|Home | About | Journals | Submit | Contact Us | Français|
Presented in part at the 44th Annual Meeting of the American Society of Cilnical Onocology, Chicago, IL, May 30-June 3 (Abstract ID: 4559)
Background: Angiogenesis has been attributed to be a well-recognized aspect of human cancer biology. As such, proteinase-activated receptor (PAR)-1, endostatin (ES) and interleukin-8 (IL-8) mediate the regulation of early-onset angiogenesis and in turn impact the process of tumor-growth and disease progression.
Patients and methods: Formalin-fixed paraffin-embedded tissues were obtained from 137 patients with localized gastric cancer at University of Southern California and Memorial Sloan-Kettering Cancer Center medical facilities. DNA was extracted and genotyping was carried out using PCR–restriction fragment length polymorphism-based protocols.
Results: In false discovery rate-adjusted univariate analysis, PAR-1 −506 ins/del (P < 0.001), ES +4349 G>A (P = 0.004), and IL-8 −251 T>A (P < 0.0001) were associated with time to tumor recurrence (TTR). Further, PAR-1 −506 ins/del and IL-8 −251 were associated with overall survival (OS). After adjusting for covariates, IL-8 remained significantly associated with TTR (adjusted P = 0.003) and OS (adjusted P = 0.049), whereas ES was significantly associated with TTR (adjusted P = 0.026).
Conclusions: Polymorphisms in PAR-1, ES, and IL-8 may serve as independent molecular prognostic markers in patients with localized gastric adenocarcinoma. The assessment of the patients’ individual risk on the basis of interindividual genotypes may therefore help to identify patient subgroups at high risk for poor clinical outcome.
Gastric adenocarcinoma is the fourth most common type of cancer and the second leading cause of cancer-related death worldwide . Despite recent improvements in the detection , surgical resection , and adjuvant (radio)-chemotherapy , the mortality of patients diagnosed with localized gastric cancer (GC) still remains high, with 5-year overall survival (OS) rates of 30% . Traditionally, surgical resection has offered the best hope for prolonged survival; however, the extent of surgery continues to be a subject of ongoing debate . Even though combined radiochemotherapy following complete gastric resection has been demonstrated to significantly improve OS and relapse-free survival , tumor recurrence after curative resection continues to be a significant problem in the management of patients with localized gastric adenocarcinoma . Therefore, the development of molecular markers of prognosis as an adjunct to traditional staging systems will not only be helpful in identifying patients who are more likely to recur, but they will also be important to select patient-specific treatment strategies with the means of tailoring a targeted and effective therapy to the molecular profile of both the patient and his or her tumor while minimizing and avoiding life-threatening toxic effects.
Neovascularization and sustained tumor angiogenesis have been studied in GC and the so-called ‘angiogenic switch’, the induction of tumor vasculature are considered hallmarks of the malignant process and are required for tumor propagation and disease progression . Although vascular endothelial growth factor (VEGF) and its downstream pathways have received considerable interest lately, pathways that regulate the switch to an angiogenic phenotype are not fully understood. Besides hypoxia-driven up-regulation of VEGF expression , a significant amount of evidence is accumulating supporting other mechanisms of neovascularization in vitro and in vivo [7, 8]. As such, tumor angiogenesis is not only dependent on tumor cells and VEGF signaling but also on platelet–endothelium interaction and interleukin-8 (IL-8) signaling. Platelets, in addition to their function in hemostasis, play an important role in wound healing and tumor growth, at least in part through the release of pro- and antiangiogenic factors; thus, platelets may modulate tumor angiogenesis by releasing not only promoters such as VEGF and epidermal growth factor (EGF) but also endostatin (ES), a potent endogenous antiangiogenic growth factor [9, 10]. In addition, recent evidence indicates that pro- and antiangiogenic proteins are segregated into different sets of α-granules within platelets. Interestingly, proteinase-activated receptors (PARs) have been shown to differentially counter-regulate the local release of VEGF and its antiangiogenic counterpart ES (ES) [8, 9]. Therefore, it has been proposed that these PARs play a pivotal role in the regulation of local and early-onset angiogenesis and in turn may influence the process of tumor growth and disease progression . In addition, IL-8 has been reported to play a major role in VEGF-independent tumor angiogenesis. As such, overexpression of IL-8 is frequently observed in GC  and induction of IL-8 preserved the angiogenic phenotype in hypoxia inducible factor 1-α (HIF1-α) deficient colon cancer cells, suggesting a critical role of IL-8 in tumor-associated angiogenesis, independent of VEGF . Based on these data, we hypothesized that functional VEGF, VEGFR2, EGF, EGFR, PAR-1, ES, IL-8 and CXRC2 polymorphisms could be associated with differences in clinical outcome in patients with localized gastric adenocarcinoma.
