Long‐term BP control and vascular health in patients with hyperaldosteronism treated with low‐dose, amiloride‐based therapy

Abstract

Whether aldosterone itself contributes directly to macro‐ or microcirculatory disease in man or to adverse cardiovascular outcomes is not fully known. We report our long‐term single‐practice experience in 5 patients with chronic hyperaldosteronism (HA, including 3 with glucocorticoid remediable aldosteronism, GRA) treated with low‐dose amiloride (a specific epithelial sodium channel [ENaC] blocker) 5‐10 (mean 7) mg daily for 14‐28 (mean 20) years. Except for 1 GRA diagnosed in infancy, all had severe resistant hypertension. In each case, BP was normal or near‐normal within 1‐4 weeks after starting amiloride and office BP’s were well controlled for 20 years thereafter. Vascular studies and 24‐hour ambulatory BP monitoring with pulse wave analysis (cardiac output, vascular resistance, augmentation index, and reflection magnitude) were assessed after a mean of 18 years as were regional pulse wave velocities, pulse stiffening ratio, ankle‐brachial index, serum creatinine, estimated glomerular filtration rate, and spot urinary albumin:creatinine ratio. All indicators were completely normal in all patients after 18 years of amiloride, and none had a cardiovascular event during the 20‐year mean follow‐up. We conclude that long‐term ENaC blockade can normalize BP and protect macro‐ and microvascular function in patients with HA. This suggests that (a) any vasculopathic effects of aldosterone are mediated via ENaC, not MR activation itself, and are fully preventable or reversible with ENaC blockade or (b) aldosterone may not play a major BP‐independent role in human macro‐ and microcirculatory diseases. These and other widely divergent results in the literature underscore the need for additional studies regarding aldosterone, ENaC, and vascular disease.

1. INTRODUCTION

Primary hyperaldosteronism (HA) due to genetic mutation (eg, glucocorticoid remediable aldosteronism, GRA), diffuse or nodular adrenal cortical hyperplasia, or adrenal adenoma often results in resistant or refractory hypertension.1 In animals and man, mineralocorticoid excess alone is not usually sufficient to elevate blood pressure; excess dietary salt intake is also required.2, 3 Nongenetic forms of HA are often treated surgically but chronic oral mineralocorticoid receptor (MR) antagonism, generally with spironolactone, has also been used to lower BP.4 It is believed that the end‐organ damage profile in patients with HA is more severe than essential hypertension,5, 6, 7 including increased prevalence of cerebral hemorrhage,8 myocardial infarction,6 left ventricular hypertrophy,6 cardiac fibrosis,9 increased arterial stiffness,10, 11, 12, 13, 14, 15, 16 microalbuminuria,17, 18 and microcirculatory wall‐to‐lumen ratio.19, 20 Attribution of HA‐associated target organ damage, however, is universally confounded by the effects of aging21 as well as the duration and degree of BP elevation (BP burden).21, 22 Furthermore, careful review of the literature suggests that the putative vasculopathic effects of aldosterone are far less clear than some investigators have suggested.

A major function of the MR is to regulate the availability of ion channels, especially ENaC, in cell membranes.23, 24, 25 A potentially useful but often forgotten drug is amiloride, a specific epithelial sodium channel (ENaC) inhibitor.23, 24, 25 Recent studies suggest that ENaC in mouse vascular endothelium and vasopressinergic brain neurons affect a wide variety of cellular functions,16, 26 so direct and indirect hemodynamic and volume‐regulatory effects of ENaC are likely. Persistent BP‐lowering effects of amiloride have been reported in short‐term studies in low‐renin hypertension24, 27 and HA,28 but there have been no long‐term or outcome trials with this agent. It is also not known whether amiloride confers specific target organ benefits other than reducing microalbuminuria, which may simply reflect good long‐term BP control.25 Because of the ongoing interest in aldosterone as a possible promoter of cardiovascular fibrosis and stiffness and the simultaneous possibility that ENaC may be involved in these processes, we report our experience with chronic amiloride therapy in a small group of patients with HA who were assessed clinically after many years for evidence of macrovascular disease (arterial stiffness) or microcirculatory disease (albuminuria and renal function). Because of the many conflicting reports in the literature, a review of related studies is also provided.

