Pharmacogenetics in Chronic Heart Failure: New Developments and Current Challenges

Abstract

The individual patient responses to chronic heart failure (HF) pharmacotherapies are highly variable. This variability cannot be entirely explained by clinical characteristics, and genetic variation may play a role. Therefore the purpose of this review is to 1) summarize the background pharmacogenetic literature for major HF pharmacotherapy classes (i.e. beta-blockers, angiotensin converting enzyme inhibitors, digoxin, and loop diuretics), 2) evaluate recent advances in the HF pharmacogenetic literature in the context of previous findings, and 3) discuss the challenges and conclusions for HF pharmacogenetic data and its clinical application.

Introduction

Large clinical trials demonstrate, on average, that pharmacotherapy significantly decreases morbidity and/or mortality due to heart failure (HF). However, the individual patient responses to HF pharmacotherapies are highly variable. For example, long-term treatment with angiotensin converting enzyme (ACE) inhibitors fails to suppress angiotensin II in as many as 15% of HF patients, and aldosterone in 38% [1]. Long-term optimal dosing of beta blockers fails to improve left ventricular ejection fraction (LVEF) >5% in as many as 43% of HF patients [2]. The maintenance dose of loop diuretics can range from no diuretic at all to >400mg furosemide equivalents [3]. Even when dosed according to age, sex, weight, renal function, and concomitant pharmacotherapy, the serum concentration of digoxin can range from 0.5ng/mL to >2.0ng/mL [4].

This wide variability in response to HF pharmacotherapies is not entirely explained by clinical characteristics, which is evident in large clinical trials where there is a similar response among most clinical sub-groups [5,6]. Genetic variation may additionally contribute to differences in drug response, the study of which is referred to as “pharmacogenetics” or “pharmacogenomics” [7]. Pharmacogenetics has proven successful in other therapeutic areas [8], but whether it can be used to improve the application of pharmacotherapies for HF remains unproven. Therefore the purpose of this review is to 1) succinctly summarize the background pharmacogenetic literature for major classes of HF pharmacotherapy, 2) critically evaluate the most current HF pharmacogenetic literature in this context, and 3) discuss the conclusions and remaining challenges to clinical application of pharmacogenetics in HF.

Beta-Blocker Pharmacogenetics: Background (Literature prior to 2010)

Compared to any other class of HF pharmacotherapy, HF pharmacogenetic studies have focused on beta blockers (BBs) much more than any other. Twenty-five studies were published between 2000 and 2009. Much BB pharmacogenetic data comes from small (n<400) observational cohorts of HF patients with systolic dysfunction. These studies are heterogeneous in many aspects (e.g. design, size, endpoint, patient population, specific BB, and the genetic variants tested), making definitive conclusions difficult. A list of genetic variants associated with BB (and other HF pharmacotherapies) response is displayed in Table 1, and the major findings are discussed herein.

Table 1.

