Association of polymorphisms in one-carbon metabolizing genes and lung cancer risk: a case-control study in Chinese population

Lung Cancer (2008) 61, 21—29

available at www.sciencedirect.com

journal homepage: www.elsevier.com/locate/lungcan

Association of polymorphisms in one-carbon metabolizing genes and lung cancer risk: a case-control study in Chinese population Hongliang Liu a,1, Guangfu Jin b,1, Haifeng Wang c, Wenting Wu a, Yanhong Liu a, Ji Qian a, Weiwei Fan a, Hongxia Ma b, Ruifen Miao b, Zhibin Hu b, Weiwei Sun c, Yi Wang c, Li Jin a,c, Qingyi Wei d, Hongbing Shen b, Wei Huang c,∗∗, Daru Lu a,∗ a

State Key Laboratory of Genetic Engineering, School of Life Sciences, Fudan University, Shanghai 200433, China Department of Epidemiology and Biostatistics, Cancer Research Center of Nanjing Medical University, Nanjing 210029, China c Department of Genetics, Chinese National Human Genome Center at Shanghai, Shanghai 201203, China d Department of Epidemiology, The University of Texas M. D. Anderson Cancer Center, Houston, TX 77030, USA b

Received 24 September 2007; received in revised form 29 November 2007; accepted 2 December 2007

KEYWORDS Lung cancer; One-carbon metabolizing genes; Polymorphism; Haplotype; Case-control study; Chinese population; MDR

∗ ∗∗ 1

Summary One-carbon metabolism facilitates the cross-talk between genetic and epigenetic processes, making it a good candidate for studying the risk of lung cancer. To investigate the role of common variants of one-carbon metabolizing genes on lung cancer risk, total 25 single nucleotide polymorphisms (SNPs) in 7 genes were genotyped among 500 incident lung cancer patients and 517 cancer-free controls. An increased risk was suggested for the variant allele carriers of MTHFR rs17037396 [odds ratio (OR) = 1.39, 95% confidence interval (CI): 1.00—1.94] and rs3753584 (OR = 1.46, 95% CI: 1.03—2.08), compared with subjects with wild homozygote, respectively, and the risk was more pronounced among older individuals (>60 years). In contrast, a decreased risk was observed for TYMS rs2853742 variant allele carriers (OR = 0.44, 95% CI: 0.19—0.99) and MTHFD rs2236225 variant allele carriers (OR = 0.76, 95% CI: 0.59—0.99). Haplotype analysis revealed that MTHFR ‘‘ACCACC’’ haplotype may contribute to the risk of lung cancer (OR = 1.49, 95% CI: 1.03—2.14, local test p value 0.032). A data mining method, multifactor dimensionality reduction (MDR), predicted a four-factor interaction model (rs1801133, rs4659731, rs2273029 and rs699517) with the lowest average prediction

Corresponding author. Tel.: +86 21 65642799; fax: +86 21 65642799. Corresponding author. E-mail addresses: [email protected] (W. Huang), [email protected] (D. Lu). These authors contribute equally to this work.

0169-5002/$ — see front matter © 2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.lungcan.2007.12.001

22

H. Liu et al. error (45.08%, p < 0.001). These findings suggest that genetic variants in one-carbon metabolizing genes might modulate the risk of lung cancer. Validation of these findings in larger studies is needed. © 2007 Elsevier Ireland Ltd. All rights reserved.

1. Introduction Lung cancer is the leading cause of cancer-related death for men and women in the world, and there were an estimated 1.35 million new cases worldwide in 2002 [1]. Although overwhelming epidemiological evidence exists that smoking is the primary risk factor for lung cancer [2], <20% of lifetime smokers develop lung cancer, suggesting that interindividual variation in genetic susceptibility [3] or habits [4], may influence lung cancer risk associated with exposure to the chemical carcinogens in tobacco. Epidemiologic studies have provided evidence that high consumption of vegetables and fruits is associated with a reduced risk of lung cancer [5—7], and dietary folate may be one of the micronutrients that provide protection against lung carcinogenesis [8]. The main function of folate is to mediate the transfer of one-carbon units in various cellular reactions, including those that are necessary for thymidine, purine and methionine synthesis [9]. A sufficient pool of thymidine and purine nucleotides is required for adequate DNA synthesis and repair, whereas methionine converted to S-adenosylmethionine, the methyl donor for various methylation reactions, is required for the maintenance of normal DNA methylation patterns. Therefore, disturbance in this pathway may lead to aberrant DNA synthesis, repair and methylation. Polymorphisms in one-carbon metabolism related genes have been studied widely among colorectal cancer [10,11], breast cancer [12—14], lung cancer [15,16] and so on. However, in most of the case-control studies on lung cancer, investigators restricted their analyses to limited functional variants, such as methylenetetrahydrofolate reductase (MTHFR 677C > T and 1298A > C) and reduced folate carrier (RFC1-80G > A) [15,17]. Little is known about the relationship of other variants or other genes in the pathway with lung cancer risk. To further investigate the role of this pathway in lung cancer, we determined the association between 25 single nucleotide polymorphisms (SNPs) in 7 folate metabolism genes and lung cancer risk among 500 incident lung cancer patients and 517 age- and gender-matched cancer-free controls.

