, 28). Therefore, we counted the number of infiltrated macrophage in NPC biopsies by IHC using the anti-CD68 antibody. As shown in Fig. 1A and C, the increased amount of infiltrated macrophages is correlated with the high MCP-1 expression levels detected in tumor cells (P = 0.012).

In vitro MCP-1 SNP−2518 A or G allele promoter assay. To further confirm if MCP-1 SNP−2518 genotype influences its expression in response to a given stimulus, we did an in vitro promoter assay in the presence of IL-1β, a potent stimulator of MCP-1 protein expression (29) and can be detected in NPC biopsies (30). A luciferase reporter gene was fused with MCP-1 promoter region (−2906 to +215, with A or G allele, respectively) isolated from NPC patients in this cohort. Compared with the A-allele–containing promoter construct (4.9-fold activation), G-allele construct had higher promoter activity (7.2-fold activation) under IL-1β stimulation (Supplementary Fig. S1; P = 0.012). Thus, results indicated that the G allele isolated from NPC biopsy showed higher enhancer activity than A allele after treated with an inflammatory cytokine.

Patient characteristics and genotypic frequencies. To evaluate the association of MCP-1 SNP−2518 with clinical outcome under current therapeutic protocols, a retrospective cohort of 411 NPC patients were subjected to clinical outcome assessment study. The patient characteristics and clinical features are summarized in Table 1. The median age at diagnosis was 47.6 years (range, 13-87), with a male-to-female ratio of 3:1. The frequencies of the MCP-1 SNP−2518 AA, AG, and GG genotypes among the studied NPC patients were 20% (82/411), 47% (194/411), and 33% (135/411), respectively. The clinicopathologic features were comparable among patient subgroups classified according to the MCP-1 −2518 genotype.

MCP-1 genotype and NPC DSS. The 5-year DSS rate across all studied patients was 75.3%. As shown by Kaplan-Meier plot, the 5-year DSS rates for patient subgroups defined by MCP-1 SNP−2518 genotype AA, AG, and GG were 71.3%, 74.5% and 78.7%, respectively (Fig. 2A). When DSS was compared among MCP-1 SNP−2518 genotypes using a log-rank test, no significant difference was observed between different subgroups of patients. We conducted a multivariate analysis of the genotype effect on DSS using the Cox proportional regression models adjusted by other prognostic factors, such as tumor stage, node stage, sex, use of chemotherapy, and age at onset. Our results indicated that DSS was significantly dependent on age at onset, primary tumor stage, and node stage, which is similar to those previously reported (31), whereas MCP-1 SNP−2518 was not an independent predictor for DSS in NPC patients (Supplementary Table S2).

MCP-1 genotype and PFS. All analyses were repeated using the date of disease progression as the end point. Kaplan-Meier plots showed that no significant difference was observed between patients of different MCP-1 SNP−2518 genotype (Fig. 2B). The Cox proportional regression model indicated that only tumor stage and node stage were independent predictors for PFS time, and MCP-1 SNP−2518 was not an independent predictor for PFS in NPC patients (Supplementary Table S2).

MCP-1 genotype and LRFS. Among the 411 NPC patients, 93 patients (22.6%) developed local recurrence after the completion of the radiotherapy. Genotypic frequencies of MCP-1 SNP−2518 AA, AG, and GG carriers were 14% (13/93), 47.3% (44/93), and 38.7% (36/93), respectively. All analyses were repeated using the date of local recurrence diagnosed as the end point. The survival curves revealed no significant difference in LRFS among different subgroups (Fig. 2C). However, in a Cox proportional regression model (Table 2) that adjusted the effect of MCP-1 SNP−2518 by other potential prognostic factors, patients of AA genotype showed a lower risk for developing local recurrence when compared with AG- and GG-genotype patients (hazard ratio, 0.53; P = 0.05).

MCP-1 genotype and DMFS. A total of 86 out of 411 (20.9%) NPC patients developed distant metastasis after the initial radiotherapy. Among the 86 patients, 24.4% (21/86), 55.8% (48/86), and 19.8% (17/86) belonged to MCP-1 SNP−2518 AA, AG, and GG genotypes, respectively. All analyses were repeated using the date of distant metastasis diagnosed as the end point. As shown in Fig. 2D, the 5-year DMFS rate of AA, AG, and GG genotype patients was 74.1%, 73.1%, and 86.4%, respectively. The presence of one or two copies of the A allele at the MCP-1 SNP−2518 strongly predicted inferior DMFS (AA versus GG, P = 0.011; and AG versus GG, P = 0.005) in NPC patients.

