The Role of α-Adducin Polymorphism in Blood Pressure and Sodium Handling Regulation May Not Be Excluded by a Negative Association Study
Abstract—The basic requirement for declaring an association study positive is that the “hypertension-favoring” allele is more frequent in hypertensive cases than in normotensive controls. However, both positive and negative associations with hypertension have been found for the same polymorphism when studied in different populations. In the present study, we addressed the question of the possible cause(s) of this discrepancy among populations by using the α-adducin polymorphism as a paradigm. Four hundred ninety hypertensives and 176 normotensives enrolled in Sassari, Italy, and 468 hypertensives and 181 normotensives enrolled in Milano, Italy, were genotyped for the α-adducin Gly460Trp polymorphism. The blood pressure response to 2 months of hydrochlorothiazide therapy could be evaluated in 143 (85 in Sassari and 58 in Milano) hypertensives with and without the 460Trp α-adducin allele. The α-adducin 460Trp allele was not significantly more frequent in hypertensives in the Sassari population but was more frequent in hypertensives than in normotensives in Milano (P=0.019). Basal plasma renin activity was lower and blood pressure fall after diuretic therapy more pronounced (P<0.01) in hypertensives carrying at least one 460Trp allele than in Gly460Gly homozygotes, irrespective of their membership in the Sassari or Milano cohort. The effect of α-adducin genotype in predicting basal plasma renin activity and blood pressure decrease with diuretic treatment is similar in Sassari and Milano, despite the lack of association of the α-adducin genotype with hypertension in Sassari.
Population association studies have been gaining prominence to unravel the genetics of complex diseases such as hypertension. The basic requirement for declaring an association study positive is that the “hypertension-favoring” allele is more frequent in hypertensive cases than in normotensive controls. However, both positive and negative associations with hypertension have been found for the same polymorphism when studied in different populations. In the present study, we addressed the question of the possible cause(s) of this discrepancy among populations with the use of the α-adducin polymorphism as a paradigm.
The human α-adducin 460Trp allele may be considered a putative hypertension-favoring allele because it may affect blood pressure by increasing renal tubular reabsorption through the activation of Na,K-ATPase. Linkage and association studies, performed with markers mapping in the region of the α-adducin locus and with the Gly460Trp α-adducin polymorphism, respectively, yielded positive results.1 Hypertensive patients carrying at least one 460Trp allele compared with those homozygous for the Gly460 wild-type allele have (1) a less-steep pressure natriuresis slope; ie, they need a higher arterial pressure to excrete the same amount of sodium after saline infusion2 ; (2) lower plasma renin activity1 ; (3) enhanced proximal tubular reabsorption3 ; and (4) a more pronounced blood pressure fall after acute sodium depletion or chronic (2 months) diuretic treatment.1 The 460Trp α-adducin allele displays a higher affinity for Na,K-ATPase than does the 460Gly allele.4 This last finding is of particular relevance, because it demonstrates the same functional protein alteration in both the rat and human “hypertensive” α-adducin variant, despite a difference in the mutation site.4 5 Therefore, these similarities in renal and protein functional abnormalities between rats and humans justify the use of data on rats as an argument to reinforce the concept that the 460Trp variant may be a hypertension-favoring allele in humans.
On the other hand, our previous report of a positive association between hypertension and the 460Trp allele has been confirmed by some6 7 but not by others.8 9 The weaknesses of such type of studies have been discussed in detail recently.10 The discrepancies of results obtained in studying different populations may be due to a number of causes, including the interaction with other gene polymorphisms that may counteract (or differentially modulate) the effect of the “culprit” allele under investigation on blood pressure. For instance, so far as α-adducin is concerned, if, in a given population, another as-yet-undetected gene variant that reduces tubular sodium reabsorption is found more frequently in the normotensives than in hypertensives, it may blunt the difference of the α-adducin 460Trp allele frequency between hypertensives and normotensives in this population.
This hypothesis implies the persistence of differences in the parameters reflecting the relationship between body sodium and blood pressure, such as plasma renin and blood pressure response to diuretics, between hypertensive patients homozygous for the 460Gly wild-type allele and those with at least 1 copy of the 460Trp allele. Therefore, these 2 parameters were measured in the hypertensives of 2 case-control studies differing for the presence or absence of an association between hypertension and the 460Trp allele.