One hundred and thirty-seven (n = 137) patients with localized (stage Ib–IV) gastric adenocarcinoma who were treated with surgery alone or surgery and adjuvant (radio)-chemotherapy, at the University of Southern California/Norris Comprehensive Cancer Center (USC/NCCC), the Los Angeles County/University of Southern California Medical Center, or the Memorial Sloan-Kettering Cancer Center/Cornell University from 1992 to 2008, were eligible for the present study. This study was conducted at USC/NCCC and approved by the Institutional Review Boards of the University of Southern California and Cornell University for Medical Sciences. Patient data were collected through chart review. All patients involved in the study signed informed consent for the analysis of molecular correlates. Detailed clinicopathological characteristics are shown in Table 1.
The polymorphisms we tested were selected by a VEGF and EGFR pathway approach with the goal of selecting genes known to modulate tumor-associated angiogenesis (Table 2). We used the following criteria to select genes for study that (i) the gene be part of a pathway for which there is a credible scientific basis to support its involvement in tumor angiogenesis; (ii) the gene has an established, well-documented genetic polymorphism; (iii) the frequency of the polymorphism is high enough that its impact on clinical outcome will be meaningful; and/or (iv) the polymorphism has some degree of likelihood to alter the function of the gene in a biologically relevant manner.
Formalin-fixed paraffin-embedded normal tissues were collected and genomic DNA was extracted using the QIAamp extraction kit (Qiagen, Valencia, CA) according to the manufacturer's protocol. The majority of the samples were tested using a PCR-based restriction fragment length polymorphism or a 5′-end 33p γATP-labeled PCR technique, as previously described [13, 16]. Primers, restriction enzymes, and annealing temperatures are listed in Table 3. For quality assurance purposes, a total of 10% positive and negative duplicate controls were matched for each polymorphism and were analyzed by direct DNA sequencing where applicable. Genotype concordance was ≥99%.
The primary end point in this study was time to tumor recurrence (TTR) in localized gastric adenocarcinoma patients, which was defined as the period from the date of diagnosis of GC to the date of first recurrence or until last contact if the patient was free of any tumor recurrence at the time of last contact. If a patient had not recurred, then TTR was censored at the time of the last follow-up or death. The secondary end point was OS that was defined as the period from diagnosis to death from any cause or the last contact if the patient was alive.
The association between each polymorphism and time to recurrence and OS was examined using Kaplan–Meier curves and log-rank test. For polymorphisms with a genotype frequency (homozygous minor allele) of <10%, the associations between genotypes and clinical outcome were analyzed in a dominant model. Otherwise, they were analyzed in a codominant or additive model. The distributions of polymorphisms across demographic characteristics were examined using Fisher's exact test. In the univariate survival analysis, the Pike estimate of relative risk (RR) and its associated 95% confidence interval (CI) was based on the log-rank test.
The Benjamini and Hochberg method was used to control the false discovery rate (FDR) of multiple testing . In the univariate analysis, an FDR-adjusted P value of <0.05 was employed to select polymorphisms as candidates for inclusion in the multivariable models. With 137 patients, we would have 80% power to detect a minimum hazard ratio ~2.0 across a range of common allele frequencies (0.2–0.5) for time to recurrence in a dominant model. For a recessive model, a minimum hazard ratio is <3.8 when the allele frequency is 0.2 and approaches 2 when the allele frequency is 0.5.
