High‐Probability Features of Primary Aldosteronism May Obviate the Need for Confirmatory Testing Without Increasing False‐Positive Diagnoses

Abstract

This retrospective review examined all primary aldosteronism (PA) adrenal vein sampling (AVS), diagnoses, and outcomes from an endocrine hypertension unit where confirmatory testing was abandoned in 2005 to determine the potential rate of false‐positive diagnoses. Patients with outcome‐verified PA (surgical patients) were compared with patients with high‐probability PA (nonsurgical but high aldosterone‐renin ratio, imaging abnormalities, and/or hypokalemia) or possible PA (nonsurgical, no features besides mild elevation of aldosterone‐renin ratio, a potential false diagnosis of PA). Of 83 patients, 58% had unilateral PA and 42% had bilateral aldosteronism. Less than 3% of the cohort showed bilateral aldosteronism without hypokalemia or computed tomographic findings, potentially representing the false‐positive PA diagnosis rate with omission of confirmatory tests in this population. In a hypertension referral unit enriched in high‐probability PA cases and where high AVS success is achieved, omission of a PA confirmatory test yields a high rate of surgical diagnosis with few potential false‐positive diagnoses.

Primary hyperaldosteronism (PA) is an increasingly recognized cause of remediable hypertension.1, 2, 3, 4 Several long‐term series have confirmed a significant benefit for both blood pressure (BP) and clinical outcomes when PA cases are identified and specifically treated.5, 6 Approximately 20% to 40% of patients with unilateral aldosteronism will be cured of their hypertension following adrenalectomy.7, 8, 9 Therefore, many centers around the world now regularly screen for PA and specialize in determination of PA subtypes that may be amenable to surgery.

The issue of sensitivity and specificity in PA testing is critical. Given the major benefits of recognition and treatment of PA, the clinician should ideally perform a test with very high sensitivity in order to avoid missing any true cases. At the same time, adrenal vein sampling (AVS) is invasive and therefore a high rate of false positive results of PA screening is unacceptable.

As may be seen in any analysis of a diagnostic receiver operating characteristic curve, the sensitivity and specificity of a quantitative test threshold are inversely related, ie, a test at maximum sensitivity will often have poor specificity and vice versa.10 This has led to the long‐standing practice of using a sensitive test for screening, followed by what is termed a confirmatory test—namely, a test that, when abnormal, is thought to have a high degree of specificity for the disease in question. The practice of confirmatory testing in PA is currently recommended by the Endocrine Society's guidelines for PA diagnosis.11

There are a variety of PA confirmatory tests used in hypertension units around the world, and the details of these have been recently reviewed.12 However, despite more than a decade of descriptive and comparative research, there is still no consensus as to what constitutes the “optimal” confirmatory approach. Attempts at comparative studies of confirmatory tests have been hampered by retrospective designs and small numbers with inconsistent use of AVS.13, 14, 15, 16

Hyperfunctional adrenal disease may be unique in that it is not easily studied with certainty. For any study of a diagnostic test to be of high quality, there are two prerequisites: (1) there must be a gold standard definitive diagnosis that must be applied to all patients for comparison, and (2) the test being studied cannot be used to determine which patients will ultimately undergo the gold standard test.17 Therefore, the true false‐negative rate of any confirmatory test in PA can never be evaluated, especially in retrospective study designs. This is further confounded by the “spectrum effect,” previously discussed with respect to adrenal disease,18 in which a test's true sensitivity and specificity cannot be definitively known except perhaps in the most obvious and severe cases as a result of uncertainty in diagnosis of milder disease that overlaps with a normal population.

The final consideration is that the clinical picture must inform the interpretation of the test results, regardless of the supposed sensitivity and specificity.19 In 2005, our Aldosteronism Interest Group concluded that studies of confirmatory tests of PA were difficult to interpret or use clinically as a result of the practical inability to know any test's true false‐negative rate and the lack of a validated pre‐test probability calculator. Because we did not want to sacrifice sensitivity in disease detection, and in concert with the knowledge that clinical context must not be discounted, we stopped performing any confirmatory tests as part of the PA workup. The current study is an audit of our experience since that change in practice.

