|Home | About | Journals | Submit | Contact Us | Français|
Individuals with mixed dyslipidemia, including high triglycerides (TGs) and low high density lipoprotein cholesterol (HDL-C), have increased risk for coronary events. We examined the effect of rare genetic variants in the APOA5 gene region on plasma HDL-C, apolipoprotein A-I (apoA-I), and TG response to fenofibric acid monotherapy and in combination with statins. The APOA5 gene region was sequenced in 1,612 individuals with mixed dyslipidemia in a randomized trial of fenofibric acid alone and in combination with statins. Student's t-test and rare variant burden tests were used to examine plasma HDL-C, apoA-I, and TG response. Rare APOA5 promoter region variants were associated with decreased HDL-C and apoA-I levels in response to fenofibric acid therapy; rare missense variants were associated with increased TG response to combination therapy. Further study is needed to examine the effect of these rare variants on coronary outcomes in this population in response to fenofibric acid monotherapy or combined with statins
Mixed dyslipidemia, which occurs frequently in individuals with insulin resistance, such as those with metabolic syndrome, is characterized by elevated triglycerides (TGs) and decreased high density lipoprotein cholesterol (HDL-C). Genome-wide association studies have demonstrated associations between common genetic variants and interindividual differences in serum TGs and HDL-C (1, 2). Rarer genetic variants, with frequency of less than 1%, discovered by DNA sequencing show a larger effect size on the respective quantitative traits (3, 4). An important potential application of common and rare variants is to improve prediction of individualized response to therapy and personalize therapy based on that information.
Fibrates, including fenofibric acid, increase HDL-C and apolipoprotein A-I (apoA-I) and decrease TGs in individuals with mixed dyslipidemia who are at increased risk for coronary heart disease (5–8). Fenofibric acid is a peroxisome proliferator-activated receptor-α (PPAR-α) agonist, and as such, its effect can be influenced by common and rare variants in genes involved in the PPAR-α pathway.
ApoA5 is an important component of TG lipolysis while inhibiting production of very low density lipoprotein (LDL). Individuals with mutations in the APOA5 gene were found to have high TG. Common single nucleotide polymorphisms (SNPs) in the APOA5 gene region were found to be associated with TG and HDL-C levels (9).
In recent studies, SNPs in the APOA5 gene region were shown to be significantly associated with both HDL-C and TG response to fibrate monotherapy or in combination with statins (10–13). In this study, we examined the association between rare genetic variants in the APOA5 gene region and change in HDL-C, apoA-I, and TG levels in response to FA alone and in combination with statins. We have included the introns, 3′ untranslated region (UTR), and promoter region of APOA5 as possible regulatory regions that may affect response to PPAR-α agonists. We hypothesized that rare genetic variants affect the response of HDL-C, apoA-I, and TGs to therapy with fenofibric acid in individuals with mixed dyslipidemia.
Our study population included European-American participants from three concurrent prospective randomized double-blind clinical trials that examined the efficacy of fenofibric acid. A detailed description of the study design has been published previously (14). Individuals with TGs ≥150 mg/dl, HDL-C <40 mg/dl in men or <50 mg/dl in women, and low density lipoprotein cholesterol (LDL-C) ≥130 mg/dl were eligible and were randomized into six groups for each trial: fenofibric acid monotherapy, lower-dose statin monotherapy, moderate-dose statin monotherapy, higher-dose statin monotherapy, FA + lower-dose statin, and Fenofibric acid + higher-dose statin. Fenofibric acid was given in a 135 mg dose in the fenofibric acid only and statin + fenofibric acid groups. Each study used a different statin: rosuvastatin, atorvastatin, or simvastatin. In all studies, participants had a 6 week washout period with no lipid-modifying therapy followed by a 12 week treatment period. Lipid measurements were obtained at the beginning and end of the treatment period. To enhance statistical power, treatment groups of the original studies were collapsed into three major therapy groups: fenofibric acid monotherapy (n = 284), statin monotherapy (n = 753), and fenofibric acid + statin combination therapy (n = 575). The distribution of statins in the fenofibric acid + statin and statin alone groups is detailed in supplementary Table I.