2. Materials and methods 2.1. Study populations The study population and subject characteristics were previously described elsewhere [18]. In brief, this was a hospital-based case-control study including 500 lung cancer patients and 517 cancer-free controls. All subjects were genetically unrelated ethnic Han Chinese and were from Nanjing city and surrounding regions in southeast China. Patients who were newly diagnosed with incident lung cancer according to the National Diagnosis Standard for Lung

Cancer, were consecutively recruited between July 2002 and November 2004 from the Cancer Hospital of Jiangsu Province, the First Affiliated Hospital of Nanjing Medical University, without the restrictions of age, sex and histology. Those who were previous cancer, metastasized cancer and previous radiotherapy or chemotherapy were excluded. The response rate for case was 90.5%. Cancer-free controls were randomly selected from 10,500 individuals participated in a community-based screening program for non-infectious diseases conducted in Jiangsu province during the same time period when the cases were recruited, with a response rate of 83.8%. These control subjects had no history of cancer and were frequency-matched to the cases on age (±5 years), sex and residential area (urban or rural areas). Each participant was scheduled for an interview after written informed consent was obtained, and a structured questionnaire was administered by interviewers to collect information on demographic data and environmental exposure history including tobacco smoking. Those who had smoked less than 1 cigarette per day and shorter than 1 year in their lifetime were defined as non-smokers, otherwise, they were considered as ever smokers. Those smokers who quit for more than 1 year were considered as former smokers. Pack—years smoked [(cigarettes per day/20) × years smoked] were calculated to indicate the cumulative smoking dose. Family history of cancer was defined as any selfreported cancer in first-degree relatives (parents, siblings or children). After interview, approximately 5-ml venous blood sample was collected from each participant. The study was approved by the institutional review boards of Fudan University.

2.2. SNP selection SNPs were selected from NCBI SNP database (http://www.ncbi.nlm.nih.gov/snp) based on: (1) polymorphism density in genes, (2) predicted function and genomic context, (3) minor allele frequency, (4) quality of validation evidence and (5) compatibility with the genotyping platform. A greedy algorithm was used during the selection process. The more detailed description about the selection method can be found in the previous paper [18,19]. Total 25 SNPs in seven one-carbon metabolism genes (MTHFR, MTR, SHMT1, TYMS, SLC19A1, MTHFD1 and FOLH1) were selected (Table 1), including seven nonsynonymous polymorphisms, nine variants located in 5 , 3 UTR or gene flanking region and nine intronic variants. Three reported SNPs in TYMS were not selected in the study: TYMS 3R > 2R (rs34743033), 1494del16 (rs16430), IVS6-68C > T (rs1059394). Based on the data of SNP500Cancer, the polymorphism 3R > 2R departured from Hardy—Weinberg equilibrium severely (p < 0.001), and the IVS6-68C > T in pacific population did not show polymorphism, only 1494del16 was complete linkage disequilibrium

The association of polymorphisms in one-carbon metabolizing genes Table 1

23

Primary information for 25 genotyped SNPs of one-carbon metabolism pathway

Gene

SNP

Base change

Location in gene region

Case/control

MAF for case

MAF for control

MAF in databasea

p value for HWE test

MTHFR

rs4846048 rs1801131 rs1801133 rs17037396 rs3753584 rs3737964

A>G A>C T>C C>T A>G G>A

3 UTR E429A, 1298A > C A222V, 677C > T 2nd intron 5 UTR 5 flanking

500/517 500/517 500/517 500/517 500/517 500/517

0.09 0.18 0.44 0.11 0.10 0.09

0.09 0.16 0.46 0.08 0.07 0.10

0.10 0.20 0.49 0.11 0.13 0.10

0.80 0.62 0.48 1.00 1.00 1.00

MTR

rs16834388 rs12138911 rs4659731 rs2297967 rs1805087 rs10925263

G>T T>C T>C G>A A>G T>C

5 flanking 5th intron 15th intron 22nd intron D919G, 2756A > G 3 UTR

500/517 500/517 500/517 500/517 500/517 500/517

0.44 0.12 0.45 0.18 0.09 0.27

0.42 0.14 0.44 0.19 0.10 0.27

0.40 0.08 0.42 0.20 0.07 0.22

0.79 0.45 0.42 0.11 0.46 0.91

SHMT1

rs2273029 rs4924750 rs2461837

C>T C>G G>A

9th intron 1st intron 1st intron

500/517 500/517 500/516

0.30 0.08 0.07

0.30 0.07 0.07

0.22 0.09 0.09

0.68 0.35 0.50

TYMS

rs2853741 rs2853742 rs11540152 rs1059393 rs699517 rs2790

T>C T>C T>C T>C T>C A>G

5 flanking 5 flanking F117L Intron 3 UTR, 1053C > T 3 UTR, 1122A > G

500/517 500/517 500/517 500/517 499/517 498/517

0.47 0.16 0.32 0.003 0.32 0.38

0.46 0.18 0.33 0.002 0.32 0.37

0.41 NA 0.00 0.04 0.30 0.35

0.93 0.29 0.42 0.95 0.27 0.45

RFC

rs914232 rs1051266

C>T G>A

4th intron H27R, 80G > A

500/516 499/504

0.49 0.49

0.49 0.48

0.49 0.50

0.93 0.93

MTHFD FOLH1

rs2236225 rs202676

C>T T>C

R653Q Ter75Q

500/517 500/517

0.22 0.32

0.25 0.35

0.22 0.26

0.13 0.50

a

MAF (minor allele frequency) in HapMap database for CHB population.

with another function SNP 1053C > T which had been included in the study.

