Pancreaticobiliary Cancers With Deficient Methylenetetrahydrofolate Reductase Genotypes

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY 2005;3:752–760

Pancreaticobiliary Cancers With Deficient Methylenetetrahydrofolate Reductase Genotypes HIROYUKI MATSUBAYASHI,*,‡ HALCYON G. SKINNER,*,§ CHRISTINE IACOBUZIO– DONAHUE,* TADAYOSHI ABE,* NORIHIRO SATO,* TAYLOR SOHN RIALL,㛳 CHARLES J. YEO,㛳 SCOTT E. KERN,¶ and MICHAEL GOGGINS*,¶,# Departments of *Pathology, ¶Oncology, 㛳Surgery, #Medicine, Johns Hopkins Medical Institutions, Baltimore, Maryland; ‡The Fourth Department of Internal Medicine, Tokyo Medical University, Tokyo, Japan; and §Department of Preventive Medicine, Feinberg School of Medicine, Northwestern University, Chicago, Illinois

Background & Aims: Methyl group deficiency might promote carcinogenesis by inducing DNA breaks and DNA hypomethylation. We hypothesized that deficient methylenetetrahydrofolate reductase (MTHFR) genotypes could promote pancreatic cancer development. Methods: First, we performed a case-control study of germline MTHFR polymorphisms (C677T, A1298C) in 303 patients with pancreatic cancer and 305 matched control subjects. Pancreatic neoplasms frequently lose an MTHFR allele during tumorigenesis; we hypothesized that such loss could promote carcinogenesis. We therefore evaluated the cancer MTHFR genotypes of 82 patients with pancreaticobiliary cancers and correlated them to genome-wide measures of chromosomal deletion by using 386 microsatellite markers. Finally, MTHFR genotypes were correlated with global DNA methylation in 68 cancer cell lines. Results: Germline MTHFR polymorphisms were not associated with an increased likelihood of having pancreatic cancer. Fractional allelic loss (a measure of chromosomal loss) trended higher in cancers with 677T genotypes than in cancers with other genotypes (P ⴝ .055). Among cancers with loss of an MTHFR allele, cancers with 677T MTHFR alleles had more deletions at folate-sensitive fragile sites (36.9%) and at tumor suppressor gene loci (68.5%) than 677C cancers (28.7% and 47.8%, P ⴝ .079 and .014, respectively). LINE1 methylation was lower in cancers with less functional 677T/TT genotypes (24.4%) than in those with 677CT (26.0%) and CC/C genotypes (32.5%) (P ⴝ .014). Conclusion: Cancers with defective MTHFR genotypes have more DNA hypomethylation and more chromosomal losses. Deficient MTHFR function due to loss of an MTHFR allele by an evolving neoplasm might, by promoting chromosomal losses, accelerate cancer development.

ancreatic ductal adenocarcinoma remains one of the deadliest cancers. The molecular pathogenesis of pancreatic cancer is increasingly well characterized. Many genetic and epigenetic alterations arise during

P

pancreatic tumorigenesis including oncogene activation by mutation (K-ras,1 BRAF), amplification (c-myc, AKT2), and inactivation of suppressor genes (p16,3 p53,4 SMAD4, BRCA2,5 STK11, hCDC4, MKK4, and Fanconi anemia genes6).2 In addition, many genes undergo silencing in pancreatic neoplasms by methylation.2,7–9 Most pancreatic carcinomas harbor chromosomal instability2,10,11; only a few harbor microsatellite instability.12 Pancreatic adenocarcinomas exhibit high levels of allelic loss, a feature that has been independently associated with poor histologic differentiation11 and poor survival.13 Smoking, aging, obesity, and diabetes mellitus are well-known risk factors of pancreatic cancer,14 and nutritional factors such as methyl group availability might also influence pancreatic carcinogenesis.14 –17 Folate deficiency lowers the concentration of S-adenosylmethionine, reducing global DNA methylation as well as synthesis of thymidine from uracil. Uracil misincorporation in place of thymidine leads to an imbalanced nucleotide pool and increased occurrence of DNA strand breaks,18 which increases genomic instability19 and is thought to contribute to cancer development. Epidemiologic studies have implicated low folate status with the development of other cancers, but this relationship is less studied in the pancreas.14,15,20,21 Methylenetetrahydrofolate reductase (MTHFR) is a key enzyme in folate metabolism that converts 5,10-methylenetetrahydrofolate to 5-methyltetrahydrofolate.22 The latter form of folate is used for the re-methylation of homoAbbreviations used in this paper: CHLC, Cooperative Human Linkage Center; CI, confidence interval; COBRA, combined bisulfite restriction analysis; FAL, fractional allelic loss; LOH, loss of heterozygosity; MTHFR, methylenetetrahydrofolate reductase; OR, odds ratio; PCR, polymerase chain reaction; RFLP, restriction fragment length polymorphism. © 2005 by the American Gastroenterological Association 1542-3565/05/$30.00 PII: 10.1053/S1542-3565(05)00359-9

