state

nature publishing group

art

Pharmacogenomics, Lipid Disorders, and Treatment Options SE Gryn1 and RA Hegele1,2 Statins form the backbone of lipid-lowering therapy in the prevention of cardiovascular disease. Numerous studies have evaluated the effect of genomics on the clinical efficacy and adverse effects of statins. Several gene variants that can be linked to either the pharmacokinetics or pharmacodynamics of statins have been identified as potentially important, although there are some discrepant findings among studies. Effect sizes are modest for lipid-lowering efficacy and perhaps somewhat larger for risk of myopathy, although results are inconsistent. Pharmacogenomics of nonstatin lipidlowering agents have not been evaluated to the same extent, given their relatively limited use, although there are some promising candidate genes for further study. Finally, with several new classes of lipid-lowering therapies soon becoming available, there may be a potential application for pharmacogenomics to identify patients ideally suited to receive—or those who should avoid—specific medications. Currently available lipid-lowering agents recommended in treatment guidelines1 include statins as first-line therapy and four classes of less commonly used second-line agents, including cholesterol absorption inhibitors (of which ezetimibe is the only agent available by prescription), fibrates, niacin-based preparations, and bile acid sequestrants. Among the lipid-lowering agents, statins are broadly effective, with overwhelming evidence for reduction of cardiovascular risk and extension of life when prescribed to appropriate patients.2 Statins are typically well tolerated and have earned the right to be the first-line class of agents for the prevention of cardiovascular disease (CVD). The clinical benefits of statins derive mainly from their ability to reduce plasma levels of low-density lipoprotein (LDL) cholesterol, which is the central player that drives the lipid component of atherogenesis,3 and which itself has pleiotropic effects on the pathogenesis of atherosclerosis.4 However, years of experience from both clinical trials and lipid clinics have shown that not all patients respond equally well to statin treatments, with respect to both their efficacy and adverse effects. This heterogeneity of response has prompted extensive investigation of factors that can modulate such interindividual differences, including pharmacogenomic factors. The ideal application of pharmacogenomics is to identify patients at risk for either a suboptimal response with respect to efficacy, or a marked adverse response to either a drug class or a specific drug. In this way, individuals who would be predicted to have an unfavorable benefit-to-risk ratio can

be identified and might gain from alternative approaches more expeditiously and without the trial-and-error process that typically accompanies initiation and maintenance of this commonly used therapy. The second-line treatments have shown more varied success with respect to reduction in CVD risk, particularly in the statin era. Outdated trials before the availability of statins demonstrated reduced CVD risk with fibrates, niacin, and bile acid sequestrants as monotherapy compared with placebo. There is no evidence that ezetimibe as monotherapy can reduce CVD risk, although it does reduce CVD risk when used in combination with statins.5 However, now that the standard of care for most high-risk patients is treatment with maximally tolerated statin doses, fibrates or niacin appear to offer no additional CVD risk reduction when used in combination with statins.6,7 For this reason, the second-line drugs are reserved as monotherapy for individuals who cannot tolerate statins or as add-on therapy for those who fail to respond adequately to statins.1 The role of target lipid levels, particularly for LDL cholesterol, is a topic of great interest, but the current controversy regarding the use of treatment targets is beyond the scope of the material to be considered here. Because of their clinical importance, we will focus mainly on statins and their complex architecture of pharmacogenomic determinants. The main type of genetic variant that has been evaluated comprises single-nucleotide polymorphisms (SNPs) in

1Department of Medicine, Schulich School of Medicine and Dentistry, Western University, London, Ontario, Canada; 2Robarts Research Institute, Schulich School of

Medicine and Dentistry, Western University, London, Ontario, Canada. Correspondence: RA Hegele ([email protected])

Received 16 March 2014; accepted 7 April 2014; advance online publication 7 May 2014. doi:10.1038/clpt.2014.82 Clinical pharmacology & Therapeutics

1

state

art

(i) genes affecting statin absorption, distribution, metabolism, and excretion (pharmacokinetic factors) and (ii) genes whose products act in physiological pathways, including cholesterol biosynthesis and lipoprotein metabolism (pharmacodynamic factors). We also discuss whether pharmacogenomics can help maximize the efficacy and minimize the adverse events related to statins among individuals at risk for CVD, for whom the treatment is potentially life saving. These understandings also extend to nonstatin treatments, including both existing alternatives and new investigational therapies under development, which we will also review. The overall outline for the drugs and candidate genes to be discussed is shown in Table 1. OVERVIEW OF GENETIC DETERMINANTS OF RESPONSE TO LIPID-LOWERING MEDICATIONS

The genetic targets described here will generally be categorized as affecting either the pharmacokinetics or pharmacodynamics of a particular medication (Table 1). Pharmacokinetic targets comprise drug metabolism enzymes, which are involved in the elimination of some lipid-lowering medications, and—perhaps most importantly—drug transporters, which may be involved in the absorption, distribution, or excretion of their substrates. Drug transporters can be involved in cellular uptake or efflux of substrates in a variety of tissues; they may be involved not only in drug elimination but also in drug distribution to the site of action or toxicity. For example, with regard to statins, hepatic transporters are of primary importance because the liver is the primary site of action and elimination of these drugs. In spite of the limited availability of data, drug transporters that affect muscle statin levels could theoretically influence risk of myopathy.

