MDM2 Promoter Polymorphism and Pancreatic Cancer Risk and Prognosis

Pancreatic cancer has an extremely poor prognosis and is the fourth leading cause of cancer death in the United States (1–5). Currently, complete surgical resection is an essential part of curative therapy. Most tumors, however, are unresectable and are treated primarily with chemotherapy and/or radiation (1, 3, 5). Recent research has shown that single nucleotide polymorphisms of genes with roles in the cell cycle play an important role in carcinogenesis (6–12) and that common polymorphisms may lead to altered susceptibility to pancreatic cancer and also affect clinical outcome (8). Mouse double minute 2 homologue (MDM2) has been identified as being amplified and overexpressed in human tumors (13, 14). It is an E3 ubiquitin ligase (15–17) that blocks the transcriptional activation of p53 (7). In the absence of stress, as occurs in normal cells, MDM2 functions as a negative regulator by binding to p53 resulting in low levels of the p53 protein. This negative feedback mechanism can be altered in several human malignancies with p53 somatic mutations (18). Dong et al. (7) have shown the overexpression of MDM2 in pancreatic tumors, which can inhibit p53-mediated transcriptional activation (7, 19).

A common MDM2 promoter polymorphism is the T→G transformation at nucleotide 309. This MDM2 309T/G promoter polymorphism has been associated with the development of a variety of tumors (6, 7, 9, 18, 20–22). Its significance in clinical outcome is not well outlined, but recent associations with prognos is have been found in lung and gastric cancer (9, 21, 23). The G allele is associated with increased affinity for Sp1 binding and higher MDM2 mRNA and protein levels, leading to diminished tumor suppressor activity of the p53 pathway (6, 17, 24–26). We examined the main effect of this genetic variant on the risk and clinical outcome of pancreatic adenocarcinoma, hypothesizing that the G allele is associated with increased risk and poorer outcome.

Study population. This study was reviewed and approved by the Institutional Review Board of the Massachusetts General Hospital and the Harvard School of Public Health. Patients were recruited prospectively at the Massachusetts General Hospital between January 2004 and December 2005. Eligible cases were patients with histologically confirmed pancreatic adenocarcinoma. Controls were recruited among healthy friends and non-blood-related family members (usually spouses) of other cancer patients. Potential controls that carried a previous diagnosis of any cancer (other than nonmelanoma skin cancer) were excluded from participation. Cases were frequency matched to controls (1:3) on age and gender. Interviewer-administered questionnaires obtained information on demographics and other significant clinical information at the time of recruitment. Over 85% of eligible cases and controls agreed to participate in the study and also provided blood sample at the time of recruitment for genotyping.

Genotyping. Peripheral blood samples were obtained from all study participants at the time of enrollment into the study. DNA was extracted from peripheral blood samples using the Puregene DNA Isolation Kit (Gentra Systems). The MDM2 polymorphism was genotyped by the 5′-nuclease assay (TaqMan) using the ABI Prism 7900HT Sequence Detection System (Applied Biosystems). Genotyping was done by laboratory personnel blinded to case and outcome status, and a random 5% of the samples was repeated to validate genotyping procedures. Blinded genotyping results were independently reviewed by two authors. Hardy-Weinberg equilibrium was tested using the χ² test in the controls and separately in the cases.

Clinical outcomes. For survival analyses, two end points were used in this study. Overall survival (OS) was the primary outcome and was calculated from the date of diagnosis to the date of last follow-up or death from any cause. Progression-free survival (PFS) was the secondary outcome and was defined as the time from the date of diagnosis to the first date of recurrence of cancer or death from any cause. We used the Response Evaluation Criteria in Solid Tumors (27). Data were collected from at least one of the following sources: (a) Massachusetts General Hospital inpatient and outpatient records, including doctors' notes, radiographic reports, and biopsy results; (b) Massachusetts General Hospital tumor registry; (c) Social Security Death Index; (d) primary physician's office; and (e) patient or family contact. Permission to contact patients and their families to obtain follow-up information was included in our original consent form.