The Cox proportional hazards regression model with stratification factors (ethnicity, T category, N category, and type of adjuvant chemotherapy) was fitted to re-evaluate the association between polymorphisms and TTR and OS considering the imbalances in the distributions of baseline characteristics. The level of significance was set to P <0.05 and P values are given for two-sided testing. Analyses were carried out using the SAS statistical package version 9.1 (SAS Institute Inc., Cary, NC).
A total of 137 patients with localized adenocarcinoma of the stomach (stage Ib–IV) were included in this analysis. The median follow-up was 3.3 years. Sixty-one of 137 patients (45%) had tumor recurrence, with a probability of 3-year recurrence of 0.52 ± 0.05. The median TTR was 2.8 years (95% CI 2.1–7.0). Fifty-five of 137 (40%) patients showed recurrent disease within the first 3 years after surgery. Forty-five of 137 patients (33%) have died and the median OS for the cohort is 4.7 years (95% CI 3.8–7.3). T category (P = 0.013), N category (P = 0.004), and type of chemotherapy (P = 0.003) were significantly associated with TTR, whereas ethnicity of the patient (P = 0.04) and type of chemotherapy (P = 0.004) were additionally associated with OS. Detailed clinicopathological characteristics are shown in Table 1. Polymorphisms of PAR-1, ES, and IL-8 were not associated with demographic (age, gender, and ethnicity), clinical [type of (radio)-chemotherapy], or pathological characteristics (tumor grade, T category, and N category). There were no differences in allelic frequencies of PAR-1, ES, and IL-8 across ethnic groups (data not shown).
Genotyping for PAR-1 −506 ins/del was successful in 136 (99%) of 137 cases. In the other one patient (1%), genotyping was not successful because of limited quantity of extracted genomic DNA. Patients with the PAR-1 −506 ins/ins or ins/del genotype had a median TTR of 1.5 years (95% CI 1.1–2.3 years), compared with 7.0 years (95% CI 2.3–14.5+ years) in patients homozygous (del/del) for the −506-deletion allele (P < 0.001, log-rank test) (Table 4). For PAR-1 −506 ins/del, the FDR-adjusted P values did meet the criteria for variable selection as a candidate predictor in the multivariable TTR (FDR-adjusted P value = 0.002) and OS model (FDR-adjusted P value = 0.027) (Figures 1A and and2A,2A, Table 4).
Genotyping for ES +4349 G>A was successful in 137 (100%) of 137 cases. Patients with the ES +4349 A/A or G/A genotype had a median TTR of 1.2 years (95% CI 0.8–3.2 years), compared with 3.7 years (95% CI 2.1–14.5+ years) in patients homozygous for the G allele (P = 0.004, log-rank test) (Table 4). For ES +4349 G>A, the FDR-adjusted P value did meet the criteria for variable selection as a candidate predictor in the multivariable TTR model (Figure 1B, Table 4) (FDR-adjusted P value = 0.015).
Genotyping for IL-8 −251 T>A was successful in 137 (100%) of 137 cases. Patients with the IL-8 −251 A/A genotype had a median TTR of 1.1 years (95% CI 0.8–1.7 years), compared with 2.2 years (95% CI 1.5–3.2 years) and 14.5+ years (95% CI: reference) for those heterozygous (T/A) and homozygous (T/T) for the 251 T allele, respectively (P < 0.0001, log-rank test) (Table 4). For IL-8 −251 T>A, the FDR-adjusted P values did meet the criteria for variable selection as a candidate predictor in the multivariable TTR (FDR-adjusted P value < 0.0001) and OS (FDR-adjusted P value < 0.0001) model (Figures 1C and and2B,2B, Table 4).