Although never systematically proven, if AVS is performed in patients without true PA, normalized aldosterone levels are expected to be relatively symmetric from both adrenals, or at least not as asymmetric as seen in the setting of an aldosterone‐producing adenoma (APA). Thus, if a large number of so‐called PA cases were actually non‐PA by virtue of high false‐positive diagnosis, one would expect to see a higher proportion of AVS results that are determined to be “bilateral.”

Methods

We conducted an institution‐approved retrospective review of all PA patients seen in a single endocrine hypertension specialty unit practice from June 2005 to July 2011. Our approach to PA diagnosis is summarized in the Figure. All patients referred to the unit had measurement of an aldosterone‐renin ratio (ARR) prior to their first appointment, irrespective of current medication used, except for mineralocorticoid antagonists, which are stopped for 6 weeks. The ARR was drawn in the seated, upright position prior to 10 AM. When reviewed in clinic by the endocrinologist, if the ARR was >550 (aldosterone in pmol/L, renin activity in ng/mL/h), the patient was counselled about a possible diagnosis of PA. At the endocrinologists' discretion, the ARR may have been repeated if it was felt to be a marginal elevation. If the patient requested further investigation to determine whether they had a surgical option, provided they were deemed to be a potential surgical candidate, then computed tomography and AVS were performed using our previously published protocol.20 Patients with resistant hypertension in whom the ARR was normal were offered a repeat testing of the ARR after manipulation of their medications to stop any β‐blockers, angiotensin‐converting enzyme inhibitors, angiotensin receptor blockers, diuretics, or mineralocorticoid receptor antagonists, with use of doxazosin and nifedipine or diltiazem as interim agents. Oral potassium supplements were given to ensure eukalemia prior to repeat testing. Those whose repeat ARR was >550 were offered imaging and AVS as per above. Patients with normal ARR were managed per usual hypertension guidelines. Based on evidence showing predictability of response to spironolactone when ARR is high,21 patients with ARR >550 who refused further investigation or who were nonsurgical candidates were diagnosed as having “possible PA” and treated with a mineralocorticoid antagonist without further testing. AVS and definitive diagnosis was thus pursued only in those who were deemed to be potential surgical candidates.

Figure 1.

Diagnostic algorithm used for diagnosis of primary aldosteronism (PA). ARR indicates aldosterone‐renin ratio; MRA, mineralocorticoid receptor antagonist; AVS, adrenal vein sampling; CT, computed tomography. ARR units are expressed as aldosterone in pmol/L and plasma renin activity in ng/mL/h.

The ARR cut‐point of 550 was chosen based on a review of prior literature regarding ARRs.22 We intentionally chose a low ARR to be used in order to maximize the sensitivity of the screening program. The computed tomographic (CT) images comprised the area from the top of the diaphragm to the iliac crests with a slice thickness of 1.5 mm at 1‐mm intervals. For the present analysis, adrenal masses on CT were defined as any reported finding of an adrenal nodule >8 mm, either unilateral or bilateral. AVS was performed using samples drawn pre‐ and post‐250 μg cosyntropin intravenous infusion with a selectivity index of >3:1 for successful catheterization and lateralization determined by aldosterone/cortisol ratio of >3:1 vs the unaffected adrenal. In our clinic, this interpretation rule has previously been shown to have 100% specificity.23 According to the placement criteria, 80 of 83 patients (96%) had successful AVS. Of the 3 patients with failed AVSs, 2 received primary medical therapy and were counted in the bilateral group and 1 still chose surgery on the basis of hypokalemia and a 14‐mm adrenal mass.

