Selective genetic advantages for users of thiazide diuretics. Is there a case for the 460Trp variant of α-adducin?

Introduction According to recent guidelines for the management of arterial hypertension, five major classes of antihypertensive agents – thiazide diuretics, calcium antagonists, angiotensin-converting enzyme (ACE) inhibitors, angiotensin receptor antagonists and β-blockers – are suitable for the initiation and maintenance of antihypertensive treatment, either alone or in combination [1]. Moreover, the Blood Pressure (BP) Lowering Treatment Trialists' Collaboration earlier confirmed significant effects of BP lowering on major cardiovascular events [2]. Although controversial, the main benefit of antihypertensive therapy does not seem to be attributed to the class or property of the molecules but to the magnitude of the achieved BP reduction during treatment [3]. Not withstanding the documented BP reducing capacity of each class of antihypertensive drugs, there are large inter-individual responses in drug metabolism, pharmacokinetics and drug efficacy over and above those induced by age, sex, renal and hepatic function. Worldwide, the majority of elderly patients with systolic hypertension cannot be brought below the 140 mmHg limit by only one drug. A number of ‘intellectual’ choices of single and combined therapies are fully recognized, but trial-and-error drug selection still is mandatory for quite a number of hypertensive patients. The former observations constitute the background and playground for the rapidly growing field of pharmacogenetics in the management of hypertension. During the last decade, a plethora of both positive and negative studies tried associating hypertension with single nucleotide polymorphisms (SNPs) of candidate genes. One of the most frequently and best-studied candidate genes is the α-adducin gene. A hypothesis of selective advantages on BP control and cardiovascular outcomes has been launched for its Gly460Trp polymorphism. Although attractive and sustained by experimental and clinical data, the validity of linkage and association of adducin with hypertension, BP control and outcome data as well as implications for pharmacogenetic tailoring of BP reducing drugs (such as diuretics) have been questioned a lot. In humans, three genes, ADD1, ADD2 and ADD3 respectively mapping to 4p16.3, 2p14-p13 and 10q24.2-24.3, encode three subunits of a heterodimeric cytoskeleton protein called adducin. The adducin protein combines one α (103 kDa) subunit with either a β (97 kDa) or a γ (90 kDa) subunit. Two polymorphisms of the α-subunit gained particular interest during the last decade: Gly460Trp (rs4961), also termed G460W, and the less studied S586C. Prevalence of the 460Trp variant shows strong regional and racial variations. In their landmark study, Cusi et al.[4] associated the 460Trp allele with a salt-sensitive form of hypertension. The basic physiological functions of the calcium and calmodulin-dependent adducin complex comprise actin polymerization, formation of the actin–spectrin lattice and activation of the renal Na+-K+ ATPase. The 460Trp allele shows a greater renal Na+-K+ pump activity in nephron segments proximal to the macula densa [5]. Deficient endocytosis of the Na+-K+ pump has been incriminated as an important factor contributing to the enhanced constitutive renal tubular sodium reabsorption in humans carrying the 460Trp allele [6,7]. Endogenous plasma ouabain positively correlates with the mutation in the adducin gene. Its role has to be clarified [8]. Deleterious effect of 460Trp on hypertension In rat models, transfer of ADD1 from the Milan hypertensive strain to the Milan normotensive strain raises BP in the latter. In humans, the situation is less clear. Confounding effects may comprise ethnicity, lifestyle and environmental factors. About half of the association studies and the majority of linkage studies applying stringent criteria found an association between the 460Trp allele and being hypertensive. None of the studies in exclusively or predominantly normotensive populations found a sound and straightforward association between 460Trp and BP level itself. If any association was observed, it had to be attributed to the small hypertensive part of these populations. A further body of evidence in both salt-sensitive and the more general form of hypertension largely came from the team of Giuseppe Bianchi in Milan [9,10]. Salt-sensitive form of hypertension Beeks et al.[11] systematically reviewed the literature from 1993 till 2003, and showed an apparent association between a salt-sensitive form of hypertension and the adducin 460Trp allele. Salt sensitivity was studied by BP response to acute salt loading, chronic diuretic treatment or standardized diet. Six (all Italian cohorts) out of the seven studies confirmed the role of 460Trp. Beeks et al.[11] added two important considerations: possibly some geographical bias as most evidence came from one research group, and the association is a proxy for low-renin hypertension as the 460Trp allele has been associated with lower renin levels [10,12]. Hypertension in general A myriad of studies has been published during the last decade on the relationship between the 460Trp allele and hypertension in the broader sense. Bianchi et al.