August 2005

cysteine to methionine, catalyzed by vitamin B12-dependent methionine synthase. If not reduced to 5-methyltetrahydrofolate by MTHFR, 5,10-methylenetetrahydrofolate can transfer its methylene group to deoxyuridine monophosphate in the synthesis of deoxythymidine monophosphate, or it might contribute to purine synthesis. The MTHFR gene is polymorphic, with single nucleotide variants at nucleotide 677 (C¡T) and nucleotide 1298 (A¡C). The 677T polymorphism encodes a thermolabile enzyme with reduced activity23,24 (more reduced than 1298C), decreasing enzyme activity by ⬃50%.25 Individuals with 677TT alleles have higher serum homocysteine concentrations than 1298CC allele carriers and have homocysteine levels more sensitive to serum vitamin B12 and folate.26 Thus, individuals with 677TT genotypes retain ⬃25%–33%23,27 of MTHFR function relative to individuals with 677CC genotypes. Reduced MTHFR function leads to a shift in the distribution of forms of folate at the expense of 5-methyltetrahydrofolate,28 elevated plasma homocysteine,22,23,29 decreased serum folate,29 and global DNA methylation.30 Thus, for a given folate level, the MTHFR 677T allele reduces methyl group availability but increases thymidine synthesis. Approximately 50%– 60% of the US population are 677CC, ⬃30%– 40% are 677CT, ⬃10% are 677TT,24,27,29,31–33 and 6%–⬃12% of the population have the 1298C polymorphism.27,31,33 Several studies have reported that carriers of germline MTHFR 677TT genotypes have altered cancer risks. Some studies have found an increase in cancer risk among 677TT carriers; others have found a decreased risk or no effect. For example, 677TT carriers have an increased risk of cervical neoplasia but a reduced risk of colorectal cancer20 and acute lymphocytic leukemia.31,34 The apparently contradicting increased or decreased risk of different cancers with MTHFR genotypes is thought to be related to the relative importance of ensuring sufficient methyl group availability versus optimal thymidine production in that cell type and also on dietary methyl group availability. The cancer risk associated with MTHFR genotype is influenced by folate and vitamin B12 status,29,35 as well as age,22 smoking,29 and alcohol consumption, with smoking and alcohol able to directly adversely affect folate metabolism probably by multiple mechanisms.22,35 We carried out a case-control analysis to determine whether defective MTHFR genotypes are more frequent in individuals who develop pancreatic cancer than in non-cancer control subjects. In addition, we hypothesized that pancreatic neoplasms with loss of heterozygosity at MTHFR, and particularly those retaining only a 677T allele, would exhibit less methylation of DNA and

MTHFR ALTERATION IN PANCREATIC CARCINOMA

753

greater chromosomal losses than cancers that had retained both MTHFR alleles. We also investigated the relationship between the MTHFR status and genomewide measurements of chromosomal losses in pancreaticobiliary cancers and determined if there was any relationship between MTHFR genotype and the level of global DNA methylation in a series of cancer cell lines.

Materials and Methods Case-Control Study Cases were 333 patients who underwent pancreaticoduodenectomy for treatment of primary pancreatic adenocarcinoma at Johns Hopkins Hospital from 1993–1999. Control subjects (identified by using the Johns Hopkins Hospital Surgical-Pathology Database, n ⫽ 333) had undergone cholecystectomy for benign gallbladder disease at Johns Hopkins Hospital through 1999. Control subjects were frequencymatched to cases by gender, self-reported ethnicity, and 5-year age group. Diabetes status was recorded for all control subjects and 270 cases and smoking status from 302 control subjects and 268 cases. Cases and control subjects with a history of cancer (except non-melanoma skin cancer) were excluded. This study was approved by the Johns Hopkins Committee for Clinical Investigation. The presurgical anesthesiologists’ assessments in the medical record provided information on diabetes, cigarette smoking, and cancer history. The duration of diabetes before pancreatic cancer diagnosis was grouped as 2 or more years, ⬍2 years, and uncertain date of onset. Cigarette smoking quantity (packs per day), duration (years), and time since quitting were recorded. Those who reported quitting within 12 months of their surgery were considered current smokers. We alternatively categorized cigarette smoking history into 2 categories, “ever smoker” and “never smoker.” Frozen or paraffin-embedded specimens of tumor-free duodenum provided DNA for case genotypes. Control genotypes were ascertained by using DNA from paraffin-embedded gallbladder samples.

Pancreaticobiliary Cancers Analyzed for Their Genetic Losses and MTHFR Status Cancer xenograft DNA36 and corresponding germline DNA (from normal duodenum) were analyzed for MTHFR genotypes from 82 patients with pancreatic or biliary cancer. These patients had their cancer surgically resected with curative intent between 1992 and 1997 as described elsewhere.37 The proportion of the cancer genome with deletions (the % of markers deleted/markers analyzed, also known as the fractional allelic loss [FAL]) was determined from each cancer by using 386 microsatellite markers (modified CHLC version 9 marker set, average spacing of 10 centimorgans) as previously described in collaboration with the Center for Inherited Disease Research.38 Seventy-four of the 82 cancers were of pancreatic origin, and 8 were of biliary origin. Sixty-one of these 74 patients were included in the population described above

754

MATSUBAYASHI ET AL

analyzed for germline MTHFR status. The other 13 patients were not excluded, but patients with cancer and control subjects were demographically matched, and so not all patients with pancreatic cancer were included in the germline MTHFR study. Clinicopathologic information on these 82 patients including smoking, alcohol history, prognosis, and cancer pathology is described in detail elsewhere.36 Seventy-eight of these 82 patients were white, and 4 were of African American ethnicity. Their average age at diagnosis was 63.6 ⫾ 11.4 years (mean ⫾ standard deviation); 43 of 82 patients (53.8%) were ever smokers. Only 6 cancers were well differentiated, 50 were moderately differentiated, and 26 were poorly differentiated.