Genetic variants associated with HF pharmacotherapy response

GENE	VARIANT	rsID	BENEFICIAL ALLELE	REFERENCES
BETA BLOCKERS
ACE	Ins/Del	rs1799752	Del	[42, 69]
ADRA1D	T1848A	rs2236554	A	[27]
	A1905G	rs709024	G	[27]
ADRA2C	Ins/Del	rs61767072	Ins	[36]
			Del	[31]
ADRB1	Ser49Gly	rs1801252	Gly	[11, 13, 70-72]
	Arg389Gly	rs1801253	Arg	[2, 10, 11, 13, 14, 17, 25, 31, 35, 37, 71-73]
			Gly	[37]
ADRB2	Gln27Glu	rs1042714	Glu	[23-25, 37]
	Thr164Ile	rs1800888	Thr	[74]
EDN1	G/A (IVS)-4	rs2071942	G	[75]
	Lys198Asn	rs5370	Lys	[75]
GRK5	Gln41Leu	rs17098707	Gln	[33, 34]
NET	T-182C	rs2242446	T	[27]
ACE INHIBITORS
ACE	Ins/Del	rs1799752	Ins	[39-42, 45]
AGTR1	A1166C	rs5186	A	[45]
CYP11B2	T-344C	rs1799998	C	[43]
ALDOSTERONE ANTAGONISTS
ACE	Ins/Del	rs1799752	Ins	[76]
ANGIOTENSIN RECEPTOR BLOCKERS
AGTR1	A1166C	rs5186	C	[77]
HYDRALAZINE/ISOSORBIDE DINITRATE
CYP11B2	T-344C	rs1799998	C	[78]
NOS3	Glu298Asp	rs1799983	Glu	[79]
*LOOP DIURETICS
ACE	Ins/Del	rs1799752	Del	[52]
ADD1	Gly460Trp	rs4961	Trp	[52]
ANP	Val32Met	rs5063	Val	[52]
	Ter152Arg	rs5065	Arg	[52]
CYP2C9	CYP2C9*1/*2/*3	n/a	*3	[46]
GNB3	C825T	rs5443	C	[52]
SCNN1B	Most frequent haplotype vs. others	n/a	Most frequent haplotype	[50]
SCNN1G	C4G	rs5723	C	[50]
SLC12A3	Gly264Ala	rs1529927	Ala	[50]

^*Data from healthy volunteers

ACE = angiotensin converting enzyme, ADD1 = alpha adducin 1, ADRA1D = alpha-1d adrenergic receptor, ADRA2C = alpha-2c adrenergic receptor, ADRB1 = beta-1 adrenergic receptor, ADRB2 = beta-2 adrenergic receptor, AGTR1 = angiotensin II receptor type 1, Arg = arginine, ANP = atrial natriuretic peptide precursor, Asn = asparagine, CYP11B2 = cytochrome P450 family 11 subfamily B polypeptide 2, CYP2C9 = cytochrome P450 family 2 subfamily C polypeptide 9, Del = deletion, EDN1 = endothelin-1, Gln = glutamine, Glu = glutamic acid, Gly = glycine, GNB3 = guanine nucleotide binding protein beta polypeptide 3, GRK5 = G-protein coupled receptor kinase 5, Ile = isoleucine, Ins = insertion, IVS = intervening sequence, Leu = leucine, Lys = lysine, Met = methionine, NET = norepinephrine transporter, NOS3 = nitric oxide synthase 3, SCNN1B = sodium channel nonvoltage-gated 1 beta, SCNN1G = sodium channel nonvoltage-gated 1 gamma, Ser = serine, SLC12A3 = solute carrier family 12 member 3, Ter = termination, Thr = threonine, Trp = tryptophan, Val = valine

Type 1 β-adrenergic receptor

Stimulation of the cardiac β-adrenergic receptors results in increased heart rate and contractility. The type 1 β-adrenergic receptor (ADRB1) is the primary target of cardiac BBs and has been the focus of the majority of BB pharmacogenetic literature. A non-synonymous variant in this gene, Arg389Gly, has been the most-studied (Table 1). This variant is common in the general population, with racial differences in its frequency (Table 2). Functionally, Arg389 has greater basal and agonist-stimulated activity compared to Gly389 [9]. Since Gly389 generally results in less ADRB1 sympathetic stimulation, it is raises the question of whether this variant is protective in BB-naïve HF patients, and that patients with Arg389 would receive greater benefit from BBs.

Table 2.

