Polymorphisms of Genes in the Lipid Metabolism Pathway and Risk of Biliary Tract Cancers and Stones: A Population-Based Case-Control Study in Shanghai, China

Biliary tract cancers, consisting of tumors of the gallbladder, extrahepatic bile ducts, and ampulla of Vater, are rare but highly fatal malignancies (1, 2). Cholesterol gallstones are the most important known risk factor for biliary tract cancers, especially gallbladder cancer (3, 4). Hyperlipidemia, characterized by increased total cholesterol, triglycerides, and low-density lipoprotein (LDL) and decreased high-density lipoprotein, has been associated with gallstones (5, 6) and with biliary tract cancer (2, 7).

Serum lipids are affected by both genetic and lifestyle factors, such as diet, body weight, and physical activity. In addition to a familial tendency, hyperlipidemia has been associated with several genetic polymorphisms in the lipid metabolism pathway (8, 9). For example, variants of the LDLR gene encoding the LDL receptor, which regulates the uptake of LDL, have been linked to familial hyperlipidemia (8, 9). Variants of the APOB and APOE genes, which encode the major carrier and binding proteins of LDL, apolipoprotein B, and apolipoprotein E, have been associated with higher serum levels of total cholesterol, LDL, apolipoprotein B, and lower high-density lipoprotein (10, 11). Also, variants of the LPL gene, which encodes the lipoprotein lipase enzyme, have been associated with higher serum LDL levels (12), and variants of ALOX5, which is involved in the synthesis of inflammatory lipid mediators, have been linked to cancer and heart disease (13, 14).

The role of genetic factors in biliary tract cancers and stones has been suggested by reports of ethnic and familial predisposition, but the genetic pathways are unclear (1, 15). Although some gene variants in the lipid metabolism pathway have been linked to gallstones (16, 17), few studies have examined their role in biliary tract cancers. In this population-based case-control study conducted in Shanghai, China, we examined the association of 12 polymorphic variants of five genes (APOB, APOE, LDLR, LPL, and ALOX5) in the lipid metabolism pathway with the risks of biliary tract stones and cancers and also assessed whether the variants were related with biliary tract cancers through their association with gallstones.

Details of the study methods have been reported elsewhere (4, 15, 18). Briefly, cancer cases were identified by a rapid reporting system established by the Shanghai Cancer Institute and 42 collaborating hospitals in Shanghai. Through this system, we identified >95% of all incident biliary tract cancer cases (International Classification of Diseases, Ninth Edition code 156) diagnosed among Shanghai residents between June 1997 and May 2001. A total of 627 incident biliary tract cancer cases (368 gallbladder, 191 extrahepatic bile duct, and 68 ampulla of Vater cases), ages between 35 and 74 years, were included in this study. To evaluate the risk of biliary stones independently of biliary tract cancer, 1,037 patients with biliary stones (774 gallstones and 263 extrahepatic bile duct stones) without a history of cancer were included and were frequency matched to the cancer cases on age (5-year groups) gender, and diagnosing hospital. Population controls, without a history of biliary tract cancer, were randomly selected from the Shanghai Resident Registry that includes records of ∼6 million Shanghai residents. We included a total of 959 population controls who were frequency matched to cancer cases on age (5-year groups) and gender. Participation rates among eligible cancer patients and control subjects were 95% and 82%, respectively.

Biliary tract cancer and biliary stone diagnoses were confirmed by a panel of clinicians, ultrasonographers, and pathologists. Seventy percent of biliary tract cancer cases were confirmed by histopathologic assessment, whereas the remaining 30% of cases, for whom we did not have histopathologic material, were confirmed using medical records, surgical reports, and imaging data, including magnetic resonance imaging, endoscopic retrograde cholangiopancreatography, or computed tomography. Biliary stone cases were confirmed by review of abdominal ultrasound, endoscopic retrograde cholangiopancreatography films, medical records, and surgical records or by pathologic material for those who underwent a cholecystectomy.