2. METHODS

Four patients were referred originally for management of resistant hypertension (BP uncontrolled on appropriate doses of at least three drug classes, including a diuretic). The fifth was the daughter of GRA1, who was the kindred proband and who received dexamethasone as a child; all 3 GRAs were described in the original work of Lifton and coworkers.29 The other 2 patients had primary HA and resistant hypertension: 1 with a large aldosterone‐producing adenoma and the other with apparent nodular hyperplasia (biopsy was declined); both refused adrenal surgery. After presentation of therapeutic alternatives, all patients agreed to the administration of amiloride monotherapy for better BP control. Within a few weeks, when it was established that amiloride had achieved this goal without causing noticeable side effects, all patients elected to remain on amiloride indefinitely. Vascular studies represent a single point in time after many years of amiloride therapy; all patients signed informed consent for noninvasive investigation of vascular function.

2.1. Patient profiles

Our initial patient, GRA1 had resistant hypertension uncontrolled by lisinopril, atenolol, and furosemide. Her condition was diagnosed by genotyping following that of GRA2, her daughter. Initial workup included an undetectable plasma renin activity and a markedly elevated 24‐hour urinary 18‐hydroxycorticosterone excretion rate (802 mcg/d)30; hypokalemia was never observed. She had received spironolactone 25 mg daily in combination with other agents for about 2 months prior to referral but had discontinued it because of abdominal discomfort and menstrual irregularities. She also had a history of estrogen receptor‐positive breast cancer, and despite no evidence that spironolactone increased her risk of recurrence, she feared the weak estrogenic effects of spironolactone and declined to continue its use. Amiloride monotherapy (10 mg daily) was immediately effective, reducing BP from 183/115 mm Hg on presentation to 120‐140/80‐90 mm Hg within 1 month. A lower dose (5 mg daily) was investigated after 3 months with no loss of BP control; on this therapy, office values ranged from about 120‐140/70‐85 mm Hg for the next 26 years, when the vascular studies were performed. She had no adverse cardiovascular events for 28 years.

GRA 2 (daughter of GRA1) was treated with dexamethasone throughout early childhood; based on her mother’s favorable response to amiloride and in consultation with the senior author (JLI), her physician switched to amiloride 5 mg daily at age 11. BP values prior to age 14 were not available, but all subsequent BP values were <120/80 mm Hg. Vascular studies were performed after 14 years on amiloride monotherapy, and she had no adverse cardiovascular events for 16 years.

GRA 3 (sister of GRA1) was also diagnosed by genotyping. She lived in a remote location, and initial biochemical testing was not performed. After telephone consultation for severe hypertension, her primary care physician started amiloride 10 mg daily; on telephone follow‐up 2 months later, she was reported to be “controlled.” Five years later, her physician retired, she temporarily lost health insurance, and she discontinued amiloride for 1 year. At that time, her office BP was 180/120 mm Hg. Reinstitution of amiloride 10 mg daily resulted in normalization of her BP within 1 month and persistent BP control thereafter (<135/85 mm Hg on multiple readings). Vascular studies were performed after 22 years of amiloride, and she had no adverse cardiovascular events during 24 years of follow‐up.

PA 1 had resistant hypertension for several years and was eventually referred for this problem. Diuretic had been avoided by the primary physician because of persistent hypokalemia (as low as 2.3 mEq/L), and for several years, she received oral potassium supplementation (20‐40 mEq daily). When she changed primary care physicians, a workup revealed a solitary adenoma of the right adrenal gland and increased venous plasma aldosterone levels (32 and 49 ng/L). Subsequent adrenal venous aldosterone sampling was uninterpretable (left adrenal = inferior vena cava; both > right adrenal). Initial office was BP 215/105 on lisinopril, nifedipine, and metoprolol; amiloride 5 mg daily was substituted for metoprolol, and amlodipine 5 mg daily was substituted for nifedipine. She requested to continue lisinopril because she believed that it was “protecting her kidneys.” Within 2 months, there was full BP control: home BP mean was about 130/70 mm Hg, and office BP thereafter was <140/75 mm Hg. Lisinopril was subsequently discontinued but beta‐blocker therapy was started for symptomatic relief of episodic palpitations. Vascular studies were performed after 18 years of amiloride. She remained in good general health for 20 years without adverse cardiovascular events.

