|Home | About | Journals | Submit | Contact Us | Français|
Common genetic variants significantly influence complex diseases such as primary biliary cirrhosis (PBC). We recently reported an association between PBC and a single nucleotide polymorphism (rs231725) of the immunoreceptor gene cytotoxic T-lymphocyte antigen 4 (CTLA4). We hypothesized that PBC risk attributed to this polymorphism might be increased by propensity to an overly robust inflammatory response. Thus, we examined its potential interaction with the commonly studied −308AG promoter polymorphism (rs1800629) of the tumor necrosis factor (TNF) gene for which the variant TNF2A allele causes increased TNF production. The polymorphisms were genotyped in 866 PBC patients and 761 controls from independent US and Canadian registries; the effects of individual single nucleotide polymorphisms (SNPs) and their interaction on PBC risk was assessed by logistic regression. The reported association of PBC with the CTLA4 “A/A” genotype was replicated in the Canadian cohort and significant for PBC risk in the combined data (odds ratio [OR], 1.68; P = 0.0005). TNF2A allele frequency was elevated in PBC patients, but only reached borderline significance using the combined data (OR, 1.21; P = 0.042). Analysis showed that TNF2A carriage was significantly increased in CTLA4 “A/A” PBC patients compared with CTLA4 “A/A” controls (39.7% versus 16.5%, P = 0.0004); no apparent increase of TNF2A carriage was noted in CTLA4 “A/G” or “G/G” individuals. Finally, interaction under a logistic model was highly significant, as TNF2A carriage in combination with the CTLA4 “A/A” genotype was present in 6.5% of PBC patients, compared with 1.7% of controls (OR, 3.98; P < 0.0001).
TNF2A amplifies the CTLA4 rs231725 “A/A” genotype risk for PBC. Although the mechanisms remain unclear, the premise that deficiency in T-cell regulation resulting in an increased risk of PBC is amplified by overexpression of an important proinflammatory cytokine provides a basis for future functional studies.
Primary biliary cirrhosis (PBC) is a rare autoimmune disease of the liver characterized by damage to and loss of the small to medium-sized intrahepatic bile ducts, leading to chronic cholestasis, development of fibrosis and cirrhosis, and ultimately end-stage liver disease.1 The initiating event appears to be loss of tolerance to a ubiquitously expressed subunit of the pyruvate dehydrogenase complex, often accompanied by the development of pyruvate dehydrogenase complex E2–specific anti-mitochondrial antibodies,2 which are detectable in some 90% of PBC patients.3 How this phenomenon is precipitated, and the mechanisms influencing the development of biliary-targeted autoimmunity and progression to clinically recognizable disease as a result, remain obscure.
Common genetic variations can result in altered gene function and significantly influence the risk of developing complex diseases such as PBC. However, the impact of individual polymorphisms is generally small, and they are present in only a fraction of the diseased population.4 The cumulative effects of such functional genetic variants, including the contribution of those that do not themselves demonstrate a significant association with disease, have the potential to appreciably modify disease risk. The identification of such interacting alleles is a daunting task but offers promise to provide additional clues to the often subtle or hidden processes underlying disease pathogenesis.
The numerous pathways responsible for maintaining self-tolerance are particularly attractive candidates for harboring genetic polymorphisms that could alter propensity toward autoimmune development.5 Cytotoxic T-lymphocyte antigen 4 (CTLA4) is a coregulatory immunoreceptor that has a broad dampening effect on T-cell responses.6–9 In this capacity, CTLA4 is thought to be a key regulator of self-tolerance, evidenced in part by CTLA4 knockout mice, which develop a severe lymphoproliferative disorder and demonstrate multi-organ autoimmunity.10,11 Genetic variants in CTLA4 have been associated with a number of autoimmune diseases, leading to its distinction as a general autoimmunity locus.12 However, despite great effort, no individual variant has been identified as the “smoking gun” for the CTLA4 genetic association with autoimmunity, and as a result the putative functional mechanism or mechanisms remain mysterious. A number of CTLA4 genetic association studies with PBC have been published, but with conflicting results.13–21 Our most recent investigation of CTLA4 genetic variation in PBC used a comprehensive, linkage disequilibrium –based tagging approach and identified a strong, novel association with a single nucleotide polymorphism (SNP) in the 3′ flanking region of the gene, rs231725, located outside of the previously investigated region.20 Interestingly, the minor “A” allele of this SNP was found to specifically tag the second most common CTLA4 haplotype, which contains a number of the previously identified and putatively functional22–25 autoimmune-associated CTLA4 variants. Moreover, a similar haplotype was also shown to be associated with PBC in a recent study from Canada that did not assess the rs231725 SNP.21 Although a functional explanation of the PBC association with rs231725 is currently lacking, this SNP remains a valid marker of the CTLA4 genetic influence on PBC risk.