Frequencies of genetic variants associated with HF pharmacotherapy response

GENE	VARIANT	rsID	MINOR ALLELE	CAUCASIAN	AFRICAN-AMERICANS
ABCB1	C3435T	rs1045642	C	45%	87% (Africans)
ACE	Ins/Del Intron 16	rs1799752	Ins	44%	43%
ADD1	Gly460Trp	rs4961	Trp	17%	11%
ADRA1D	T1848A	rs2236554	A	46%	12% (Africans)
	A1905G	rs709024	A	38%	70% (Africans)
ADRA2C	Ins/Del 322-325	rs61767072	Del	4%	43%
ADRB1	Ser49Gly	rs1801252	Gly	17%	25%
	Arg389Gly	rs1801253	Gly	27%	38%
ADRB2	Gly16Arg	rs1042713	Arg	40%	50% (Africans)
	Gln27Glu	rs1042714	Glu	42%	20%
	Thr164Ile	rs1800888	Ile	2%	<2% (Africans)
AGTR1	A1166C	rs5186	C	25%	5%
ANP	Val32Met	rs5063	Met	4%	2%
	Ter152Arg	rs5065	Ter	14%	43% (Africans)
CYP11B2	T-344C	rs1799998	C	43%	30%
CYP2C9	*2	N/A	N/A	14%	<1%
	*3	N/A	N/A	11%	<1%
EDN1	G/A (IVS)-4	rs2071942	A	29%	17% (Africans)
	Lys198Asn	rs5370	Asn	30%	17%
GNB3	C825T	rs5443	T	39%	91% (Africans)
GRK5	Gln41Leu	rs17098707	Leu	2%	24%
NET	T-182C	rs2242446	C	25%	15% (Africans)
NOS3	Glu298Asp	rs1799983	Asp	22%	7% (Africans)
SCNN1B	Most frequent haplotype	N/A	n/a	65%	n/a
SCNN1G	C4G	rs5723	G	28%	24% (Africans)
SLC12A3	Gly264Ala	rs1529927	Ala	8%	<1% (Africans)

Consistent in the literature, patients possessing the Arg allele have greater LVEF improvement in response to BBs than those possessing Gly389. This comes from several prospective and retrospective studies totaling 569 patients with a variety of HF etiologies, ethnicities, and BBs [10-13]. There is also evidence to support the influence of Arg389Gly on the survival benefit from BBs. The most convincing is a large (n=1040) pharmacogenetic sub-study [14] of the Beta-Blocker Evaluation of Survival Trial (BEST) [15]. Patients homozygous for Arg389 had a statistically significant improvement in survival with bucindolol compared to placebo (HR=0.62; p=0.03), whereas Gly389 carriers did not (HR=0.90; p=0.57). It is argued whether the results for bucindolol can be applied to other BBs because of its unique pharmacologic properties [16]. While these results were replicated in a prospective, observational study of 201 HF patients treated with metoprolol or carvedilol [17], other larger cohort studies have not found this association [18]. A pharmacogenetic sub-study of MERIT-HF [5] also did not find an association of Arg389Gly with survival benefit regardless of treatment (metoprolol CR/XL or placebo) [19]. However this last study did not test BB effect within genotype groups (as was done in BEST), which may help explain the discordant results.

Type 2 β-adrenergic receptor

Although not the primary target of BBs, the type 2 β-adrenergic receptor (ADRB2) is present in myocardium, can mediate inotropic response, and while ADRB1 is down-regulated the expression of ADRB2 is unchanged in the failing heart [20]. The most-studied variant in ADRB2 is Gln27Glu. Functionally, Glu27-containing ADRB2 is resistant to agonist-promoted desensitization compared to Gln27 [21,22]. This suggests that Gln27 genotype is associated with less sympathetic output relative to Glu27, but the clinical pharmacogenetic literature is inconsistent. Several small studies showed a favorable LVEF response for patients carrying Glu27 compared to patients homozygous for Gln27 [23-25]. However, four other small studies failed to show a significant association [11,12,26,27]. Although, these included β1-selective BBs, which might limit the ability to detect an interaction with ADRB2 variants. In terms of survival benefit, one study showed a survival difference by genotype among BB-treated HF patients [28], but several large cohort studies have not demonstrated an association [18,26,29]. Notably, the majority of patients in these studies were treated with BB, limiting the ability to examine true pharmacogenetic interactions.