The study populations consisted of 490 essential hypertensive patients and 176 normotensive controls enrolled at the Clinica Medica, Sassari University Medical School, Sassari, Italy, and 468 essential hypertensive patients and 181 normotensive controls enrolled at the Division of Nephrology, San Raffaele Hospital, Milano, Italy. All of the subjects were unrelated. The patients and controls selected in Milano are an expansion of our previously published sample.1 In particular, the number of hypertensives in the Milano sample was increased from 282 to 468 (≈40% more), whereas that of the normotensives was increased from 151 to 181 (≈17% more) compared with those in the previous article.1 We decided to increase the numbers of subjects in the Milano sample according to those reached in Sassari, because we wanted to have 2 samples with an equal number of cases and controls. Selection criteria for the hypertensive and normotensive samples selected in Sassari were the same as those of the Milano sample already published.1 Both in Milano and Sassari, great care was taken to enroll only cases and controls who were able to trace their origin (North Sardinia or northern Italy) back to at least their grandparents. Although ethnic stratification is practically impossible in Sardinia because there is practically no migration, it may be relevant in Milano. However, we should have been able to avoid major ethnic stratification in Milano also, because the major migrations from the south of Italy occurred 30 to 40 years ago.
Response to Diuretic Treatment
As mentioned above, the response to 2 months’ treatment, the first with 12.5 mg/d hydrochlorothiazide PO and the second with a double dose of the same drug, is an extension of a previously published study.1 One hundred sixty mildly to moderately hypertensive outpatients were studied after informed, written consent was obtained from each participant and after approval by the Ethics Committee of the 2 universities (95 of them selected in Sassari and 65 in Milano). Blood pressure was measured with a sphygmomanometer and, for each patient, all measurements were made by the same investigators (P.S., C.T.) at the same time of day (8 to 10 am). Ten consecutive measurements were recorded after the patient had been clinically examined and after a 10-minute rest in the supine position. The last 3 blood pressures were averaged and used for the analysis. Thirty patients who had been taking antihypertensive medications discontinued them for at least 4 months. In general the therapy withdrawal was spontaneous, and in the majority of patients it occurred during the summer because of the blood pressure reduction during this season. We decided to perform such an unusually long period of therapy withdrawal because we wanted to avoid any residual carryover effect. During this period, blood pressure was monitored every 2 to 3 weeks. Ten patients were withdrawn during the washout phase because of diastolic blood pressure persistently >105 mm Hg (n=3) or <95 mm Hg (n=4) or withdrawal of informed consent (n=3). Preliminary data from some of these patients (26 selected in Milano and 32 in Sassari) have already been published.1 Of that original sample, 14 were previously treated thus only 6 more previously treated patients were included in the present analysis.
Blood and urine samples for plasma renin activity, urinary sodium and potassium, creatinine clearance, and urinary volume were collected on the last visit before starting the treatment. Body mass index (BMI) was computed as body weight (kg)/height squared (m2). Systolic, diastolic, and mean blood pressures were used in the analysis. Mean blood pressure was computed as diastolic pressure plus one third of the pulse pressure.
Genotyping for α-Adducin Polymorphism
All individuals were genotyped for the α-adducin polymorphism as previously described.1
All clinical parameters are expressed as mean±SEM. Blood pressure and anthropometric variables were compared with Student’s t test. Sex ratios and allele-disease association were tested by χ2. To test for possible false-positives of the association between the α-adducin polymorphism and hypertension, logistic regression was also applied, entering as the dependent variable the genotype for α-adducin (contrasting Gly460Gly versus Gly460Trp+Trp460Trp), hypertensive status and sex as categorical covariates, and age and BMI as continuous covariates.
Blood pressure change after diuretic treatment was analyzed with multivariate ANOVA (MANOVA) for repeated measures, with 1 fixed factor within patients (blood pressure at 2 times, ie, before and after diuretic treatment) and 2 fixed factors between patients (genotype at 2 levels, Gly460Gly and Gly460Trp+Trp460Trp, and ethnicity at 2 levels, Milano and Sassari). To also take into account the effect of possible confounders originating in anthropometric differences between the hypertensives studied in Sassari and those studied in Milano, BMI, sex, and age were also entered as covariates in the model. Statistical analysis was performed with SPSS statistical software.
The main clinical characteristics of the 2 hypertensive and control populations are summarized in Table 1⇓. Owing to the study design in both populations, the controls were older than the hypertensives. BMI was larger in hypertensives than in controls as already shown and discussed.11 However, the Sassari population as a whole had a higher BMI, and the Milano population had a greater proportion of men than women.