When we analyzed PAR-1 −506 ins/del (adjusted P value = 0.57), ES +4349 G>A (adjusted P value = 0.026), and IL-8 −251 T>A (adjusted P value = 0.003) jointly, stratified by ethnicity, T category, N category, and type of adjuvant chemotherapy, polymorphisms in IL-8 and ES remained significantly associated with TTR. In addition, IL-8 −251 T>A remained significantly associated with OS (adjusted P value = 0.049) (Table 5). We observed consistent patterns in the associations between PAR-1 −506, ES, and IL-8 −251 polymorphisms and clinical outcome within subsets of patients by race, T category, N category, and type of chemotherapy (data not shown). Patients with 0 unfavorable alleles (PAR-1 −506 insertion allele, ES +4349 A allele, IL-8 A allele) were at lowest risk to develop tumor recurrence (95% CI: 1 reference), compared with patients displaying 1 (RR 9.74, 95% CI 1.22–77.6) and ≥2 (RR 13.7, 95% CI 1.68–111.0) unfavorable alleles, who were more likely to develop tumor recurrence (adjusted P > 0.0001) (Table 5, Figure 1D).
We did not observe statistically significant associations between other tested genes involved in tumor angiogenesis pathway and TTR and OS (Table 4).
Tumor angiogenesis is driven locally by the release of pro- and antiangiogenic factors, such as VEGF and ES. It is well established that platelets stimulate endothelial cells in culture and can promote the assembly of capillary-like structures in vitro [10, 22]. Accordingly, the proximity to and interaction with the endothelium allows platelets to profoundly influence tumor development . As such, platelet–endothelium interaction modulates the angiogenic balance of the tumor and it has been suggested that platelets may contribute to these angiogenesis-dependent processes at least in part through the local release of pro- and antiangiogenic factors [10, 24]. A recent study by Italiano et al.  demonstrated that two distinct populations of α-granules containing endogenous angiogenic regulatory proteins are present in platelets and indicate that this subcellular organization is facilitating the differential release of these proteins in response to platelet–endothelial interaction.
Activation of platelets by thrombin is mediated through the cleavage of PARs, a receptor family of G-coupled proteins with seven transmembrane proteins . Four distinct PARs have been identified, with PAR-1, PAR-3, and PAR-4 acting as receptors for thrombin. Interestingly, tumors may modulate PAR activity and its own angiogenic properties through the release of proteases that can either activate or disarm these receptors. In fact, a recent study by Ma et al.  revealed that PAR-1 and PAR-4 act in a counter-regulatory manner to influence the local release of VEGF and ES from platelets and as such may profoundly influence tumor neovascularization. Activation of the thrombin/PAR-1 receptor axis triggers multiple signaling pathways that result in endothelial cell survival, apoptosis, and angiogenesis [26, 27]. In addition, overexpression of PAR-1 mRNA and protein was found to correlate with invasive properties of colorectal cancer, breast cancer and, GC [7, 28]. DNA sequence variations within the PAR-1 gene may modulate PAR-1 production and/or activity. Among those, a 13-bp insertion/deletion polymorphism (PAR-1 −506 ins/del; rs11267092) has been identified, repeating the preceding 5′-CGGCCGCGGGAAG-3′ sequence at position −506 within the promoter region of the PAR-1 gene . Interestingly, a potential recognition site (5′-GCGGGAAGC-3′) for activating Ets protein (erythroblastosis virus protein, family of transcription factors) has been identified within the polymorphic 13-bp sequence at position −506 . Ets proteins play a major role in the formation of the transcriptional initiation complex in genes lacking the canonical TATA box sequence, such as the PAR-1 gene promoter . Because the PAR-1 −506 ins/del polymorphism duplicates a putative binding site for Ets-1/Ets-2 and is located within the active part of the promoter, it may regulate the transcription of the PAR-1 gene . Higher serum levels of ES, a potent endogenous inhibitor of angiogenesis, induce the regression of tumor growth and angiogenesis in vivo . Furthermore, Down syndrome patients, who have higher serum levels of ES due to three instead of two gene copies, have a decreased incidence of solid tumors, including gastric adenocarcinoma . ES +4349 G>A (rs12483377), a polymorphism located within exon 42, has been associated (A/A genotype) with increased prostate cancer risk and impaired function of ES . In our study, ‘high-expression’ variants of PAR-1 −506 ins/del (ins/ins) and ‘low-expression’ variants of ES +4349 G>A were found to be significantly associated with TTR in FDR-adjusted univariate analysis (Table 4, Figure 1A and B). In addition, ES +4349 G/A showed to be an independent prognostic marker for TTR after adjustment for covariates in the multivariable model. Our findings demonstrate for the first time that PAR-1 −506 ins/del and ES +4349 G>A may be important prognostic factors for GC progression.