Adrenergic receptor alpha-2C

The alpha-2C (ADRA2C) adrenergic receptor is presynaptic, auto-inhibiting norepinephrine release. A deletion variant in ADRA2C, causing a loss of amino acids 322-325, results in the loss of normal auto-inhibitory function and increased norepinephrine [30]. A potential interaction between ADRB1 Arg389, the ADRA2C deletion, and BB response was prospectively studied in 54 HF patients with systolic dysfunction [31]. The deletion-carriers had a greater improvement in LVEF compared to insertion homozygotes (+6% versus +1%; p=0.045). Synergy between the ADRB1 and ADRA2C variants was supported by the magnitude of results. No association was found in 80 IDC patients [27], but synergy with ADRB1 Arg389 was not tested.

G-protein coupled receptor kinase 5

The function of the G-protein receptor kinases (GRK) is to desensitize ligand-occupied G-protein coupled receptors such as β-adrenergic receptors [32]. GRK5 is abundant in the heart and a Gln41Leu variant in this gene has been studied in vitro, with the Leu41 version more effectively desensitizing agonist-stimulated responses compared to Gln41 [33]. Because the Gln41 subtype should have more active β-adrenergic receptors, it has been proposed that Leu41 is protective in BB-naïve HF patients, but these patients may be less responsive to BB.

In an observational study of 375 African-Americans with HF, only patients homozygous for Gln41 had significantly improved transplant-free survival with BB (HR=0.22; p<0.001) [33]. There was no difference in outcome for patients carrying Leu41 with or without BB (HR=0.78; p=0.53). Similar results were found in another study featuring 711 African-American HF patients [34]. Among all of the African-American patients, the BB treatment effect did not reach statistical significance (HR=0.698; p =0.1). However in a genetic sub-group of ADRB1 Gly389 homozygous/GRK5 Gln41 homozygous African-Americans, BBs were associated with marked mortality benefit (HR=0.385; p=0.012). Interestingly, when the African-Americans and Caucasians were matched by ADRB1 and GRK5 genotypes and BB treatment their survival times were similar. This suggests that genetic variants, rather than race, are the major factor contributing to the apparent differences in BB treatment effect between Caucasians and African-Americans.

Beta Blocker Pharmacogenetics: Recent Advances (Publication Year ≥2010)

As evidenced by the heterogeneous and observational nature of the background HF BB pharmacogenetic literature, this field is still in an early stage. The four most recent HF BB pharmacogenetic studies support previous insights, but they also demonstrate some unexpected results.

A small but intriguing study in 93 HF patients [35] studied a panel of both pharmacokinetic and pharmacodynamic genetic variants relevant to BBs (ADRB1[Arg389Gly], CYP2D6, and UGT1A1) and assessed them using a multidimensional BB response criteria. CYP2D6 and UGT1A1 are highly polymorphic metabolic enzymes for which carvedilol is a substrate, and metoprolol is mainly metabolized by CYP2D6. They defined BB response as meeting at least three out of five criteria: the 1) duration and 2) tolerability of dose titration, 3) an increase in NYHA functional class, 4) LVEF, or 5) 6-minute walk distance. There was no association between the panel of genetic variants and their BB response criterion, but there was a weak relationship between carvedilol dose and Arg389Gly status (p=0.012). Gly389 patients reached higher doses, perhaps indicating greater BB responsiveness in those with Arg389.

A recent prospective study of 183 patients with HF and three previously studied genetic variants (ADRB1[Arg389Gly] and ADRB2[Arg16Gly and Gln27Glu]) [25] had findings consistent with the background BB pharmacogenetic literature. The increase in LVEF after carvedilol tended to be greater in ADRB1 Arg389 homozygous (+7.8±7.6%) and heterozygous patients (+9.0±11.4%) compared to those homozygous for Gly389 (+4.1±7.6%) (p=0.0847). Subjects homozygous for ADRB2 Glu27 showed a greater increase in the LVEF (+13.0 ±12.2%), compared to both heterozygous (+7.1 ±8.1%) and Gln27 homozygous patients (+8.3 ±11.4%) (p=0.022). In multivariable analysis, cause of HF, systolic blood pressure, dose of carvedilol, and Gln27Glu genotype were significant correlates of LVEF improvement after carvedilol treatment. Notably, the ADRB1 Arg389Gly genotype was not independently informative in this dataset.