On the other hand, pharmacodynamic targets encompass the molecular entities or downstream effectors related to the mechanism of action—or toxicity—of the medication. In this review, genes encoding components of lipid metabolism pathways will be briefly described, although we attempt to focus on SNPs that have an effect on drug response independent of effects on the baseline lipid profile. Many of the genes in these metabolic pathways were discovered due to their causative role in human disorders of lipoprotein metabolism.8 CURRENT STATE OF PHARMACOGENOMICS IN THE PREDICTION OF STATIN EFFICACY

Numerous studies have examined the impact of pharmacogenomics on the efficacy of statins, evaluating such readouts as lipid-lowering effect, surrogate markers of cardiovascular risk, and hard clinical outcomes (Table 2). These findings have already been extensively reviewed.9–11 Genes related to either the pharmacokinetics or the pharmacodynamics of statin drugs have been studied in detail, and some genetic variants associated with underlying disease states have also been associated with the efficacy of statins. In this section, we will focus on the results of several recently published genetic substudies of various large clinical trials of statin use, as well as on selected smaller pharmacogenetic studies, and will put them into a mechanistic context. It is important to recognize that SNPs may directly possess—or act as markers of—functional consequences that are specific for some statins rather than others, or even for high vs. low doses of the same agent, which could potentially explain the discrepant findings among studies. Other reasons for discrepancies include varying allele frequencies among populations, which affects statistical

Table 1  Overview of candidate genes involved in the response to lipid-lowering medications sorted by mechanism of action of the gene product Candidate genes (references) Statins   Pharmacokinetics     Drug metabolism

CYP2C8, CYP2C9, CYP2C19, CYP2D6, CYP3A4, CYP3A5, UGT1A1, UGT1A3, and UGT2B7 (refs. 17,71)

    Uptake transporters

SLCO1A2, SLCO1B1, SLCO1B3, SLCO2B1, and SLC15A1 (refs. 14,16,18,71)

    Efflux transporters

ABCB1, ABCB11, ABCC2, and ABCG2 (refs. 15–17,71)

  Pharmacodynamics     Drug efficacy

HMGCR, APOE, LDLR, PCSK9, CETP, LPA, CLMN, and DNAJC5B (refs. 15–17,20–25,31–33)

    Drug toxicity

COQ2, RYR1, PYGM, CPT2, DMPK, AMPD1, ATP2B1, LPIN1, HTR3B, HTR7, AGTR1, NOS3, (ref. 38) and GATM (ref. 58)

Ezetimibe   Pharmacokinetics

SLCO1B1 (ref. 72)

  Pharmacodynamics

NPC1L1 (refs. 58,59,73,74)

Fibrates   Pharmacokinetics

CYP3A4, UGTs (ref. 75)

  Pharmacodynamics

PPARA, APOA1, and APOA5 (refs. 54–56)

Niacin   Pharmacokinetics

None identified

  Pharmacodynamics

DGAT2 (ref. 10)

2

www.nature.com/cpt

state power, or the absence of key SNPs from some studies, even in extensive genome-wide association studies (GWASs). PHARMACOKINETIC DETERMINANTS OF STATIN EFFICACY

Several statins are metabolized by genetically polymorphic classic drug-metabolizing enzymes (Table 2). For instance, simvastatin, lovastatin, and atorvastatin are all metabolized primarily by cytochrome P450 (CYP)3A4, whereas fluvastatin is metabolized primarily by CYP2C9. The knowledge of these pathways has been important in explaining drug–drug interactions seen with the use of individual statins and other classes of drugs. However, genetic polymorphisms in CYP enzymes seem to have minimal and inconsistent associations with respect to predicting statin efficacy in a particular individual. The effect sizes of the SNPs are typically small and often not statistically significant. Thus, the potential clinical value of genetic polymorphisms in CYP enzymes with respect to predicting clinical efficacy of statins remains unclear, despite years of study.9 By contrast, genetic variation among drug transporters, particularly those related to hepatic uptake and efflux, appears to be a more robust determinant of statin efficacy. The associations are more consistently significant and show stronger effect sizes across studies. For instance, the liver is the primary site of action of statins, and nearly all the statins have been identified as substrates of the organic anion transporting polypeptide 1B1 (OATP1B1) hepatic uptake transporter encoded by the SLCO1B1 gene, although to a variable extent.12 OATP1B1 was originally designated as OATP-C, and the complete characterization of the genetic polymorphisms of the gene product, together with functional validation, was completed almost 15 years ago.13 Some of the originally defined variants in SLCO1B1 have now proven to be consistent determinants of response to statins. For instance, genetic substudies of large prevention trials that studied rosuvastatin (the Justification for the Use of Statins in Primary Prevention: An Intervention Trial Evaluating Rosuvastatin, JUPITER), pravastatin (the Prospective Study of Pravastatin in the Elderly at Risk, PROSPER), and simvastatin (the Heart Protection Study, HPS) all found that the SLCO1B1 c.521 T>C variant (rs4149056), which is associated with reduced cell surface expression and decreased transporter activity, is significantly associated with attenuated LDL cholesterol lowering, albeit with a small magnitude of effect.14–16 Furthermore, in a genetic substudy of participants in a safety study of atorvastatin (the Atorvastatin Comparative Cholesterol Efficacy and Safety Study, ACCESS), this same SLCO1B1 SNP, together with rs4149080, was associated with altered high-density lipoprotein (HDL) cholesterol response to atorvastatin. In addition, another SLCO1B1 SNP, namely rs4149036, was associated with changes in triglyceride (TG) but not LDL cholesterol levels.17 Moreover, the SLCO1B1 c.463C>A variant (rs11045819) was associated with enhanced lowering of LDL cholesterol with use of simvastatin in HPS and with efficacy of fluvastatin, as previously demonstrated,18 although in this case it was linked with the c.388A>G variant (rs2306283) within the larger SLCO1B1*14 haplotype; this variant/haplotype is thought to increase transporter function. Clinical pharmacology & Therapeutics