PA2 had resistant hypertension but no hypokalemia; office BP was 172/98 mm Hg while taking hydrochlorothiazide 25 mg, amlodipine 10 mg, and valsartan 320 mg daily. Elevated 24‐hour urinary aldosterone was documented on 2 occasions (22.1 and 23.2 mcg/d, with corresponding urinary sodium values of 248 and 157 mg/d). Imaging studies did not reveal any adenoma or significant glandular enlargement. After discussion of potential side effects of spironolactone, the patient opted to have amiloride 5 mg daily added to his prior regimen. One month later, office BP was 142/98 mm Hg and amiloride was increased to 10 mg daily; 2 months later, the patient’s BP was 132/92 mm Hg and valsartan was discontinued. For the next 5 years, office BP remained borderline controlled (mean office BP < 145/95 mm Hg). Combination HCTZ/spironolactone (25/25 mg daily) was substituted for HCTZ and office BP values over the next 6 years ranged from <130/70 to about 145/90 mm Hg. Vascular studies were performed after 12 years of amiloride, and he continued taking the drug for 14 years without adverse cardiovascular events.

2.2. BP monitoring, arterial function, and hormonal studies

Each patient had arterial function and 24‐hour ambulatory BP monitoring studies after standard seated clinic BP values (mean of 3) were determined by oscillometry (Omron905CP).

Arterial stiffness and ankle‐brachial index were assessed on 1 occasion by measuring foot‐to‐foot (2‐point) pulse wave velocities (PWV, Colin VP1000) in three different arterial regions: the mixed carotid‐femoral region (cfPWV), the aorta (heart‐femoral [hf] PWV), and peripheral muscular arteries (femoral‐ankle [fa] PWV). We and others have noted a scalar problem with “raw” PWV values determined by Colin VP1000 that was created by the manufacturer’s incorrect algorithms for arterial path length.31 Based on our data from direct measurements in a reference cohort of patients (n = 40), we developed simple correction factors (corr) that were applied to the uncorrected (uncorr) values: hfPWV_corr = 0.8 hfPWV_uncorr, cfPWV_corr = 0.7 cfPWV_uncorr, and faPWV_corr = 1.1 faPWV_uncorr. Because PWV values are also highly dependent on age and systolic BP, we created normative models for expected age‐ and SBP‐adjusted PWV values. These were derived from a reference population (n = 162, age 16‐91) constructed to mimic closely the strong age‐BP relationship found in the US NHANES III cohort.32, 33 The predictive model equations were:

$${\text{hfPWV}}_{\text{corr}}=-\text{271}+\text{8}.\text{6}\times \text{age}+\text{5}.\text{1}\times \text{systolic BP(cm}/\text{sec)}(r^{\text{2}}=0.\text{76}).$$

$${\text{cfPWV}}_{\text{corr}}=-\text{439}+\text{11}\times \text{age}+\text{6}.\text{3}\times \text{systolic BP(cm}/\text{sec)}(r^{\text{2}}=0.\text{61}).$$

$${\text{faPWV}}_{\text{corr}}=\text{238}+\text{2}.\text{4}\times \text{systolic BP}+\text{4}.\text{9}\times \text{diastolic BP}+\text{2}.0\times \text{age(cm}/\text{sec)}(r^{\text{2}}=0.\text{58}).$$

All three models were independent of gender, heart rate, height, or weight (excluded variables) in each case.

