Genetic and intermediate phenotypic susceptibility markers of gastric cancer in Hispanic Americans: A case-control study
Hispanics are the largest nonwhite ethnic group in the US population, and they have higher incidence and mortality rates for gastric cancer (GC) than whites and Asians. Studies have identified several genetic susceptibility loci and intermediate phenotypic biomarkers for GC in whites and Asians. No studies have evaluated genetic susceptibility and intermediate phenotypic biomarkers in Hispanics.
In a case-control study of 132 Hispanic patients with GC (cases) and a control group of 125 Hispanics (controls), the authors evaluated the association of 5 single nucleotide polymorphisms (SNPs) that predispose whites and/or Asians to GC and of 2 intermediate phenotypic markers in peripheral blood leukocytes, ie, telomere length and mitochondrial DNA (mtDNA) copy number, with the GC risk.
The variant C allele of the reference SNP rs2294008 in the PSCA gene was associated with a significantly reduced risk of GC (per allele-adjusted odds ratio [aOR], 0.51; 95% confidence interval [CI], 0.33-0.77; P = .002). Leukocyte mtDNA copy numbers were significantly lower in GC cases (mean ± standard deviation, 0.91 ± 0.28) than in controls (1.29 ± 0.42; P < .001). When individuals were dichotomized into high and low mtDNA copy number groups based on the median mtDNA copy number value in the controls, those who had a low mtDNA copy number had a significantly increased risk of GC (aOR, 11.00; 95% CI, 4.79-25.23; P < .001) compared with those who had a high mtDNA copy number. Telomere length was not associated significantly with the risk of GC (aOR, 1.21; 95% CI, 0.65-2.27; P = .551).
Hispanics share certain genetic susceptibility loci and intermediate phenotypic GC biomarkers with whites and Asians and may also have distinct genetic susceptibility factors. Cancer 2014;120:3040–3048. © 2014 American Cancer Society.
Gastric cancer (GC) is the fourth most common malignancy and the second leading cause of cancer death worldwide. The incidence rate of GC varies geographically and ethnically. Approximately 70% of GCs in the world occur in developing countries, and the highest incidence and mortality is reported in Eastern Asian countries.[1, 2] Epidemiologic studies have demonstrated that a high intake of salt, tobacco smoking, and Helicobacter pylori infection increase the risk of GC.[3-5]
There have been many association studies assessing genetic susceptibility to GC. Genome-wide association studies (GWAS) have identified several single nucleotide polymorphisms (SNPs) as loci for susceptibility to GC in Asian and/or Caucasian populations, including SNPs in prostate stem cell antigen (PSCA) (8q24.2); mucin 1, cell surface associated (MUC1) (1q22); phospholipase C, epsilon 1 (PLCE1) (10q23); zinc finger and BTB domain containing 20 (ZBTB20) (3q13.31); and protein kinase, AMP-activated, alpha 1 catalytic subunit (PRKAA1) (5p13.1).[6-14] The first GWAS on GC was performed in Japan. Two linked reference SNPs, rs2976392 and rs2294008, in the PSCA gene on 8q24.2 were associated with diffuse-type GC in Japanese and Korean populations. This association has been replicated in other Asian and Caucasian populations.[6, 8-10] The same group later identified 2 SNPs, rs2070803 and rs4072037, in MUC1 that were associated significantly with diffuse-type GC risk. A GWAS of noncardiac GC in Chinese populations identified 2 susceptibility loci at 5p13.1 (PRKAA1, rs13361707) and 3q13.31 (ZBTB20, rs9841504). Another GWAS in Chinese populations demonstrated that rs2274223 (PLCE1, 10q23) was associated significantly with the risk of GC.