MTHFR 677 and MTHFR 1298 Polymorphism Analysis DNA was extracted from tissues by using standard methods. The MTHFR 677 and 1298 polymorphisms were analyzed by using a combination of methods. For the casecontrol study, real-time polymerase chain reaction (PCR) was performed by using allele-specific probes (available at http:// pathology2.jhu.edu/pancreas/MTHFR). To confirm real-time genotype information, DNAs from ⬃1/2 of the subjects in the case-control study were randomly selected for confirmation by using cycle sequencing and/or by PCR–restriction fragment length polymorphism (RFLP).24,27 PCR-RFLP analysis was performed by using previously reported primers24,27 and with HinfI for MTHFR 67724 and MboII for MTHFR 1298.27 The 82 pancreaticobiliary cancers were genotyped by using both PCR-RFLP analysis and cycle sequencing with standard methods.

Analysis of Loss of Heterozygosity at MTHFR Allelic loss at the MTHFR locus was determined by using 5 microsatellite markers, 7.4 mega basepairs (Mbp) telomeric to 5.6 Mbp centromeric to MTHFR. Loss of heterozygosity (LOH) at the MTHFR locus could be assigned in 81 of 82 cancers. Presumptive LOH was called at MTHFR in cancer cell lines if only one allele was seen for 3 or more consecutive microsatellite markers overlapping the MTHFR locus, because without LOH, heterozygosity would likely be in at least one highly polymorphic microsatellite marker.

Analysis of Genetic Deletions at Fragile Sites and Tumor Suppressor Loci Because folate deficiency is thought to predispose deletions at fragile sites, we analyzed the correlation between MTHFR genotype and chromosomal losses at folate-sensitive fragile sites and at 12 tumor suppressor loci inactivated in pancreatic cancers.39 The microsatellite markers analyzed for chromosomal deletions in pancreaticobiliary cancers are available at http://pathology2.jhu.edu/pancreas/MTHFR.

Measurement of Global DNA Methylation Combined bisulfite restriction analysis (COBRA)40 was used to analyze the level of methylation of LINE1 repet-

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 8

itive elements in 105 samples of DNA from normal duodenum, 45 pancreaticobiliary cancer xenografts, and 68 cancer cell lines by using previously reported primers and thermal cycle conditions.40 Each DNA sample was assayed twice, and results were averaged. In our hands, the intra-assay variation of the COBRA LINE1 assay was ⱕ3%.

Statistical Analysis For the case-control analysis, we computed odds ratios (ORs) from contingency tables to measure the association between MTHFR genotype and the presence or absence of pancreatic cancer. Demographic and clinicopathologic variables were jointly classified by MTHFR genotype and case/ control status, and differences in their distributions were compared by ␹2 test or Mann–Whitney U test. Differences in FAL, fragile site loss, and LINE1 methylation levels by MTHFR genotype were analyzed by the Kruskal–Wallis test and multivariate linear regression adjusting for potential confounders. Confounding was assessed by stratifying contingency tables by potential confounding factors and by evaluating changes in logistic regression coefficients when including/excluding potential confounders. We defined evidence for confounding as a 10% or greater change in the estimate of the log OR. We assessed statistical interaction on the multiplicative scale by likelihood ratio tests in the multivariate analysis.

Results Germline MTHFR Genotypes in Patients With Pancreatic Cancer and Control Subjects There was no significant difference in the prevalence of MTHFR 677T and in 1298C genotypes in patients with pancreatic cancer compared with control subjects. MTHFR genotypes did vary with ethnicity. African Americans had a lower prevalence of both MTHFR 677T and 1298C polymorphisms than did whites (P ⫽ .004 in total) (Table 1), and this was true for both control subjects (P ⫽ .004) and patients (P ⫽ .013) (data not shown). This lower prevalence of MTHFR polymorphisms among African Americans has been previously reported.33 The distribution of MTHFR genotypes in our control subjects was similar to that previously reported in case-control studies.24,31,34 A history of smoking and of diabetes mellitus was associated with increased odds of having pancreatic cancer (for smoking: OR, 1.83; 95% confidence interval [CI], 1.28 –2.58, P ⫽ .0004 and for diabetes: OR, 2.02; 95% CI, 1.36 –3.01, P ⫽ .0005). Patients with pancreatic cancer smoked more cigarettes and for longer duration than control subjects (period: 17.5 ⫾ 20.5 years vs 6.3 ⫾ 14.3 years; quantity: .6 ⫾ .8 pack per day vs .3 ⫾ .7 pack per day) (P ⬍ .0001 for period and P ⫽ .0002 for quantity). Diabetes was also of longer duration in pa-

August 2005

MTHFR ALTERATION IN PANCREATIC CARCINOMA

755

Table 1. Odds of Pancreatic Cancer and of Ethnicity by Germline MTHFR Genotype No. of polymorphic nucleotides of MTHFR

677

1298

Control

Case

4 3 3 2 2 2 1 1 0

TT TT CT TT CT CC CT CC CC

CC AC CC AA AC CC AA AC AA

0 5 0 31 64 21 71 71 42

0 3 4 40 54 32 57 77 36

aP

MTHFR genotype

No. of patients

Odds ratio (confidence interval)

0.60 (0.14–2.53) 1.34 (0.82–2.11) 0.82 (0.55–1.22) 1.60 (0.90–2.84) 0.76 (0.51–1.13) 1.12 (0.78–1.63) 0.84 (0.52–1.36)

No. of African Americans (%)a 0 (-) 1 (12.5) 0 (0) 2 (2.8) 4 (3.4) 0 (0) 0 (0) 9 (6.1) 12 (15.4)