The study was approved by the Institutional Review Board of Baylor College of Medicine, and informed consent was obtained by Abbott Laboratories.
The Sanger method was used to sequence the APOA5 gene including exons, introns, and promoter regions. Bidirectional sequencing of the APOA5 gene was performed at the Human Genome Sequencing Center at Baylor College of Medicine using intron-based exon-specific primers. Polymerase chain reactions (PCRs) were performed in 8 μl containing 10 ng of genomic DNA, 0.4 μM oligonucleotide primers, and 0.7× Qiagen® PCR HotStar Taq Master Mix containing buffer and polymerase. Cycling parameters were 95° for 15 min, 95° for 45 s, 60° for 45 s, and 72° for 45 s for 40 cycles, followed by a final extension at 72° for 7 min. After thermocycling, 5 μl of a 1:15 dilution of Exo-SAP was added to each well, and reactions were incubated at 37°C for 15 min prior to inactivation at 80° for 15 min. Reactions were diluted by 0.6×, and 2 μl were combined with 5 μl of 1/64 Applied Biosystems® BigDye™ sequencing reaction mix and cycled as above for 25 cycles. Reactions were precipitated with ethanol, resuspended in 0.1 mM EDTA, and loaded on Applied Biosystems 3730XL sequencing instruments using the Rapid36 run module and 3xx base-caller. SNPs were identified using SNPdetector software (15). Identified mutations were verified by bidirectional resequencing of the original DNA sample.
The APOA5 gene was covered by a contiguous overlapping set of 14 amplicons. The most upstream hg19 coordinate is 116,664,060 and the coordinate of the transcription start site is 116,663,137. Exon 1 starts at that coordinate (116,663,137) and extends to 116,663,096; it is untranslated and could therefore be thought of as the upstream end of the 5′ UTR. The Exon 2 start coordinate is at either 116,662,611 or 116,662,608 depending on isoform. The translation start site (ATG codon) is at 116,662,576. The distance from the upstream end of the most upstream amplicon to the translation start site is 1,484 bp which was defined as the promoter for the purpose of the current analysis.
Collapsing approaches (“burden tests”) were used for the analyses. Within a gene, we classified rare variants as intronic, missense, synonymous, and promoter variants. Minor allele frequency was calculated by the following formula: (count of heterozygotes + 2 × homozygotes)/(2 × total counts).
Burden tests used in the analyses included fixed allele-frequency threshold, variable allele-frequency threshold, and frequency-weighted approaches (16–18). Computational prediction of functional effects (e.g., PolyPhen scores or conservation scores) was incorporated in those analyses so the variants could be weighted according to their inferred functional importance. For descriptive purposes, participants were classified in a “rare” group if a rare allele was present at any of the variant sites in a studied gene or gene region; otherwise, participants were classified in a “nonrare” group. Descriptive statistics, including sample size, mean, and standard deviation, were calculated for both rare and nonrare groups and were compared by t-test. Response to therapy was defined as the difference between baseline and posttherapy levels (e.g., ΔHDL-C = final HDL-C − baseline HDL-C). All responses were adjusted for age, sex, and baseline lipid or lipoprotein levels. We obtained residuals (e.g., rHDL-C) from the linear regression of ΔHDL-C = sex + age + diabetes status + body mass index (BMI) + baseline HDL-C and baseline TGs. Statistical analyses using either Student's t-test or burden tests were conducted for the adjusted trait difference (e.g., ΔHDL-C), which included the trait's residual and the mean trait difference before and after therapy (e.g., ΔHDL-C = rHDL-C + mean ΔHDL-C).
At baseline, plasma concentrations of HDL-C, apoA-I, and TGs were not statistically significantly different among the fenofibric acid + statin, statin monotherapy, and fenofibric acid monotherapy treatment groups. As expected, there were significant differences between males and females for baseline HLD-C and apoA-1 based on the inclusion criteria (HDL-C <40 for males, HDL-C <50 for females). There were no, or borderline, differences for baseline TGs. Average age and BMI were higher in women compared with men (differences of 2.8–4.5 kg and 1.1–1.46 kg/m2 respectively, depending on the treatment group). Although there was a trend toward higher frequency of diabetes in women, this was not statistically significant (Table 1).