In addition to identifying genetic susceptibility loci, recent studies have also identified intermediate phenotypic biomarkers for GC, including telomere length and mitochondrial DNA (mtDNA) copy number in peripheral blood leukocytes, and both the leukocyte telomere length (LTL) and the mtDNA copy number have high genetic heritability.[15-17] Telomeres are nucleoprotein complexes at the ends of chromosomes and consist of short, repetitive sequences (TTAGGG in humans) and a set of specialized proteins.[18-20] Telomere shortening and telomerase activation are hallmarks of human cancers. LTL has a genetic heritability of 70% to 80% as estimated by classic twin studies and a meta-analysis.[15, 16] LTL has been evaluated as cancer susceptibility biomarker in many case-control and cohort studies.[21-24] Two retrospective case-control studies have suggested that short LTL is associated with increased GC risk in Chinese and whites[25, 26]; however, in a recent large, prospective study among whites, no significant association was observed. Mitochondria play important roles in cellular energy metabolism, proliferation, and apoptosis.[27, 28] Numerous reports have demonstrated that mtDNA is mutated and the mtDNA copy number is increased or decreased in human tumor tissues, depending on the specific tumor type.[29-37] A classic twin study estimated that the leukocyte mtDNA copy number has a heritability rate of 65%. Several recent studies have indicated that the leukocyte mtDNA copy number is associated with altered cancer risks in a cancer type-specific manner.[17, 22, 38, 39] In terms of GC, several studies have reported mtDNA mutation or depletion in tumor tissue[32, 36, 37]; and, in a recent prospective study in a Chinese population, a significant association was observed between a low leukocyte mtDNA copy number and an increased risk of GC in blood drawn within 2 years before cancer diagnosis.
Hispanics are the largest nonwhite ethnic group in the United States and account for approximately 17% of the US population. Although GC incidence and mortality rates are declining in the United States, the rate among Hispanics remains twice the rate among non-Hispanic whites. We were interested in identifying common and unique genetic factors for GC in Hispanics, Caucasians, and other ethnicities. To date, no studies have evaluated the association of SNPs, LTL, and mtDNA copy number with the risk of GC among Hispanics. In the current study, we used a pilot case-control study design to evaluate 5 GWAS-identified SNPs, LTL, and mtDNA copy number in relation to GC risk in Hispanics. To the best of our knowledge, this is the first cancer association study of GC in Hispanic Americans.
MATERIALS AND METHODS
Study Population and Epidemiologic Data
In this study, patients (n = 132) with pathologically confirmed gastric adenocarcinoma were recruited from The University of Texas MD Anderson Cancer Center from 1998 to 2011. All patients (cases) were newly diagnosed and previously untreated before enrollment. There were no age, sex, or cancer stage restrictions on recruitment. Healthy controls (n = 125) were recruited in collaboration with Kelsey-Seybold Clinic, the largest multispecialty health organization in the Houston metropolitan area. All cases and controls were Hispanics. Control participants without a history of cancer other than nonmelanoma skin cancer were matched to cases by age (±5 years) and sex and were recruited during the same period as the cases. The study was approved by the MD Anderson and Kelsey-Seybold institutional review boards. All study participants signed an informed consent before participation in the study. Epidemiologic data, including demographics, smoking and alcohol use history, family cancer history, body weight, height, and history of hypertension, were collected. At the end of the interview, a 40-mL blood sample was drawn and delivered to the laboratory for leukocyte isolation and DNA extraction. GC was staged according to the American Joint Committee on Cancer (AJCC) tumor-lymph node-metastasis (TNM) classification system (AJCC Cancer Staging Manual, seventh edition, 2010).
SNP Selection and Genotyping
We selected 5 GWAS-identified, GC-predisposing SNPs[7, 12, 14, 43] in PSCA (rs2294008), PLCE1 (rs2274223), PRKAA1 (rs13361707), ZBTB20 (rs9841504), and MUC1 (rs4072037). Genomic DNA was extracted from peripheral blood using QIAmp DNA extraction kit (Qiagen, Valencia, Calif) and was genotyped using Taqman genotyping assays with the ABI 7900HT Sequence Detection System (Life Technologies, Grand Island, NY) according to the manufacturer's protocol. Each run included negative controls (water) and 5% of samples as replicates. Concordance was 100%. All SNPs were in Hardy-Weinberg equilibrium (P > .05).