2.3. Genotyping assays SNPs were genotyped using the Illumina Genotyping Facility by combining with Illumina Golden GateTM assay, SentrixTM array matrices and SherlockTM scanner technology (Illumina Corp., Foster City, CA), at Chinese National Human Genome Center at Shanghai, China. More detailed description for the Illumina Genotyping Facility are available in our published paper [18] and the HapMap website (http://www.hapmap.org/downloads/protocols overiview. html).

2.4. Statistical analyses The HWE test was done for each SNP among controls using Fisher probability test statistic, as implemented in the software package SNPassoc [20]. The single locus association was estimated by computing the odds ratios and 95% confidence intervals with adjustment for age, sex, family cancer history, smoking status and square root of pack—years of smoking using SNPassoc. We also evaluate the interactions of haplotypes and SNPs with gender (male,

female), dichotomized age (≤60 years, >60 years); trichotomized cumulative smoking doses (non-smokers, light smokers: 1—30 pack—years, heavy smokers: >30 pack—years) under the codominant and dominant genetic model. The issue of multiple-test was controlled by using 1000-fold permutation test. Haploview was used to measure the linkage disequilibrium between SNPs and construct the haplotype blocks [21]. PHASE 2.0 was used to infer the haplotype frequencies based on the observed genotypes for each gene [22]. Assessment of multi-locus interactions was carried out by using the nonparametric multifactor dimensionality reduction (MDR) software [23]. Briefly, this method includes a combined cross-validation/permutation-testing procedure that minimizes false-positive results that might otherwise result from multiple examinations of the data. The method involves several steps: in the first step, the data were divided into a training set (consisting of 9/10 of the data) and an independent testing set (consisting of the remaining 1/10 of the data) as part of cross-validation. In the second step, a set of n factors (in this case, SNPs) were selected. In steps 3 and 4, the n SNPs and their possible multifactor classes are represented in n dimensional space, e.g., for two SNPs with three genotypes each, there are nine possible two-locus—genotype combinations. The ratio for the number of cases to the number of controls was calcu-

24 lated within each multifactor class. Each multifactor class in n dimensional space was then labeled as ‘‘high risk’’ if the case to control ratio met or exceeded a threshold (for example, 1.0), or as ‘‘low risk’’ if that threshold was not exceeded, thus reducing the n dimensional space to one dimension with two levels (low risk and high risk). In the fifth step, the model that gave the lowest misclassification error (error in classifying cases and controls based on high risk or low risk in the training set) was selected for each set of n SNPs. In step six, a prediction error (error in classifying disease status in the testing set) was estimated for each model selected in step five, as a cross-validation procedure. Steps 1—6 were repeated 10 times using a random seed number. We did this entire 10-fold cross-validation procedure 10 times, using different random seed numbers, to reduce the chance of observing spurious results due to chance divisions of the data. Empirical p values were based on the number of prediction errors estimated among the 1000 simulations that were as small as or smaller than the observed prediction errors. The best MDR model was selected as the one with the minimal prediction and also high cross-validation consistency (CVC). All statistical analyses were done with R 2.4.1 package (http://CRAN.R-project.org/) if not mentioned specially.

3. Results The 500 lung cancer cases and 517 cancer-free controls appeared to be adequately matched on age and sex. The mean age was 59.3 years (±10.4 years) for the cases and 60.0 years (±10.3 years) for the controls (p = 0.66); 386 (77.2%) cases and 400 (77.4%) controls were male (p = 1.00). Smoking and family history of cancer in the first-degree relatives were significant risk factors for lung cancer. Specifically, light smokers had a 1.39-fold (95% CI: 1.04—1.87) and heavy smokers had a 2.75-fold (95% CI: 2.01—3.78) increased risk, compared with non-smokers. In addition, 122 (24.4%) cases and 87 (16.8%) controls reported a family history of cancer in first-degree relatives, and the difference was associated with a 1.60-fold (95% CI: 1.17—2.17) increased risk for lung cancer. Of the 500 cancer patients, 229 were adenocarcinomas, 141 were squamous cell carcinomas, 34 were small cell carcinomas and 96 were large cell, mixed cell, or undifferentiated carcinomas. All SNPs were in Hardy—Weinberg equilibrium (p > 0.05) among controls (Table 1) and most SNPs’ minor allele frequencies in our healthy controls were similar to those published in Chinese population in HapMap database. And this suggested strong representation of controls to general population to some extent. The MAF of rs11540152 and rs1059393 were different from those reported in HapMap database, which may reflect either region differences or frequency bias due to small sample sizes. Single locus analysis revealed an increased risk for the variant allele carriers of MTHFR rs17037396 (OR = 1.39, 95% CI: 1.00—1.94) and rs3753584 (OR = 1.46, 95% CI: 1.03—2.08), compared with wild homozygote, respectively, and a decreased risk was observed for TYMS rs2853742 allele carriers (OR = 0.44, 95% CI: 0.19—0.99) and MTHFD rs2236225 (R653Q) variant allele carriers (OR = 0.76,