When a Cox proportional regression model was used to assess the prognostic significance of MCP-1 SNP−2518 for an NPC patient to develop distant metastasis, the results (Table 3) revealed that, in addition to tumor stage and node stage, NPC patients with the AA and AG genotype had a higher risk for developing metastasis compared with patients with GG genotype (AA versus GG: hazard ratio, 2.21; P = 0.017; AG versus GG: hazard ratio, 2.23; P = 0.005). These results suggest that in addition to the tumor stage and node stage, the MCP-1 SNP−2518 A allele was an independent predictor for higher risk of developing distant metastasis in NPC patients after treatment.

Our data indicate that a single nucleotide polymorphism, −2518G/A, in the regulatory region of MCP-1 gene, is an independent genetic risk predictor for developing distant metastasis after complete treatment in NPC patients. It is based on the correlation studies of a retrospective cohort of 411 NPC patients, collected over 10 years, with MCP-1 frequencies and disease outcomes. Patients with AA or AG genotype that have relatively lower MCP-1 expression are more prone to distant metastasis in the first 3 years after treatment than those with the GG genotype. Previous studies and our data confirm that the MCP-1 expression level is indeed associated with genotype. The MCP-1 expression level is likely to be regulated through the binding of transcription factors. A nuclear factor complex containing IRF-1 has shown greater affinity to the A allele than the G allele (32). Although the effect of IRF-1 binding is unclear, AG or GG genotype patients showed slightly higher MCP-1 level. Consistent with the above results, our in vitro MCP-1 promoter-reporter assay showed that the construct with the G allele has ∼1.5-fold higher activity than its counterpart A allele under IL-1β stimulation. Hence, genetic background may be involved in the predetermination of the expression level of MCP-1.

No significant difference was found among three genotypes in DSS and PFS. However, our statistical analysis suggested that among these NPC patients that had recurrence, patients with AA genotype tend to develop distant metastasis, whereas patients with GG genotype tend to develop local recurrence. Therefore, MCP-1 SNP−2518 genotype cannot foretell the risk of NPC relapse, but it may predict, if relapse does occur, which genotype may associate with a particular cancer phenotype. Because the present work is focused on the study of a candidate gene, future search of new and more SNPs related to NPC risk by whole genome association study may be needed. In addition, this study is limited by its retrospective nature and the patients collected in a single institute; validation studies by others are awaited.

A recent report suggested that the tumor microenvironment plays a critical role in the control of local tumor recurrence and distant metastasis (33). Antitumor immune response can act as an environmental stress, which may select the specific cancer cell clones with metastatic potential (34). MCP-1 is a key player in innate and adaptive immunity, which mediates macrophage recruitment to inflammation sites and modulates T helper cell polarization (35). As shown in Fig. 1C, the number of CD68⁺ macrophages infiltrated in the NPC tumor mass is correlated with the MCP-1 expression level, which can be modulated by the MCP-1 genotype. As a result, the NPC tumor microenvironment may be altered due to the different MCP-1 levels, which is governed by genetic background. However, how MCP-1 may lead to different cancer phenotypes such as distant metastasis and local recurrence is still obscured.

Host genetic factors are likely to influence cancer susceptibility and outcome. The role of germ line polymorphisms in modulating the outcome or metastatic potential of tumor cells has been proposed (36). For example, in human breast cancer, the −313 G/A polymorphism located in the regulatory region of the SIPA1 gene has been strongly associated with axillary node involvement at the time of diagnosis (37). Here, we report that MCP-1 SNP−2518 genotype seems to be a novel candidate for NPC distant metastasis, although the detailed mechanism is still unclear. Because overexpression of MCP-1 has been reported in a variety of cancers, it is worth testing if the MCP-1 SNP−2518 genotypes can predict the metastasis potential of patients with other cancers.

Despite its radiosensitive nature, NPC is characterized by its high metastatic potential; 20% to 25% of the patients develop distant metastasis after initial treatment in our hospital. Current clinical prognostic indicators for NPC patients include TNM stage (31), histopathologic classification (38), age of onset (39), cumulative dose of radiotherapy (40), and pretherapy circulating EBV DNA load (41–43). Elevated plasma EBV DNA in combination with the Unio Internationale Contra Cancrum staging data improves risk discrimination in early-stage disease (44). The practical value of the genetic predictor, such as MCP-1 SNP−2518, should be validated by other cohorts and also judged in the light of plasma EBV DNA level.

MCP-1 induction through NF-κB activation has been reported in HHV8-infected endothelial cells (45) and in EBV-infected monocytes (46). EBV-encoded latent membrane protein 1 (LMP1), which expressed in NPC tumor cells, can activate NF-κB signaling (47). It would be interesting to investigate if EBV infection or LMP1 expression in NPC tumor cells can play a role in MCP-1 induction.

With the increasing emphasis on personalized medicine and the life quality of cancer patients, the assistance of genetic metastasis predictor may provide a simple and reproducible assay for clinical outcome prediction, which can facilitate an appropriate allocation of adjuvant therapy.