Determination of mtDNA Copy Number by Real-Time Quantitative Polymerase Chain Reaction
The relative mtDNA copy number was determined using a quantitative real-time polymerase chain reaction (PCR)-based method, as previously reported, with some modifications.[38, 44, 45] The PCRs were performed using the ABI 7900 Sequence Detection System. Briefly, 2 pairs of primers were designed and used in the 2 steps of relative quantification. One primer pair was used for the amplification of the mitochondrial subunit ND1 gene (MT-ND1). The primer sequences were: forward primer (ND1-F), 5′-CCCTAAAACCCGCCACATCT-3′; and reverse primer (ND1-R), 5′-GAGCGATGGTGAGAGCTAAGGT-3′. Another primer pair was used for the amplification of the single-copy nuclear gene human globulin (HGB). Those primer sequences were: forward primer (HGB-1), 5′-GTGCACCTGACTCCTGAGGAGA-3′; and reverse primer (HGB-2), 5′-CCTTGATACCAACCTGCCCAG-3′. During the first step, we determined the ratio of mtDNA copy number to HGB copy number for each sample using standard curves, which was proportional to the mtDNA copy number in each sample. The calibrator DNA was a genomic DNA sample from a healthy control participant. The ratio for each sample was then normalized to a calibrator DNA standardize between different runs. A standard curve of a diluted reference DNA, 1 calibrator DNA, and 1 negative control were included in each run. The PCR mixture (14 μL) consisted of 1 × SYBR Green Mastermix (Applied Biosystems; Foster City, Calif), 215 nmol/L ND1-R (or HGB-1) primer, 215 nmol/L ND1-F (or HGB-2) primer, and 4 ng of genomic DNA. The thermal cycling conditions were 1 cycle at 95°C for 10 minutes followed by 40 cycles at 95°C for 15 seconds and at 60°C (for mtDNA amplification) or 56°C (for HGB amplification) for 1 minute. For each standard curve, a reference DNA sample was serially diluted 1:2 to produce a 7-point standard curve between 0.3125 ng and 20 ng of DNA. Each sample was assayed in duplicate on a 384-well plate. The PCRs for mtDNA and HGB were performed on separate 384-well plates. The coefficient of determination (R2) value for each standard curve was ≥0.99. The acceptable standard deviations (SDs) were set at 0.25 for the cycle of threshold (Ct) values.
Telomere Length Assessment by Real-Time PCR
Relative LTL was measured using a modified version of a real-time PCR method.[46, 47] The ratio of the telomere repeat copy number to a single gene (HGB) copy number was determined for each sample using standard curves (the derived ratio was proportional to the overall telomere length). The PCR reaction mixture (14 μL) for the telomere amplification contained 1 × SYBR Green Master Mix, 200 nmol/L Tel-1 primer, 200 nmol/L Tel-2 primer, and 5 ng of genomic DNA. The PCR reaction mixture (14 μL) for HGB gene amplification contained 1 × SYBR Green Master Mix, 200 nmol/L HGB-1 primer, 200 nmol/L HGB-2 primer, and 5 ng of genomic DNA. The thermal cycling conditions for the telomere amplification were 95°C for 10 minutes followed by 40 cycles at 95°C for 15 seconds and at 56°C for 1 minute; and, for the HGB amplification, the conditions were 95°C for 10 minutes followed by 40 cycles at 95°C for 15 seconds and at 58°C for 1 minute. The telomere and HGB PCRs were done on separate 384-well plates. During each run, a calibrator DNA sample, negative controls (water), and a standard curve were included. For the standard curve, a reference DNA sample was serially diluted 1:2 to generate a 6-point standard curve between 0.625 ng and 20 ng of DNA in each reaction. The R2 correlation for each standard curve was ≥0.99. The acceptable SD was set at 0.30 for the Ct values. The laboratory personnel were blinded to case and control status for all the above experiments.
For statistical analyses, we used the STATA statistical software package (version 10.1; Stata Corporation, College Station, Tex). Differences in the distribution of host characteristics between patients and control participants were assessed using the Pearson chi-square test for categorical variables (sex and ethnicity) and the Student t test for continuous variables. Unconditional multivariate logistic regression was applied to estimate the odds ratio (OR) and 95% confidence interval (CI), adjusting for possible confounding by age, sex, smoking status, and body mass index (BMI). mtDNA content and LTL were categorized in 2 groups using the median values in the control group as cutoff points. We used Hosmer and Lemeshow goodness-of-fit tests to assess the fit of the model in logistic regression analysis, and all data fit the models. Hardy-Weinberg equilibrium for the genotypes was tested using a goodness-of-fit chi-square test. For each SNP, we evaluated its association with cancer risk using either a dominant model or an additive model. The Bonferroni-corrected significance level was .01 (for 5 SNPs). All statistical tests were 2-sided, and statistical significance was set at P < .05.
