Oral salt loading combined with postural stimulation tests for confirming and subtyping primary aldosteronism

Introduction

Primary aldosteronism (PA) is one of the most common causes of secondary hypertension [1]. The prevalence is often underestimated as it is a spectrum of disease that ranges from completely asymptomatic to overt PA [2]. Common causes of PA include aldosterone-producing adenoma (APA) or bilateral adrenal hyperplasia (BAH). Patients with PA have been reported to have a higher risk of cardiovascular events and renal impairment [3]. Early diagnosis and treatment can minimize the adverse effects of PA.

The treatment of PA depends on its subtype. Patients with APA can be treated with unilateral adrenalectomy, whereas mineralocorticoid receptor antagonists, such as spironolactone, can be used in all patients with PA. Yet, patients with APA undergoing surgical treatment have been shown to have better cardiovascular outcomes compared to those treated medically [4], 5].

According to the Endocrine Society guideline [6], PA subtyping can be achieved using adrenal computed tomography (CT) and adrenal venous sampling (AVS). Despite being the gold standard for PA subtyping, AVS has several limitations [7]. It is invasive, and it requires experienced radiologists for catheterization. The direct connection of the right adrenal veins with the inferior vena cava makes catheterization difficult. Even when catheterization is successful, the interpretation of AVS results may sometimes fall into the grey zone. Cases with asymmetric cortisol secretion between the two adrenal glands could lead to a misleading AVS result [8]. Therefore, different studies have explored alternative methods for subtyping PA, including imaging [9], [10], [11], [12] and biochemical tests [13], 14]. A prospective study has even shown that using abdominal imaging for PA subtyping might not be inferior to AVS results [15]. Another prospective study also proposed that the use of [¹¹C]metomidate scan could achieve non-inferior accuracy for PA subtyping compared to AVS [16]. This challenged the value of AVS for PA management.

Non-invasive biochemical tests might be a potential aid for PA subtyping. The oral salt loaded postural stimulation test (PST) for PA subtyping was first proposed in 1973, based on a sample size of 11 APA and 7 BAH patients [17]. Patients were required to check for plasma renin, aldosterone, and potassium after (i) overnight bed rest and fasting and (ii) 4 h after mobilization. A drop in aldosterone levels suggests APA. This test is not widely used in clinical settings owing to its questionable diagnostic accuracy [18]. Recently, the potential of non-salt loaded PST for PA subtyping with or without imaging has been evaluated by different studies, but with variable performance [19], 20]. In these studies, biomarkers were measured mainly by immunoassay, and salt loading, which eliminates aldosterone rise due to sodium depletion [6], was not a prerequisite for the PST. Optimising PST might improve the diagnostic accuracy for PA subtypes.

Salt-loaded 24-h urinary aldosterone (UALD) was one of the recommended confirmatory tests for PA in the Endocrine Society guideline [6]. In 2003, Abdelhamid et al. [21] has proposed the use of urine tetrahydroaldosterone (UTHA), the major metabolite of aldosterone, as a screening test for PA. However, its use for PA confirmation has not yet been evaluated. At that time, urinary metabolites were still measured by radioimmunoassays. With the advance of technology, UALD and UTHA could be accurately and simultaneously measured in urine with LC-MS/MS [22]. Its potential use in PA workup should be explored.

UALD measured by LC-MS/MS in our centre, has been available since 2013. Therefore, oral salt loading with UALD measurement has contributed to the diagnostic algorithm in our centre. Meanwhile, the increase in the number of PA patients but limited radiological resources lengthened the queuing time for imaging or AVS. Therefore, in our centre, a non-invasive protocol, the balance study, consisting of an oral salt loading test followed by PST and 24-h UALD collection, was adopted. This protocol aimed to help confirm PA by salt loading UALD and provide preliminary insight for the PA subtyping by PST. In this study, we conduct a retrospective evaluation to ascertain if additional biomarkers measured in this protocol could assist in diagnosing and subtyping PA. We also aim to determine the cut-offs for potential biomarkers using LC-MS/MS.

Materials and methods

The Chinese University of Hong Kong–NTEC Clinical Research Ethics Committee has approved this observational retrospective study (CRE-2022.500) to evaluate the clinical utility of the different biomarkers used in the balance study. Due to its retrospective observational nature and the use of anonymised data obtained during routine clinical practice, no written consent from the patients was required.

Results

Baseline characteristics of the study cohort

Figure 1 shows the data extraction algorithm. There was a total of 402 patients diagnosed with either PA or EH. PA patients had a significantly higher screening PALD, ARR, UALD, and UTHA, and a significantly lower plasma potassium concentration than EH patients. However, no significant difference was observed in the screening PRA (Supplementary Table S1). Comparing the screening parameters of APA and BAH patients (Supplementary Table S2), the baseline plasma potassium concentrations in APA patients were significantly lower than in BAH patients. The screening PRA was lower, and ARR was higher in APA patients than in BAH patients.