art

Efflux transporters within the adenosine triphosphate (ATP)– binding cassette (ABC) family also show consistent associations with aspects of statin pharmacokinetics, and potentially with clinical efficacy, presumably via regulation of biliary excretion. For instance, ABCG2, encoding ABCG2, also known as breast cancer resistance protein, has a common c.421C>A polymorphism (rs2231142) that has been associated with increased systemic exposure to rosuvastatin in pharmacokinetic studies, as well as enhanced lipid-lowering effects,19 probably due to increased hepatic exposure. In the JUPITER genetic analysis, another ABCG2 SNP, rs2199936, which is in linkage disequilibrium (LD) with rs2231142 (r2 = 0.81), showed significant association at the genome-wide level for enhancement of the LDL cholesterol–lowering effect of rosuvastatin.15 Furthermore, ABCB1, encoding multidrug resistant protein 1 (MDR1), also known as P-glycoprotein, is expressed in numerous tissues and has a triallelic SNP, namely c.G2677T/A (rs2032582) that was identified as relevant to plasma lipid response in the ACCESS trial in patients taking atorvastatin, as well as in the Pravastatin or Atorvastatin Evaluation and Infection Therapy— Thrombolysis in Myocardial Infarction 22 (PROVE-IT TIMI 22) trial in patients with acute coronary syndrome, although the effect was statistically significant only in the pravastatin arm.17,20 Furthermore, in PROVE-IT TIMI 22, an additional ABCB1 SNP, namely c.C3435T (rs1045642), was not independently associated with LDL cholesterol response to either statin, although haplotype analysis that incorporated both ABCB1 SNPs demonstrated a significant association with response to pravastatin. Finally, in HPS, the T allele of the intronic SNP rs2002042 in ABCC2 was associated with a small increase in response to simvastatin, consistent with the potential importance of this transporter.16 PHARMACODYNAMIC DETERMINANTS OF STATIN EFFICACY

The other main category of genetic variants governing statin efficacy encompasses those that affect gene products that act within lipid metabolism pathways (Table 2). Although these SNPs have the potential to alter the pharmacodynamics of statin drugs, many of them also affect baseline lipid levels as well as CVD risk, and data demonstrating independent associations with statin response are so far sparse. Because 3-hydroxy-3-methylglutaryl-coenzyme A reductase (HMGCR) is the molecular target of statins, the HMGCR gene would logically seem to be the first candidate gene to discuss with respect to pharmacodynamics. In this regard, the HMGCR SNPs rs17244841 and rs17238540 are in strong LD with each other as part of the H7 haplotype and are more common in African-American populations than in other populations. These HMGCR SNPs and the H7 haplotype were associated with reduced LDL cholesterol response to pravastatin in the Pravastatin Inflammation/CRP Evaluation (PRINCE) study and to simvastatin in the Cholesterol and Pharmacogenetics (CAP) trial.21,22 Furthermore, in candidate gene analysis of patients taking pravastatin in the Cholesterol and Recurrent Events (CARE), West of Scotland Coronary Prevention Study (WOSCOPS), and PROSPER studies, those who carried the 3

state

art

Table 2  Genetic substudies of large statin clinical trials with significant findings related to statin efficacy Study Thompson et al.17; ACCESS

Sample size 2,735

Thompson et al.32; TNT

1,984 GWAS

Statin (dose) Atorvastatin (10–80 mg) (approximately half of study population) Atorvastatin (10–80 mg)

5,745 Candidates Atorvastatin (10 mg)

Mega et al.20; PROVE-IT TIMI 22

Atorvastatin (80 mg) vs. pravastatin (40 mg)

Chasman et al.21; PRINCE

1,536 On pravastatin

GWAS

+6.595

Candidate

None confirmed: (rs7412 not on platform) APOE: rs7412 (6.0E-30) PCSK9: rs11591147 (2.86E-04) LPA: rs10455872 (6.13E-09 for CARDS/ASCOT, 1.2E-09 including PROSPER) APOE: rs445925 (2.22E-16) rs4420638 (1.01E-11) APOE: Atorvastatin rs7412 (8.3E-06) rs429358 (4.4E-02) rs769449 (2.2E-02) Pravastatin rs7412 (6.5E-05) rs429358 (2.3E-02) rs769449 (4.7E-02) ABCB1 (pravastatin): rs2032582/rs1045642 Non G–C vs. G–C haplotype (3.0E-04) HMGCR: rs17244841 (0.05 corrected, in Caucasians) rs17238540 (0.03 corrected, in Caucasians) SLCO1B1: rs4149056

GWAS

DNAJC5B: rs13279522 (4.8E-07)

GWAS

ABCG2: rs2199936 (2.1E-12) rs1481012 (1.7E-15) LPA: rs10455872 (3.5E-09) (5.0E-15) APOE: rs7412 (5.8E-19) PCSK9: rs17111584 (5.8E-04) rs11591147 (3.1E-04) SLCO1B1: rs4149056 (7.7E-05) rs4363657 (4.0E-05) rs12317268 (4.1E-06, 1.3E-04 adjusted) LDLR: rs11668477 (1.8E-03, 4.2E-02 adjusted)

GWAS

Replication with pravastatin (40 mg)

Pravastatin (40 mg)

Akao et al.14; PROSPER

5,411

Shiffman et al.23; CARE/WOSCOPS/ PROSPER

First-stage GWAS: Pravastatin (40 mg) 1,065 Second-stage: 13,784 6,989 (3,523 On Rosuvastatin (20 mg) statin, 3,466 on placebo); European ancestry only

Chasman et al.15; JUPITER

Effect size (additional LDL cholesterol reduction per allele unless otherwise specified) +3.5% +3.0% (var/var vs. WT)

Candidate

Deshmukh et al.31; 2,702 From CARDS/ASCOT CARDS/ASCOT with replication in PROSPER 2,550 From PROSPER 1,507

Genetic Significant genetic variants analysis design (P value) Candidate APOE: rs7412 (1.51E-03) ABCB1: rs2032582 (1.26E-03)

Pravastatin (40 mg)

Candidate

Candidate

Candidate

+6.148 From CARDS patients: approximately −3.5% Approximately +6.0% Approximately −10%

+7.89% −2.89% −3.40% +8.41% −3.68% −4.76%

−10.5% −6.4% −6.4% WT 37.0% reduction Het 36.0% reduction Var 31.8% reduction Hazard ratio for CHD 0.65/copy

+5.23 mg/dl absolute +5.1% −6.23 mg/dl absolute −6.8% +5.1% −4.33 mg/dl absolute +4.5% −2.6% −2.8% −3.2%