Characteristics of the Study Population
In total, 132 Hispanic patients with GC and 125 control participants were included in this study. The characteristics of the study population are summarized in Table 1. The cases and controls were matched on age (mean ± SD, 58.19 ± 13.58 years vs 55.45 ± 12.75 years; P = .097) and sex (P = .514). There was a significant difference in smoking status and BMI between cases and controls (P < .001 and P = .030, respectively). On the basis of TNM staging, approximately half of cases (50.76%) had early stage disease (stage I-II), and another half (49.24%) had late-stage disease (stage III-IV). Twenty-one patients (15.91%) had cardiac GC, and the remaining patients had noncardiac GC.
Table 1. Host and Clinical Characteristics of Patients With Gastric Cancer and Controls
|Age: Mean ± SD, y||58.19 ± 13.58||55.45 ± 12.75||.097|
|Age distribution, y|| || || |
|<40||11 (8.33)||13 (10.40)|| |
|40–59||58 (43.94)||65 (52)|| |
|60–70||34 (25.76)||26 (20.80)|| |
|≥70||29 (21.97)||21 (16.80)||.436|
|Sex|| || || |
|Men||74 (56.06)||65 (52)|| |
|Women||58 (43.94)||60 (48)||.514|
|Smoking status|| || || |
|Never||77 (58.33)||43 (34.40)|| |
|Ever||55 (41.67)||82 (65.60)||< .001|
|Pack-years: Mean ± SD||24.15 ± 36.43||20.90 ± 22.74||.530|
|BMI, kg/m2|| || || |
|<25||43 (32.58)||20 (18.52)|| |
|25–30||57 (43.18)||50 (46.30)|| |
|≥30||32 (24.24)||38 (35.19)||.030|
|Stagea|| || || |
|I-II||67 (50.76)|| || |
|III-IV||65 (49.24)|| || |
|Subsite|| || || |
|Cardiac||21 (15.91)|| || |
|Noncardiac||111 (84.09)|| || |
|Grade|| || || |
|1||0 (0)|| || |
|2||43 (32.58)|| || |
|3||89 (67.42)|| || |
GC Risk Associated With Individual SNPs
Among the 5 evaluated SNPs, after adjusting for age, sex, smoking status, and BMI, only rs2294008 in PSCA was associated significantly with the risk of GC. The variant C allele of rs2294008 was associated with a significantly reduced risk of GC (per allele-adjusted odds ratio [aOR], 0.51; 95% CI, 0.33-0.77; P = .002) (Table 2). Compared with individuals who had the wild-type TT genotype, the aORs for those who had 1 variant allele (TC genotype) and 2 variant alleles (CC genotype) were 0.58 (95% CI, 0.30-1.12; P = .106) and 0.25 (95% CI, 0.11-0.58; P = .001), respectively (Table 3). This association remained significant after adjusting for the other 4 SNPs and for the 2 biomarkers (LTL and mtDNA copy number; aOR, 0.55; 95% CI, 0.32-0.93; P = .027).The other 4 SNPs were not associated significantly with GC risk (Table 2). In stratified analyses based on smoking status, PSCA rs2294008 was associated significantly with GC susceptibility in both never-smokers (OR, 0.50; 95% CI, 0.27-0.93; P = .027) and ever-smokers (OR, 0.54; 95% CI, 0.30-0.98; P = .044) (Table 3). There was no significant interaction between this SNP and any variables (data not shown).