Ambulatory 24‐hour BP and pulse wave analysis was performed using a Mobil‐O‐Graph (IEM), with BP determinations every 20 minutes. Mean ambulatory hemodynamic indicators (cardiac output, systemic resistance, augmentation index, reflected wave magnitude) were reported for 24‐hour, daytime, and nighttime values (based on sleep diaries). From the 24‐hour BP data, we calculated an additional arterial function indicator, the pulse stiffening ratio (PSR), which quantitatively describes the nonlinear responses of arteries to variations in pulse volume and the linear sensitivity of an artery to variation in distending pressure (Table 3).34 A stepwise multiple linear regression model for normative PSR values was derived from a separate ambulatory BP monitoring reference cohort (n = 76):

$$\text{PSR}=\text{1}.\text{14}+0.0\text{1}\times \text{systolic BP}+0.00\text{4}\times \text{age}-0.0\text{17}\times \text{diastolic BP}(r^{\text{2}}=0.\text{36}).$$

Plasma aldosterone and renal function studies were performed at the same time as ambulatory monitoring studies.

Statistical analyses were based on paired t tests, including the direct comparison of mean model‐predicted and mean observed values for all vascular stiffness‐related variables.

3. RESULTS

Table 1 summarizes basic demographic, laboratory, and pharmacological information for each case. Blood pressure values before amiloride treatment on standard triple therapy (4 of 5 individuals) demonstrate hypertension refractory to 3 or more antihypertensive drugs and also demonstrate rapid response to amiloride monotherapy within 1 month. Table 2 demonstrates 24‐hour ambulatory BP and hemodynamic monitoring variables after long‐term amiloride (mean follow‐up 18 years). All indicators fell within the normal range. Table 3 demonstrates that macro‐ and microvascular health indicators (central and peripheral pulse wave velocities, pulse stiffening ratio, and ankle‐brachial index) as well as microcirculatory indicators (serum creatinine, estimated glomerular filtration rate, and urinary albumin:creatinine ratio) fell within the normal or expected range during long‐term amiloride therapy. None of the patients experienced any cardiovascular, cerebrovascular, or renal adverse events during the follow‐up period (mean 20 years).

Table 1.

Demographic, clinical, and laboratory values

Patient	GRA1	GRA2	GRA3	PA1	PA2	Mean (SD)
Current age (yr)	69	30	53	84	65	58 (20)
Gender	F	F	F	F	M	‐‐
BP pre‐amiloride therapy (mmHg)	183 115	N/A	180 120	215 105	172 98	188 (19) 110 (9.9)
Duration of amiloride therapy (yrs)
Prior to vascular study	26	14	22	18	12	18 (5.7)
Total follow‐up	28	16	24	20	14	20 (5.6)
Amiloride daily dose (mg)	5	5	5/10	5	10	7 (2)
Most recent office BP (mmHg)	120 74	114 71	107 66	145 62	124 83	122 (14) 71 (8.0)
Most recent spot urinary aldosterone (mcg/g Cr)	11	‐‐	27	111	24	43 (46)
Most recent serum K (mEq/L)	4.1	4.0	4.2	3.8	4.3	4.1 (0.2)

Most recent office blood and urinary values occurred at the time of vascular studies or within a few months.

Abbreviations: GRA, glucocorticoid remediable hyperaldosteronism; PA, primary hyperaldosteronism.

Table 2.

24‐h ambulatory BP and hemodynamic variables after long‐term amiloride‐based therapy