Gallstone status was assessed in nearly all biliary tract cancer cases and population controls. Among cancer cases, gallstones were identified by self-reported history, surgical reports, or imaging results from magnetic resonance imaging, endoscopic retrograde cholangiopancreatography, computed tomography, or ultrasound. Among population controls, gallstones were identified by self-reported history or by abdominal ultrasound among those who gave consent for the procedure (85% of all population controls).

Information on demographic characteristics, medical histories, and lifestyle factors was obtained through in-person interviews conducted by trained interviewers using a structured questionnaire. Cases were interviewed within 3 weeks of diagnosis. Weight and height were measured at the time of interview. The response rate for interviews was >95% among cases and 82% among controls. Five percent of the study subjects were randomly selected for reinterview 3 months after the initial interview to assess the reproducibility of responses. The concordance of responses to key questions between the original and follow-up interviews was >90%.

Five candidate genes were selected for their role in lipid metabolism, their effects on serum lipid levels, and their potential effects on biliary disease pathogenesis. Single nucleotide polymorphisms (SNPs) were selected based on their putative functional significance, having a variant allele frequency of at least 5% among Asians as reported by National Cancer Institute's SNP500Cancer Database (http://snp500cancer.nci.nih.gov; ref. 19) and a validated assay at the National Cancer Institute Core Genotyping Facility (http://cgf.nci.nih.gov/home.cfm). The following 12 SNPs in five genes were selected: APOB [EX26-3573T>C (rs676210), IVS23-79T>C (rs673548), IVS6+360C>T (rs520354), EX4+56C>T (rs1367117)], APOE [IVS1+69C>G (rs440446)], LDLR [IVS9-30C>T (rs1003723), EX10+55G>A (rs5930), EX15-80G>A (rs5927), IVS17-42A>G (rs6413504), EX18+88G>A (rs14158)], LPL [IVS5-540C>T (rs263)], and ALOX5 [IVS3+100G>A (rs2029253)]. In total, we included seven SNPs in introns and five in exons (two nonsynonymous, two synonymous, and one noncoding 3′-untranslated region). Overnight fasting blood samples were collected from all subjects who gave consent (>80%). Genomic DNA was extracted from buffy coat by the standard phenol chloroform method at the National Cancer Institute laboratory. Genotyping was conducted at the Core Genotyping Facility using the TaqMan assay (Applied Biosystems; http://snp500cancer.nci.nih.gov). The genotyping failure rate among all samples was <2%. The quality and potential misclassification of the genotyping results were assessed by evaluating 20 duplicate DNA samples from four quality-control subjects (80 total samples) that were randomly placed within the same reaction plates used for study subjects. The concordance between duplicate samples was >99%.

The final analysis included subjects who completed the interview and for whom we had DNA, which totaled 406 incident biliary tract cancer cases (235 gallbladder, 125 extrahepatic bile duct, and 46 ampulla of Vater), 880 biliary stone cases (663 gallstone and 217 bile duct), and 779 healthy controls. To make the appropriate case-control comparisons, for all analyses, gallbladder cancer cases were compared with controls without a history of cholecystectomy (n = 730), bile duct cancer cases and ampulla of Vater cancer cases were compared with all controls (n = 779), and biliary stone cases were compared with controls without biliary stones (n = 586). Distributions of selected characteristics [gender, age group, education status, cigarette smoking status, alcohol drinking status, body mass index (BMI), hypertension, diabetes, and gallstone status (among cancer cases)] were evaluated among cases and controls; characteristics with statistically significant different distributions between cases and controls (Fisher's exact P < 0.05) were examined for their associations with SNPs among controls.