The frequency of the α-adducin 460Trp allele was not significantly different between hypertensives and controls in the Sassari population (Table 2⇓), but it remained higher in hypertensives than in normotensives in Milano (P=0.019), as previously reported.1 No significant deviation from Hardy-Weinberg equilibrium was present in hypertensives and normotensives both in Milano and Sassari (data not shown).
Interference of Other Clinical Variables on the Results of the Association Study
Owing to the study design, the control subjects were older than the hypertensives. Also, differences in BMI and sex distribution were present between groups. Because such differences may have contributed to contrasting findings (significant allele-disease association in Milano versus no allele-disease association in Sassari), we performed several analyses to test for this possibility and were able to exclude it.11
The effect of age was tested in 2 ways. Ninety-eight hypertensive patients from Sassari and 60 from Milano were >60 years old at the time of blood sampling, but they were included in the sample because their hypertension had been diagnosed many years before (when they were <60). The frequency of the 460Trp allele in this subgroup of hypertensives was not different when compared with that of younger hypertensives in both Milano and Sassari (P=0.66 and P=0.28, respectively). However, because the sample of hypertensives >60 was indeed relatively small, the analysis was repeated by using as a cutoff the median of age of hypertensives (49 years for Milano and 50 years for Sassari). In this case also, no difference was found, indicating that there was no selective loss of the 460Trp allele with increasing age in Milano as well as in Sassari. The same kind of analysis was also repeated for BMI by using as a cutoff the median of BMI, and again none of the comparisons resulted in significant differences.
The effect of sex may have been an important confounding variable, because the number of hypertensive females was less than half the number of hypertensive males in the Milano sample, whereas sex distribution was much better balanced in the other 3 groups. However, also in this case, when tested as before, sex did not appear to play any significant role in determining the relative frequencies of the allele distribution in the subsamples.
Finally, because multiple univariate comparisons may not be as sensitive in clarifying the role of confounders in determining spurious allele-disease association, logistic regression was also applied, with the genotype for α-adducin as the dependent variable, hypertensive status and sex as categorical covariates, and age and BMI as continuous covariates. Confirming the results of the allelic association, the presence of at least one 460Trp allele in the genotype was also significantly associated with hypertension only in Milano, whereas no association was found in Sassari. None of the other variables (sex, age, and BMI) gave a significant contribution to the model (as well as their interactions; not shown), as summarized in Table 3⇓.
Response to Diuretic Treatment According to α-Adducin Genotype
Clinical characteristics of the Sassari and Milano hypertensives are summarized in Table 4⇓. Seven patients did not complete the study because of hypokalemia (2 to 3.5 mmol/L, n=4) or other side effects such as epigastralgia (n=1) and dizziness (n=2). Only the 143 patients who completed the study were considered for statistical analysis. Because 460Trp homozygotes are rare, the 5 homozygous patients that we found were included in the 460Trp heterozygote group. The Sassari hypertensives were older, with a higher BMI and urinary potassium excretion than in Milano hypertensives. The latter had a higher proportion of men than women, but sex did not influence blood pressure response to diuretic treatment (P=0.133).
Basal plasma renin activity was lower in hypertensives bearing at least one 460Trp allele than in 460Gly homozygotes, irrespective if they were of Sassari or Milano origin (the Figure⇓, lower panel). Average mean blood pressure decreased after diuretic treatment significantly more in hypertensives bearing at least one 460Trp allele than in the 460Gly homozygotes (F=13.57, P<0.001). The overall mean blood pressure decrease was slightly more pronounced in the Sassari than in the Milano group (F=3.39, P=0.068), but that because of α-adducin genotype was the same in both cohorts (F=0.02, P=0.92; the Figure⇓, upper panel; actual data of mean blood pressure before and after treatment are also reported in Table 4⇑). However, when sex, BMI, and age were also entered in the repeated-measures MANOVA model as covariates, the slightly though nonsignificantly larger mean blood pressure decrease in Sassari than in Milano was no longer evident (F decreased from 3.39 to 1.06, and P increased from 0.068 to 0.306).
Interference of Previous Antihypertensive Treatment With the Response to Thiazide Diuretics
Despite the long pharmacological washout (4 months), one can never exclude a carryover effect. Although we do not have any suggestion that α-adducin polymorphism played any role in the carryover, we excluded that possibility by simply reanalyzing our sample after we excluded those 20 previously treated patients and obtained the same results (data not shown).