Table 2. Association Between 5 Single Nucleotide Polymorphisms and Gastric Cancer Risk
|No. with WW/WV/VV genotypes|| || || || || |
|MAF|| || || || || |
|Dominant model|| || || || || |
|OR (95% CI)a||1.05 (0.59–1.87)||0.72 (0.41–1.27)||0.88 (0.48–1.63)||0.81 (0.46–1.43)||0.47 (0.25–0.87)|
|Additive model|| || || || || |
|OR (95% CI)a||1.01 (0.67–1.52)||0.94 (0.59–1.50)||0.91 (0.54–1.53)||0.80 (0.50–1.26)||0.51 (0.33–0.77)|
Table 3. Logistic Regression Analysis of Significant Single Nucleotide Polymorphism rs2294008 Stratified by Smoking Status
|No. of cases (%)||49 (37.69)||64 (49.23)||17 (13.08)||—|
|No. of controls (%)||32 (25.60)||63 (50.40)||30 (24)||—|
|OR [95% CI]a||1.00 [Ref]||0.58 [0.30–1.12]||0.25 [0.11–0.58]||0.51 [0.33–0.77]|
|Smoking status|| || || || |
|Never-smokers|| || || || |
|No. of cases/controls||28/10||40/20||9/13||—|
|OR [95% CI]a||1.00 [Ref]||0.81 [0.31–2.08]||0.22 [0.07–0.76]||0.50 [0.27–0.93]|
|Ever-smokers|| || || || |
|No. of cases/controls||21/22||24/43||8/17||—|
|OR [95% CI]a||1.00 [Ref]||0.47 [0.19–1.19]||0.31 [0.09–1.02]||0.54 [0.30–0.98]|
GC Risk Associated With mtDNA Copy Number
The leukocyte mtDNA copy number was significantly lower in patients with GC (mean ± SD, 0.91 ± 0.28 mtDNA copies) than in controls (mean ± SD, 1.29 ± 0.42 mtDNA copies; P < .001) (Table 4). In stratified analysis according to sex (men and women), age (<60 years and ≥60 years), smoking status (never and ever), and BMI (<25 kg/m2, 25-30 kg/m2, and ≥30 kg/m2), mtDNA copy numbers were significantly lower in cases than in controls for all strata (Table 4).
Table 4. Mitochondrial DNA Copy Number and Leukocyte Telomere Length by Host Characteristics in All Participants
|mtDNA copy number|| || || || || |
|Overall||103||0.91 ± 0.28||124||1.29 ± 0.42||< .001|
|Sex|| || || || || |
|Men||58||0.86 ± 0.28||65||1.17 ± 0.34||< .001|
|Women||45||0.98 ± 0.27||59||1.42 ± 0.45||< .001|
|Age, y|| || || || || |
|<60||59||0.96 ± 0.26||77||1.33 ± 0.37||< .001|
|≥60||44||0.84 ± 0.30||47||1.21 ± 0.48||< .001|
|Smoking status|| || || || || |
|Never-smokers||60||0.94 ± 0.28||43||1.31 ± 0.52||< .001|
|Ever-smokers||43||0.87 ± 0.29||81||1.28 ± 0.36||< .001|
|BMI, kg/m2|| || || || || |
|<25||36||0.97 ± 0.29||20||1.48 ± 0.46||< .001|
|25–30||44||0.89 ± 0.27||50||1.14 ± 0.30||< .001|
|≥30||23||0.85 ± 0.29||37||1.45 ± 0.47||< .001|
|LTL|| || || || || |
|Overall||103||0.86 ± 0.48||124||0.95 ± 0.32||.092|
|Sex|| || || || || |
|Men||58||0.96 ± 0.47||65||0.97 ± 0.36||.966|
|Women||45||0.73 ± 0.47||59||0.93 ± 0.27||.006|
|Age, y|| || || || || |
|<60||59||0.97 ± 0.53||77||0.99 ± 0.32||.824|
|≥60||44||0.71 ± 0.37||47||0.89 ± 0.30||.012|
|Smoking status|| || || || || |
|Never-smokers||60||0.83 ± 0.47||43||0.93 ± 0.29||.187|
|Ever-smokers||43||0.91 ± 0.50||81||0.96 ± 0.33||.512|
|BMI, kg/m2|| || || || || |
|<25||36||0.98 ± 0.48||20||0.88 ± 0.25||.422|
|25–30||44||0.84 ± 0.50||50||0.98 ± 0.33||.103|
|≥30||23||0.73 ± 0.42||37||0.90 ± 0.33||.082|
Next, we performed an unconditional logistic regression analysis adjusting for age, sex, smoking status, and BMI to assess the association between mtDNA copy number and GC risk (Table 5). When individuals were dichotomized into high and low groups based on the median (50th percentile) mtDNA copy number value in the controls, individuals in the low mtDNA copy number group had a significantly increased risk of GC (aOR, 11.00; 95% CI, 4.79-25.23; P < .001) compared with those in the high mtDNA copy number group (Table 5). In stratified analyses based on smoking status, the mtDNA copy number retained a significant association with GC susceptibility in never-smokers and ever-smokers (P < .001) (Table 5). In an interaction analysis, we did not observe a significant interaction of mtDNA copy number with smoking status (P = .799).