Plasma levels of HDL-C and apoA-I increased and TGs decreased in response to therapy in all treatment groups. As expected, the increases in HDL-C and apoA-I were greatest in the fenofibric acid + statin combination group and least with statin monotherapy. A similar pattern was observed for the decrease in TGs following therapy (Table 2).
Women had better HDL-C response to therapy than men in the groups receiving fenofibric acid, either alone or in addition to statins, but not in the statin alone treatment group. However, better response for women did not persist for apoA-I therapy response in the fenofibric acid only group, but did so for the fenofibric acid + statin group. A possible explanation of the differences in HDL-C response between men and women could be related to the difference in baseline HDL-C and apoA-I; in addition, women had somewhat higher baseline BMIs and were older than men; this may have made them better responders for therapy with fenofibric acid, either with or without statins, but not for therapy without fenofibric acid as in the statin alone. Comparison of response to therapy between men and women is detailed in supplementary Table II.
On the basis of location within the APOA5 gene and, in the case of the exon variants, whether there was an inferred change in the protein amino acid sequence, we classified the rare DNA variants into six classes: APOA5 3′ UTR, APOA5 intron, APOA5 promoter, APOA5 missense, and APOA5 synonymous rare variants. Only variants with a frequency of <0.01 were included in the analysis. Identified rare variants, their genomic location, frequency in our population study, frequency in a public database if ever reported (1000 Genomes, Exome sequencing project, or other database), and number of individuals who carry each variant in our study are detailed in supplementary Table III. A total of 58 rare variants were identified, of which 27 were previously reported in an external database. Allele frequency ranged between 0.0003 and 0.004. There were eight participants who had two rare variants, and two participants who had three variants.
We used several approaches to analyze potential rare variant associations because each method may have certain advantages depending on the actual genetic structure of the trait.
Pooled analysis of rare genetic variants in the APOA5 promoter demonstrated significant associations with decreases in HDL-C (Table 3) and apoA-I (Table 4) in response to FA therapy, in contrast to the expected increases. The adjusted mean change in HDL-C following fenofibric acid monotherapy for participants with rare variants was −4.6 mg/dl; for participants without rare variants, the adjusted mean change in HDL-C following fenofibric acid monotherapy was +6.2 mg/dl. Comparison of these mean changes using Student's t-test (P < 0.0001) or by the fixed-allele frequency threshold method (P = 0.0001) showed highly statistically significant differences. Interestingly, all participants in the fenofibric acid monotherapy group with rare variants in the APOA5 promoter region had either a decrease or a very small increase in HDL-C in response to therapy (Fig. 1). For the statin monotherapy and combination therapy groups, adjusted mean HDL-C changes in response to therapy were not significantly different in participants with versus participants without rare variants in the APOA5 promoter region. A similar pattern was observed for plasma apoA-I response to therapy; the adjusted mean change in apoA-I in response to fenofibric acid monotherapy was −2.3 mg/dl in participants with rare variants in the APOA5 promoter region versus +11.6 mg/dl in participants without rare variants in the APOA5 promoter region; the difference between these means was borderline statistically significant by Student's t-test (P < 0.06) and the fixed allele-frequency threshold method (P = 0.056). There were no significant associations between other classes of APOA5 rare variants and HDL-C response to fenofibric acid therapy.
We detected a significant effect of missense variants in APOA5 on TG response to therapy (Table 5). The adjusted mean change in TGs in response to fenofibric acid + statin therapy was −228 mg/dl in participants with rare missense variants in the APOA5 gene, compared with −143 mg/dl in participants without rare variants in the APOA5; the difference was statistically significant using both the Student's t-test (P = 0.002) and fixed allele-frequency threshold method (P = 0.004). In the fenofibric acid monotherapy treatment group, adjusted mean change in TGs was not significantly different between participants with rare missense variants (−167 mg/dl) and those without (−102 mg/dl), however, only two participants had rare missense variants in that treatment group. In contrast to the analysis of HDL-C response to therapy, there was no significant association between the adjusted mean change in TGs following fenofibric acid monotherapy and rare variants in the region of the APOA5 promoter (Table 5). A significant difference in adjusted mean change in TGs in response to fenofibric acid monotherapy was found in participants with versus without intronic rare variants in the APOA5 gene, however, only two individuals had intronic rare variants in that treatment group.