H. Liu et al. 95% CI: 0.59—0.99) (Table 2 ). Interaction effects of gene—environment only exist between the MTHFR polymorphisms and dichotomized age factor (Table 2), and as shown in Table 3, the risk of variants was more pronounced among older individuals (>60 years). As shown in Table 4, haplotype analysis revealed that MTHFR ‘‘ACCACC’’ haplotype might contribute to the risk of lung cancer (OR = 1.49, 95% CI: 1.03—2.14, local test p value 0.032). MDR software was used to assess potential gene—gene interactions on lung cancer risk with all of the 25 SNPs genotyping data. As shown in Table 5, the fourfactor model including MTHFR rs180133 (677C > T), MTR rs4659731, SHMT1 rs2273029 and TYMS rs699517 (1053C > T) had the highest CVC (95%) and the lowest prediction error (45.08%) among all the possible combinations of four SNPs. The prediction error was statistically significant, with an empirical p < 0.001 based on 1000 permutations.

4. Discussion In this lung cancer case-control study in a Chinese population, we investigated, for the first time, the role of multiple common variants in one-carbon metabolism genes in the susceptibility to lung cancer. For the first time, we identified four single variants in one-carbon metabolism genes (rs17037396, rs3753584, rs2853742 and rs2236225) and one MTHFR haplotype ‘‘ACCACC’’ might contribute to the risk of lung cancer. Multi-locus interaction analysis revealed a significant interaction among polymorphisms in four key enzymes (MTHFR, MTR, SHMT1 and TYMS) which resulted in a 2.8-fold increase of lung cancer risk. At present, most studies involved in the association of one-carbon metabolizing genes and lung cancer susceptibility have focused on two putative functional SNPs in MTHFR (677C > T and 1298A > C) and the results were inconsistent [17,24—26]. Recently, a Central Europe population study with large sample size observed a moderate effect of MTHFR C677T on lung cancer risk and a possible effect modified by folate intake [16]. However our study did not found the association between MTHFR C677T and lung cancer risk by single locus analysis. The reasons for the difference are unclear, but difference in circulating folate levels between different populations may account for this [16,27]. Based on the previous study, there was a significant gene-exposure interaction between MTHFR C677T and folate intake, and the functional effect of the polymorphism may be influenced by folate availability, which, in turn, may have a bearing on the association of the polymorphism with lung cancer risk. This could account for the different findings between different populations. Because our study did not adjust the folate intake, validation of these findings in larger studies with folate intake information is needed. By MDR method, we found that four-factor model included MTHFR rs180133, MTR rs4659731, SHMT1 rs2273029 and TYMS rs699517 had the highest CVC (95%) and the lowest prediction error (45.08%). Although this prediction error is far from a perfect 0%, it is an important improvement from a prior 50% chance in predicting

Adjusted OR (95% CI) for the association between SNPs in one-carbon metabolism genes and the risk of lung cancer

Gene

SNP

MTHFR

MTR

SHMT1

Genotype wt/wt

wt/varb

var/var

var carrier

p for interaction by gender

p for interaction by age

p for interaction by pack—years

rs4846048

Cases/controls OR (95% CI)a

421/424 1.00 (Ref)

71/88 0.86 (0.60—1.22)

8/5 1.69 (0.52—5.52)

79/93 0.90 (0.64—1.27)

0.19

0.02

0.58

rs1801131

Cases/controls OR (95% CI)

341/364 1.00 (Ref)

141/142 1.08 (0.81—1.44)

18/11 2.15 (0.97—4.79)

159/153 1.15 (0.87—1.52)

0.09

0.24

0.72

rs1801133

Cases/controls OR (95% CI)

157/149 1.00 (Ref)

245/265 0.89 (0.66—1.19)

98/103 0.97 (0.67—1.41)

343/368 0.91 (0.69—1.21)

0.27

0.31

0.97

rs17037396

Cases/controls OR (95% CI)

395/433 1.00 (Ref)

97/81 1.31 (0.93—1.85)

8/3 3.92 (0.98—15.58)

105/84 1.39 (1.00—1.94)

0.17

0.03

0.79

rs3753584

Cases/controls OR (95% CI)

408/446 1.00 (Ref)

87/69 1.40 (0.98—2.01)

5/2 3.95 (0.72—21.72)

92/71 1.46 (1.03—2.08)

0.39

0.03

0.54

rs3737964

Cases/controls OR (95% CI)

420/416 1.00 (Ref)

72/96 0.78 (0.55—1.10)

8/5 1.66 (0.51—5.42)

80/1012 0.82 (0.58—1.15)

0.15

0.03

0.45

rs16834388

Cases/controls OR (95% CI)

155/174 1.00 (Ref)

251/249 1.20 (0.89—1.60)