[10] outstandingly reviewed 63 of these studies. The conclusions of studies in Caucasian, African American, Chinese and Japanese populations were contradictory. Unequivocal evidence supporting an association was lacking. Things have not changed a lot since 2001 when Bianchi and Cusi [9] ended a review with the rhetorical question: ‘Is the glass half full or half empty?’ Also, in the recent Dutch Erasmus Rucphen Family study (a genetically isolated population), the ADD1 polymorphism explained only 0.2% of the systolic BP variance, 0.1% of diastolic BP variance and systolic BP was not highest for the 460Trp allele [13]. Of interest, the association of 460Trp with ion exchange seems much closer than with BP levels. Carriers of the 460Trp variant have a faster Na+-K+-Cl− cotransport, Na+-H+ countertransport and Na+-K+ pump function and reduced erythrocyte sodium content [14]. Clinical outcomes according to the 460Trp allele The Gly460Trp polymorphism has been proposed as a potential determinant of cardiovascular outcome (myocardial infarction or stroke) and renal disease. Once again, the group of Bianchi was pioneering some of the first studies. In their review [10], they considered the evidence that relates mutations in the adducin gene with cardiovascular and renal disorders as ‘overwhelming’. In 2005, 12 out of 16 studies found that the ADD1 polymorphism alone (or in combination with that of ACE, see below) was associated with stroke, coronary events, renal disease or vascular dysfunction. At that time, the well taken conclusion from the Bianchi review was: ‘When context is taken into account, the impact of adducin in hypertension and its related disorders is clear’. Several more recently published papers further emphasized the role of the 460Trp allele in determining outcome, although others pointed to a more balanced, mixed or even contrasting message. At least there is room for some concern with a number of the earlier and more recent conflicting association studies as straightforward interpretation is hampered by the cross-sectional and retrospective design of most of these studies. The evidence for increased risk of coronary heart disease and stroke Morrison et al.[15] showed in the Atherosclerosis Risk in Communities (ARIC) study that the ADD1 460Trp variant carried a multiadjusted risk of 2.30 [hazard ratio 95% confidence interval (CI) 1.20–4.42] for coronary heart disease (CHD) and of 2.61 [95% CI 1.27–5.37] for peripheral arterial disease. The study of Morrison et al.[15] in hypertensive patients should be weighted against reports in hypertensive and other populations showing no association for the 460Trp allele in the prediction of CHD or myocardial infarction (MI) [16–18]. Again in the prospective ARIC study, Morrison et al.[19] found no association either for clinical or for sub-clinical stroke. Of interest, an opposing protective effect on stroke has been reported in the study of Psaty et al.[16]. In controls, 4.8% carried the 460Trp variant against only 0.9% in stroke patients (P = 0.02). The large Rotterdam Study [20] in the elderly revealed a significant association of ischemic stroke and MI in contrast to a nonsignificant association for hemorrhagic stroke. The most recent prospective cohort of more than 15 000 initially healthy Dutch women [18] shows a mixed (more balanced?) picture. The risk of ischemic stroke in the variant allele was 3.9 times higher than in patients with the common genotype in a dominant inheritance model. The study indicated that α-adducin might act in the cause of stroke. Although essentially independent of BP values, the risk was particularly expressed in women with systolic hypertension. Another striking finding in that study was the absence of an association for MI and CHD. In depth pharmacogenetic evaluations of these studies are discussed below. Powered mortality data are scarce. At Belgian population level [17], although nonsignificant for MI and stroke, the hazard ratio for total mortality associated with 460Trp versus Gly460 was 2.30 in stage 2 systolic hypertension. In some studies, the effect of 460Trp on stroke was independent of BP or phases of hypertension. It may point towards other underlying mechanisms or epistatic effects of other genes (such as ADD2, ACE). These effects may even be of greater importance, but are out of the scope of this editorial comment as is the role of 460Trp in different kinds of renal disease [10]. Beneficial pharmacogenetic interaction of 460Trp with thiazide diuretics The confusion did not diminish when pharmacogenetics entered the playground. The rationale for a pharmacogenetic approach in adult hypertensive patients is that thiazide diuretics, in the presence of constitutive enhanced tubular sodium reabsorption, should trigger fewer counterregulating mechanisms, leading to an ‘optimal’ BP effect. Thus, a clinically relevant interaction between 460Trp and the mode of action of thiazide diuretics seems plausible. Essentially, these interactions should be explored for two endpoints: BP level and outcome variables. Both a greater fall in BP and a differential effect on endpoints are expected for the 460Trp variant. In this issue of the Journal of Hypertension, van Wieren-de Wijer et al.