In this study, we identified a number of novel rare APOA5 gene promoter variants that were associated with a paradoxical HDL-C response to fenofibric acid therapy; participants with these variants had either a decrease or no significant change in HDL-C in response to fenofibric acid monotherapy. A similar pattern was observed for apoA-I response to fenofibric acid therapy, suggesting that these variants affect the cholesterol content as well as the apolipoprotein component of the HDL particle.
The APOA5 promoter has previously been shown to have an important effect on gene transcription and expression (19). Vu-Dac et al. (19) identified a number of regions in the APOA5 promoter that affected gene expression, including a specific peroxisome proliferator-response element (PPRE). Other studies have shown the importance of a common APOA5 promoter SNP, −1131T→C, for APOA5 expression in vitro (20), association with TGs, HDL-C, coronary disease in vivo (21, 22), and response to fibrate therapy (23). Thus, the importance of the APOA5 gene promoter as a modulator of APOA5 expression and function is well established. Although the variants identified in our study were not located in the PPRE region of the APOA5 promoter, it is possible that they disrupt APOA5 promoter activation in response to fenofibric acid and may actually have an inhibiting effect on gene expression when interacting with fenofibric acid.
The possible clinical utility of the rare promoter variants identified in this study relates to fenofibric acid therapy in a specific study population; individuals with mixed dyslipidemia, such as the participants in this study, are at high risk for coronary heart disease events. Evidence from a number of previous clinical trials suggests that in individuals with mixed dyslipidemia, fibrate therapy may improve coronary outcomes. For instance, monotherapy with gemfibrozil, another fibrate and PPAR-α agonist, reduced coronary events in the Helsinki Heart Study in men with non-HDL-C ≥200 mg/dl at baseline (24) and in the Veterans Affairs High-Density Lipoprotein Cholesterol Intervention Trial in men with HDL-C ≤40 mg/dl (25). The same pattern was observed with fenofibrate in the Fenofibrate Intervention and Event Lowering in Diabetes (FIELD) study (26) subgroup with mixed dyslipidemia (TGs ≥150 mg/dl and HDL-C <40 mg/dl in men and <50 mg/dl in women), who had a reduction in risk for cardiovascular events (26). Thus, it is possible that the rare APOA5 promoter variants identified in this study, which were associated with decreases or no change in HDL-C and apoA-I levels, may actually attenuate the coronary heart disease reduction effect of fenofibric acid in the mixed dyslipidemia population.
As a PPAR-α agonist, fenofibric acid modulates multiple downstream targets including lipoprotein lipase, ATP-binding cassette transporter-1, apoA-I, apoA-II, and apoC-III; however, the exact pathway by which PPAR-α agonists increase lipoprotein lipase and apoA-I levels is not completely understood (27). A number of studies have examined the effect of missense common SNPs on response to fenofibrate therapy. Lai et al. (11) showed an association of a single SNP, rs3135506, in APOA5 with HDL-C and TG response to fenofibric acid therapy. In our study, a greater reduction in TGs was observed for participants who had rare missense variants compared with those who did not in the fenofibric acid + statin combination therapy group and a trend toward a difference in the fenofibric acid monotherapy group.
Our study has a number of limitations. First, it was a clinical trial that included only individuals with mixed dyslipidemia which are estimated to be ~21% of the adult population (28), and this had a considerable effect on the number of participants. However, the study had a specific single gene-related hypothesis that was based on preliminary published studies describing common SNPs in the APOA5, including SNPs in the promoter region, which were shown to effect fibrate therapy response in multiple studies (10, 11, 23, 29).