The associations between SNPs and biliary tract stone and cancer risk were estimated by odds ratios (OR) and 95% confidence intervals (95% CI) using unconditional logistic regression. Risk estimates were calculated for the heterozygous and homozygous variant genotypes relative to the common homozygous genotype, as well as for the presence and absence of the variant allele, using indicator variables in the regression model. An initial model was adjusted for age, and additional models were further adjusted for gender, BMI, waist-to-hip ratio (WHR), cigarette smoking, alcohol drinking, hypertension, diabetes, and gallstone status (cancer risk only) to evaluate potential confounding by these factors. Associations between SNPs and biliary stones and cancers were assessed using the linear test of trend for the number of copies of the variant allele (0, 1, 2) assuming an additive model, or with the Wald's χ² test for the presence or absence of the variant allele (0, 1). In addition, the likelihood ratio test was used to formally test for multiplicative interactions between the above-mentioned characteristics and the SNPs on stone and cancer risk. Although gallstones may be an intermediate step in the causal pathway between genetics and biliary tract cancer, it is also possible that genetic variants may affect cancer risk independent of stones, especially for bile duct and ampulla of Vater cancers, which have a much lower prevalence of stones than gallbladder cancer. Thus, in subsequent analysis, we examined the association between SNPs and biliary tract cancer after adjustment for biliary stones.

To assess the overall gene effects on biliary tract cancer and stone risk, the Simes global test was used to calculate a summary P value for each gene for which we examined multiple SNPs (APOB and LDLR). The Simes global test uses the P_trend values of each SNP and accounts for multiple SNP comparisons within each gene by controlling for the gene-wide type I error rate (20, 21). Similarly, the Simes global test was used to obtain an overall pathway P value adjusting for all 12 SNPs and controlling for the family-wise error rate.

We also examined the associations between the haplotypes of the APOB and LDLR genes and the risks of biliary stones and cancers. Among population controls, linkage disequilibrium between markers was assessed by calculating pairwise Lewontin's D′ and r² using Haploview version 3.11 (22). Haplotypes were reconstructed for cases and controls separately from genotype data using Haplostats in R version 2.0.1 (23), which uses an expectation maximization algorithm to calculate maximum likelihood estimates of haplotype frequencies while taking into account phase ambiguity (24). Associations between haplotypes (>1% frequency) and the risks of biliary tract stones and cancer were evaluated by computing OR and 95% CI, using Haplostats, assuming an additive model, and using the most common haplotype as the referent category. Global differences in haplotype frequencies between cases and controls were assessed for each gene using the Score test in Haplostats (23).

Selected characteristics of the study subjects are shown in Table 1. The frequency of selected characteristics did not differ considerably between the three subgroups of controls (all controls, controls without cholecystectomy, and controls without biliary stones). Biliary stone cases were more likely to be women, younger, overweight, and diabetic, but less likely to drink alcohol or have hypertension, compared with controls without stones. Gallbladder cancer cases were predominantly women (72.3%), whereas bile duct cancer cases were more common in men (60%). In general, biliary cancer cases were more likely to have a higher WHR (>0.92) and gallstones than controls. Bile duct and ampulla of Vater cancer cases were more likely to ever smoke cigarettes and drink alcohol but less likely to have hypertension, whereas gallbladder cancer cases were more likely to be obese (41.3%) and diabetic (13.2%) compared with controls.

The risks of biliary stones and cancers in relation to the 12 SNPs in the lipid metabolism pathway are shown in Table 2. Among population controls, the minor allele frequencies of the 12 SNPs ranged from 5% to 39%, and the genotype frequencies showed no deviation from the expected Hardy-Weinberg equilibrium proportions (P > 0.05). Of the 12 SNPs examined, only the APOE IVS1+69C>G marker was significantly associated with biliary stones. Specifically, carriers of the G allele had a 1.3-fold risk of biliary stones (95% CI, 1.03-1.60; P_trend = 0.02) compared with those having the CC genotype. Variants in the APOE, APOB, and LDLR genes were associated with either bile duct or ampulla of Vater cancers, whereas none of the markers examined was significantly associated with gallbladder cancer. Carriers of the G allele of the APOE IVS1+69C>G marker had a 2.0-fold risk of bile duct cancer (95% CI, 1.32-3.11; P_trend = 0.003) and a 2.5-fold risk of ampulla of Vater cancer (95% CI, 1.21-5.08; P_trend = 0.002) compared with those having the CC genotype. We also found that carriers of the T allele of the APOB IVS6+360C>T marker and the T allele of the LDLR IVS9-30C>T marker each had a 1.5-fold risk of bile duct cancer compared with those having the CC genotype of each marker. In addition, carriers of the GA genotype of the LDLR EX18+88G>A marker had a 2.7-fold risk of ampulla of Vater cancer relative to those having the GG genotype. The associations between the APOE, APOB, and LDLR gene variants with stone and cancer risks were independent of gender, BMI, WHR, hypertension, diabetes, cigarette smoking, and alcohol drinking, whereas the associations with cancer were independent of gallstones.