94/94 1.16 (0.80—1.69)

345/343 1.19 (0.90—1.56)

0.39

0.20

0.39

rs12138911

Cases/controls OR (95% CI)

390/384 1.00 (Ref)

100/126 0.82 (0.60—1.12)

10/7 1.43 (0.52—3.91)

110/133 0.85 (0.63—1.15)

0.19

0.65

0.90

rs4659731

Cases/controls OR (95% CI)

147/168 1.00 (Ref)

255/254 1.25 (0.93—1.67)

98/104 1.10 (0.76—1.59)

353/349 1.20 (0.91—1.59)

0.33

0.11

0.70

rs2297967

Cases/controls OR (95% CI)

338/346 1.00 (Ref)

141/147 1.00 (0.75—1.33)

21/24 0.89 (0.48—1.67)

162/171 0.98 (0.75—1.29)

0.85

0.28

0.19

rs1805087

Cases/controls OR (95% CI)

415/419 1.00 (Ref)

79/95 0.84 (0.60—1.19)

6/3 2.17 (0.52—8.99)

85/98 0.89 (0.63—1.24)

0.78

0.27

0.31

rs10925263

Cases/controls OR (95% CI)

255/273 1.00 (Ref)

216/207 1.04 (0.80—1.36)

29/37 0.83 (0.49—1.42)

245/244 1.01 (0.78—1.31)

0.46

0.47

0.47

rs2273029

Cases/controls OR (95% CI)

248/249 1.00 (Ref)

209/223 0.95 (0.72—1.24)

43/45 0.97 (0.60—1.57)

252/268 0.95 (0.74—1.23)

0.68

0.51

0.32

rs4924750

Cases/controls OR (95% CI)

432/441 1.00 (Ref)

61/75 0.78 (0.53—1.14)

7/1 7.02 (0.82—60.30)

68/76 0.86 (0.59—1.24)

0.80

0.36

0.46

rs2461837

Cases/controls OR (95% CI)

431/443 1.00 (Ref)

64/72 0.86 (0.59—1.26)

5/1 3.65 (0.40—32.95)

69/73 0.90 (0.62—1.31)

0.68

0.54

0.70

The association of polymorphisms in one-carbon metabolizing genes

Table 2

25

26

Table 2 (Continued ) Gene

TYMS

SNP

c

Genotype wt/wt

wt/varb

var/var

var carrier

p for interaction by gender

p for interaction by age

p for interaction by pack—years

rs2853741

Cases/controls OR (95% CI)

138/148 1.00 (Ref)

251/259 1.03 (0.76—1.40)

111/110 1.08 (0.75—1.56)

362/369 1.05 (0.78—1.39)

0.83

0.97

0.73

rs2853742

Cases/controls OR (95% CI)

346/353 1.00 (Ref)

145/144 1.07 (0.80—1.42)

9/20 0.44 (0.19—0.99)

154/164 0.98 (0.75—1.30)

0.78

0.75

0.36

rs11540152

Cases/controls OR (95% CI)

230/239 1.00 (Ref)

222/219 1.03 (0.78—1.35)

48/59 0.83 (0.53—1.29)

270/278 0.98 (0.76—1.28)

0.64

0.41

0.95

rs699517

Cases/controls OR (95% CI)

227/243 1.00 (Ref)

225/215 1.10 (0.84—1.44)

47/59 0.84 (0.54—1.31)

272/274 1.05 (0.81—1.35)

0.46

0.36

0.83

rs2790

Cases/controls OR (95% CI)

194/202 1.00 (Ref)

232/249 0.97 (0.74—1.28)

72/66 1.10 (0.73—1.65)

304/315 1.00 (0.77—1.30)

0.21

0.51

0.86

rs914232

Cases/controls OR (95% CI)

129/136 1.00 (Ref)

252/257 1.04 (0.76—1.42)

119/123 1.05 (0.73—1.51)

371/380 1.04 (0.78—1.40)

0.30

0.66

0.25

rs1051266

Cases/controls OR (95% CI)

127/137 1.00 (Ref)

250/250 1.08 (0.79—1.47)

122/117 1.14 (0.80—1.64)

372/367 1.10 (0.82—1.47)

0.36

0.66

0.35

MTHFD

rs2236225

Cases/controls OR (95% CI)

303/383 1.00 (Ref)

171/208 0.77 (0.59—1.01)

23/26 0.69 (0.37—1.27)

194/234 0.76 (0.59—0.99)

0.16

0.09

0.80

FOLH1

rs202676

Cases/controls OR (95% CI)

230/222 1.00 (Ref)

220/228 0.95 (0.72—1.25)

50/67 0.75 (0.49—1.16)

270/295 0.91 (0.70—1.17)

0.70

0.40

0.66

SLC19A1

a b c

All of the OR (CI 95%) were adjusted for age, gender, family history of cancer, smoking status and square root of pack—years. wt, wild-type; var, variant. TYMS rs1059393 were not included in the table as its MAF was just 0.003.