VEGF-driven tumor angiogenesis has been recognized to play a fundamental role in GC pathogenesis. However, progression into advanced and recurrent gastric adenocarcinoma has also been associated with activation of VEGF-independent cascades, such as the IL-8-signaling pathway . As such, IL-8 participates in the pathogenesis and maintenance of several human malignancies, including GC . IL-8 exerts its potent angiogenic properties on endothelial cells through interaction with its receptors chemokine receptor (CXCR) 1 and CXCR2 [32, 33] and induction of IL-8 preserved the angiogenic phenotype in HIF1-α-deficient colon cancer cells, suggesting a critical role of IL-8 in tumor-associated angiogenesis, independently of VEGF . Interestingly, GC cells derived from surgical specimens of human GCs overexpress IL-8 compared with corresponding normal mucosa, and the level of IL-8 mRNA expression has been shown to directly correlate with the vascularity of the tumor. Furthermore, Kitadai et al.  demonstrated that transfection of GC cells with the IL-8 gene enhanced their tumorigenic and angiogenic potential in the gastric wall of nude mice . Polymorphisms in the IL-8 gene were studied for associations with inflammatory diseases and cancers . Among these, the IL-8 −251 T>A polymorphism (rs4073), located −251 nucleotides relative to the transcriptional start site of the gene, affects its expression. In fact, the variant A allele of rs4073 has been associated with increased IL-8 plasma levels  and further functional studies including promoter assays confirmed the functional significance of the IL-8 −251 T>A polymorphism . Recently, the IL-8 −251 A allele was reported to be associated with an increased risk of developing GC [35, 36]. As previously demonstrated by our group, IL-8 −251 T>A and its receptor CXCR-1 were associated with TTR and progression-free survival in colorectal [13, 37] and ovarian cancer . In our study, high-expression variants of IL-8 −251 T>A (A/A) were found to be significantly associated with TTR and OS in both FDR-adjusted univariate and multivariable analysis (Tables 4 and and5,5, Figures 1C and and2B),2B), supporting our hypothesis that increased angiogenic potential is critical for GC tumor relapse.
In summary, our study did not show significant associations of VEGF- and EGF-signaling genes and clinical outcome. This may be explained by a low ‘disease-specific’ significance of these genetic variants in this particular malignancy and/or clinical setting. Nevertheless, consistent with our hypothesis, high-expression variants of PAR-1 −506 ins/del (ins/ins) and IL-8 −251 T>A (A/A), as well as low-expression variants of ES +4349 G>A (A/A) were associated with decreased TTR in FDR-adjusted univariate analysis. After adjusting for covariates in the multivariable models, IL-8 −251 T>A and ES +4349 G>A were independently associated with TTR. In addition, IL-8 −251 T>A was independently associated with OS, suggesting that the assessment of the patients’ angiogenic potential on the basis of PAR-1, ES, and IL-8 genotypes might further enhance targeted treatment not only by the identification of patients at high risk but also by identifying novel targets for antiangiogenic treatment strategies. Interestingly, targeted agents against PAR-1 and IL-8 as well as recombinant ES are in different stages of preclinical and clinical development. Larger and prospective biomarker-embedded clinical trials are needed to confirm our preliminary findings.
National Institutes of Health (5 P30CA14089-27I); Dhont Family Foundation.
This work was carried out in the Sharon A. Carpenter Laboratory at USC/NCCC.
H-JL declares following conflicts of interest: Consultant of Genentech Inc. and Response Genetics Inc. Other authors have no conflicts of interest to declare.