Table 5. Risk of Gastric Cancer Associated With Mitochondrial DNA Copy Number and Leukocyte Telomere Length
|mtDNA copy number|| || || || || |
|High||13 (12.62)||62 (50)||1.00 [Ref]|| || |
|Low||90 (87.38)||62 (50)||11.00 [4.79–25.23]||< .001||.430|
|Smoking status|| || || || || |
|Never-smokers|| || || || || |
|High||9 (15)||22 (51.16)||1.00 [Ref]|| || |
|Low||51 (85)||21 (48.84)||14.99 [4.37–51.35]||< .001||.204|
|Ever-smokers|| || || || || |
|High||4 (9.30)||40 (49.38)||1.00 [Ref]|| || |
|Low||39 (90.70)||41 (50.62)||9.31 [2.67–32.45]||< .001||.559|
|LTL|| || || || || |
|Long||45 (43.69)||62 (50)||1.00 [Ref]|| || |
|Short||58 (56.31)||62 (50)||1.21 [0.65–2.27]||.551||.325|
|Smoking status|| || || || || |
|Never-smokers|| || || || || |
|Long||26 (43.33)||18 (41.86)||1.00 [Ref]|| || |
|Short||34 (56.67)||25 (58.14)||1.30 [0.49–3.44]||.594||.239|
|Ever-smokers|| || || || || |
|Long||19 (44.19)||44 (54.32)||1.00 [Ref]|| || |
|Short||24 (55.81)||37 (45.68)||1.05 [0.43–2.50]||.926||.262|
GC Risk Associated With Relative LTL
The LTL in GC cases (mean ± SD, 0.86 ± 0.48 LTL) and controls (mean ± SD, 0.95 ± 0.32 LTL) did not differ significantly (P = .092). In stratified analysis according to sex (men and women) and age (<60 years and ≥60 years), LTL was significantly shorter in cases than in controls among women and older individuals, which should be interpreted with caution because of the small numbers of individuals in each stratum (Table 4).
We performed unconditional logistic regression analysis adjusting for age, sex, smoking status, and BMI to assess the association between LTL and the risk of GC (Table 5). When individuals were dichotomized into long and short LTL groups based on the median (50th percentile) LTL value in the controls, the short LTL group was not associated significantly with the risk of GC (aOR, 1.21; 95% CI, 0.65-2.27; P = .551) compared with the long LTL group (Table 5).
In this case-control study, we investigated the association of 5 SNPs, leukocyte mtDNA copy number, and relative LTL with the risk of GC in a Hispanic population. To the best of our knowledge, this is the first molecular epidemiologic study evaluating the role of these factors in the risk of GC among Hispanics.
Among the 5 SNPs that we evaluated, only rs2294008 in the PSCA gene was significant in Hispanics. This SNP initially was identified as a susceptibility allele for diffuse-type GC in a Japanese population, and those results were replicated later in other Asian and Caucasian populations.[6, 10, 49, 50] In the current study, for the first time, we demonstrated that this SNP is also a susceptibility allele among Hispanics, and the direction of the association is consistent with that observed in other ethnic groups. PSCA (8q24.2) encodes a cell surface antigen. PSCA is expressed in epithelial cells, such as those in the prostate and stomach.[52-54] The rs2294008 is a missense SNP that alters the start codon of PSCA. It is notable that the risk allele (T) exists as the major allele in Japan, whereas it is a minor allele in other ethnic groups. In all of the ethnic groups examined to date, the T allele is the risk allele, and the C allele is protective. The biologic function of PSCA is still not clear. The other 4 SNPs—MUC1:rs4072037, PLCE1:rs2274223, ZBTB20:rs9841504, and PRKAA1:rs13361707—were associated with GC risk in Asian populations,[11-14, 43] and MUC1:rs4072037 was confirmed in Caucasians.[55, 56] In our current study, these 4 SNPs were not associated significantly with the risk of GC in Hispanics. These data indicate that Hispanics have common and distinct genetic susceptibility loci for GC. Because of the limited sample size, we could only test SNPs that were identified from other ethnicities. To identify Hispanic-specific SNPs, much larger sample sizes will be needed.