A diagnosis of PA was considered confirmed if supported by the combination of lateralized AVS, surgical pathology showing adenoma +/− hyperplasia, postoperative clinical response (hypertension resolution or marked improvement), and postoperative normalization of ARR and potassium. A diagnosis of “high‐probability PA” was considered for patients in whom AVS was nonlateralized yet had either an adrenal mass and/or history of hypokalemia (serum potassium ≤3.5 mmol/L) in concert with their high ARR. Given that specificity increases (by definition) with increasingly abnormally high ARR, high‐probability PA was also assigned to patients without imaging findings or hypokalemia if their ARR was higher than twice the ARR cutoff used for screening (ie, >1100). A diagnosis of “possible PA” was reserved for patients with nonlateralizing AVS who lacked imaging findings or hypokalemia and whose ARR was less than twice the ARR screening cutoff. In all groups, the clinical response was defined as “hypertension resolution” for those with sustained BP <140/90 mm Hg off all medications, “marked improvement” for those with sustained BP <140/90 mm Hg on fewer medications than presentation, and “minimal improvement” for those with <10% change in BP after either surgery or mineralocorticoid antagonist use. However, clinical response to a mineralocorticoid receptor antagonist was not used as part of the case definitions because it is not specific to PA‐related hypertension.

Laboratory Procedures

Plasma aldosterone was measured by solid‐phase I¹²⁵ radioimmunoassay (RIA) using the Siemens Coat‐A‐Count Aldosterone assay (PITKAL‐10; Siemens, Los Angeles, CA). Mean intra‐assay imprecision coefficients of variation (CVs) were 13.0%, 6.8%, and 6.8% for aldosterone levels of 115 pmol/L, 665 pmol/L, and 1610 pmol/L, respectively. Plasma renin activity (PRA) was measured by RIA of angiotensin I in the presence of reagents that inhibit angiotensin‐converting enzyme and angiotensinases using the GammaCoat Plasma Renin Activity I¹²⁵ kits (DiaSorin, Stillwater, MN). Mean interassay imprecision CVs were 21.3%, 12.5%, and 17.9% for PRA levels of 0.93 ng/mL/h, 4.8 ng/mL/h, and 19.5 ng/mL/h, respectively. Where PRA was undetectable, the lowest reporting limit of 0.1 ng/mL/h was used to compute the ARR to avoid overinflation of the result.

Plasma cortisol was measured by an automated electrochemiluminescence assay on the Roche Elecsys E170 (Roche Diagnostics, Mississauga, Ontario, Canada). Mean intra‐assay imprecision CVs were 6.4%, 4.0%, and 3.2% for cortisol levels of 110 nmol/L, 624 nmol/L, and 908 nmol/L, respectively.

Descriptive statistics were performed for the entire cohort along with comparisons of population demographics between the lateralized and nonlateralized cohorts. A Mann‐Whitney test was used for comparison between nonparametrically distributed groups using a 2‐sided test with α=0.05 for significance. A chi‐square test compared the difference between categorical variables. All computations were performed using Analyse‐it software, 2011, (Analyse‐it Software, Leeds, United Kingdom).

Results

Population Going to AVS

Table 1 demonstrates the demographics and diagnostic details for the entire 83‐patient cohort who underwent AVS for PA during the study period. In the total population, the mean number of antihypertensive medications at presentation was 2.66, and 67.5% had hypokalemia (serum potassium <3.5 mmol/L). The median ARR was >3000. In total, 14% had an ARR between 550 and 1100, 27% had an ARR between 1101 and 2750, and 59% had an ARR >2750 (500% above the screening cutoff). At imaging, 66% of patients were thought to have an adrenal mass present, with a mean size of 16 mm. Of 83 AVS procedures, 80 (96.3%) were technically successful according to the cortisol results. There were no complications from AVS procedures.

Table 1.