+2.1%

Table 2  Continued on next page 4

www.nature.com/cpt

state

art

Table 2  Continued Study Barber et al.33; CAP/PRINCE/TNT

Sample size 3,936 Caucasians

Statin (dose)

Genetic analysis design

Significant genetic variants (P value) CLMN: rs8014194 (3.9E-06, stronger association for TC lowering) APOC1: rs4420638 (4.2E-03)

GWAS Simvastatin (40 mg), pravastatin (40 mg), and atorvastatin (10 mg)

Krauss et al.22; CAP 922

Simvastatin (40 mg)

Candidate

HMGCR: rs17238540 (7.0E-03) rs17244841 (1.0E-02) rs6453131 (4.0E-03 for ApoB, NS for ΔLDL cholesterol)

Hopewell et al.16; HPS

Simvastatin (40 mg)

GWAS

None confirmed

Candidate

LPA: rs3798220 (7.1E-05) rs10455872 (8.1E-28) APOE: rs7412 (4.8E-18) rs4803750 (8.5E-06) rs2075650 (7.8E-05) rs4420638 (6.4E-06) ABCC2: rs2002042 (8.2E-05) SLCO1B1: rs4149056 (5.0E-08) rs11045819 (2.4E-05) CELSR2/PSRC1/SORT1: rs646776 (6.7E-03)

3,895 GWAS (+14,810 replications) 18,705 For candidate studies

Effect size (additional LDL cholesterol reduction per allele unless otherwise specified) 84% Posterior probability of association with TC lowering, only 16% for LDL cholesterol 34% Posterior probability of association with LDL lowering −0.20 mmol/l −0.18 mmol/l AA: −0.266 g/l ApoB;AB: −0.291 g/l ApoB;BB: −0.311 g/l ApoB

−2.30% −3.15% +2.55% +1.22% −0.82% −0.96% +0.65% −1.15% +0.92% +0.47%

ACCESS, Atorvastatin Comparative Cholesterol Efficacy and Safety Study; ASCOT, Anglo-Scandinavian Cardiac Outcomes Trial; CAP, Cholesterol and Pharmacogenetics; CARDS, Collaborative Atorvastatin Diabetes Study; CARE, Cholesterol and Recurrent Events, CHD, coronary heart disease; GWAS, genome-wide association study; HPS, Heart Protection Study; JUPITER, Justification for the Use of Statins in Primary Prevention: An Intervention Trial Evaluating Rosuvastatin; LDL, low-density lipoprotein; PRINCE, Pravastatin Inflammation/CRP Evaluation; PROSPER, Prospective Study of Pravastatin in the Elderly at Risk; PROVE-IT TIMI 22, Pravastatin or Atorvastatin Evaluation and Infection Therapy— Thrombolysis in Myocardial Infarction 22; TC, total cholesterol; TG, triglyceride; TNT, Treating to New Targets; WOSCOPS, West of Scotland Coronary Prevention Study; WT, wild type.

minor allele of HMGCR rs17238540 (or rs16872523, which was in strong LD with rs17238540) had increased CVD events consistent with reduced efficacy of pravastatin.23 However, these effects were not replicated in other trials that examined these SNPs, such as ACCESS (atorvastatin), PROSPER (pravastatin), and JUPITER (rosuvastatin). However, because the minor allele frequencies were low in these study samples, they were generally underpowered to detect efficacy effects.15,17,24 Other HMGCR SNPs have been studied, but none has been consistently associated with statin response. Another pathway that provides candidate genes for association with statin efficacy is the LDL receptor–mediated endocytosis mechanism, which includes the LDL receptor itself (encoded by LDLR), together with processing proteins such as proprotein convertase subtilisin kexin 9 (PCSK9) and LDL receptor–associated protein (LDLRAP1, also known as ARH). In general, genetic variants in proteins related to LDL receptor function are more consistently associated with baseline lipids and CVD risk than response to therapy. For instance, genetic analyses in the PROSPER study showed that (i) the LDLR SNPs C44857T and A44964G were associated with baseline LDL cholesterol levels and were modestly associated with LDL cholesterol lowering and CVD risk reduction with pravastatin treatment in men17; and (ii) functionally important PCSK9 SNPs—p.P46L and p.E670G—were strongly associated with Clinical pharmacology & Therapeutics

baseline lipid profiles but were not associated with response to pravastatin.25 More recently, researchers have turned their attention to SNPs in the KIF6 gene encoding kinesin-like protein 6 as a determinant of response to statins. KIF6 is part of the kinesin superfamily of proteins that mediate intracellular transport of organelles, protein complexes, and mRNAs. The association with response to statins was suggested based on findings from candidate gene association studies in which the 719Arg allele of KIF6 (rs20455) was associated with increased coronary heart disease risk in the Atherosclerosis Risk in Communities (ARIC) study and in the placebo arms of two pravastatin trials, namely CARE and WOSCOPS.26,27 Since these initial reports, KIF6 SNPs have been examined in numerous studies and have been the subject of meta-analysis several times. It would appear from more recent studies that KIF6 SNPS are only minimally associated with clinically relevant traits, including response to statins, a recent trend that may reflect earlier publication bias in the reported positive findings.28 A recent analysis by Peng et al.29 demonstrated that KIF6 rs20455 was significantly associated with increased baseline CVD risk, as well as enhanced efficacy of statin drugs in reducing risk in prospective clinical trials involving mainly Caucasian patients. 29 On the other hand, multiple case–­ control studies, primarily of non-Caucasian patients, showed 5