⫽ .004 by Mann–Whitney U test.

tients with pancreatic cancer (1.1 ⫾ 4.8 years) than in control subjects (.4 ⫾ 2.6 years) (P ⫽ .035). Current smokers showed an increased likelihood of developing pancreatic cancer compared with former smokers (OR: 1.58; 95% CI, 1.00 –2.50; P ⫽ .049). However, there was no significant interaction or confounding between these factors and MTHFR genotype in association with pancreatic cancer (Table 2). Genetic Deletions in Pancreatic Cancers by Their MTHFR Genotypes Because for many diseases any adverse effect of defective germline MTHFR genotypes is greater in individuals with low dietary folate intake,25,35 many individuals with defective MTHFR genotypes might not manifest disease. This might be particularly true for a disease that evolves as a result of multiple genetic and epigenetic alterations such as pancreatic cancer. If so, an adverse effect of defective MTHFR on pancreatic cancer development might be more easily detected by examining pancreatic cancers for consequences of defective MTHFR rather than for the occurrence of pancreatic cancer. We also hypothesized that acquired genetic defects in folate metabolism in an evolving pancreatic neoplasm (such as deletion of an MTHFR allele) could

influence the risk of developing pancreatic cancer or the behavior of the cancer by increasing the occurrence of chromosomal losses as the tumor evolves into a pancreatic cancer. We therefore determined the percentage of genetic deletions in the genomes of 82 pancreaticobiliary cancers, stratifying them by their MTHFR genotype, which included identifying whether the cancer had lost an MTHFR allele by LOH. Three measurements of genome-wide deletions were used: FAL, the percentage of losses at folate-sensitive fragile sites, and the percentage of losses at tumor suppressor gene loci. The MTHFR genotypes that result in the least MTHFR activity are MTHFR 677T (one allele lost by LOH) and MTHFR 677TT (MTHFR 677CC has 100% function relative to other 677 genotypes, 677CT: ⬎65%, 677C with LOH: 50%, 677TT: ⬃30% [25%–33%], and 677T with LOH: 15%).24,25,27 The MTHFR 1298C allele does not affect MTHFR function as much as 677T, so cancer FAL data were stratified by 677 genotypes.27 Pancreatic cancers were just as likely to evolve LOH at the MTHFR locus whether the cancer retained a 677C allele or a T allele. This is not surprising because an immediate selective growth advantage would not be expected from losing one MTHFR allele over another. FAL was differ-

Table 2. Epidemiologic Data of Control Subjects and Patients

Diabetes Smoke

% (no. of sample) Period (y) % of ever smoker (no. of samples) % of current smoker (no. of samples) Period (y) Quantity (pack per day)

Control subjects (N ⫽ 305)

Patients (N ⫽ 303)

Odds ratio

95% confidence interval

17% (51/305) 0.4 ⫾ 2.6 49% (149/302)

29% (78/270) 1.1 ⫾ 4.8 64% (172/268)

2.02

1.36–3.02

1.83

1.31–2.58

.0005 .0349 .0004

16% (47/302)

26% (70/268)

1.91

1.26–2.90

.0018

6.3 ⫾ 14.3 0.3 ⫾ 0.7

17.5 ⫾ 20.5 0.6 ⫾ 0.8

P value

⬍.0001 .0002

756

MATSUBAYASHI ET AL

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 8

Table 3. MTHFR 677 Genotype and Pancreatic Cancer FAL (N ⫽ 81)a MTHFR 677 genotype

No. of patients

FAL

T C TT CT CC

7 16 9 19 30

21.3% ⫾ 6.9%b 17.7% ⫾ 7.2%c 14.8% ⫾ 3.1%d 12.6% ⫾ 5.6%e 14.8% ⫾ 6.6%f

aOne pancreatic cancer case was excluded because LOH analysis at MTHFR was not informative. b–fP ⫽ .055 by Kruskal–Wallis test. b vs dP ⫽ .023. c vs fP ⫽ .025 by Student t test.

ent among the groups. FAL was highest in cancers with the 677T genotype (that is, 677T with loss of the other allele) (FAL: 21.3% ⫾ 6.9%), followed by cancers with 677C genotypes (17.7% ⫾ 7.2%), TT (14.8% ⫾ 3.1%), CC (14.8% ⫾ 6.6%), and CT (12.6% ⫾ 5.6%) (P ⫽ .055 by Kruskal–Wallis test). Cancers with 677T had higher FAL than the group of cancers with all other genotypes (P ⫽ .012 by Student t test). Although LOH at MTHFR could be predicted to be associated with higher FAL, these data also indicate that defective MTHFR function could have contributed directly to a higher percentage of genetic deletions during cancer development because the cancers with the least functioning MTHFR genotype (677T) had the highest FAL (Table 3). Among cancers with LOH at MTHFR, those with 677T had higher FAL than those with 677C genotypes, but this difference was not statistically significant. Consistent with our case-control analysis of germline polymorphisms, we did not find any relationship between an individual’s germline MTHFR genotype and the level of FAL in the patient’s pancreaticobiliary cancer. Deletions at fragile sites and tumor suppressor genes and MTHFR genotype. To further evaluate