H. Liu et al.

The association of polymorphisms in one-carbon metabolizing genes Table 3

27

Adjusted OR (95% CI) for the stratified analysis of MTHFR polymorphisms by dichotomized age

SNP

Genotype

Age ≥ 60a

Age < 60a Case

Control

OR (95% CI)b

Case

Control

OR (95% CI)b

rs4846048

AA AG GG AG + GG

218 41 3 44

254 45 5 50

1.00 1.11 0.75 1.08

(Ref) (0.69—1.79) (0.17—3.33) (0.68—1.70)

203 30 5 35

170 43 — 43

1.00 (Ref) 0.62 (0.36—1.06) — 0.72 (0.43—1.21)

rs1801131

AA AC CC AC + CC

176 75 11 86

208 86 10 96

1.00 1.05 1.46 1.09

(Ref) (0.72—1.54) (0.59—3.66) (0.76—1.58)

165 66 7 73

156 56 1 57

1.00 1.11 8.88 1.23

(Ref) (0.72—1.72) (1.03—76.81) (0.80—1.89)

rs1801133

CC TC TT TC + TT

90 116 56 172

87 153 64 217

1.00 0.73 0.84 0.76

(Ref) (0.49—1.08) (0.52—1.35) (0.52—1.10)

67 129 42 171

62 112 39 151

1.00 1.16 1.20 1.17

(Ref) (0.74—1.81) (0.67—2.15) (0.76—1.79)

rs17037396

GG GA AA GA + AA

209 47 6 53

243 58 3 61

1.00 0.94 2.76 1.02

(Ref) (0.60—1.47) (0.64—11.88) (0.66—1.56)

186 50 2 52

190 23 — 23

1.00 (Ref) 2.18 (1.25—3.80) — 2.29 (1.32—3.99)

rs3753584

TT TC CC TC + CC

214 45 3 48

250 52 2 54

1.00 1.00 2.30 1.04

(Ref) (0.64—1.58) (0.34—15.38) (0.67—1.63)

194 42 2 44

196 17 — 17

1.00 (Ref) 2.53 (1.36—4.73) — 2.69 (1.45—5.00)

rs3737964

CC TC TT TC + TT

218 41 3 44

248 51 5 56

1.00 0.96 0.73 0.94

(Ref) (0.60—1.52) (0.17—3.25) (0.60—1.47)

202 31 5 36

168 45 — 45

1.00 (Ref) 0.60 (0.35—1.01) — 0.69 (0.42—1.15)

a b

The age was dichotomized by 60 years according to the mean age of controls. Adjusted for gender, family history of cancer, smoking status and square root of pack—years.

lung cancer status. The prediction error was statistically significant, with an empirical p < 0.001 based on 1000 permutations. The result suggested that the identified multifactor interaction is unlikely due to chance. The exact biological mechanisms of these four SNPs jointly affecting lung cancer risk remain unclear; however previous studies about the biological pathway were informative in understanding the potential roles of the interaction pattern. These four genes are all key enzymes and interact with each other in the one-carbon metabolism pathway [28]. The flux of deoxynucleotides for DNA synthesis is directly controlled by TYMS. SHMT1 regulates the availability of 5,10-MeTHF to act as substrate for MTHFR. MTHFR catalyzes the irreversible conversion of 5,10-methylenetetrahydrofolate (5,10-MeTHF) to 5-methyltetrahydrofolate (5-MeTHF), the major circulating form of folate which acts as a methyl donor for S-adenosylmethionine production. 5-MeTHF, the product of the MTHFR reaction, is a substrate for MTR. This study provides a possible framework for a potential SNP—SNP interaction-based model in one-carbon metabolism pathway for lung cancer risk. The significance of genotypes’ main effects in our study was subtle. This may due to that genetic variants in onecarbon metabolism pathway might just have moderate impact on the risk of lung cancer and our sample size

was not large enough to identify significant associations of the effect in different strata in subgroup analyses. Further more, except for tobacco smoking, other factors, such as dietary component and occupational exposure, might interact with variants in one-carbon metabolism pathway or act as potential confounders. Unfortunately, information on these factors in our case-control study was not available. It would be interesting to investigate interactions between genotypes and these risk factors in future studies. The key limitation of this study is the lack of an assessment of folate status. According to the previous studies, there exist seasonal and gender differences in folate status among Chinese people [27,29]. People living in south China have a lower prevalence folate concentration deficiency (6.2% for plasma folate and 4.0% for blood cell) defined as plasma folate <6.8 nmol/L (3 ␮g/L) or red blood cell folate <363 nmol/L (160 ␮g/L) [27]. Considering objects in the study were mainly coming from south China (Jiangsu province), the folate status may have limited influence on the association between SNPs and lung cancer susceptibility. In the study, we also matched the controls to the cases on age, sex and residential area to control the population heterogeneous. Although the sample size of current study was only moderate, the genotyping rates in our study were very high (99.8%) for each locus, and the haplotype-based anal-

28

H. Liu et al.

Table 4

Associations between one-carbon metabolism pathway common haplotypes and lung cancer risk

Gene

Haplotypea

Total freq

Case freq

Control freq

OR (95% CI)

p valueb

MTHFR

AATGTC AACGTC GCCGTT ACCACC Rarec

0.425 0.374 0.087 0.076 0.037

0.415 0.378 0.084 0.089 0.034

0.439 0.367 0.091 0.064 0.040

1.00 1.00 0.99 1.49 0.97

(Ref) (0.82—1.23) (0.71—1.38) (1.03—2.14) (0.57—1.63)