The ADRA2C insertion/deletion was tested in the genetic sub-study of BEST (n=1040) [36]. In contrast to previous data, which indicated that the ADRA2C deletion was associated with improved LVEF response after BB, this large and adequately powered study found no differential effect on LVEF by ADRA2C genotype. Interestingly, interaction with ADRB1 Arg389Gly was not tested. Even more surprising was that this study found that the insertion allele, and not the deletion, was associated with enhanced survival benefit from BB. For the all-cause mortality end point, bucindolol produced a strong tendency toward significance (p=0.025) for a reduction in mortality by 30% in the insertion homozygotes, while there was a non-significant (p=0.79) 9% increase in mortality in the bucindolol-treated deletion-carriers. There are several possible explanations for the discrepancy in results; the previous small studies could be false positive associations, or this could be a bucindolol-specific interaction.

A recent retrospective study of 586 HF patients examined differential pharmacogenetic interactions between carvedilol and metoprolol [37]. The investigators combined two genotypes, ADRB1 Arg389-homozygous and ADRB2 Gln27-carrier, and compared these patients to all others in terms of time-to-death from any cause. They found a significant interaction between genotype group and carvedilol treatment (p=0.003), but no interaction with metoprolol (p=0.61). In patients treated with carvedilol, survival was lower for Arg389/Gln27 group than the remaining patients (HR=2.30). Because two different variants defined these groups and the fact that one is associated with favorable BB response while the other is not, these results are difficult to put in context of the existing literature. Another concern is that there may have been negative results for metoprolol because the genotype groups were partly defined by an ADRB2 genotype, and metoprolol is β1-specific.

Overall these more recent results, like the pre-existing BB pharmacogenetic literature, are provocative but require validation in large, prospective clinical trials of genetic-guided BB therapy. There is sufficient evidence to support this approach, for bucindolol as well as the currently approved agents such as metoprolol and carvedilol, and this represents the most pressing challenge for BB pharmacogenetics in HF. Other areas that remain unclear are whether additional yet unidentified genes should be the focus of future research, whether other genes are relevant to BB effectiveness, and the interaction of race with genetics. The current candidate gene list revolves strictly around the receptor, and whether other genetic loci may directly modify response or interact with the current candidates is unknown. As pointed out above, all of the current candidate variants have frequencies that differ significantly by ancestry (Table 2), which raises the issue of both genetic and non-genetic confounding factors. Therefore additional investigation is required for confidence in these associations and the potential application to non-Caucasian populations.

ACE Inhibitor Pharmacogenetics: Background (Literature prior to 2010)

Our extensive searches reveal only six pharmacogenetic studies of ACE inhibitors in HF patients from 1998 to 2008. Like the BB literature, most of the ACE inhibitor pharmacogenetic literature in HF patients comes from small observational studies (n<200) that are heterogeneous in design. Not surprisingly, the literature has focused on the gene encoding the target of these agents, ACE. A 287 base pair insertion/deletion in intron 16 of ACE accounts for half of the variance in serum ACE levels [38]. The deletion results in significantly higher ACE levels [38]; therefore it was postulated that HF patients possessing the deletion would require a higher dose of ACE inhibitor to achieve the same inhibition. Most of the clinical pharmacogenetic studies are consistent with the functional effects, with respect to mean arterial pressure [39], aldosterone escape [40], serum ACE activity [41], and survival [42], but the association with LVEF is less clear [43,44].