In the present case-control study, we found no association of the 460Trp α-adducin allele with hypertension in a large population from Sassari. Conversely, we confirmed a positive association in the enlarged population of Milano.1 In our previous article,1 hypertensives with the 460Trp allele displayed a lower plasma renin activity and a more pronounced blood pressure fall with chronic diuretic treatment compared with the patients homozygous for the 460Gly allele. These findings are confirmed in this enlarged population of Milano and Sassari, despite the discrepancy in the results of the association study.
Because the sample size and the allele frequencies were similar in the 2 populations, we may assume that a power factor was not responsible for the discrepant results. The lack of association of the 460Trp allele with hypertension in Sassari may depend on 2 groups of causes: (1) Because of the weak effect of the 460Trp allele on hypertension in the overall population of patients, the detection of a positive association may be heavily dependent on a mild case-control difference in confounding factors, such as population stratification, environment, lifestyle, age, BMI, and sex. Tables 1⇑ and 4⇑ show differences between the 2 populations in BMI, sex, and potassium intake that may be relevant in this regard. However, as in our previous analysis,1 11 here we excluded an influence of BMI, age, and sex on the α-adducin allele distribution. (2) A different frequency between Sassari hypertensives and normotensives of another gene variant, affecting either the constitutive effect of the 460Trp allele on tubular reabsorption or on the sequence of events linking the rate of tubular reabsorption to arterial hypertension, is also possible.
The rate of tubular reabsorption is determined by the interaction of many cellular proteins, besides other extracellular factors (physical, hemodynamic, hormonal, etc). We already know several proteins whose genetic variants may affect tubular reabsorption differently.12 It is then conceivable that the faster rate of tubular reabsorption, which may be responsible for a genetic form of hypertension, may be achieved through different molecular genetic mechanisms in different populations.
It should be recalled that in all positive association studies, the putative causal allele is indeed more frequent in hypertensives, but it is also found in a substantial number of normotensives. This is obviously due to the multifactorial nature of essential hypertension and means that, besides other mechanisms, 1 or more alleles of some other gene(s) must be present and counteracting the effect of the allele associated to hypertension in a substantial number of individuals who will eventually remain normotensive, despite the presence of the hypertensive allele.
Our hypothesis is that in Sassari (but not in Milano), more people remain normotensive despite the presence of the 460Trp allele because more also have the counteracting allele of 1 or more as-yet-undetected other genes. According to the theory of evolution by bricolage,13 it is possible to postulate that different populations “use” the alleles available to achieve the modulation of the peculiar cellular function needed to optimize their biological fitness for their particular environment. Selective pressure may favor the genetic mechanisms devoted to body sodium conservation, either by decreasing the frequency of alleles that depress the rate of reabsorption or by increasing the frequency of those that enhance it. This mechanism may affect the distribution of any given allele among cases and controls but will not affect its intrinsic specific effect on cell functions. If the 460Trp α-adducin allele increases tubular reabsorption, then the lack of difference between cases and controls may simply be due to some other gene variant that counteracts the effect of the 460Trp allele on tubular reabsorption, whose frequency could differ between Sassari hypertensives and normotensives.
Because there is no reason to postulate that this counteracting allele is in linkage disequilibrium with α-adducin, it is most likely inherited independently of α-adducin. This implies that such an allele is distributed evenly between the different α-adducin genotypes, allowing detection of the 460Trp allele effect on sodium retention in hypertensives. In fact, within the hypertensive patients of Sassari, the 460Trp α-adducin allele remains associated with lower renin and a greater blood pressure fall after chronic diuretic treatment, implying a similar effect on renal sodium handling of the 460Trp allele in both Milano and Sassari.
May the α-adducin 460Trp allele still be considered a hypertension-favoring allele that increases blood pressure in the Sassari population? The answer to this question may be either negative or positive according to the type of study considered. It is negative if we consider the association study but positive if we consider that Sassari hypertensives bearing the 460Trp allele have low renin and a larger blood pressure fall in response to diuretic treatment.
A hypertension-favoring allele may act in 2 ways: (1) It may produce a dysfunction in the physiological mechanisms regulating sodium balance and blood pressure; therefore, some abnormality of these mechanisms may also be detectable in normotensive subjects. (2) It may hinder compensatory mechanisms that normally counteract the development of hypertension. The example of the progressive reduction of renal function with age may be illuminating in this regard. At least some subtypes of essential hypertension may be caused by the inability of the homeostatic mechanisms controlling sodium balance to keep blood pressure low, despite the age-dependent reduction in renal function. The alleles that cause this form of hypertension are present from birth, but hypertension and altered sodium balance develop only in adults, when larger-than-due sodium retention ensues and causes low-renin hypertension. The recent article by Kamitaniet al,14 wherein no significant α-adducin–related difference in renin/sodium metabolism in a sample of young, normotensive individuals could be detected (although a trend toward lower plasma renin activity values increasing the dose of 460Trp alleles was also seen in this sample), make the latter hypothesis more likely.