While the mechanism of the CTLA4 association with PBC and other autoimmune disorders remains obscure, its low penetrance suggests that other factors contributing to autoimmune susceptibility may play a supporting interactive role in the development of disease. For instance, we previously reported a modest interaction between a putatively protective CTLA4 haplotype and a purported autoimmune risk allele of the programmed cell death-1 gene affecting the risk for and features of PBC.18 Identification of additional genetic variants interacting with CTLA4 to modify the risk of PBC will strengthen the evidence for CTLA4 involvement with disease and may help to unravel the complex mechanisms underlying the loss of self-tolerance and development of autoimmunity in PBC.
Tumor necrosis factor (TNF) alpha is an important cytokine with diverse and intricate biological function that is primarily focused on the promotion of a strong inflammatory response.26 In this regard, common genetic polymorphisms leading to increased expression of TNF may contribute to amplified and sustained inflammation, which could increase the likelihood for loss of self-tolerance and induction of autoimmunity, especially in those otherwise genetically predisposed. Although the autoimmune influence of such TNF variants is illustrated by weak associations with diseases such as systemic lupus erythematosus and rheumatoid arthritis,26 their effect on PBC development remains obscure, despite previous efforts.19,27–29 Moreover, examination of the contribution of TNF promoter polymorphisms in the context of other autoimmune genes remains largely unexplored. To this end, we hypothesized that carriage of a functional SNP located in the TNF promoter, commonly referred to as −308A/G (rs1800629), for which the minor “A” allele (hereafter referred to as TNF2A) has been shown to result in increased TNF expression,30–32 might augment the underlying genetic risk for PBC indicated by the offending genotype of CTLA4 rs231725. This study aimed to replicate the CTLA4 association with PBC in an independent cohort and to test our hypothesis by examining its potential interaction with the TNF2A allele.
The 866 PBC patients and 761 controls included in this study derive from two large previously described North American PBC registries, the Mayo Clinic PBC Genetic Epidemiology Registry and Biospecimen Repository33 and the Toronto Canadian PBC registry, which includes patients collected from multiple medical centers across Canada. 34 In all instances, PBC diagnosis was based on standard medical criteria, including evidence of persistent biochemical cholestasis (lasting greater than 6 months) without other known liver disease, compatible liver histopathology, and detectable anti-mitochondrial antibodies in serum. PBC patients with negative or unknown history of anti-mitochondrial antibodies and those of nonwhite race were excluded from this study to minimize potential genetic confounders between the groups. Demographic characteristics of the Mayo and Toronto patient and control groups are shown in Table 1.
Informed consent was obtained from all study participants. The registries and current study conform to the ethical guidelines of the 1975 Declaration of Helsinki, and were approved by the Institutional Review Board of the Mayo Clinic or by the local ethics committees of centers contributing to the Toronto cohort.
Genotyping of the Mayo specimens was performed by commercially available or custom-designed TaqMan allelic discrimination assays (Applied Biosciences, Carlsbad, CA) using an Applied Biosciences 7300 real-time polymerase chain reaction system. The Toronto specimens were genotyped using the Sequenom MassArray iPLEX genotyping platform (Sequenom, San Diego, CA).
Disease association with individual and interacting SNPs was analyzed by means of logistic regression to determine statistical significance, odds ratios (OR), and 95% confidence intervals (CI); multiple inheritance models were evaluated. Correction for multiple tests was not performed because of the selective nature and goals of this inquiry. Significance of comparisons between groups was assessed using chi-squared analysis on 2 × 2 contingency tables. All analyses were two-sided, and completed using SAS, version 9.13 (SAS, Cary, NC).