Statistical analysis. Demographic and clinical information were compared across genotype using Pearson χ² tests (for categorical variables) and t tests (for continuous variables), where appropriate. Unconditional logistic regression was used to assess the association between MDM2 and pancreatic cancer risk, adjusting for age (continuous variable), gender, indicator variables for smoking status (nonsmoker, ex-smoker, current smoker), a continuous variable for cumulative smoking exposure (pack-years), and an indicator variable for alcohol intake (yes, no). Where indicated, the odds ratios (OR) and 95% confidence intervals (95% CI) for the risk of pancreatic cancer were calculated from these models. The associations between polymorphism status and survival outcomes in cases were estimated using the method of Kaplan-Meier and assessed using the log-rank test. We restricted this analysis to patients with complete follow-up information and good performance status (Eastern Cooperative Oncology Group 0-2; ref. 28). Cox proportional hazard models were used to adjust for potential confounders. We created indicator variables for performance status (1 if performance status = 2; 0 if performance status = 0, 1), stage (1 if stage = III, IV; 0 if stage = IB, IIA, IIB), and treatment (1 if treatment = surgery plus chemoradiation; 0 if treatment = surgery, chemotherapy, or chemoradiation). Any covariate associated with OS in the univariate analysis with P < 0.10 was included in the final multivariable model. All statistical testing was done at the two-sided 0.05 level, and SAS software version 9.1 (SAS Institute) was used.

Patient characteristics. There were a total of 123 cases and 372 controls. The distribution of demographic and clinical characteristics is summarized in Table 1. We restricted our clinical outcomes analysis to 109 patients with complete follow-up information and Eastern Cooperative Oncology Group performance status 0 to 2. Distribution of characteristics for cases with clinical outcomes data was similar to the entire set of cases used in the case control analysis. The MDM2 -309T/G polymorphism was in Hardy-Weinberg equilibrium when tested in the controls (P = 0.48) and cases (P = 0.11). MDM2 genotype frequency distributions were significantly different between cases and controls (P = 0.01). There were 53 deaths and 84 individuals progressed during follow-up. Median follow-up time for OS was 17 months. Median OS was 24 months and median PFS was 7 months.

Table 2 summarizes the results of the multiple logistic regression models on all cases and controls. The adjusted ORs (95% CI) of pancreatic cancer associated with the MDM2 T/G and G/G genotypes compared with T/T were 1.89 (1.20-2.99) and 2.07 (1.03-4.16), respectively (after adjusting for age, gender, smoking status, and pack-years of smoking). We also adjusted for alcohol drinking in the subgroup of patients with complete alcohol information, and similar results were observed [MDM2 T/G and G/G genotypes compared with TT were 1.86 (95% CI, 1.13-3.04) and 1.56 (95% CI, 0.74-3.27)]. When the T/G and G/G genotypes were combined, the G allele of the MDM2 -309T/G polymorphism was associated with almost a 2-fold increased risk of pancreatic adenocarcinoma risk [MDM2 -/G genotypes compared with TT, adjusted OR (95% CI), 1.93 (1.25-2.99)].

In crude analyses, the G allele was associated with a trend for decreased OS and PFS (log-rank test P = 0.07 and 0.11, respectively; Table 3; Figs. 1 and 2). In Cox proportional hazards models with the wild-type T/T genotype as the reference category and adjusting for stage, treatment, and performance status, both the heterozygous T/G and the homozygous G/G genotype were associated with decreased PFS [adjusted hazard ratio (HR; 95% CI), 1.67 (0.98-2.84) for T/G and 2.28 (1.11-4.71) for G/G] and OS [2.64 (1.23-5.67) for T/G and 3.12 (1.22-7.94) for G/G; Table 4]. When the T/G and G/G genotypes were combined, the presence of the G allele resulted in 2- to 3-fold increase in the risk of cancer progression and a corresponding decrease in survival: PFS [adjusted HR (95% CI), 1.77 (1.06-2.96)] and OS [2.75 (1.31-5.75)].

To the best of our knowledge, we are the first to report a significant association between the MDM2 -309T/G promoter polymorphism and both risk and prognosis of pancreatic adenocarcinoma. In the case-control analysis, we found that the G allele of the MDM2 polymorphism was associated with a 2-fold risk of pancreatic cancer. We also found that this allele was associated with 2- to 3-fold decrease in OS (and PFS) in our case series analysis after adjusting for performance status, stage, and treatment. Identification of this significant risk and prognostic factor may be useful in predicting individuals at higher risk of developing pancreatic adenocarcinoma as well as having poorer outcomes once the diagnosis has been made.