The rat studies have clearly shown an epistatic interaction between the polymorphisms of α-, β-,15 and γ-adducin (G.B. et al, unpublished observations, 1999) in determining the final blood pressure level. Despite extensive studies on 20 normotensives and 20 hypertensives with single-strand conformational polymorphism techniques (N.G. et al, unpublished observations, 1999), no polymorphism has been detected in humans for β- or γ-adducin cDNA. Of course, these preliminary negative findings do not exclude the possibility that a still-unknown polymorphism in these 2 genes may be present in different populations and, because of their epistatic interaction, that they may influence the assessment of the pressor role of the α-adducin polymorphism.
In conclusion, the major finding of the present article is that the effect of α-adducin genotype in predicting basal plasma renin activity and blood pressure decrease with diuretic treatment is similar in Sassari and in Milano, despite no association of α-adducin genotype with hypertension in Sassari. Our findings would prompt the search for a difference in the distribution between Sardinian hypertensives and normotensives of an allele of another gene counteracting the effect of α-adducin.
This work was supported in part by the Ministero Università e Ricerca Scientifica of Italy (ex MPI 60%; years 1996 to 1998 to N.G., D.C., and G.B. and grant 9806154140 001 to G.B.), by Sigma Tau/MURST, National Research Project on Genetic and Molecular Analysis of Physiologic and Pathologic Response of Endocellular Receptors, and by Telethon grant No. E.C. 516.
- Received April 19, 1999.
- Revision received April 27, 1999.
- Accepted May 26, 1999.
Cusi D, Barlassina C, Azzani T, Casari G, Citterio L, Devoto M, Glorioso N, Lanzani C, Manunta P, Righetti M, Rivera R, Stella P, Troffa C, Zagato L, Bianchi G. α-Adducin polymorphism in primary hypertension: linkage and association study; relationship to salt sensitivity. Lancet. 1997;349:1353–1357.
Manunta P, Burnier M, D’Amico M, Buzzi L, Maillard M, Barlassina C, Lanella G, Cusi D, Bianchi G. Adducin polymorphism affects renal proximal tubule reabsorption in hypertension. Hypertension. 1999;33:694–697.
Ferrandi M, Slardi S, Tripodi G, Barassi P, Rivera R, Manunta P, Goldshlager R, Ferrari P, Bianchi G, Karlish SJD. Interaction between adducin and Na,K-ATPase: differential effect of hypertension related human and rat adducin polymorphisms. Am J Physiol. In press.
Castellano M, Barlassina C, Muiesan M, Beschi M, Cinelli A, Rossi F, Rizzoni D, Cusi D, Agabiti-Rosei E. α-Adducin gene polymorphism and cardiovascular phenotypes in a general population. J Hypertens. 1997;15(pt 2):1707–1710.
Kato N, Sugiyama T, Nabika T, Morita H, Kurihara H, Yazaki Y, Yamori Y. Lack of association between the α-adducin locus and essential hypertension in the Japanese population. Hypertension. 1998;31:730–733.
Schork NJ. Genetically complex cardiovascular traits: origins, problems, and potential solutions. Hypertension. 1997;29(pt 2):145–149.
Casari G, Barlassina C, Cusi D, Zagato L, Muirhead R, Righetti M, Nembri P, Amar K, Gatti M, Macciardi F, Binelli G, Bianchi G. Association of the α-adducin locus with essential hypertension. Hypertension.. 1995;25:320–326.
Lifton RP. Molecular genetics of human blood pressure variation. Science. 1996;272:676–680.
Kamitani A, Wong ZY, Fraser R, Davies DL, Connor JM, Foy CJ, Watt GC, Harrap SB. Human α-adducin gene, blood pressure, and sodium metabolism. Hypertension. 1998;32:138–143.
Bianchi G, Tripodi G, Casari G, Salardi S, Barber BR, Garcia R, Leoni P, Torielli L, Cusi D, Ferrandi M, Pinna LA, Baralle FE, Ferrari P. Two point mutations within the adducin gene are involved in blood pressure variation. Proc Natl Acad Sci U S A. 1994;91:3999–4003.