state

art

no such association, suggesting possible ethnicity dependence of the effects of this SNP on CVD risk.29 An interesting outcome related to KIF6 genotyping was reported by Charland et al.30 in the Additional KIF6 Risk Offers Better Adherence to Statins (AKROBATS) trial, in which patients who received genetic testing were more likely to remain adherent to statin medication, although the study was nonrandomized and thus potentially biased. APOE encodes the defining apolipoprotein (apo) E of verylow-density lipoprotein, which is the metabolic precursor for LDL. APOE genetic variants have been studied for almost 4 decades and have been identified as being significantly associated with baseline lipid profiles and statin efficacy in multiple studies, more recently at the genome-wide level, when APOE SNPs were included on the GWAS platform. With respect to statin response, carriers of the minor allele at rs7412 or other SNPs in LD with this SNP, corresponding to the APOE ε2 isoform, have consistently demonstrated an enhanced LDL cholesterol–lowering effect of statins in several of the trials cited above—ACCESS, Treating to New Targets (TNT), HPS, Collaborative Atorvastatin Diabetes Study (CARDS)/AngloScandinavian Cardiac Outcomes Trial (ASCOT)/PROSPER, PROVE-IT TIMI 22, and JUPITER, although the association with coronary heart disease events is uncertain.15–17,20,31,32 When Shiffman et al. analyzed CARE, WOSCOPS, and PROSPER/Pharmacogenetic study of Statins in the Elderly at risk (PHASE), it was suggested that patients taking pravastatin who were homozygous for APOE E3/3 had significantly greater risk reduction, which seems to contradict the results from CARDS and ASCOT. By contrast, the APOE E4 (rs4420638) allele was associated with a smaller LDL cholesterol response to statins in some of these trials, as well as in the analysis by Barber et al.33 of the CAP, PRINCE, and TNT studies, although no clear association with CVD outcomes was identified. APOE polymorphisms are also well known to affect baseline lipid levels, as demonstrated, for instance, by Trompet et al.34 in patients from PROSPER/PHASE, with a replication cohort from CARE and WOSCOPS, as well as PROVE-IT TIMI 2220; therefore, it may be difficult to dissociate the effect of APOE polymorphisms on baseline lipids from treatment-related effects. Lipoprotein(a) is an LDL-like particle that is associated with CVD risk independently of other risk factors. The defining protein of lipoprotein(a) is apo(a), which is encoded by the LPA gene on chromosome 6q. The LPA rs10455872 SNP was identified by a GWAS of patients from the CARDS and ASCOT trials as being significantly associated with attenuated LDL cholesterol–lowering effect when given with atorvastatin and pravastatin; the finding was replicated in the PROSPER study. This SNP was also significantly associated at the genome-wide level with response to rosuvastatin in the JUPITER study.15,31 There was also a lack of association between rs10455872 and risk of cardiovascular events, which could be explained either by inadequate statistical power or potentially by preserved cardiovascular risk reduction with the statin despite (artifactually) higher LDL cholesterol in individuals with elevated lipoprotein(a).31 6

Finally, in the previously mentioned analysis of CARE, WOSCOPS, and PROSPER by Shiffman et al.,23 the GWAS identified rs13279522 in DNAJC5B, with the minor allele further reducing coronary heart disease events in pravastatin-treated patients. This may relate to the pleiotropic cardiovascular effects of statins because there was no association with lipid lowering and because the gene encodes a protein not known to be related to lipid metabolism.23 In summary, genetic variation in drug disposition pathways, particularly drug transporters, may alter the pharmacokinetics of statins, with subsequent alterations in hepatic exposure to the drug, thus affecting lipid-lowering efficacy. Statins are pharmacologically heterogeneous in terms of their disposition pathways, so specific enzymes or transporters, or even specific genetic variants in them, may not be relevant to all medications in the class. Thus, it is not surprising that associations may be specific to a particular agent. Furthermore, any effect of a particular SNP on a complex trait, such as lipid lowering, is expected to be small, and any association with a more distal phenotype, such as cardiovascular outcomes, is expected to be even smaller. However, compared with association between statin efficacy and genetic variation affecting components of lipid metabolism pathways, the associations between statin efficacy and genetic variants affecting lipid metabolism have been less consistent and more tenuous. PHARMACOGENOMICS IN THE PREDICTION OF ADVERSE EFFECTS OF STATINS

Statin intolerance is a common clinical concern (Table 3). Among the numerous side effects that have been attributed to statins, only muscle-related effects, increases in transaminases, and worsening of glycemia have been cited as clinically relevant in a recent critical review.35 Intolerance is important because adherence to statin therapy after 1 year was estimated at only ~50%.36 Muscle-related side effects are cited as the most common cause for discontinuation or switching,37 suggesting prevention of these effects can perhaps improve the adherence, persistence, and, ultimately, effectiveness of statins in the general population. Among the adverse effects of statins, musclerelated effects ranging from myalgia to elevated creatine kinase levels to (fortunately) extremely rare rhabdomyolysis and renal failure have garnered the most notoriety among patients and health-care providers. These have been most extensively studied for genetic associations, but an ongoing concern has been the lack of consensus for the precise criteria that comprise a clinical diagnosis of statin-related myopathy and even for the precise definition of this condition.35 As with pharmacogenomic determinants of statin efficacy, determinants of adverse effects of statins include gene variants that alter the pharmacokinetics of statins, resulting in increased systemic—and muscle—exposure, as well as gene variants that alter muscle physiology (i.e., “toxicodynamics”). Again, variants related to pharmacokinetics such as drug transporters can have different substrate specificity among statin drugs, and genetic associations may not be universally applicable across the class. It should be recognized that www.nature.com/cpt

state

art

Table 3  Key genetic studies of statin adverse events Study

Sample size

SEARCH Collaborative 85 Myopathy patients, Group39; SEARCH, HPS 90 controls in SEARCH

Genetic analysis design

Statin (dose) Simvastatin (80 mg in SEARCH, 40 mg in HPS)

Significant genetic variants (P value)

Effect size

GWAS

SLCO1B1: rs4363657 (4.0E-09 in SEARCH)

Odds ratio for myopathy: 4.3/copy of minor allele

Candidate

rs4149056 (2.0E-09)

Odds ratio for myopathy: 4.5/copy of minor allele

rs4149056 (4.0E-03)

Relative risk: 2.6/copy of minor allele

21 cases, 16,643 controls in HPS

Replication

Mangravite et al.49; Marshfield cohort, SEARCH (initial discovery in CAP)