In 107 idiopathic dilated cardiomyopathy patients receiving 2.5 years of ACE inhibitor therapy [43], the LVEF improvement was similar among ACE genotypes, but another study of 168 HF patients with systolic dysfunction [44] found that deletion-carriers had a greater LVEF improvement after ACE inhibitor (deletion-carriers LVEF +8.8 %; insertion homozygotes -1.73%; p=0.01). The discordance in results may be due to population-specific effects. In contrast, the relationship between the ACE variant and survival benefit from ACE inhibitors appears more clear based on the largest ACE inhibitor pharmacogenetic study in HF patients (n=479) [42]. There was a dose-dependent relationship between the ACE insertion/deletion and transplant-free survival. After a median follow-up of 33 months, patients on low-dose ACE inhibitors (≤50% of target dose) had poorer transplant-free survival associated with the deletion allele (RR for deletion homozygotes=2.07;p=0.03), and this was exaggerated in patients who were also not receiving a BB. However high-dose ACE inhibitor (>50% of target dose), with or without concomitant BB, eliminated the adverse effect of the ACE deletion. Although the deletion was associated with poorer transplant-free survival, it seemed that deletion homozygotes benefitted the most from ACE inhibitor and BB therapy.

ACE Inhibitor Pharmacogenetics: Recent Advances (Publication Year ≥2010)

There is little ACE inhibitor pharmacogenetic literature recently, with only one study published [45] within the last 2 years. This study is consistent with the previous findings on survival, and extends this to HF patients with preserved LVEF. This study enrolled 285 HF patients followed for ~7 years for all-cause mortality [45]. The deletion allele was associated with higher mortality in patients not receiving an ACE inhibitor (HR=2.23; p=0.008), but this impact was reduced among patients receiving an ACE inhibitor (HR=1.64; p=0.20). Acknowledging the limitation that the ACE pharmacogenetic literature comes entirely from observational cohorts, the sum of these data suggest that HF patients that are deletion-carriers may need higher dose ACE inhibition to achieve similar outcomes compared to insertion homozygotes. While at this point it seems unlikely that renewed interest will come to ACE inhibitor pharmacogenetics, a variety of intriguing points remain such as whether other genes are important, whether adverse events (e.g. angioedema and hyperkalemia) can be predicted based on genetics, or whether genetics can help guide combinations of therapies (e.g. BB +ACE inhibition vs. BB alone vs. ACE inhibition alone).

Pharmacogenetics of Loop Diuretics: Background (Literature prior to 2010)

There have been six studies investigating the association between genetic variation and loop diuretic response published from 2004 to 2008. Five of the studies were performed in healthy volunteers (HV), and one small study included patients with HF. Collectively, these studies demonstrated that genetic variants involved in the metabolism (CYP2C9), uptake (SLC22A11 and SLCO1B1), and action (SLC12A3, SCNN1B, and SCNN1G) of the loop diuretics can influence their pharmacokinetics and pharmacodynamics.

In 10mg torsemide single dose HV studies, the decreased function CYP2C9*3 allele had significant effects on torsemide pharmacokinetics and modest effects on pharmacodynamics [46]. The total oral clearance of torsemide was about 3-fold lower in CYP2C9 *3/*3 subjects compared to *1/*1 subjects. Sodium and chloride excretion were ~25% higher in carriers of one CYP2C9*3 allele after torsemide administration [46]. HVs homozygous for the two most frequent haplotypes of SLC22A11 (gene encoding the organic anion transporter 4 [OAT4]) had an 80% difference in the renal clearance of torsemide [47], and HVs homozygous for 521T, heterozygous, and homozygous for 521C in SLCO1B1 (gene encoding the organic anion transporting polypeptide 1B1 [OATP1B1]) had torsemide oral clearances estimated as 62, 46, and 41mL/min, respectively (p<0.001) [48]. Taken together, variants in CYP2C9, OATP1B1, OAT1, and OAT4 explain nearly 50% of torsemide pharmacokinetic variation [49].