Because of gender differences in the incidence of biliary stones and biliary tract cancers, we evaluated the gender-specific effects of the APOE, APOB, and LDLR markers for which we found significant main effects (Table 3). The effects of the APOE IVS1+69C>G marker were limited to men, with male carriers of the G allele having a 1.7-fold risk of biliary stones (95% CI, 1.19-2.36), 1.8-fold risk of gallbladder cancer (95% CI, 1.04-3.26), a 3.7-fold risk of bile duct cancer (95% CI, 1.99-6.96), and a 4.1-fold risk of ampulla of Vater cancer (95% CI, 1.36-12.38). A statistically significant interaction between gender and the APOE IVS1+69C>G marker was seen for biliary stones (P_interaction = 0.03) and bile duct cancer (P_interaction = 0.01). The effect of the APOB IVS6+360C>T marker was also limited to men, with male carriers of the T allele having a 2-fold risk of bile duct cancer (95% CI, 1.21-3.40), although the interaction between gender and the APOB IVS6+360C>T marker was not statistically significant (P_interaction = 0.09). The effects of these two SNPs among men were independent of BMI, WHR, hypertension, diabetes, cigarette smoking, alcohol drinking, and gallstones (for biliary cancer). No gender differences were seen in the association between the LDLR markers, IVS9-30C>T and EX18+88G>A, and the risks of biliary stones or cancers.

Using the Simes test to adjust for multiple SNP comparisons within the APOB and LDLR genes, we found that bile duct cancer had a borderline association with APOB (P_Simes = 0.08) and a nonsignificant association with LDLR (P_Simes = 0.20). Also, adjusting for all 12 SNPs, the combined effect of these gene variants was significantly associated with bile duct cancer (P_Simes = 0.04). Among men, but not women, the associations of bile duct cancer with the APOB gene (P_Simes = 0.01) and all variants examined in the lipid metabolism pathway (P_Simes = 0.001) remained significant.

Of the four APOB SNPs examined in the study, strong linkage disequilibrium was present among the APOB EX26-3573T>C and IVS23-79T>C markers (D′ = 1.0) and the APOB IVS6+360C>T and EX4+56C>T markers (D′ = 0.81). Thus, we inferred common haplotypes separately for these two sets of APOB SNPs. Two major haplotypes (frequency ≥ 1.0%) were inferred from the APOB EX26-3573T>C and IVS23-79T>C markers (TT = 74.8%, CC = 24.7%) and three from the APOB IVS6+360C>T and EX4+56C>T markers (CC = 80.3%, TT = 11.8%, and TC = 5.8%; Table 4). Consistent with our single marker results, the APOB TT haplotype, which contains the risk-conferring alleles of the IVS6+360C>T and EX4+56C>T markers, was associated with a 1.6-fold risk of bile duct cancer (95% CI, 1.07-2.33) compared with the most frequent CC haplotype. The other haplotype was associated with a significant risk of ampulla of Vater cancer (OR, 3.10; 95% CI, 1.12-8.61); however, this estimate was based on very small number of cases (n = 5). Although the Omnibus score test was significant for the APOB (EX26-3573T>C and IVS23-79T>C) and LDLR haplotypes, these P values may reflect the small numbers of the uncommon haplotypes because none of the common haplotypes was associated with biliary tract cancers.