Meta-analysis of two cohorts, 172 myopathy cases total, 4,241 controls

Atorvastatin, simvastatin, pravastatin in Marshfield cohort (various doses), and simvastatin in SEARCH (80 mg)

Candidate

GATM: rs1719247 (7.0E-04)

Odds ratio for myopathy 0.60/copy of minor allele

Voora et al.40; STRENGTH

99 cases out of 452 total

Simvastatin (20–80 mg), atorvastatin (10–80 mg), and pravastatin (10–40 mg)

Candidate

SLCO1B1: rs4149056 (0.03)

Odds ratio for composite adverse event 1.7/copy of minor allele

Donnelly et al.41; GO-DARTS

816 cases, 1,275 controls

Multiple (primarily simvastatin, Candidate atorvastatin, and pravastatin)

SLCO1B1: rs4149056 (0.0427)

Odds ratio for intolerance 2.05

rs2306283 (0.026)

Odds ratio for intolerance 0.71

CAP, Cholesterol and Pharmacogenetics; GO-DARTS, Genetics of Diabetes Audit and Research Study in Tayside Scotland; GWAS, genome-wide association study; HPS, Heart Protection Study; SEARCH, Study of the Effectiveness of Additional Reductions in Cholesterol and Homocysteine; STRENGTH, Statin Response Examined by Genetic Haplotype Markers.

pharmacogenomic studies of adverse effects typically involve a case–control design, and even when nested into a large clinical trial or cohort study, the resulting sample size is much smaller compared with the studies of statin efficacy discussed above. This makes discovery and replication of findings difficult because the quality of the evidence is not as high as that with assessment of genetic determinants of efficacy. PHARMACOKINETIC DETERMINANTS OF ADVERSE EFFECTS OF STATINS

As with assessment of genetic determinants of interindividual differences in efficacy, most genetic variants related to statin metabolism have not consistently been associated with risk of myalgia, myopathy, or other adverse effects (Table 3).38 Some in vitro data exist, and some clinical data suggest the potential importance of genetic variation for the statins that undergo metabolism.38 As with genetic determinants of statin efficacy, genes encoding drug transporters—especially SLCO1B1— appear to be the most important factors for statin intolerance. Because it encodes a hepatic uptake transporter, dysfunction of the SLCO1B1 gene product would increase systemic (including muscle) exposure, which is probably a contributor to myopathy. Most prominently, the GWAS substudy of the Study of the Effectiveness of Additional Reductions in Cholesterol and Homocysteine (SEARCH) trial brought attention to the strong association between SLCO1B1 rs4363657, which was itself in strong LD with c.521T>C (rs4149056), discussed above, and clinical myopathy, defined as pathologically elevated serum creatine kinase in patients taking 80 mg simvastatin daily.39 In the SEARCH sample, the odds ratio of biochemical statinrelated myopathy was 4.5 per copy of the C allele and was 16.9 for CC homozygotes compared with wild-type carriers. These findings were replicated in the HPS study sample in patients Clinical pharmacology & Therapeutics

taking 40 mg simvastatin daily, although the relative risk was only 2.6 per allele, suggesting a possible gene–statin dose interaction.39 Furthermore, the SLCO1B1 c.388A>G variant (rs2306283, mentioned above) appeared to reduce the risk of myopathy, as would be expected because it increases hepatic uptake of the drug.39 Similar findings involving milder manifestations of the adverse effects of statins were demonstrated in the Statin Response Examined by Genetic Haplotype Markers (STRENGTH) and Genetics of Diabetes Audit and Research Study in Tayside Scotland (GO-DARTS) studies, which collectively involved a range of statins including simvastatin, atorvastatin, and pravastatin at different doses.40,41 Using a strict biochemical definition of myopathy, however, Brunham et al.42 replicated the association of rs4149056 with simvastatin, although not with atorvastatin-induced myopathy; there were similar drug-specific findings in the Rotterdam study, which evaluated the need for a dose decrease or statin switching.43 Moreover, there was no association between either SLCO1B1 rs4363657 or rs4149056 SNPs and myalgia in patients taking rosuvastatin in the JUPITER trial,44 again highlighting the importance of considering the context of a specific drug, specific dose, or other variables that might affect pharmacokinetics. The weight of evidence certainly confirms the importance of SLCO1B1 genotype as determining the individual response to simvastatin over all other statins, which subsequently led to a Clinical Pharmacogenetics Implementation Consortium guideline on the topic.45 Other data on the importance of genetic variation in drug transporters on adverse effects have been limited, although Becker et al.46 showed increased risk in dose reduction or statin switching in carriers of ABCC2 −24C>T (rs717620) in patients on simvastatin in the Rotterdam study. 7

state

art

ABCB1 genotype might be relevant as well.38 However, there is no universal consensus on whether SLCO1B1 genotyping should be included as part of routine clinical assessment to guide statin use because the strongest association was seen with 80 mg simvastatin daily, a drug and dose that are not commonly used clinically. Furthermore, even with an increased odds ratio for myopathy of ~17 for homozygotes, who represent ~2% of most populations, at least 85% of homozygotes did not develop myopathy on 80 mg simvastatin. PHARMACODYNAMIC DETERMINANTS OF ADVERSE EFFECTS OF STATINS