whether MTHFR genotype influenced a cancer’s chromosomal losses, we examined other measures of chromosomal loss. We hypothesized that if cancers with defective MTHFR function were indeed liable to undergo more genetic deletions, they would also have higher levels of fragile site loss at folate-sensitive sites. This was indeed the case. The average fragile site loss at 20 known folate-sensitive fragile sites was higher in individuals with 677T genotypes (36.9% ⫾ 10.5%) than in those with any other genotype including cancers with 677C genotypes (P ⫽ .039) (677C: 28.7% ⫾ 9.4%, 677TT: 27.4% ⫾ 11.7%, 677CT: 23.5% ⫾ 11.9%, 677CC: 28.2% ⫾ 15.3%). Among cancers with LOH at MTHFR, those with 677T genotypes had more fragile

site losses genome-wide than cancers with 677C genotypes (P ⫽ .079). These data suggested that MTHFR genotypes could influence fragile site loss independent of the presence of LOH at the MTHFR locus. Average fragile site loss was also more common in smokers than in non-smokers (P ⫽ .038). Among the pancreaticobiliary cancers as a whole, LOH occurred to a variable extent at the fragile sites ranging from 3.3% at 2p11 to 87.9% at 9p21. The most commonly deleted fragile site was the FRAXB locus at Xp22.3. Markers adjacent to the FRAXB locus were informative as to the presence or absence of LOH in 57 of the cancer DNAs. Ten of 21 cancers with LOH at MTHFR had loss of one FRAXB allele, and there was no significant difference in the occurrence of FRAXB loss in cancers with MTHFR 677T versus those with 677C (3/6, 50% vs 7/15, 48%). There was also no difference in the prevalence of loss of a FRAXB locus in cancers with LOH at MTHFR (48%) than in cancers without LOH at MTHFR (47%). But loss of one FRAXB allele was more frequent in cancers with 677TT alleles (75%, 6/8) compared with those with 677CC (6/24) ⫹ CT (5/14) alleles (29%, 11/28, P ⫽ .014). This suggests that in cancers without MTHFR LOH, cancers with the more defective TT genotypes have more fragile site loss than those with less defective genotypes. We next examined whether deletions at known tumor suppressor gene loci were more common in cancers carrying defective MTHFR alleles (some of these loci were also fragile sites). We quantified the percentage of LOH at the loci of 12 tumor suppressor genes targeted for inactivation in pancreatic cancers including p53, p16, DPC4, MKK4, FHIT, FancG, hCDC4, FancC, TGF␤R2, Alk5, BRCA2, and STK11 (LOH at these loci averaged 20% or more in the cancers). The purpose of this analysis was to obtain an overall measure of genetic deletions known to be relevant to an evolving neoplasm. The overall prevalence of LOH at these 12 suppressor loci was highest in cancers with 677T genotypes (68.5%) compared with other genotypes including 677C (47.8%) (P ⫽ .014). As with the folate-sensitive fragile site data, these results suggest that defective MTHFR genotypes influence chromosomal losses independent of the presence of LOH at the MTHFR locus. Similarly, cancers with T genotypes (677T ⫹ 677TT) had a higher percentage of deletions at tumor suppressor gene loci (60.4%) than those with 677C (677C and 677CC) genotypes (49.0%) (P ⬍ .05). Overall, these results predict that if cancers with the more defective MTHFR genotypes (677T and 677TT) have more chromosomal losses, then a neoplasm that

August 2005

MTHFR ALTERATION IN PANCREATIC CARCINOMA

757

Table 4. Association Between LOH at MTHFR and Demographic Factors, Clinicopathologic Factors, and FAL MTHFR LOH Factors

(⫺)

(⫹)

P value

Age (y) Gender (M:F) Habit Alcohol Smoker rate Smoking year Ethnicity White African American 1-y prognosis Tumor size (mm) Histology, well:moderate:poor Lymph node metastasis Rate of positive case No. of positive node/case FAL

63.3 ⫾ 11.0 27:31

63.7 ⫾ 12.4 12:11

.889 .648

22.2% (12/54) 47.4% (27/57) 9.3 ⫾ 17.8

19.0% (4/21) 72.7% (16/22) 17.2 ⫾ 17.4

.763 .043 .164

55:3

22:1

.185

66.7% (32/48) 30.2 ⫾ 13.3 4:42:12

40.9% (9/22) 35.4 ⫾ 10.5 2:8:13

.042 .097 .009

82.8% (48/58) 3.2 ⫾ 3.1 14.1% ⫾ 5.9%

73.9% (17/23) 3.4 ⫾ 4.2 18.8% ⫾ 7.2%

.367 .828 .003

undergoes deletion of an MTHFR allele could accelerate the number of chromosomal losses it undergoes during tumorigenesis. Clinicopathologic correlations by pancreatic cancer MTHFR genotypes. We examined the relationship

between smoking status, FAL, and MTHFR genotype. Cigarette smoke contains mutagens that cause genomic instability,41,42 and smokers have been shown to have higher FAL in their cancers than non-smokers.38 FAL was higher in ever smokers (17.1% ⫾ 6.4%) than in never smokers (13.6% ⫾ 6.4%) (P ⫽ .018). LOH at the MTHFR locus was more than twice as frequent in ever smokers than in never smokers (37.2% [16/43] vs 16.7% [6/36], P ⫽ .043), raising the possibility that smoking could influence MTHFR activity in a cancer by facilitating LOH at MTHFR. Among patients whose cancers had LOH at MTHFR, there was no difference in the number of smokers with 677C versus 677T genotypes. Thus, at least with our sample size we did not find evidence that genomic differences observed among cancers with 677C versus 677T genotypes could be explained by smoking status. LOH at MTHFR was more frequent in poorly differentiated cancers, and patients whose resected cancers had LOH at MTHFR had a reduced 1-year survival (Table 4). Interestingly, poorly differentiated cancers also have higher FALs (17.9% ⫾ 7.7%) than moderately (14.4% ⫾ 5.6%) or well (12.5% ⫾ 5.7%) differentiated cancers (moderately vs poorly: P ⫽ .028), suggesting that the relationship between MTHFR LOH and poor tumor differentiation reflects an influence of factors related to high FAL. Our study was not powered to determine