In this population-based study, variants in lipid metabolism genes, including APOE, APOB, and LDLR, were associated with excess risks of biliary stones and cancers, although most of the effects were limited to men. The APOB TT haplotype, containing the risk-conferring alleles of the APOB IVS6+360C>T and EX4+56C>T markers, was associated with an excess risk of bile duct cancer, supporting further a relationship between APOB variants and bile duct cancer. These results, although needing confirmation, suggest that polymorphic variants in genes involved in lipid metabolism may play a role in the etiology of biliary stones and biliary tract cancers.

Our finding of an association between the APOE IVS1+69 marker and biliary stones corroborates previous studies relating gallstones to APOE gene polymorphisms. Several population-based studies of APOE have reported higher risks of gallstones among carriers of the e4 allele, a variant with a single amino acid change, compared with non-e4 carriers (25, 26). The APOE e4 allele has been associated with high cholesterol content in gallstones, faster crystallization, and frequent stone recurrence (26). Another study reported an excess risk of gallstones mainly among APOE e2 allele carriers who also had lower serum LDL levels (27). Because apolipoprotein E is a major component of lipoproteins and plays a critical role in the transport of cholesterol to the liver, our findings, together with previous studies, suggest that the APOE gene may predispose to biliary stones through effects on serum lipid levels (28).

The observed effects of the APOE IVS1+69 marker on bile duct and ampulla of Vater cancer risks were independent of gallstones, suggesting that mechanisms in addition to stones may be involved in biliary carcinogenesis. Apart from its role in lipid metabolism, APOE has important effects on the inflammatory response. Specifically, serum apolipoprotein E acts as a modulator of proinflammatory cytokines, including tumor necrosis factor-α and interleukin-1β (29, 30), which are expressed in biliary tract tumor tissue (31, 32). In addition, the inhibitory effect of apolipoprotein E on endothelial and tumor cell proliferation suggests its possible involvement in angiogenesis, tumor cell growth, and metastasis (33, 34). Although the APOE gene has not been investigated previously in biliary tract cancer, it has been related to increased risk of other malignancies, including breast, colon/rectum, and prostate (35-37).

In our study, the effects of the APOE IVS1+69C>G marker on biliary stones and cancer were limited to men, suggesting a gender-specific role of APOE. These findings are noteworthy because previous studies have noted stronger positive associations between the APOE e4 allele and triglyceride levels among men than women (38, 39), raising the possibility that the metabolic effect of APOE may contribute to the higher incidence of bile duct and ampullary cancers reported among men than women (1).

Of the four APOB SNPs examined, the APOB IVS6+360C>T marker was associated with an excess risk of bile duct cancer, independently of gallstones and the APOE IVS1+69C>G marker. Supporting this observation, we also found a relation between bile duct cancer and the APOB TT haplotype containing the at-risk allele of the APOB IVS6+360C>T marker. The mechanism by which APOB confers risk to bile duct cancer through a pathway independent of gallstones is unclear, but it may be linked to its relationship with serum apolipoprotein B, the major carrier and binding protein of LDL (40, 41). Variants of APOB have been linked to higher levels of serum apolipoprotein B (27, 42), which are correlated with higher levels of serum LDL. The resulting elevation in oxidized LDL may trigger an increase in reactive oxygen species (43, 44), DNA damage, activation of proto-oncogenes, or inactivation of tumor suppressor genes (45), all of which may play a role in biliary tract carcinogenesis (46).

Independent of gallstones and the APOE and APOB markers, two of the five LDLR markers, LDLR IVS9-30C>T and LDLR EX18+88G>A, were associated with biliary cancers. The mechanisms are unclear but may relate to the role of the LDL receptor in the uptake of LDL by the liver and extrahepatic tissue. Decreased activity of the LDL receptor has been associated with increased circulating levels of LDL, triglycerides, and total cholesterol (47) as well as proinflammatory cytokines, including tumor necrosis factor-α, interleukin-1, and interleukin-6, which are expressed in biliary tract cancer tissue (31, 32).