Figure 1:

Flow diagram of data extraction. APA: aldosterone-producing adenoma; BAH: bilateral adrenal hyperplasia; EH: essential hypertension; PA: primary aldosteronism.

As for the balance study, salt-loaded PRA, PALD, and ARR performed in erect and supine postures showed a statistically significant difference between the PA and EH groups. The same was observed between the APA and BAH groups. Salt-loaded UALD and UTHA were significantly higher in the PA group compared to the EH group, and they were significantly higher in APA patients than BAH patients (Tables 1 and 2).

Table 1:

Comparison of parameters measured in the balance study between patients with primary aldosteronism (PA) and essential hypertension (EH).

	PA (n=310)	EH (n=92)	p-Value
24-h salt-loaded urine
Aldosterone, nmol/day	62 (57–66)	24 (21–26)	<0.001
Tetrahydroaldosterone, nmol/day	293 (272–320)	115 (101–132)	<0.001
Plasma: Supine
Renin activity, ng/mL-h	0.17 (0.15–0.20)	0.23 (0.17–0.40)	0.026
Aldosterone, pmol/L	423 (400–456)	160 (143–175)	<0.001
Aldosterone-renin ratio, pmol/L/ng/mL-h	2,073 (1,828–2,553)	624 (438–843)	<0.001
Plasma: Erect
Renin activity, ng/mL-h	0.40 (0.35–0.44)	0.61 (0.42–0.82)	0.002
Aldosterone, pmol/L	540 (497–574)	289 (247–338)	<0.001
Aldosterone-renin ratio, pmol/L/ng/mL-h	1,226 (1,049–1,538)	439 (305–608)	<0.001

1. p-Value <0.05 are bolded. Numerical results are expressed in median (95 % confidence interval).

Table 2:

Comparison of parameters measured in the balance study between patients with unilateral aldosterone-producing adenoma (APA) and bilateral adrenal hyperplasia (BAH).

	APA (n=97)	BAH (n=105)	p-Value
24-h salt-loaded urine
Aldosterone, nmol/day	75 (61–81)	57 (50–64)	0.016
Tetrahydroaldosterone, nmol/day	329 (281–360)	267 (237–296)	0.009
Plasma: Supine
Renin activity, ng/mL-h	0.13 (0.10–0.17)	0.26 (0.17–0.39)	<0.001
Aldosterone, pmol/L	605 (546–689)	343 (276–392)	<0.001
Aldosterone-renin ratio, pmol/L/ng/mL-h	4,448 (2,769–6,713)	1,107 (871–1,872)	<0.001
Plasma: Erect
Renin activity, ng/mL-h	0.28 (0.22–0.38)	0.75 (0.50–0.99)	<0.001
Aldosterone, pmol/L	373 (306–467)	595 (542–665)	<0.001
Aldosterone-renin ratio, pmol/L/ng/mL-h	1,359 (902–1,888)	829 (613–1,201)	0.015

1. p-Value <0.05 are bolded. Numerical results are expressed in median (95 % confidence interval).

Diagnostic performance of different markers for confirming PA

The diagnostic performance of different markers for confirming PA was evaluated in all patients with PA and EH. Supine or erect ARR, supine or erect PRA, and erect PALD had areas under the curve (AUC) of less than 0.8 (Figure 2A). UALD, the marker recommended by the Endocrine Society for PA confirmation [6], achieved an AUC of 0.894. The derived cut-off was >38 nmol/day, with a sensitivity (SN) of 82 % and a specificity (SP) of 91 %. The markers with a similar AUC and their respective cut-off values included: UTHA (AUC 0.875, cut-off > 160 nmol/day; SN=86 %; SP=74 %), the sum of UALD and UTHA (SUM; AUC 0.884, cut-off >259 nmol/day; SN=72 %; SP=91 %), and supine PALD (AUC: 0.902, cut-off >273 pmol/L; SN=76 %; SP=92 %).

Figure 2:

(A) The receiver operating characteristic curves for different analytes in confirming primary aldosteronism in 310 patients with primary aldosteronism and 92 patients without primary aldosteronism. (B) The receiver operating characteristic curves of different analytes for primary aldosteronism subtyping for 97 patients with aldosterone-producing adenoma and 105 patients with bilateral adrenal hyperplasia.