Three renal sodium reuptake transporters are the primary targets of the loop diuretics: NKCC2 (gene: SLC12A1), NCC (gene: SLC12A3), and ENaC (genes: SCNN1A, SCNN1B, and SCNN1G) [50]. A three-period crossover study was performed in 97 HVs using single oral doses of bumetanide (2mg), furosemide (80mg), and torsemide (10mg) to evaluate the influence of variation in these on diuretic response [50]. There were three significant associations with the 24-hour excretion, and this was consistent among the three loop diuretics: 1) Subjects homozygous for Ala264 in SLC12A3 excreted an average 32% more chloride and 42% more potassium compared to those homozygous for Gly264. 2) Subjects homozygous for the most frequent haplotype in SCNN1B excreted 24% more volume, 13% more sodium, and 13% more chloride compared to subjects without the most frequent haplotype. 3) Subjects homozygous for G4 at a synonymous C4G substitution in SCNN1G excreted 23% less volume and 24% less calcium compared to subjects homozygous for the C4 allele.

The loop diuretic pharmacogenetic data discussed to this point involves single-dose studies performed in HVs, but the data in HF patients receiving steady-state dosing appears consistent with the HVs. In a small, open-label pharmacokinetic study of 24 patients receiving stable doses of 10mg daily torsemide (n=18 with arterial hypertension and n=6 with HF) [51], the primary endpoint was area under the plasma concentration-time curve during the 24-hour dosing interval at steady state (AUC_24,ss). CYP2C9 genotype, SLCO1B1 genotype, and gender independently predicted AUC_24,ss. Similar to HVs, HF patients with the CYP2C9 *1/*3 genotype had a mean AUC_24,ss 46% greater than those with the *1/*1. Patients heterozygous for T521C in SLCO1B1 had a 38% increase in AUC_24,ss compared to patients homozygous for 521T (no 521C homozygotes found).

Pharmacogenetics of Loop Diuretics: Recent Advances (Publication Year ≥2010)

Despite the interesting associations above, whether these differences in pharmacokinetic parameters across genotypes translate into clinically meaningful differences in drug effectiveness in patients with HF remains unknown. One recent study attempted to answer this question. This was a randomized, single-blind, three-arm, triple-crossover study in 95 HVs [52] who received a single oral dose of bumetanide (2mg), furosemide (80mg), and torsemide (10mg) at 2 weeks intervals. Together, eight genetic variants had an impact of 20, 15, 10, and 23% on the variation in the urinary excretion of sodium, volume, potassium, and calcium. Thus, genetic variation seems to importantly impact not only pharmacokinetics of loop diuretics but also their clinical effect. The incorporation of genetic data may help in determining diuretic dosing, though the clinical situation where this would be necessary is not obvious. Important questions which remain to be addressed are whether genetic variation can predict worsening renal function associated with chronic diuretic therapy, or even the risk of recurrent hospitalization.

Digoxin Pharmacogenetics: Background (Literature prior to 2010)

Digoxin is a narrow therapeutic index drug with recommended serum digoxin concentration (SDC) being 0.7-0.9 ng/mL [53]. The pharmacokinetics, and therefore SDC, may be affected by genetic variation. Indeed, a study in monozygotic and dizygotic twins estimated the genetic component contributing to digoxin AUC_0-12 to be 89% [54]. The seminal paper investigating the influence of genetic variation on digoxin pharmacokinetics was published in 2000 [55], and it has been followed by 12 more studies with inconsistent results.