Genotype frequencies and significant findings of the single SNP association analyses are shown in Table 2. The CTLA4 rs231725 “A/A” genotype was found to be significantly associated with PBC (recessive model) in the Mayo (OR, 1.61; 95% CI, 1.06–2.46; P = 0.026), Toronto (OR, 1.75; 95% CI, 1.15–2.66; P = 0.007), and combined (OR, 1.68; 95% CI, 1.25–2.25; P = 0.0005) study sets, in agreement with our previous report.20 Moreover, rs231725 was associated with PBC under the additive model in the Toronto (OR, 1.29; 95% CI, 1.06–1.57; P = 0.01) and combined (OR, 1.22; 95% CI, 1.06–1.40; P = 0.006) data, but not significant in the Mayo group. However, the additive significance appears to be dependant on the increase in rs231725 “A/A” reflected in the recessive model findings, suggesting the allelic effect of the CTLA4 SNP on PBC risk to be weak at best. Frequency and carriage of the rs1800629 TNF2A allele was slightly elevated in PBC patients compared with controls in each study group. This difference was of borderline significance under the additive model in the combined data set (OR, 1.21; 95% CI, 1.01–1.45; P = 0.042) but was not significant in either of the individual groups. Both SNPs were found to be in Hardy-Weinberg equilibrium.
To begin assessing the potential interaction between the CTLA4 and TNF SNPs, carriage of the TNF2A allele within the various CTLA4 rs231725 genotype groups was determined for the combined Mayo-Toronto study set (Fig. 1). As is clearly seen, TNF2A carriage is significantly increased in the PBC patients with the rs231725 “A/A” genotype compared with the controls (39.7% PBC versus 16.5% controls, chi-squared = 12.73; P = 0.0004), whereas TNF2A carriage was quite similar between patients and controls with CTLA4 “A/G” and “G/G” genotypes. We further addressed this finding by examining the effect of the TNF rs1800629 SNP on risk of PBC in the three separate CTLA4 genotype groups using logistic regression. Not surprisingly, TNF2A was found to be significantly associated with PBC among CTLA4 “A/A” participants (dominant model OR, 3.34; 95% CI, 1.69–6.63; P = 0.0002; additive model OR, 3.24; 95% CI, 1.72–6.10; P = 0.0001), whereas the TNF SNP was not associated with PBC in either the CTLA4 “A/G” or “G/G” participants.
In addition to analyzing the effect of the TNF2A allele on risk of PBC within the CTLA4 genotype groups, we addressed the putative interaction in the context of the overall contribution of the CTLA4/TNF genotype combination, because this approach allows for determination of the extent to which the interaction contributes to PBC risk. Overall frequencies of CTLA4 rs231725 TNF rs1800629 genotypes and the significant results of statistical analyses are presented in Table 3. In the logistic model, interaction was defined as carriage of the TNF2A allele (in other words, “A/A” or “A/G” genotype) in combination with the implicated PBC risk CTLA4 “A/A” genotype. In the combined data set, this accounted for 6.5% of PBC patients and 1.7% of controls, and the results were highly significant (OR, 3.98; 95%CI, 2.16–7.33; P < 0.0001).
In this report, we replicate our previous finding of the association between PBC and rs231725, an SNP in the 3′ flanking region of CTLA4,20 using an independent case-control cohort from Canada. Furthermore, we demonstrate an interaction between the TNF and CTLA4 genes that increases the risk of PBC above that of the individual gene effects. Specifically, carriage of the TNF overexpressing TNF2A allele is significantly increased in PBC patients who have the PBC risk associated CTLA4 rs231725 “A/A” genotype compared with CTLA4 “A/A” controls. Susceptibility to autoimmune disorders is thought to be the result of interaction between multiple genetic factors regulating the threshold of autoreactivity.35 However, there remains little empirical evidence to back this assertion despite the recent profusion of genome-wide data for a number of these disorders, including PBC.34 Such a shortfall may attributable in part to lack of attempt to identify or report such interacting alleles among these large data sets, considering the extensive ongoing effort to validate individual SNP associations, most of which still elude functional explanation.36 Moreover, these interacting combinations are apt to be rare in the studied populations even when relatively common SNPs contribute the main effect, and thus very large populations would be needed to detect them at a genomewide significance level, even considering the potentially higher odds ratios. Until the time that meaningful genomewide approaches to the study of gene interaction effects on disease risk become more widely applied and technically feasible, candidate-gene–based inquiries are warranted and may help to inform future efforts.