Patient	GRA1	GRA2	GRA3	PA1	PA2	Mean (SD)
Systolic BP (mm Hg)
24‐h mean	111	111	103	118	124	113 (8.0)
Daytime mean	115	116	105	122	134	118 (11)
Nighttime mean	103	101	99	113	117	107 (7.9)
Nighttime decrease (“systolic dipping”)	−12	−15	−6	−9	−17	−12 (−4.7)
Diastolic BP (mm Hg)
24‐h mean	59	68	69	60	77	67 (7.4)
Daytime mean	62	74	70	64	91	72 (12)
Nighttime mean	55	57	64	54	68	60 (6.1)
Nighttime decrease (“diastolic dipping”)	−7	−17	−6	−10	−23	−13 (−7.2)
Other hemodynamic variables from 24‐h monitoring (24‐h means)
Heart rate (b/min)	69	86	62	63	66	69 (9.8)
Cardiac output (L/min)	4.3	4.6	4.0	4.5	5.2	4.5 (0.4)
Total vascular resistance (mm Hg.sec/mL)	1.2	1.2	1.3	1.2	1.2	1.2 (0.04)
Augmentation index@75 (%)	35	20	33	24	16	26 (8.2)
Reflection magnitude (%)	69	57	70	69	68	67 (5.4)
Pulse pressure (mm Hg)	52	43	34	58	46	47 (9.1)

Abbreviations as in Table 1.

Table 3.

Macro‐ and microvascular health indicators after long‐term amiloride‐based therapy

Patient		GRA1	GRA2	GRA3	PA1	PA2	Mean (SD)
Macrovascular indicators (arterial stiffness and stiffening, ABI)
hfPWVcorr (cm/sec)	Observed	858	470	821	1141	869	832 (238)
	Predicted	874	479	696	1156	877	817 (258)
	P =						0.62
cfPWVcorr (cm/sec)	Observed	947	560	792	1239	698	847 (260)
	Predicted	989	497	774	1354	1002	925 (317)
	P =						0.29
faPWVcorr (cm/sec)	Observed	959	826	905	1113	985	958 (106)
	Predicted	993	858	916	1049	1062	926 (87)
	P =						0.47
Pulse Stiffening Ratio	Observed	1.5	0.99	1.1	1.3	0.94	1.16 (0.23)
	Predicted	1.4	1.2	1.3	1.6	1.3	1.36 (0.15)
	P =						0.07
Ankle‐Brachial Index (ABI)		1.2	0.93	1.0	1.2	1.2	1.1 (0.12)
Microvascular indicators
Serum Creatinine (mg/dL)		1.0	0.79	0.90	0.86	1.17	0.94 (0.15)
eGFR (mL/min)		59	90	70	67	66	70 (12)
Urinary Albumin:Creatinine (mg/g)		3	5	13	28	27	15 (12)

See also Table 1.

Abbreviations: cf, carotid‐femoral; corr, corrected; fa, femoral‐ankle; hf, heart‐femoral; p (observed vs predicted, by paired t test)PWV, pulse wave velocity; SD, standard deviation.

4. DISCUSSION

The rationale for using amiloride in HA arose from our initial patient with GRA, who could not tolerate spironolactone. The success she experienced with amiloride persuaded her daughter and sister to request the same treatment, which was then offered to 2 other patients with HA who did not wish to have surgery. Although our results are preliminary rather than definitive, they are highly consistent internally and demonstrate that ENaC blockade with low‐dose amiloride‐based therapy can be highly effective for BP control in chronic HA. By all available measures, amiloride‐based therapy is also highly effective in protecting against macro‐ and microvascular complications. It is also noteworthy that none of these patients experienced any overt cardiovascular events during two decades of follow‐up, especially given admonitions that patients with HA experience more end‐organ damage and adverse cardiovascular events.5, 6, 7 These results further suggest that all benefits are directly attributable to ENaC blockade specifically rather than to the degree of persistent MR activation, since aldosterone remained elevated throughout. Alternatively, chronic aldosterone excess does not cause macro‐ or microcirculatory disease in man the same way it does in experimental animals.