The majority of the digoxin pharmacogenetic literature has focused on a common C3435T variant in ABCB1 (Table 2). ABCB1 encodes for P-glycoprotein (P-gp), an efflux protein for which digoxin is a substrate. In the seminal paper, subjects homozygous for 3435T (n=5) had >2-fold lower expression of P-gp in the duodenum compared to subjects homozygous for 3435C (n=6; p=0.056). As would be expected, subjects homozygous for 3435T had 38% higher SDC than subjects homozygous for 3435C (p=0.006). Acknowledging this small sample size, these results have been consistent with six subsequent studies in HVs [56-61], as well as a population-based study of 195 “digoxin-users” (HF status not reported) [62]. Despite this seemingly consistent line of evidence, some contrasting data has arisen. Another study of 39 HF patients found no difference in steady-state SDC among C3435T genotypes, which was consistent with another study in 50 HVs [63] as well as a meta-analysis [64]. Unfortunately, this small HF study did not control for differences in renal function, which can vary widely among patients with HF.

Overall the preponderance of evidence favors an impact of genetic variation on digoxin pharmacokinetics. Despite this, whether the difference in digoxin pharmacokinetics by C3435T genotypes is clinically meaningful in patients with HF is not established. Adding to this complexity is that the association of the C3435T genotype with digoxin pharmacokinetics may depend on ethnicity. Two studies in Japanese subjects found a reverse association, in which the 3435C genotype had higher SDC [65,66].

Digoxin Pharmacogenetics: Recent Advances (Publication Year ≥2010)

Only a single study, with a unique post-mortem SDC study design (n=112), was identified in our searches regarding digoxin pharmacogenetics in the past 2 years. This study’s results are consistent with the notion that ABCB1 3435T confers higher SDC [67], but adds to the existing data by suggesting that the interactions with C3435T may be gender-specific. There was a relationship between the frequency of 3435T allele and post-mortem SDC, but surprisingly the results were driven by females. If true, this relationship could help explain previous data demonstrating higher mortality in women treated with digoxin versus placebo due to differences in SDC between women and men [68]. Validation of this finding in adequately sized, prospective, gender-specific cohorts is needed. If validated, one could envision using genotype to identify patients at higher risk of toxicity who should not receive digoxin, receive reduced dosing, or more intense drug level monitoring.

Future Challenges

While there are many gaps in the HF pharmacogenetic knowledge base, limiting its current clinical application, we have already learned a great deal from the relatively modest body of HF pharmacogenetic literature. For example, while genetic variant functional/mechanistic effects need to be demonstrated to truly establish causation, this is not sufficient as these associations do not always translate into clinical effects. Another important insight is that there can be synergy/interaction between multiple genetic variants as is the case for ADRB1 Arg389Gly and ADRA2C insertion/deletion with LVEF response to BBs. Even more daunting is the complexity and specificity of some pharmacogenetic associations. HF pharmacogenetic associations may be race-specific (e.g. GRK5 Gln41Leu and BB response in African-Americans), dose-specific (e.g. ACE insertion/deletion and low- and high-dose ACE inhibitors), gender-specific (e.g. ABCB1 C3435T and SDC in women), and drug-specific (e.g. ADRB1/ADRB2 and response to carvedilol but not metoprolol).

Despite the fact that pharmacogenetics has been in existence for decades, the number of studies on HF therapies are still relatively small and we are yet in the early stages of this part of the field. This foundation has yielded the important insights above and also provided numerous improvements in approach both in terms of genotyping and analysis. This has set the stage for accelerated advances moving forward. At this point, some of the key knowledge gaps include: 1) Lack of foundational pharmacogenetic data regarding angiotensin receptor blockers, aldosterone antagonists, and hydralazine/isosorbide dinitrate; 2) Investigation into acute or intravenous HF pharmacotherapies; 3) Better understanding of multi-variant / multi-drug combinations on pharmacogenetics. Most importantly, what is broadly needed in order to make real progress are prospective intervention clinical trials where a genetic-guided approach is compared to empiric therapy. These are required to validate proposed associations and establish that a specific response to the genetic information can improve treatment outcomes. Because HF is a fatal and common disease requiring polypharmacy, any information (even genetic) that could improve HF pharmacotherapies would give profound patient and public health benefit.