whether MTHFR LOH was an independent predictor of differentiation or survival in a multivariate analysis; such studies generally require several hundred patients.43 Global DNA Methylation by MTHFR Genotype We next determined whether MTHFR genotype influenced overall levels of DNA methylation in cancer genomes. We suspected that defective MTHFR function could lead to less global DNA methylation. We first used COBRA to compare the percentage methylation of LINE1 elements (a CpG-rich repeat sequence abundant throughout the genome)40 in germline versus cancer DNA. We randomly selected 105 of the 303 patients with pancreatic cancer analyzed for their germline MTHFR status and analyzed their germline LINE1 methylation status by using their non-neoplastic germline duodenal DNA. LINE1 methylation was also measured in 68 cancer cell lines (41 breast, 24 pancreas, and 3 colon). We predicted that the relationship between LINE1 methylation and MTHFR status would not differ by site of origin of cancer, and so we included additional cancer cell lines to increase sample size. Methylation was significantly higher in the germline DNA (48.2%) than in the cancer cell lines (29.0%) (P ⬍ .0001). Germline LINE1 methylation did not differ by MTHFR genotype or by demographic factors. However, LINE1 methylation in the 68 cancer cell lines varied by MTHFR 677 genotype. LINE1 methylation was lowest in the cancers with the 677T genotypes: LINE1 methylation in 677CC/677C cancers: 32.5% (677CC,

758

MATSUBAYASHI ET AL

Figure 1. Correlation between MTHFR 677 genotype and % of LINE1 methylation. The average of LINE1 methylation level was decreased from 677CC/C genotype (32.5% ⫾ 10.3%), via 677CT (26.0% ⫾ 8.9%) to 677TT/T (24.4% ⫾ 8.5%). White circle, pancreatic cancer cell line; gray circle, breast cancer cell line; black, colon cancer cell line. *P ⫽ .014 by Kruskal–Wallis test.

n ⫽ 20: 32.9%; 677C, n ⫽ 14: 32.0%), in 677CT cancers (n ⫽ 24): 26.0%, and in 677TT/677T cancers: 24.4% (677TT, n ⫽ 6: 24.5%; 677T, n ⫽ 4: 24.2%) (P ⫽ .014 by Kruskal–Wallis test). A similar trend was observed among the pancreatic cancer cell lines alone but was not statistically significant (677CC/C: 36.3%, 677CT: 30.7%, 677TT/T: 29.6%, P ⫽ .207) (Figure 1).

Discussion In this study we found that pancreaticobiliary cancers with the most defective MTHFR genotypes have significantly more chromosomal losses. Cancers with only one MTHFR 677T allele (the 2nd allele lost by LOH) had the highest percentage of chromosomal losses by 3 related measurements: their FAL (Table 3), their percentage of deletions at folate-sensitive fragile sites, and their percentage of alleles lost at tumor suppressor loci. The differences in the measures of genetic deletion between the cancers with the most defective and least defective MTHFR genotypes were modest, indicating that MTHFR function and folate status is but one of the many factors influencing genetic stability in an evolving neoplasm. Cancers with 677T or 677TT alleles also had lower LINE1 methylation in their genomes (Figure 1). These results suggest that neoplasms with severely defective MTHFR function might evolve greater genetic deletions perhaps promoted by global DNA hypomethylation.19,25 These results are consistent with other

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 8

studies indicating the role of low folate and low vitamin B12 on the development of global DNA hypomethylation19 and genomic instability.20,30,35 Among cancers with LOH, those with 677T had more chromosomal losses at folate-sensitive fragile sites and at tumor suppressor loci than cancers with 677C genotypes. Thus, because it was cancers with the most defective MTHFR genotypes (those with 677T alleles) and not merely those cancers with LOH at MTHFR that were more prone to genetic deletions and hypomethylation, our results indicate defects in MTHFR function promote some of these chromosomal losses. This being the case, our results also predict that somatic deletion of an MTHFR allele could also accelerate cancer development in a neoplasm with already compromised folate pathways (such as carrying a defective MTHFR 677T or lacking in methyl groups). We did find LOH at MTHFR was correlated with genome measures of chromosomal loss such as FAL, but these are not independent variables. To address the role of MTHFR function we focused on analyzing the cancers that had LOH at MTHFR and among this group determining whether chromosomal losses were greater in cancers with the more defective 677T MTHFR genotypes than in those with 677C genotypes. Loss of an MTHFR allele (located at 1p36) occurs commonly during tumorigenesis (28% of cancers in this study). An increased propensity of MTHFR-deficient neoplasms to undergo genetic deletions would be analogous to a caretaker defect that arises when neoplasms inactivate DNA repair genes. Our results are consistent with several previous studies implicating low folate status or defective folate metabolism with DNA strand breaks19 and genomic instability mediated through DNA hypomethylation.25 Indeed, p53-deficient animals hypomorphic for DNMT1 are more likely to get sarcomas,44 and folate/ methyl group deficiency in laboratory animals appears to facilitate genetic deletions.19 However, genomic DNA hypomethylation does not always facilitate cancer development because in Apc min mice DNMT1 hypomorphs have a reduced risk of gastrointestinal neoplasia.45 Our studies also predict that one mechanism by which other causes of impaired folate metabolism, such as low vitamin B12 or low folic acid intake, accelerate tumorigenesis is to increase the susceptibility of neoplasms with severely defective MTHFR genotypes to undergo genetic deletions. In conclusion, we have demonstrated that defective cancer MTHFR genotypes are associated with higher levels of chromosomal losses and reduced levels of LINE1 element methylation.