Pharmacogenomic variants that alter pharmacodynamics at the muscle level—perhaps termed more appropriately as “toxicodynamics”—have also been associated with statin-related adverse effects in that tissue (Table 3). For instance, Oh et al.47 demonstrated an association between SNPs in COQ2, encoding coenzyme Q2, and statin-associated myopathy, which was replicated in a larger study by Ruaño et al.48 The association of COQ2 with statin myopathy was not statistically significant in the SEARCH study, although the allele frequencies were not reported, and the study could have been underpowered to detect such an association because only 85 cases and 90 controls were finally studied.39 A recent finding of major impact was the discovery of the GATM gene encoding glycine amidinotransferase, a rate-limiting enzyme of hepatic creatine synthesis, as a key determinant of statin-induced myopathy.49 The authors studied global gene expression profiles from lymphoblastoid cell lines of patients who had a history of statin-induced myopathy. They found six loci at which SNPs were associated with a significant effect of statins on gene expression. GATM was the only gene associated with lipid metabolism, with effects on expression of APOE and lipoprotein lipase. GATM knockdown led to reduced sterol-mediated regulation of a range of genes in hepatocytes, including HMGCR, SREBP2, and LDLR. Moreover, the GATM locus was associated with incidence of statin-induced myotoxicity in two separate populations. From these experiments, GATM would appear to act as a functional link between statin-mediated lowering of cholesterol and susceptibility to statin-induced myopathy.49 This very promising new pharmacogenomic development urgently needs to be followed up and replicated. Other genetic variants related to muscle diseases predisposing individuals to statin myopathy include rare mutations that cause clinical conditions defined by defective muscle function, such as malignant hyperthermia (RYR1), McArdle’s disease (PYGM), carnitine palmitoyltransferase II deficiency (CPT2), and myotonic dystrophy (DMPK), among several others reviewed in detail by Needham et al.38 It has been proposed that statin myopathy is more common in rare carriers of heterozygous mutations in this group of genes, which in the homozygous state result in clinical neuromuscular syndromes in the absence of statin intake.50 An anti-HMGCR antibody–mediated necrotizing myopathy has recently been described as a distinct, although rare, 8

mechanism for statin-associated myopathy. Incidence is estimated at two per million annually, although this may represent a sizable proportion of those with an autoimmune myopathy. This condition often persists after discontinuation of statin therapy and has been associated with certain polymorphisms in the human leukocyte antigen (HLA) locus, specifically the HLA-DRB1*11:01 allele, which has been strongly associated with anti-HMGCR autoimmune myopathy in patients of various ethnic backgrounds, whereas HLA-DQA1 and HLA-DQB6 are potentially protective.51 To summarize, increased systemic—and thus muscle— exposure to statins and their metabolites contributes to the risk of muscle side effects. Reliable identification of variables that affect the pharmacokinetics of a particular statin may prompt the selection of either a different statin with alternative mechanisms of distribution into the liver and clearance or simply a lower dose. On the other hand, patients with musclesusceptibility genes or those at risk for an anti-HMGCR autoimmune necrotizing myopathy may potentially be intolerant to many statins, even at the lowest doses. However, integrating this pharmacogenomic understanding into clinical algorithms or recommendations for dosing or switching is far removed from clinical application and remains the topic of ongoing investigations. CLINICAL APPLICATION OF PHARMACOGENOMICS DETERMINANTS OF STATIN RESPONSE: A DOSING ALGORITHM

There is emerging evidence that knowledge of genetic determinants of response to statins can be translated into a clinical algorithm to guide dosing decisions for statins based on individual genotypes. The proof of concept was provided by DeGorter et al.52, who prospectively studied a cohort of 299 patients taking various doses of atorvastatin or rosuvastatin, with further validation in a retrospective cohort. They identified dose, patient age, and pharmacogenetic variants in SLCO1B1 (for atorvastatin and rosuvastatin) and ABCG2 (for rosuvastatin only) as being significant predictors of statin plasma concentration in a multiple linear regression analysis. They found no association between plasma statin and lathosterol concentrations, an intermediate marker of cholesterol synthesis, modulated by statin effects. Furthermore, there was no difference in plasma rosuvastatin concentrations in patients who did or did not achieve their LDL cholesterol target level. Again, increased plasma statin concentrations and systemic exposure may not reflect hepatic exposure, particularly in the setting of dysfunctional or reduced expression of hepatic uptake transporters, which might explain the lack of incremental pharmacodynamic effect. On the other hand, increased systemic exposure could predispose individuals to muscle side effects, and DeGorter et al. developed an algorithm to predict the maximum dose of atorvastatin or rosuvastatin expected to result in plasma drug concentrations C polymorphism is associated with dose decrease or switching during statin therapy in the Rotterdam Study. Pharmacogenet. Genomics 24, 43–51 (2014). 44. Danik, J.S., Chasman, D.I., MacFadyen, J.G., Nyberg, F., Barratt, B.J. & Ridker, P.M. Lack of association between SLCO1B1 polymorphisms and clinical myalgia following rosuvastatin therapy. Am. Heart J. 165, 1008–1014 (2013). 45. Wilke, R.A. et al.; Clinical Pharmacogenomics Implementation Consortium (CPIC). The clinical pharmacogenomics implementation consortium: CPIC guideline for SLCO1B1 and simvastatin-induced myopathy. Clin. Pharmacol. Ther. 92, 112–117 (2012). 46. Becker, M.L. et al. Genetic variation in the ABCC2 gene is associated with dose decreases or switches to other cholesterol-lowering drugs during simvastatin and atorvastatin therapy. Pharmacogenomics J. 13, 251–256 (2013). 47. Oh, J., Ban, M.R., Miskie, B.A., Pollex, R.L. & Hegele, R.A. Genetic determinants of statin intolerance. Lipids Health Dis. 6, 7 (2007). 48. Ruaño, G. et al. Mechanisms of statin-induced myalgia assessed by physiogenomic associations. Atherosclerosis 218, 451–456 (2011). 49. Mangravite, L.M. et al. A statin-dependent QTL for GATM expression is associated with statin-induced myopathy. Nature 502, 377–380 (2013). 50. Baker, S.K., Vladutiu, G.D., Peltier, W.L., Isackson, P.J. & Tarnopolsky, M.A. Metabolic myopathies discovered during investigations of statin myopathy. Can. J. Neurol. Sci. 35, 94–97 (2008). 51. Mohassel, P. & Mammen, A.L. The spectrum of statin myopathy. Curr. Opin. Rheumatol. 25, 747–752 (2013). 52. DeGorter, M.K. et al. Clinical and pharmacogenetic predictors of circulating atorvastatin and rosuvastatin concentrations in routine clinical care. Circ. Cardiovasc. Genet. 6, 400–408 (2013). 53. Aslibekyan, S., Straka, R.J., Irvin, M.R., Claas, S.A. & Arnett, D.K. Pharmacogenomics of high-density lipoprotein-cholesterol-raising therapies. Expert Rev. Cardiovasc. Ther. 11, 355–364 (2013). 54. Frazier-Wood, A.C. et al. The PPAR alpha gene is associated with triglyceride, low-density cholesterol and inflammation marker response to fenofibrate intervention: the GOLDN study. Pharmacogenomics J. 13, 312–317 (2013). 55. Aslibekyan, S. et al. Variants identified in a GWAS meta-analysis for blood lipids are associated with the lipid response to fenofibrate. PLoS One 7, e48663 (2012). 56. Lai, C.Q. et al. Fenofibrate effect on triglyceride and postprandial response of apolipoprotein A5 variants: the GOLDN study. Arterioscler. Thromb. Vasc. Biol. 27, 1417–1425 (2007). 57. Brautbar, A. et al. Rare APOA5 promoter variants associated with paradoxical HDL cholesterol decrease in response to fenofibric acid therapy. J. Lipid Res. 54, 1980–1987 (2013). 58. Hegele, R.A., Guy, J., Ban, M.R. & Wang, J. NPC1L1 haplotype is associated with inter-individual variation in plasma low-density lipoprotein response to ezetimibe. Lipids Health Dis. 4, 16 (2005). 59. Simon, J.S. et al. Sequence variation in NPC1L1 and association with improved LDL-cholesterol lowering in response to ezetimibe treatment. Genomics 86, 648–656 (2005).