Diagnostic performance of different markers for subtyping of PA

The diagnostic performance of different markers for subtyping PA was also evaluated in APA and BAH patients. No single marker (i.e., supine and erect PRA, supine and erect ARR, supine and erect PALD, UALD, or UTHA) could achieve an AUC greater than 0.8, with the highest being supine PALD with an AUC of 0.79. Therefore, the combination of different markers, namely supine and erect PRA, supine and erect ARR, and supine and erect PALD, were assessed. While the difference or ratio of PRA could reach an AUC of less than 0.7, and that of ARR could reach less than 0.8, we observed that the difference or ratio of erect and supine PALD could achieve an AUC of 0.90 (Figure 2B). The corresponding cut-offs to suggest APA would be a difference between erect and supine PALD less than +19 pmol/L (SN=84 %, SP=87 %), or at most a 5 % increase of erect PALD compared to supine PALD. Despite the similar AUC values, the AUC for supine PALD was significantly lower than that of the erect-to-supine PALD difference or ratio (p-value =0.019).

We applied the proposed cut-offs for PA confirmation and subtyping to our original study subjects (Figure 3). If at least one of the criteria was fulfilled, it was counted as positive. For PA confirmation, the overall SN reached 94 % with an SP of 76 %. There would be 17 patients (4 % of total) who have the diagnosis of PA missed, in which 2 of them were APA patients. For PA subtyping, 32 patients (16 %) would have incorrect subtyping, and 14 of them who should have BAH would be wrongly classified as APA, and 18 APA patients would be wrongly regarded as BAH and missed the chance for lesion localisation.

Figure 3:

Flow diagram of patients’ diagnosis based on proposed cut-offs.

Discussion

Making a diagnosis of PA and its subtype is a constant challenge. Our study demonstrated that the balance study could be a non-invasive approach for PA confirmation and aiding subtyping. The major findings from our study included 1) combining different markers in addition to UALD for PA confirmation could minimise the chance of missing PA patients, and 2) the difference or the ratio between erect and supine PALD might help subtype PA.

Several markers, including UALD, UTHA, SUM, and supine PALD after salt loading, have been evaluated for their diagnostic utility in confirming PA. Our findings suggest that UTHA and UALD, measured by LC-MS/MS, exhibit comparable diagnostic performance (as reflected by similar AUC values). Incorporating UTHA into the diagnostic panel for PA could provide additional confirmatory evidence without significantly increasing the laboratory workload, given that LC-MS/MS can simultaneously quantify both markers.

Among the proposed markers for PA confirmation, fasting supine PALD in the balance study could achieve an AUC greater than that of UALD. Unlike usual sitting PALD or ARR for PA screening, it is specifically tailored to eliminate variabilities associated with posture and dietary salt intake. In this study, the AUC of screening ARR for PA confirmation was only 0.677, while fasting supine PALD in the balance study was 0.902. Interestingly, the fasting supine ARR could only achieve an AUC 0.791. This probably could be accounted by the pathophysiological nature of PA, which is the autonomous secretion of aldosterone. The autonomous increase in aldosterone would be the first abnormality observed, which later suppresses the secretion of renin by feedback mechanism. As the patients included in balance study were based on both clinical and laboratory findings, a spectrum of the disease would be expected, including those with PRA not yet suppressed. Therefore, the sensitivity for ARR (both screening sitting or salt-loaded supine ones) might be inferior to that of supine salt-loaded fasting PALD for confirming PA.

The recommended cut-off for UALD based on a mass spectrometry method by the Endocrine Society [6] is more than 33 nmol/day, which is lower than our proposed cut-off of 38 nmol/day in our study. Our proposed cut-off for UALD of 38 nmol/day would miss 54 PA cases, while using a UALD cut-off of 33 nmol/day would miss 43 cases (i.e., 11 cases less). However, the additional use of supine PALD could detect seven out of the 11 PA cases. Therefore, using both UALD and supine PALD could maximize the number of PA patients diagnosed. On the other hand, 19 EH patients would be regarded as PA if the UALD cut-off of 33 nmol/day was adopted. Our proposed UALD cut-off would only wrongly diagnose 7 EH patients as PA. If UTHA is also considered, 4 cases would be correctly diagnosed as EH instead. Therefore, considering UALD and UTHA results might help rule out PA. Therefore, a combination of markers for diagnosing PA should be considered. To the best of our knowledge, this is the first study to suggest the potential use of UTHA and supine PALD for confirming PA.

Approaches recommended in the Endocrine Society guidelines for PA confirmation include oral sodium loading test, saline infusion test, captopril challenge test, and fludrocortisone suppression test [6]. There was paucity of evidence to prove that any one of the PA confirmation tests could be superior to the rest [24]. The captopril challenge test has lower sensitivity and specificity than other confirmatory tests [6]. Saline infusion is a more invasive method for salt loading as it requires intravenous access. The fludrocortisone suppression test requires frequent monitoring of plasma potassium levels, as fludrocortisone, a potent mineralocorticoid agonist, can suppress potassium significantly [6]. Compared to plasma confirmation tests, which check PRA and PALD at a particular time point with a particular posture, urinary tests might be more convenient because they do not have restrictions in posture and diet. Therefore, the oral salt loading test might be a good choice for PA confirmation.