In reviewing in vitro and in vivo investigations, and despite substantial opinion to the contrary,35, 36, 37, 38 it is not at all clear that aldosterone itself or MR stimulation per se inevitably causes premature macro‐ or microcirculatory disease. On the pro‐side, aldosterone stimulates cardiac fibroblasts in vitro to increase collagen syntheses,37 an effect that can be at least partially ameliorated by MR blockade with spironolactone.10, 11, 37 Also in rats, MR antagonists (spironolactone and eplerenone) reduce cardiac procollagen markers.39 However, the effects of MR antagonism in vivo on BP, vascular collagen content, and arterial stiffness (PWV) are less consistent.10, 11 Humans with high plasma aldosterone or overt HA tend to have stiffer arteries (higher PWV values),40, 41 but it must be stated that these individuals also tend to have a higher lifetime BP burden, which itself may be the primary explanation for high arterial stiffness or microcirculatory disease patterns. Other confounders of the aldosterone‐hypertension‐vascular disease relationship include aging itself and importantly, dietary salt intake.42, 43 Perhaps more troubling is the timing of reported vascular benefits of MR antagonism. For example, arterial stiffness in patients with HA has been reported to decrease within a few weeks after adrenalectomy—but so does the BP.35 Such reductions in arterial stiffness are fully attributable to the reduction in arterial distending pressure alone; arterial stiffness is the tangent slope of an artery’s exponential pressure‐volume curve and is thus intrinsically BP‐dependent. Furthermore, substantial reductions in large artery stiffness within a few weeks are not consistent with the known kinetics of human aortic remodeling, a slow process that takes place over months or years.44 With respect to microcirculatory damage, confounding by arterial pressure is also a major issue. HA in humans is believed to promote glomerular capillary hypertension, hyperperfusion, and microalbuminuria,16, 17, 18, 19 and MR antagonism is known to reduce albuminuria10, 11, 35, 37, 38, 40, 41, 45, 46, 47 but a similar argument could be made for both hypertension and diabetes. In both of these conditions, BP itself plays a major role in glomerular hemodynamics and renal function. If all of these cases, glomerular capillary hyperperfusion and hypertension are important harbingers of focal glomerulosclerosis and progressive renal failure48, 49, 50 and blood pressure control is extremely important.

Amiloride, an inhibitor of ENaC, is largely a forgotten drug that may have important under‐appreciated benefits. ENaC expression is directly dependent on MR activation,24, 51, 52 and present results are consistent with the notion that ENaC is the final common pathway of MR activation. In addition to blocking sodium transport, amiloride inhibits sodium‐hydrogen exchange and has a wide variety of other cellular effects,16, 26 including blunting of angiotensin II‐dependent aldosterone release in cultured human adrenal cells.53 In man, amiloride therapy causes an initial increase in plasma aldosterone, possibly via reflex activation of the renin‐angiotensin system in response to acute BP‐lowering or volume contraction.27, 54 Chronically, there is no consistent effect of amiloride on plasma aldosterone.28, 55 Amiloride may be particularly effective in lowering BP in African Americans (another form of low‐renin hypertension)24 but further studies are needed. In addition to its well‐known renal tubular location, the presence of ENaC in vascular endothelial cells16 and brain26 has the potential to stimulate much further investigation.24, 51 ENaC has not yet been described in fibroblasts, so amiloride may not directly affect fibroblast collagen synthesis but its indirect effects are under investigation as previously noted.

There are obvious weaknesses of the present observational report but our results are highly consistent internally and the population is unique. In addition to the small sample size, we have no time or case‐matched controls but we compared the observed arterial stiffness parameters in this cohort to the corresponding values predicted by the application of a model for each arterial stiffness parameter derived from a normative reference population without HA. Thus, a control group was not necessary. Overall, it is hoped that these observations will spur additional investigation.

5. SUMMARY

Long‐term low‐dose amiloride‐based therapy in patients with HA was highly effective in controlling BP and maintaining normal macrocirculatory (arterial stiffness and stiffening, ankle‐brachial index) and microcirculatory (glomerular filtration and urinary albumin excretion) function. Whether these benefits are attributable to superior BP control, are specific to ENaC blockade, or involve both mechanisms remains to be determined.

CONFLICT OF INTEREST

The authors have nothing to disclose. This work was unfunded.

Izzo JL Jr., Hong M, Hussain T, Osmond PJ. Long‐term BP control and vascular health in patients with hyperaldosteronism treated with low‐dose, amiloride‐based therapy. J Clin Hypertens. 2019;21:922–928. 10.1111/jch.13567