Population Demographics of Study Cohort

Parameter	Lateralized (n=48)	Nonlateralized (n=35)	P Value (Lateralized vs Nonlateralized)
Female sex, %	41.7	51.4	.27
Age, y	49 (41–56)	48 (39–54)	.51
BMI, kg/m2	30.5 (25.4–35.1)	30.3 (25.3–33.1)	.63
Race (white/black/Asian), %	73/8/19	71/23/6	.96
SBP, mm Hg	149 (135–159)	145 (130–158)	.67
DBP, mm Hg	90 (82–101)	94 (80–103)	.86
Mean medications, No.	2.52 (1.5)	2.86 (1.54)	.32
CT abnormality, %	88	34	<.001
CT abnormality, mm	16 (12–23)	17 (13–22)	.82
Potassium, mmol/L	3.05 (2.60–3.46)	3.40 (3.2–3.8)	<.001
Hypokalemia, %	77.0	54.2	.06
Renin in patients with diagnosis made at first measurement, ng/mL/h	0.11 (0.1–0.18)	0.11 (0.1–0.19)	.89
Aldosterone in patients with diagnosis made at first measurement, pmol/L	646 (417–924)	402 (267–729)	.01
ARR in patients with diagnosis made at first measurement	5070 (2259–8123)	2847 (1557–4347)	.03
Renin in patients with diagnosis made after medication adjustment, ng/mL/h	0.42 (0.23–0.62) (n=16)	0.34 (0.17–0.48) (n=8)	.37
Aldosterone in patients with diagnosis made after medication adjustment, pmol/L	526 (265–1003) (n=16)	308 (260–450) (n=8)	.24
ARR in patients with diagnosis made after medicine adjustment	1314 (877–2241) (n=16)	1134 (707–2251) (n=8)	.71
Overall renin for diagnosis, ng/mL/h	0.18 (0.1–0.36)	0.13 (0.1–0.29)	.49
Overall aldosterone for diagnosis, pmol/L	646 (387–980)	402 (271–567)	.007
Final ARR for diagnosis	3490 (1318–7067)	2580 (1214–3291)	.32
Histology, APA/APA‐H/UAH, % (N=40 [2])	65/23/12
Lost to follow‐up, %	6	6
Follow‐up, months post‐AVS	8.0 (5–14)	11 (5–18)	.22
Outcome: cure, No. (%)[Link]	17 (43.5)	n/a
Outcome: marked improvement, % (No.)[Link]	53.8 (21)	91.4
Outcome: marginal improvement, % (No.)	2.6 (1)	9.6

Abbreviations: APA, aldosterone‐producing adenoma; APA‐H, aldosterone‐producing adenoma with surrounding hyperplasia; BMI, body mass index; CT, computed tomography; DBP, diastolic blood pressure; n/a, not applicable; SBP, systolic blood pressure; UAH, unilateral adrenal hyperplasia. Values are expressed as median–interquartile range unless otherwise indicated. ^aAmong lateralized patients, 7 chose primary medical therapy despite lateralizing adrenal vein sampling (AVS), 1 underwent surgery but was lost to follow‐up, and 1 chose no therapy and was lost to follow‐up. ^bNonlateralized patients' outcomes determined after the addition of an aldosterone antagonist.

Patients With Lateralization by AVS

A total of 58% of all patients with AVS were found to lateralize to one adrenal (Table 1). The patients with lateralization did not differ from those with nonlateralization in terms of age, sex, racial background, body mass index, BP, number of medications, or ARR. The lateralized group had a lower mean initial potassium level and, as a group, a higher proportion of hypokalemia compared with the nonlateralized group. In addition, 88% of the patients in the lateralized group had a positive CT finding compared with just 34% of patients in the nonlateralized group (P<.001). There were 40 of 48 patients in the lateralized group who went on to unilateral adrenalectomy. Of those who refused surgery, one was lost to follow‐up after AVS and 7 chose primary medical therapy instead of surgery. Their specific data are shown in Table 2 where it may be seen that all 8 could at least be classified as having high‐probability PA as a result of a combination of very high ARR, hypokalemia, and adrenal masses seen on CT with concordant lateralization on AVS. In the surgical group, 65% were found to have an isolated APA, 23% had APA with surrounding hyperplasia, and 12% had pure adrenal hyperplasia. After a median postoperative follow‐up of 8 months, 34.1% had hypertension resolution, 61.4% had marked improvement, and 4.5% had marginal improvement. Patients with minimal improvement had APA on histology, one of whom had an additional contralateral adrenal mass with persistent hypokalemia and high ARR and was presumed to be probable bilateral APA but treated medically thereafter. Of 25 patients who had a postoperative ARR measured, 23 (92%) were normal (<550 pmol/L), one was borderline elevated but with very low aldosterone (106 pmol/L), and one was a “nonresponder” whose ARR was grossly elevated despite histology confirming solitary adenoma, as detailed above.