Urine samples were collected in an inpatient setting according to our protocol, which might have minimized the chance of incomplete collection and a falsely low UALD result. However, patients with chronic renal failure, congestive cardiac failure, or uncontrolled hypertension might not be able to tolerate oral salt loading. Meanwhile, overnight admission would increase the medical cost. Therefore, considering a combination of biomarkers proposed in this study for PA confirmation might improve diagnostic accuracy and cost effectiveness.

For PA subtyping, no single marker alone provided a satisfactory AUC. By comparing the absolute or percentage changes in PALD from a supine to an erect position, an AUC of 0.90 was achieved. To suggest APA, an absolute change of erect PALD from supine less than +19 pmol/L or an absolute percentage change of <+5 % could achieve an almost 90 % sensitivity and specificity. Our proposed cut-offs aim to minimise the number of false positive APA cases so that the chances of unnecessary adrenalectomy would be minimised. However, AVS or imaging would be required for localisation of the lesion before adrenalectomy could be offered. Table 3 summarizes the performance of the erect/supine PALD ratios in different publications. Compared to the PST protocols used in other publications [19], 20], 25], our protocol did not include cortisol. The higher AUC in our study for the erect/supine PALD ratio might be explained by the salt loading before testing, as well as the use of LC-MS/MS for analytes measurement, similar to the case in saline infusion test [26].

We propose using the balance study as this non-invasive protocol can achieve PA confirmation and subtyping simultaneously with a reasonable specificity and sensitivity, as compared to the conventional approaches for PA subtyping (i.e. AVS and computed tomography of adrenals). This could not be a complete replacement to the orthogonal diagnostic algorithm for PA, especially for subtyping, as it could not achieve localisation. However, with reference to the results of balance study, clinicians could prioritise the work-up for localisation of those having a higher chance of PA, which could streamline the workflow and detect more APA patients earlier, given that imaging and AVS are the usual bottleneck for PA subtyping.

From the decision tree suggested based on our cut-offs, for PA confirmation, the overall sensitivity was 94 % and the specificity was 76 %. Compared to salt loaded UALD alone (SN=82 %, SP=91 %), this approach could diagnose more PA patients for further workup. As for PA subtyping, the accuracy was 84 %, which deemed satisfactory as a non-invasive guide for prioritising patients for APA localisation. However, as it could never provide laterality, it could not replace AVS or other localisation modalities.

Our study has several strengths. The biochemical analyses were performed by LC-MS/MS, which has a lower chance of interference and higher accuracy than immunoassays. The sample size in our study was sufficient to demonstrate the usefulness of the balance study.

Our study has some limitations. The primary limitation of this study is its retrospective design, which introduces potential sources of bias, particularly selection bias. This is partly reflective of real-world clinical practice, as only patients with a high probability of PA were included, and definitive subtyping was not possible for approximately one-third of the cohort. Supplementary Table S3 compares patients with and without a determined subtype. The duration of hypertension and the older age in the latter group might explain the reluctance to invasive treatment. Furthermore, the data collection was incomplete in some cases, including the baseline screening parameters, as some were measured either by external laboratories or using different methodologies (such as radioimmunoassay vs. LC-MS/MS). Additionally, the effect of renal impairment on UALD or UTHA was not explicitly assessed, as patients with significant renal impairment would usually not undergo urinary hormonal assessment in routine practice. Also, our combined protocol did not evaluate the effect of adrenocorticotropic hormone (ACTH) on aldosterone release, especially in patients with APA. Evaluating changes in plasma cortisol concentrations may help improve the accuracy of PST. As shown in a study on PST by Fuss et al. [20], using the same cut-off of a 28 % drop in PALD when a patient changed from a supine to an erect position, excluding cases without a cortisol drop of at least 10 %, would raise the sensitivity from 36.4 to 51.4 % with the same specificity of 100 %, and increase the AUC of erect/supine PALD ratio from 0.72 to 0.77. However, there are some studies adopting PSTs without measuring cortisol [19], 27]. The effect of ACTH or cortisol on aldosterone secretion remained uncertain [28].

In summary, we demonstrated that the balance study could help PA confirmation and provide direction on PA subtyping in a non-invasive manner. Using UTHA and supine PALD could complement UALD for PA confirmation, whereas the erect/supine PALD ratio could assist in PA subtyping. These results might improve the diagnostic process for PA, especially in cases with equivocal AVS results or in regions where AVS is not readily available. More studies are warranted to assess the effects of ACTH and cortisol on PA subtyping in PST.