In addition, an approach taken in several published studies examining rare variants of quantitative traits was to examine the extremes of the distribution which is expected to be enriched for rare alleles affecting the trait. Similarly, the current study was designed to examine only individuals with elevated TGs and low HDL-C, and elevated LDL-C, the mixed dyslipidemia phenotype. These individuals are expected to be enriched for rare variants in genes affecting lipid metabolism including the APOA5 gene region.
Thus, although a larger sample size would have been beneficial, for the targeted hypothesis in this study and specific population of mixed dyslipidemia that is a fraction of the adult population enriched for rare variants, the current sample size and study design were able to further support the role of the APOA5 gene in fibrate therapy response by showing the role of rare variants in the response to fenofibric acid therapy.
Another limitation concerns the distribution of the various statins in the statin alone and statin + fenofibric acid treatment groups which limits the understanding of the specific effects of each statin type and dose. However, our major finding was in the group receiving fenofibric acid alone which was homogenous with regards to its therapy.
It is important to mention that there was a gradual increase in response to therapy (for levels of HDL-C, apoA-1, and TGs) based on the type of therapy (statin alone, combination therapy, fenofibric acid alone). There was a modest response to statins alone, a much better response to fenofibric acid alone, and some additional improvement for the statin + fenofibric acid combination compared with fenofibric acid alone. This pattern is in agreement with previous studies examining response to statin and fibrate therapy, suggesting that our study does follow the therapy response pattern described in other clinical trials. Thus, although different statins were used, the overall effect was similar to other studies examining statins and their combination with fibrates.
In conclusion, the present study is the first to show that rare variants in the APOA5 gene promoter region are associated with a paradoxical decrease in plasma HDL-C and apoA-I in response to fenofibric acid therapy, whereas rare missense variants in the APOA5 gene were associated with greater reduction in TGs in response to the combination of fenofibric acid and statin. These findings may be partly responsible for the interindividual differences in response to fibrate therapy and may be associated with the effect of fibrate monotherapy, or in combination with statins, on coronary outcomes in individuals with mixed dyslipidemia. As sequencing becomes more affordable and available in the clinical setting, rare variants such as those identified in this study may help to personalize fibrate therapy.
The authors wish to acknowledge the contributions of the following individuals within the Human Genome Sequencing Center at Baylor College of Medicine: Richard Gibbs (director), Donna Muzny, Lora Lewis, Humeira Akbar, Shannon Gross, Robert Ruth, and Kyle Chang. The authors thank Marie Fleisner of the Marshfield Clinic Research Foundation for editorial assistance in preparing this manuscript.
DNA extraction and genotyping were funded by Abbott Laboratories, Abbott Park, IL. S.S.V. is supported by a Department of Veterans Affairs Health Services Research and Development Career Development Award. A.B., M.B., F.C., J.B., and S.S. have declared no potential conflicts of interest. S.S.V. received a research grant (which ended December 2011) from Merck and was on the speakers’ bureau for Abbott (discontinued November 2010). R.A.H. has been on the speakers’ bureau for Merck, Abbott, and AstraZeneca and a consultant for Merck and Genzyme. C.M.B. has received grant/research support (all paid to institution, not individual) from Abbott, Amarin, AstraZeneca, Bristol-Myers Squibb, GlaxoSmithKline, Genentech, Kowa, Merck, Novartis, Roche, Sanofi-Synthelabo, Takeda, National Institutes of Health, American Diabetes Association, and American Heart Association; is a consultant for Abbott, Adnexus, Amarin, Amylin, AstraZeneca, Bristol-Myers Squibb, Esperion, Genentech, GlaxoSmithKline, Idera Pharma, Kowa, Merck, Novartis, Omthera, Pfizer, Resverlogix, Roche, Sanofi-Synthelabo, and Takeda; is on the speakers’ bureau for Abbott, GlaxoSmithKline, and Merck; and has received honoraria from Abbott, Adnexus, Amarin, Amylin, AstraZeneca, Bristol-Myers Squibb, Esperion, Genentech, GlaxoSmithKline, Idera Pharma, Kowa, Merck, Novartis, Omthera, Resverlogix, Roche, Sanofi-Synthelabo, and Takeda.
[S]The online version of this article (available at http://www.jlr.org) contains supplementary data in the form of three tables.