Table 2.

Patients With Lateralized PA Who Refused Surgery

Patient	ARR (Normal <550)	CT Mass, mm	Initial Potassium, mmol/L	AVS Ratio/Normalized Aldosterone (Dominant vs Nondominant Adrenal)
1	2070	23	4.8	5.7:1
2	1509	26	4.1	3.1:1
3	1369	14	2.9	12.3:1
4	1079	15	3.9	9.1:1
5	4630	None	3.1	3.7:1
6	3940	21	3.3	9.5:1
7	1467	13	2.4	20.0:1
8	1321	None	4.0	9.9:1

Abbreviations: ARR, aldosterone‐renin ratio; AVS, adrenal vein sampling; CT, computed tomography; PA, pulmonary aldosterone.

Patients With Nonlateralization

Despite lack of evidence of APA/lateralization, 54% of this group had hypokalemia and 34% had an adrenal mass according to CT with an ARR that did not differ from patients who did have lateralization (P=.29). After a median follow‐up of 11 months, the addition of an aldosterone antagonist resulted in a BP <140/90 mm Hg in 91% of patients, with just 9% showing marginal improvement. Considering the “whole patient” context of the nonlateralized group, demonstrated in Table 3, it was found that 66% had either a positive CT finding or hypokalemia (in this group, 23% had both features) and just 34% had neither CT findings nor low potassium levels. In patients who lacked either a CT finding or hypokalemia, the mean ARR was 2860 pmol/L and therefore only 8% of the entire cohort went on to AVS with neither CT findings or hypokalemia and an ARR <2750 pmol/L (5 times greater than the stated screening cutoff). Using the case definitions of high‐probability PA and possible PA/possible false‐positive PA, it may be seen that only 3 patients in the entire cohort could retrospectively have their diagnosis of PA called into question by virtue of nonlateralizing AVS and lacking any other features that would raise a clinician's pre‐test likelihood for a true PA diagnosis. Alternately, there were only 3 patients for whom one might retrospectively recognize that there could be no final proof of PA diagnosis, which, known in advance, might have avoided AVS testing.

Table 3.

Clinical Context of Patients With Nonlateralizing AVS

High‐Probability Feature	High‐Probability PA/AVS Clearly Justifieda	Possible PA/Potential False‐Negative Diagnosis/Possibly Unnecessary AVS[Link]
Patients, No.	32	3
CT unilateral adrenal mass, %	34	0
Bilateral CT adrenal masses, %	3	0
Mean size of adrenal mass, mm	16
Hypokalemia, %	54	0
Both adrenal mass and hypokalemia, %	23	0
Either adrenal mass or hypokalemia, %	66	0
Median ARR, range	3334 (550–26,250)	972 (911–1100)
ARR >100% above screening cutoff (1100), %	91	0

Abbreviation: AVS, adrenal vein sampling. ^aHigh‐probability primary aldosterone (PA) defined as aldosterone‐renin ratio (ARR) >550 + hypokalemia and/or adrenal mass or ARR >1100 with or without potassium/computed tomographic (CT) abnormalities. ^bPossible PA defined as ARR <1100 and no potassium or imaging abnormalities.