Our observation that the effect of lipid metabolism genes varied by biliary cancer subsite is consistent with the etiologic heterogeneity suggested by subsite differences in the epidemiologic (1, 2, 48) and molecular characteristics of biliary tumors (46, 49, 50). In our study, the effects of lipid metabolism genes on biliary tumors were most consistently associated with bile duct cancer despite the fact that gallstones are most closely linked to gallbladder cancer. Reasons for this apparent anomaly are unclear, but it is possible that in addition to gallstones and lipid metabolism the etiology of gallbladder cancer involves several major risk factors, including obesity and reproductive variables, which may have obscured detection of a modest genetic effect.

As with most epidemiologic studies measuring the effect of many candidate genes, false-positive findings due to multiple comparisons is a concern. To minimize this potential bias, we implemented rigorous procedures for statistical adjustment. For each gene, we used the Simes global test to adjust for multiple SNPs and found that the associations of APOB and LDLR with bile duct cancer persisted. The significant association between the APOB TT haplotype and bile duct cancer also persisted after adjustments for multiple comparisons. Although the Simes P value obtained for all 12 SNPs suggests an effect of lipid metabolism genes on bile duct cancer, it is possible that the observed associations may be due to linkage disequilibrium with causative SNPs that were not evaluated in this study, particularly because the functional significance of the SNPs examined is unknown and because the three SNPs with significant associations were in noncoding regions. For example, moderate to high linkage disequilibrium has been reported previously between APOE IVS1+69C>G and the markers responsible for the e2 allele (51, 52), suggesting that the observed effects of APOE IVS1+69C>G may be due to the e2 allele. Nevertheless, the totality of our data suggests that APOE, APOB, and LDLR genes and possibly other genes in the lipid metabolism pathway may play an important etiologic role in biliary tract cancers, especially of the bile duct.

Several strengths of this study should be noted. (a) Selection bias was minimal due to the population-based design and very high case ascertainment (>95%) and high response rates (>85%). (b) Misclassification of cancer and stone cases was also minimal due to the detailed review of pathology and clinical data. (c) The nearly complete assessment of gallstone status among cancer cases and controls made it possible to evaluate cancer risk while controlling for gallstones. (d) Misclassification of genotypes was unlikely given the good quality and purity of the extracted DNA, the 99% concordance between duplicate samples, and the 99% reproducibility of genotyping results among quality-control samples. (e) We were able to adjust for multiple comparisons and provide confirmation of our single-locus results.

However, our study had some limitations. (a) We may have missed the effect of some important markers, because gene coverage was limited and SNP selection was not based on complete sequencing data. (b) The functional effect of the SNPs examined was unknown and may actually represent markers for other functional SNPs that are in linkage disequilibrium. (c) Although our study of biliary tract cancer is the largest to date, some small sample sizes limited the statistical power to evaluate certain main effects, especially of ampullary cancer, as well as gene-gene and gene-environment interactions. (d) Our findings may have limited generalizability due to the homogeneous Chinese study population, although an effect of population stratification would be minimal.

In conclusion, this population-based study in Shanghai showed that, among men, APOE was associated with the risk of biliary stones and biliary tract cancers, whereas APOB was associated with bile duct cancer. In addition, LDLR was related to the risk of bile duct cancer in both sexes. Further studies with greater statistical power and broader coverage of genes in the lipid metabolism pathway and other pathways are needed to identify the causal gene variants and mechanisms involved in the formation of biliary stones and cancers.

We thank the collaborating surgeons and pathologists in Shanghai for assistance in patient recruitment and pathology review; Chia-Rong Cheng, Lu Sun, and Kai Wu (Shanghai Cancer Institute) for coordinating data and specimen collection; and Shelley Niwa (Westat) for support with study and data management.