Our results are consistent with a growing body of evidence supporting a deleterious role of the G allele of this polymorphism on both the risk and the outcome. The G allele is associated with increased affinity for Sp1 binding and higher MDM RNA and protein levels, leading to diminished tumor suppressor activity of the p53 pathway (17, 24–26). Due to this and the significant role of p53 tumor suppression in the development and progression of cancer, we have postulated that the homozygous variant G/G would be associated with the highest risk and worst survival outcome in pancreatic cancer. Predisposition to cancers has also been shown in mice made to overproduce MDM2 (29). Dong et al. (7) reported that mdm2 gene amplification has been discovered in 36% of all types of sarcomas and 10% of well-differentiated glioma as well as esophageal cancer, neuroblastoma, and anaplastic astrocytoma.

The MDM2 polymorphism has been studied extensively with regards to the risk and survival outcome of several other cancers. The G allele was found to be associated with increased susceptibility to lung, ovarian, and gastric carcinomas and with increased risk of tumor formation in patients with Li-Fraumeni syndrome and sporadic soft-tissue sarcoma (9, 21, 22, 30). However, results were null for breast cancer (18). The same allele leads to poorer survival in gastric and cancers (9, 23). In a study of 410 cases and 438 controls, Ohmiya et al. (9) showed that the risk of sporadic gastric carcinoma was significantly increased when the MDM2 homozygous variant G/G was compared with the combination of T/T and T/G [OR (95% CI), 1.45 (1.02-2.05)]. The homozygote allele also correlated with extragastric tumors (including liposarcoma, breast carcinoma, and adrenal and brain tumors as observed in Li-Fraumeni syndrome) complicated with gastric carcinoma. In outcome analysis among stages IB to IV patients, those with G/G genotype had poor OS [3.16 (95% CI, 1.22-8.20), adjusted for stage, age, sex, extragastric tumor, and concomitant chemotherapy]. In our recent study of 383 early-stage lung cancer patients, Heist et al. (23) found that the G/G genotype (compared with T/T) was associated with worse OS [1.57 (95% CI, 1.03-2.40)] after adjusting for age, sex, stage, histology, and pack-years smoking.

The present study found that MDM2 309T/G independently predicts both risk and outcome in pancreatic cancer. The study, however, is limited by lacking some covariate information, such as alcohol exposure. Nonetheless, subgroup analysis adjusted for alcohol produced similar results. We also lacked comprehensive pancreatitis and diabetes information to adjust in our models, although each of these risk factors is unlikely to explain the strong associations observed. Our analysis was also limited to a single polymorphism, although one is an excellent candidate given published data. Further studies of MDM2 and other polymorphisms in this cell cycle pathway are warranted due to the complex nature of the p53 pathway and the fact that multiple proteins and enzymes work in tandem to maintain cell integrity. It is plausible that other polymorphisms play a significant role in predicting risk and outcome for pancreatic cancer. Fewer than half of all patients had tissue blocks, with the majority having received fine-needle aspirations. Therefore, we were not able to correlate our polymorphism data with gene or protein expression of p53 or MDM2 or p53 mutations. Dong et al. (7) measured p53 and MDM2 protein expression among 59 patients with invasive ductal carcinoma of the pancreas and obtained positive p53 protein expression in 40 of 59 patients (67.8%) and positive MDM2 protein expression in 17 of 59 patients (28.8%). Median survival time was lowest for patients positive for both MDM2 and p53 (group comparison done by log-rank test P < 0.05). However, correlations in other studies have resulted in conflicting results (23), further raising the question of whether tissue immunohistochemistry is the most appropriate means of assessing p53 pathway dysregulation in our patient samples.

In conclusion, we identified that the risk and survival outcome of pancreatic cancer is associated with the MDM2 SNP309 polymorphism. The results agree with the previously reported oncogenic function of the variant MDM2 309T/G promoter polymorphism that leads to overexpression of Mdm2 protein and the inactivation of p53 suppressor function and suggest a role of this pathway in both pancreatic etiology and prognosis.

No potential conflicts of interest were disclosed.