August 2005

MTHFR ALTERATION IN PANCREATIC CARCINOMA

References 1. Matsubayashi H, Watanabe H, Yamaguchi T, et al. Multiple K-ras mutations in hyperplasia and carcinoma in cases of human pancreatic carcinoma. Jpn J Cancer Res 1999;90:841– 848. 2. Hansel DE, Kern SE, Hruban RH. Molecular pathogenesis of pancreatic cancer. Annu Rev Genomics Hum Genet 2003;4:237– 256. 3. Wilentz RE, Geradts J, Maynard R, et al. Inactivation of the p16 (INK4A) tumor-suppressor gene in pancreatic duct lesions: loss of intranuclear expression. Cancer Res 1998;58:4740 – 4744. 4. Redston MS, Caldas C, Seymour AB, et al. p53 mutations in pancreatic carcinoma and evidence of common involvement of homocopolymer tracts in DNA microdeletions. Cancer Res 1994; 54:3025–3033. 5. Goggins M, Schutte M, Lu J, et al. Germline BRCA2 gene mutations in patients with apparently sporadic pancreatic carcinomas. Cancer Res 1996;56:5360 –5364. 6. van der Heijden MS, Yeo CJ, Hruban RH, et al. Fanconi anemia gene mutations in young-onset pancreatic cancer. Cancer Res 2003;63:2585–2588. 7. Ueki T, Toyota M, Sohn T, et al. Hypermethylation of multiple genes in pancreatic adenocarcinoma. Cancer Res 2000;60:1835–1839. 8. Sato N, Fukushima N, Maitra A, et al. Discovery of novel targets for aberrant methylation in pancreatic carcinoma using highthroughput microarrays. Cancer Res 2003;63:3735–3742. 9. Matsubayashi H, Sato N, Fukushima N, et al. Methylation of cyclin D2 is observed frequently in pancreatic cancer but is also an age-related phenomenon in gastrointestinal tissues. Clin Cancer Res 2003;9:1446 –1452. 10. Kimura M, Furukawa T, Abe T, et al. Identification of two common regions of allelic loss in chromosome arm 12q in human pancreatic cancer. Cancer Res 1998;58:2456 –2460. 11. Han HJ, Yanagisawa A, Kato Y, et al. Genetic instability in pancreatic cancer and poorly differentiated type of gastric cancer. Cancer Res 1993;53:5087–5089. 12. Yamamoto H, Itoh F, Nakamura H, et al. Genetic and clinical features of human pancreatic ductal adenocarcinomas with widespread microsatellite instability. Cancer Res 2001;61:3139 – 3144. 13. Yatsuoka T, Sunamura M, Furukawa T, et al. Association of poor prognosis with loss of 12q, 17p, and 18q, and concordant loss of 6q/17p and 12q/18q in human pancreatic ductal adenocarcinoma. Am J Gastroenterol 2000;95:2080 –2085. 14. Risch HA. Etiology of pancreatic cancer, with a hypothesis concerning the role of N-nitroso compounds and excess gastric acidity. J Natl Cancer Inst 2003;95:948 –960. 15. Stolzenberg-Solomon RZ, Albanes D, Nieto FJ, et al. Pancreatic cancer risk and nutrition-related methyl-group availability indicators in male smokers. J Natl Cancer Inst 1999;91:535–541. 16. Stolzenberg-Solomon RZ, Pietinen P, Barrett MJ, et al. Dietary and other methyl-group availability factors and pancreatic cancer risk in a cohort of male smokers. Am J Epidemiol 2001;153: 680 – 687. 17. Skinner HG, Michaud DS, Giovannucci EL, et al. A prospective study of folate intake and the risk of pancreatic cancer in men and women. Am J Epidemiol 2004;160:248 –258. 18. Blount BC, Mack MM, Wehr CM, et al. Folate deficiency causes uracil misincorporation into human DNA and chromosome breakage: implications for cancer and neuronal damage. Proc Natl Acad Sci U S A 1997;94:3290 –3295. 19. Pogribny IP, Basnakian AG, Miller BJ, et al. Breaks in genomic DNA and within the p53 gene are associated with hypomethylation in livers of folate/methyl-deficient rats. Cancer Res 1995; 55:1894 –1901. 20. Ma J, Stampfer MJ, Christensen B, et al. A polymorphism of the methionine synthase gene: association with plasma folate, vita-

21.

22.

23.

24.

25.

26.

27.

28.

29.

30.

31.

32.

33.

34.

35.

36.

37. 38.