— 0.980 0.960 0.032 0.890

MTR

TTGGTA GTAGTG GTAATA GCAGCA TCGGTA GTGGTA Rarec

0.388 0.271 0.179 0.092 0.033 0.023 0.015

0.399 0.272 0.172 0.088 0.027 0.023 0.019

0.378 0.268 0.186 0.096 0.038 0.022 0.012

1.00 0.92 0.87 0.89 0.76 0.87 1.38

(Ref) (0.73—1.16) (0.68—1.13) (0.64—1.24) (0.46—1.27) (0.46—1.64) (0.61—3.14)

— 0.500 0.300 0.490 0.300 0.660 0.440

TYMSd

ATTTC ATTTT GTCCT GCCCT GTTTT Rarec

0.337 0.178 0.154 0.148 0.126 0.057

0.344 0.172 0.166 0.139 0.126 0.053

0.332 0.183 0.143 0.157 0.126 0.060

1.00 0.93 1.10 0.87 1.01 0.83

(Ref) (0.71—1.22) (0.83—1.46) (0.66—1.15) (0.75—1.38) (0.53—1.29)

— 0.620 0.520 0.330 0.930 0.400

SHMT1

CCC TCC CGT Rarec

0.626 0.299 0.072 0.004

0.628 0.295 0.072 0.005

0.626 0.300 0.072 0.003

1.00 0.97 0.92 2.24

(Ref) (0.79—1.18) (0.64—1.31) (0.38—13.05)

— 0.730 0.630 0.370

SLC19A1

CG TA Rarec

0.507 0.486 0.007

0.502 0.487 0.011

0.512 0.485 0.003

1.00 (Ref) 1.04 (0.87—1.25) 3.09 (0.84—11.43)

— 0.640 0.091

a b c d

The SNPs are shown as their orders in Table 1. The p value was calculated by using Pearson Chi-Square test. ‘‘Rare haplotypes’’ consists of haplotypes with a frequency <2%. TYMS rs1059393 were not included in the table as its MAF was just 0.003.

yses provided a sufficient power to detect the association between SNPs and lung cancer risk. In conclusion, we found four single variants in one-carbon metabolism genes and one MTHFR haplotype ‘‘ACCACC’’ may contribute to the risk of lung cancer. Using the MDR method to explore the gene—gene interactions on lung cancer risk, we have identified a best four-factor interaction model which could significantly predict lung cancer risk. Validation of these findings in larger studies with folate intake information is needed.

Table 5

Conflict of interest statement None.

Acknowledgements This work was supported in part by the China National Key Basic Research Program Grants 2002CB512902 (to D. Lu & H. Shen), 2002BA711A10 and 2004CB518605 (to W. Huang),

Results from MDR analysis

Number of factors considered

Best candidate model

Average crossvalidation consistency (%)

Average classification error (%)

Average prediction error (%)

1 2 3 4 5

rs3753584 rs2297967, rs4059731, rs1801133, rs1801133,

58 50 53 95 71

47.49 45.33 42.48 37.69 33.06

52.27 51.68 51.44 45.08a 49.77

a

rs202676 rs2273029, rs699517 rs4659731, rs2273029, rs699517 rs4659731, rs2273029, rs699517, rs2790

p < 0.001 based on 1000 permutations.

The association of polymorphisms in one-carbon metabolizing genes National Outstanding Youth Science Foundation of China 30425001 (to H. Shen), and National ‘‘211’’ Environmental Genomics Grant (to D. Lu).