[21] report a nested case-control study in the PHARMO database from the Netherlands. PHARMO is a population-based large registry covering drug dispensing of about two million inhabitants. The study question was whether the Gly460Trp polymorphism of α-adducin modifies the effect of thiazide diuretics on the risk of nonfatal MI. The authors compared 613 patients with first MI and at least one prescription of an antihypertensive drug with 3627 controls matched for a number of factors but not hospitalized for MI. Among patients with the 460Trp variant, risk of MI was similar for users of thiazide versus users of other drugs. Among patients with the wild type, the risk of MI was lower in users of diuretics. However, the interaction term between use of diuretics and Gly460Trp for MI was not significant. The authors concluded that the α-adducin gene does not play an important role in modifying the risk of nonfatal MI associated with the use of thiazide diuretics. The paper of van Wieren-de Wijer et al.[21] is not the first negative study on the interaction of diuretics with the Gly460Trp polymorphism. Some of the coauthors of the van Wieren-de Wijer et al. study [21] were involved in genetic studies performed in cohorts in the Netherlands. They also documented lack of interaction. In the elderly of the Rotterdam Study [22], the Gly460Trp polymorphism did not modify the difference in BP levels among patients who used diuretics. Another recent study by Schelleman et al.[23] in the same Rotterdam study failed to document an interaction of Gly460Trp with the use of diuretics on the prevention of stroke or MI. In addition, in a study of African American and white adults, Turner et al.[24] concluded that no relationship exists between adducin genotypes and BP responses to diuretics. From a historical point of view, however, the story of pharmacogenetic studies on diuretics and adducin was initially promising. Three early Italian studies [4,25,26] partly favored an effect of the 460Trp allele on the BP/diuretic effect. The excellent design in newly discovered and never-treated hypertension may have contributed to the positive results. In the study of Psaty et al.[16], there was a selective pharmacogenetic advantage of diuretics in preventing the combined outcome (stroke or MI) in hypertensive patients carrying 460Trp who took diuretics versus other therapies for similar BP reductions. It did not hold true for the Gly460 allele. The most recent study in this field is a substudy of the INVEST-GENES [27] that analyzed 5661 patients with hypertension and coronary artery disease. Significant association of cardiovascular outcome with 460Trp variant was limited to blacks. The effect of diuretic use on the risk of cardiovascular outcomes did not vary by 460Trp carrier status. Study limitations and future perspectives First, relevant endpoint analyses are mandatory. The lack of data on mortality and stroke is the major limitation of the study of van Wieren-de Wijer et al.[21]. A study design on the interaction of diuretics and Gly460Trp polymorphism (a hypertension-related polymorphism) on an outcome measure neglecting stroke (a hypertension-related outcome) is remarkable. Secondly, the epistatic effects from other genes (such as ACE I/D) should be further explored in outcome studies of patients treated with diuretics or ACE inhibitors. There might be a role for genome-wide scanning and environmental factors. Thirdly, adducin acts as a heterodimer and usually only one adducin unit (α) is dealt with. Thus, the interactions among the three subunits as well as tissue-specific splicing isoforms should be considered in the pharmacogenetic approach. Conclusion The 460Trp allele of α-adducin has been associated in several studies with hypertension and increased risk of cardiovascular events. The conclusions from the 2005 review by Bianchi et al.[10] most likely should be modified to some extent, but are not rejected by recent publications that are at odds. There are at least good theoretical arguments and a few convincing clinical observations that the 460Trp variant beneficially modifies the risk in users of thiazide diuretics. Does the study of van Wieren-de Wijer et al.[21] provide the readership with a definite quintessentially negative answer to this pharmacogenetic paradigm? Clearly the answer is no, and further clarifications are required. Clinicians are still far away from sound recommendations on which patients will benefit most from thiazide diuretics (or some other class of antihypertensive drugs) based on pharmacogenetic insights from the adducin gene. It is not excluded that studying other genes and polymorphisms might be more rewarding. Nonetheless, the pharmacogenetic approach is and will remain a major research topic in clinical pharmacology. Prospective, controlled studies to demonstrate added benefits of incorporating genotype-guided therapy have already been carried out with success or are planned in other fields of medicine such as in warfarin therapy. In the same line, further reinforced efforts are needed to come to a better tailored drug therapy of essential hypertension leading to a better BP control, fewer adverse effects, better drug compliance, a more cost-effective treatment of hypertension and ultimately to fewer cardiovascular events. Is there a pharmacogenetic case for the 460Trp variant of α-adducin? For the time being, the hypothesis may appear rejected, but its ‘de-novo resubmission’ is strongly encouraged.