759

min B12, homocyst(e)ine, and colorectal cancer risk. Cancer Epidemiol Biomarkers Prev 1999;8:825– 829. Silverman DT, Swanson CA, Gridley G, et al. Dietary and nutritional factors and pancreatic cancer: a case-control study based on direct interviews. J Natl Cancer Inst 1998;90:1710 –1719. Heijmans BT, Boer JM, Suchiman HE, et al. A common variant of the methylenetetrahydrofolate reductase gene (1p36) is associated with an increased risk of cancer. Cancer Res 2003;63: 1249 –1253. Engbersen AM, Franken DG, Boers GH, et al. Thermolabile 5,10methylenetetrahydrofolate reductase as a cause of mild hyperhomocysteinemia. Am J Hum Genet 1995;56:142–150. Frosst P, Blom HJ, Milos R, et al. A candidate genetic risk factor for vascular disease: a common mutation in methylenetetrahydrofolate reductase. Nat Genet 1995;10:111–113. Lathrop Stern L, Shane B, Bagley PJ, et al. Combined marginal folate and riboflavin status affect homocysteine methylation in cultured immortalized lymphocytes from persons homozygous for the MTHFR C677T mutation. J Nutr 2003;133:2716 –2720. Bailey LB, Duhaney RL, Maneval DR, et al. Vitamin B-12 status is inversely associated with plasma homocysteine in young women with C677T and/or A1298C methylenetetrahydrofolate reductase polymorphisms. J Nutr 2002;132:1872–1878. van der Put NM, Gabreels F, Stevens EM, et al. A second common mutation in the methylenetetrahydrofolate reductase gene: an additional risk factor for neural-tube defects? Am J Hum Genet 1998;62:1044 –1051. Bagley PJ, Selhub J. A common mutation in the methylenetetrahydrofolate reductase gene is associated with an accumulation of formylated tetrahydrofolates in red blood cells. Proc Natl Acad Sci U S A 1998;95:13217–13220. Ulvik A, Evensen ET, Lien EA, et al. Smoking, folate and methylenetetrahydrofolate reductase status as interactive determinants of adenomatous and hyperplastic polyps of colorectum. Am J Med Genet 2001;101:246 –254. Friso S, Choi S, Girelli D, et al. A common mutation in the 5,10-methylenetetrahydrofolate reductase gene affects genomic DNA methylation through an interaction with folate status. Proc Natl Acad Sci U S A 2002;16:5606 –5611. Wiemels JL, Smith RN, Taylor GM, et al. Methylenetetrahydrofolate reductase (MTHFR) polymorphisms and risk of molecularly defined subtypes of childhood acute leukemia. Proc Natl Acad Sci U S A 2001;98:4004 – 4009. Heijmans BT, Gussekloo J, Kluft C, et al. Mortality risk in men is associated with a common mutation in the methylenetetrahydrofolate reductase gene (MTHFR). Eur J Hum Genet 1999;7:197–204. Keku T, Millikan R, Worley K, et al. 5,10-Methylenetetrahydrofolate reductase codon 677 and 1298 polymorphisms and colon cancer in African Americans and whites. Cancer Epidemiol Biomarkers Prev 2002;11:1611–1621. Skibola CF, Smith MT, Kane E, et al. Polymorphisms in the methylenetetrahydrofolate reductase gene are associated with susceptibility to acute leukemia in adults. Proc Natl Acad Sci U S A 1999;96:12810 –12815. Pufulete M, Al-Ghnaniem R, Leather AJ, et al. Folate status, genomic DNA hypomethylation, and risk of colorectal adenoma and cancer: a case control study. Gastroenterology 2003;124:1240 –1248. Hahn SA, Seymour AB, Hoque AT, et al. Allelotype of pancreatic adenocarcinoma using xenograft enrichment. Cancer Res 1995; 55:4670 – 4675. Yeo CJ. The Whipple operation: is a radical resection of benefit? Adv Surg 2003;37:1–27. Iacobuzio-Donahue CA, van der Heijden MS, Baumgartner MR, et al. Large-scale allelotype of pancreaticobiliary carcinoma provides quantitative estimates of genome-wide allelic loss. Cancer Res 2004;64:871– 875.

760

MATSUBAYASHI ET AL

39. Arlt MF, Casper AM, Glover TW. Common fragile sites. Cytogenet Genome Res 2003;100:92–100. 40. Yang AS, Estecio MR, Doshi K, et al. A simple method for estimating global DNA methylation using bisulfite PCR of repetitive DNA elements. Nucleic Acids Res 2004;32:e38. 41. Godschalk R, Nair J, van Schooten FJ, et al. Comparison of multiple DNA adduct types in tumor adjacent human lung tissue: effect of cigarette smoking. Carcinogenesis 2002;23:2081–2086. 42. Sozzi G, Pierotti MA. When smoke gets in your genes. Nat Med 1998;4:1119 –1120. 43. Tascilar M, Skinner HG, Rosty C, et al. The SMAD4 protein and prognosis of pancreatic ductal adenocarcinoma. Clin Cancer Res 2001;7:4115– 4121. 44. Eden A, Gaudet F, Waghmare A, et al. Chromosomal instability

CLINICAL GASTROENTEROLOGY AND HEPATOLOGY Vol. 3, No. 8

and tumors promoted by DNA hypomethylation. Science 2003;300:455. 45. Eads CA, Nickel AE, Laird PW. Complete genetic suppression of polyp formation and reduction of CpG-island hypermethylation in Apc(Min/⫹) Dnmt1-hypomorphic Mice. Cancer Res 2002; 62:1296 –1299.

Address requests for reprints to: Michael Goggins, MD, The Johns Hopkins Medical Institutions, 632 Ross Building, 720 Rutland Avenue, Baltimore, Maryland 21205-2196. e-mail: [email protected]; fax: (410) 614-0671. Supported by the National Cancer Institute (CA91968, CA62924, CA90709) and Sankyo Foundation of Life Science (to H.M.).