References [1] Parkin DM, Bray F, Ferlay J, Pisani P. Global cancer statistics, 2002. CA Cancer J Clin 2005;55(2):74—108. [2] Shields PG. Molecular epidemiology of smoking and lung cancer. Oncogene 2002;21(45):6870—6. [3] Spitz MR, Wei Q, Dong Q, Amos CI, Wu X. Genetic susceptibility to lung cancer: the role of DNA damage and repair. Cancer Epidemiol Biomarkers Prev 2003;12(8):689—98. [4] Glade MJ. Food, nutrition, and the prevention of cancer: a global perspective. American Institute for Cancer Research/World Cancer Research Fund, American Institute for Cancer Research, 1997. Nutrition 1999;15(6):523—6. [5] Darby S, Whitley E, Doll R, Key T, Silcocks P. Diet, smoking and lung cancer: a case-control study of 1000 cases and 1500 controls in South-West England. Br J Cancer 2001;84(5):728—35. [6] Goldbohm RA, Voorrips LE. Epidemiology of nutrition and lung cancer. Nestle Nutr Workshop Ser Clin Perform Programme 2000;4:23—35 [discussion 35-27]. [7] Mannisto S, Smith-Warner SA, Spiegelman D, Albanes D, Anderson K, van den Brandt PA, et al. Dietary carotenoids and risk of lung cancer in a pooled analysis of seven cohort studies. Cancer Epidemiol Biomarkers Prev 2004;13(1):40—8. [8] Shen H, Wei Q, Pillow PC, Amos CI, Hong WK, Spitz MR. Dietary folate intake and lung cancer risk in former smokers: a case-control analysis. Cancer Epidemiol Biomarkers Prev 2003;12(10):980—6. [9] Stover PJ. Physiology of folate and vitamin B12 in health and disease. Nutr Rev 2004;62(6 Pt 2):S3—12 [discussion S13]. [10] Hubner RA, Muir KR, Liu JF, Sellick GS, Logan RF, Grainge M, et al. Folate metabolism polymorphisms influence risk of colorectal adenoma recurrence. Cancer Epidemiol Biomarkers Prev 2006;15(9):1607—13. [11] Koushik A, Kraft P, Fuchs CS, Hankinson SE, Willett WC, Giovannucci EL, et al. Nonsynonymous polymorphisms in genes in the one-carbon metabolism pathway and associations with colorectal cancer. Cancer Epidemiol Biomarkers Prev 2006;15(12):2408—17. [12] Chou YC, Wu MH, Yu JC, Lee MS, Yang T, Shih HL, et al. Genetic polymorphisms of the methylenetetrahydrofolate reductase gene, plasma folate levels and breast cancer susceptibility: a case-control study in Taiwan. Carcinogenesis 2006;27(11):2295—300. [13] Zhai X, Gao J, Hu Z, Tang J, Qin J, Wang S, et al. Polymorphisms in thymidylate synthase gene and susceptibility to breast cancer in a Chinese population: a case-control analysis. BMC Cancer 2006;6:138. [14] Grieu F, Powell B, Beilby J, Iacopetta B. Methylenetetrahydrofolate reductase and thymidylate synthase polymorphisms are not associated with breast cancer risk or phenotype. Anticancer Res 2004;24(5B):3215—9.

29

[15] Shen M, Rothman N, Berndt SI, He X, Yeager M, Welch R, et al. Polymorphisms in folate metabolic genes and lung cancer risk in Xuan Wei, China. Lung Cancer 2005;49(3):299— 309. [16] Hung RJ, Hashibe M, McKay J, Gaborieau V, SzeszeniaDabrowska N, Zaridze D, et al. Folate related genes and the risk of tobacco-related cancers in Central Europe. Carcinogenesis 2007. [17] Shi Q, Zhang Z, Li G, Pillow PC, Hernandez LM, Spitz MR, et al. Sex differences in risk of lung cancer associated with methylene-tetrahydrofolate reductase polymorphisms. Cancer Epidemiol Biomarkers Prev 2005;14(6):1477—84. [18] Hu Z, Jin G, Liu H, Sun W, Wang H, Shao M, et al. Genetic variants in MGMT and risk of lung cancer in Southeastern Chinese: a haplotype-based analysis. Hum Mutat 2007; (February). [19] Hu Z, Xu L, Shao M, Yuan J, Wang Y, Wang F, et al. Polymorphisms in the two helicases ERCC2/XPD and ERCC3/XPB of the transcription factor IIH complex and risk of lung cancer: a case-control analysis in a Chinese population. Cancer Epidemiol Biomarkers Prev 2006;15(7):1336—40. [20] Gonzalez JR, Armengol L, Sole X, Guino E, Mercader JM, Estivill X, et al. SNPassoc: an R package to perform whole genome association studies. Bioinformatics 2007;23(5):644—5. [21] Gabriel SB, Stephen SF, Nguyen H, Moore JM, Roy J, Blumenstiel B, et al. The structure of haplotype blocks in the human genome. Science 2002;296(5576):2225—9. [22] Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 2001;68(4):978—89. [23] Hahn LW, Ritchie MD, Moore JH. Multifactor dimensionality reduction software for detecting gene—gene and gene—environment interactions. Bioinformatics 2003;19(3):376—82. [24] Honecker F, Mayer F, Stoop H, Oosterhuis J, Koch S, Bokemeyer C, et al. Xeroderma pigmentosum group A protein and chemotherapy resistance in human germ cell tumors. Lab Invest 2003;83(10):1489—95. [25] Shen M, Berndt SI, Rothman N, DeMarini DM, Mumford JL, He X, et al. Polymorphisms in the DNA nucleotide excision repair genes and lung cancer risk in Xuan Wei, China. Int J Cancer 2005;116(5):768—73. [26] Shen H, Spitz MR, Wang LE, Hong WK, Wei Q. Polymorphisms of methylene-tetrahydrofolate reductase and risk of lung cancer: a case-control study. Cancer Epidemiol Biomarkers Prev 2001;10(4):397—401. [27] Hao L, Ma J, Stampfer MJ, Ren A, Tian Y, Tang Y, et al. Geographical, seasonal and gender differences in folate status among Chinese adults. J Nutr 2003;133(11):3630—5. [28] Lucock M. Folic acid: nutritional biochemistry, molecular biology, and role in disease processes. Mol Genet Metab 2000;71(1—2):121—38. [29] Gao X, Yao M, McCrory MA, Ma G, Li Y, Roberts SB, et al. Dietary pattern is associated with homocysteine and B vitamin status in an urban Chinese population. J Nutr 2003;133(11): 3636—42.