The biological concept behind our current hypothesis is quite simple: among those individuals with a genetically encoded defect in the suppressive capabilities of CTLA4 (rs231725 risk), the inherited propensity to a more robust inflammatory response (TNF2A allele) might significantly increase the risk of PBC. Indeed, the strong interaction we report suggests this may be the case. However, the actual mechanisms governing the PBC risk-amplifying effect of this statistical gene–gene interaction remain unknown and will be difficult to determine. For instance, the extent of CTLA4 function in immunity is widespread, and its characterization remains incomplete even under normal circumstances.9 Although the suppressive effect of CTLA4 on T-cell activity is widely acknowledged to include its key role in determining the function of regulatory T-cells,37,38 the numbers of which are reduced in some PBC patients and their immediate family members,39 any suggestion that this is the primary mechanism behind the susceptibility to PBC would be speculative at best. Complicating matters is the significant linkage disequilibrium of the PBC-associated CTLA4 rs231725 SNP extending into the promoter region of the downstream gene inducible T-cell costimulator (ICOS),21 which itself plays an important role in T-cell activation,40,41 leaving open the possibility of variation in the activity of both CTLA4 and ICOS as the effector of the rs231725 PBC risk.
The TNF effect on the CTLA4-associated risk of PBC also may be more complex than our hypothesis envisions. In addition to its ability to induce inflammation, TNF also influences many other biological processes such as apoptosis, phagocytosis, and cell differentiation. 26 In this regard, overexpression of TNF in response to potential disease-prompting environmental stimuli (such as infectious agents or toxins) might elicit a number of detrimental effects on the biliary epithelia, leading to an excess of local antigen and the inappropriate presentation of pyruvate dehydrogenase complex E2. Perpetuation of such circumstances could then promote the loss of self-tolerance and development of PBC-specific autoimmunity; a process that would be largely dependant on the underlying CTLA4 genetic risk. Moreover, the TNF2A allele is correlated with particular extended human leukocyte antigen haplotypes that may harbor potentially functional variation affecting different immune mechanisms. All told, fine mapping of the PBC association with CTLA4, determination of human leukocyte antigen haplotypes associated with TNF2A carriage in context of the interaction, as well as elegant multigene functional studies will be necessary to definitively clarify the subtlety of the mechanisms at play.
In light of the publication of the first genomewide association study of PBC,34 our study may seem quite basic. We simply performed an interaction analysis between polymorphisms of two of the most highly studied genes involved with immunity. However, our finding serves to illustrate that meaningful insight can still be gained from the hypothesis-driven investigation of candidate genes in the genomewide era. In fact, neither of the SNPs we studied was present on the chip used for the Canadian PBC genomewide association study.34 Better understanding and cataloging of the complex genetic propensity to PBC development will provide us the background on which identification of the environmental instigators of disease is possible, potentially leading to refinements in disease classification and a more individualized approach to the prevention and treatment of PBC. In conclusion, we have demonstrated that carriage of the TNF2A allele appears to amplify the risk for PBC attributed to the previously identified and now replicated CTLA4 rs231725 “A/A” genotype. Although the precise mechanisms remain unclear, the simple premise that an increased risk of PBC attributable to deficient T-cell regulation is amplified by overexpression of an important proinflammatory cytokine is reasonable and provides a basis for future genetic and functional studies.
Supported by grants to K.N.L. from the NIH (K23 DK68290, RO3 DK078527, and RO1 DK80670); American Gastroenterological Association (AGA), and Foundation for Digestive Health and Nutrition (FDHN); American Liver Foundation; Palumbo Charitable Trust; and the Mayo Clinic College of Medicine.
Potential conflict of interest: Nothing to report.