We observed that a lower mtDNA copy number was associated with a significantly increased risk of GC. Several previous studies have reported that mtDNA copy numbers were reduced significantly in GC tumor tissues compared with normal tissues,[32, 36, 37] consistent with our observation. In a previous, prospective, nested case-control study in a Chinese population, Liao et al reported no overall association between leukocyte mtDNA copy number and the risk of GC; however, a significant association was observed between low mtDNA copy number and the risk of GC in blood drawn within the 2 years before GC diagnosis. Our study is the first to demonstrate a significant association in Hispanics. Future studies will be needed to investigate the association of leukocyte mtDNA copy number with the risk of GC in different ethnicities. Mitochondria provide most of the adenosine triphosphate to cells through an oxidative phosphorylation process.[57, 58] Prior studies reported that the reduction or deletion of mtDNA results in a deficiency in oxidative phosphorylation, increases reactive oxygen species production, and oxidizes less pyruvate and nicotinamide adenine dinucleotide, leading to excessive lactate production; and it increases the expression of prosurvival proteins. These events may initiate and promote cancer development.
We did not observe a significant association between LTL and GC risk. Two previous retrospective case-control studies demonstrated that LTL was associated significantly with an increased risk of GC in Chinese and whites[25, 26]; however, a large prospective study failed to replicate the significant association in whites. Our result was consistent with that from the latter prospective study. It has been demonstrated that retrospective case-control studies tend to produce a spurious or inflated association between LTL and cancer risk if they are not well designed. In the current study, all patients with GC were newly diagnosed and previously untreated, which should limit the effect of disease status and treatment on LTL. Future studies will be needed to confirm the lack of a significant association between LTL and GC risk in Hispanics and other ethnicities.
There are a few limitations to our study. First, because of the low incidence of GC and the small overall population size and young age distribution of Hispanics, we could not obtain more Hispanic patients with GC, which limited our power to detect modest associations. Larger studies are needed to confirm our findings. Second, as a retrospective case-control study, reverse causation is a concern for intermediate phenotypic biomarkers (mtDNA copy number and LTL). We only included newly diagnosed patients before treatment. However, we cannot rule out the effect of disease state on these biomarkers. Future prospective studies are warranted to confirm our results. Third, because this was a retrospectively assembled population, some relevant confounders were not documented and could not be adjusted; for example, H. pylori infection is a major risk factor for GC, but we only had this information in cases, not in controls. Future studies need to include relevant confounders, particularly H. pylori. Finally, the ethnic information was based on self-report, and there may be potential misclassification. It would be ideal to use ancestry-informative markers to confirm the ethnicity of Hispanics.
In summary, to our knowledge, this is the first case-control study to examine the association of SNPs, mtDNA copy number, and LTL with the risk of GC among Hispanics. We observed that rs2294008 in the PSCA gene was associated with the risk of GC in Hispanics. A low leukocyte mtDNA copy number was associated significantly with an increased risk of GC, but short LTL was not. Our data suggest that GC in Hispanics has both common and distinct biology compared with GC in other ethnic groups. Further studies are needed to validate our findings and to identify Hispanic-specific GC susceptibility loci.
This study was supported by grant CA130821 from the National Cancer Institute, by an MD Anderson Cancer Center start-up fund to J.G., by an MD Anderson Cancer Center Research Trust to X.W., and by MD Anderson Cancer Center institutional support for the Center for Translational and Public Health Genomics.
CONFLICT OF INTEREST DISCLOSURES
The authors made no disclosures.