12

60. Kim, D.S. et al. Novel gene-by-environment interactions: APOB and NPC1L1 variants affect the relationship between dietary and total plasma cholesterol. J. Lipid Res. 54, 1512–1520 (2013). 61. Berthold, H.K., Laaksonen, R., Lehtimäki, T., Gylling, H., Krone, W. & GouniBerthold, I. SREBP-1c gene polymorphism is associated with increased inhibition of cholesterol-absorption in response to ezetimibe treatment. Exp. Clin. Endocrinol. Diabetes 116, 262–267 (2008). 62. Shitara, Y., Hirano, M., Sato, H. & Sugiyama, Y. Gemfibrozil and its glucuronide inhibit the organic anion transporting polypeptide 2 (OATP2/ OATP1B1:SLC21A6)-mediated hepatic uptake and CYP2C8-mediated metabolism of cerivastatin: analysis of the mechanism of the clinically relevant drug-drug interaction between cerivastatin and gemfibrozil. J. Pharmacol. Exp. Ther. 311, 228–236 (2004). 63. Seithel, A. et al. The influence of macrolide antibiotics on the uptake of organic anions and drugs mediated by OATP1B1 and OATP1B3. Drug Metab. Dispos. 35, 779–786 (2007). 64. Neuvonen, P.J., Niemi, M. & Backman, J.T. Drug interactions with lipid-lowering drugs: mechanisms and clinical relevance. Clin. Pharmacol. Ther. 80, 565–581 (2006). 65. Pang, J., Chan, D.C. & Watts, G.F. Critical review of non-statin treatments for dyslipoproteinemia. Expert Rev. Cardiovasc. Ther. 12, 359–371 (2014). 66. Thomas, G.S., Cromwell, W.C., Ali, S., Chin, W., Flaim, J.D. & Davidson, M. Mipomersen, an apolipoprotein B synthesis inhibitor, reduces atherogenic lipoproteins in patients with severe hypercholesterolemia at high cardiovascular risk: a randomized, double-blind, placebo-controlled trial. J. Am. Coll. Cardiol. 62, 2178–2184 (2013). 67. Cuchel, M. et al.; Phase 3 HoFH Lomitapide Study investigators. Efficacy and safety of a microsomal triglyceride transfer protein inhibitor in patients with homozygous familial hypercholesterolaemia: a single-arm, open-label, phase 3 study. Lancet 381, 40–46 (2013). 68. Goldberg, A.S. & Hegele, R.A. Cholesteryl ester transfer protein inhibitors for dyslipidemia: focus on dalcetrapib. Drug Des. Devel. Ther. 6, 251–259 (2012). 69. Lee, P. & Hegele, R.A. Current Phase II proprotein convertase subtilisin/ kexin 9 inhibitor therapies for dyslipidemia. Expert Opin. Investig. Drugs 22, 1411–1423 (2013). 70. Stein, E.A., Honarpour, N., Wasserman, S.M., Xu, F., Scott, R. & Raal, F.J. Effect of the proprotein convertase subtilisin/kexin 9 monoclonal antibody, AMG 145, in homozygous familial hypercholesterolemia. Circulation 128, 2113–2120 (2013). 71. Whirl-Carrillo, M. et al. Pharmacogenomics knowledge for personalized medicine. Clin. Pharmacol. Ther. 92, 414–417 (2012). 72. Oswald, S. et al. Disposition of ezetimibe is influenced by polymorphisms of the hepatic uptake carrier OATP1B1. Pharmacogenet. Genomics 18, 559–568 (2008). 73. Pisciotta, L. et al. Effect of ezetimibe coadministered with statins in genotype-confirmed heterozygous FH patients. Atherosclerosis 194, e116–e122 (2007). 74. Wang, J., Williams, C.M. & Hegele, R.A. Compound heterozygosity for two nonsynonymous polymorphisms in NPC1L1 in a non-responder to ezetimibe. Clin. Genet. 67, 175–177 (2005). 75. Miller, D.B. & Spence, J.D. Clinical pharmacokinetics of fibric acid derivatives (fibrates). Clin. Pharmacokinet. 34, 155–162 (1998).

www.nature.com/cpt

Pharmacogenomics, lipid disorders, and treatment options.

Statins form the backbone of lipid-lowering therapy in the prevention of cardiovascular disease. Numerous studies have evaluated the effect of genomic...
386KB Sizes 1 Downloads 3 Views