Marginal Positives: Patients With Screening ARR Between 550 pmol/L and 1100 pmol/L

Table 4 shows the individual data for all patients whose final medication‐adjusted ARR was above the screening cut‐point but was considered just slightly positive and <100% above the case definition cutoff. A total of 83% of these patients could still be identified as having a high pre‐test probability of true PA since 50% had a concomitant adrenal mass on CT, 58% had hypokalemia, and 33% had both. Only 2 (2.4% of entire cohort) patients underwent AVS with a marginal positive ARR, no hypokalemia, no CT findings, and subsequent “bilateral” AVS. Excluding the 1 patient who was lost to follow‐up after AVS, 6 of these “marginal positives” had lateralized disease and all 6 benefitted from surgical therapy.

Table 4.

Patients With Marginally High ARR (550–1100)

Patient	Sex	Age, y	Potassium, mmol/L	Aldosterone, pmol/L	PRA, ng/mL/h	ARR (Normal <550)	CT	Highest AVS Ratio	Pathology or Diagnosis	BP Medication at Presentation vs Final	Surgical Outcome[Link]
1	Male	33	3.8	787	0.86	915	Normal	2.9	BAH (no surgery)	2 vs 2
2	Male	49	3.4	262	0.32	818	Normal	Failed	BAH (no surgery)	6 vs 4
3	Female	55	3.4	196	0.32	613	Left, 14 mm	Failed	Unilateral hyperplasia	1 vs 0	1
4	Male	47	3.1	422	0.57	740	Normal	1.1	BAH (no surgery)	3 vs 3
5	Male	54	3.6	437	0.40	1092	Normal	1.1	BAH (no surgery)	2 vs 1
6	Male	38	3.6	394	0.65	606	Left, 11 mm	7.3	Unilateral hyperplasia	4 vs 2	2
7	Male	69	4.0	421	0.43	979	Normal	41.9	Unilateral hyperplasia	5 vs 2	2
8	Male	41	3.1	572	0.75	762	Normal	2.7	Unilateral hyperplasia	4 vs 1	2, eukalemic
9	Female	44	2.8	127	0.18	705	Right, 15 mm	31.5	APA	2 vs 1	2, eukalemic, ARR 52
10	Female	40	3.1	233	0.29	803	Left, 18 mm	98.0	APA	2 vs 1	2, eukalemic, ARR 38
11	Male	65	3.9	313	0.29	1079	Left, 15 mm	9.1	Lost to follow‐up
12	Male	64	3.4	142	0.21	676	Left, 25 mm	1.1	BAH (no surgery)	3 vs 2

Abbreviations: APA, aldosterone‐producing adenoma; ARR, aldosterone‐renin ratio; AVS, adrenal vein sampling; BAH, bilateral adrenal hyperplasia; CT, computed tomography; PRA, plasma renin activity. Outcome: 1 = normal blood pressure (BP) off all medications, 2 = BP <140/90 mm Hg on fewer medications, 3 = minimal change.

Published Prevalence of Lateralized and Nonlateralized PA

Table 5 summarizes recent published reports of nonlateralized PA prevalence from other specialized hypertension units with aggressive PA screening and routine confirmatory testing. In some centers, and particularly in older reports, AVS usage has not always been uniformly applied, which limits any kind of direct comparison between hypertension units. There is a wide variation in the population prevalence of high‐probability features such as hypokalemia (22%–92%) and adrenal masses on CT (44%–96%). However, when examining modern series with uniform AVS performance, it may be seen that our rate of nonlateralization is very similar to other centers where confirmatory testing is routine, even with similar rates of contextual factors such as low potassium level and abnormal CT findings. Thus, compared with other reports, despite omission of a confirmatory diagnostic step, there is no apparent undue increase in nonlateralizing (and potentially false‐positive) diagnoses of PA.

Table 5.

Reported Prevalence of Bilateral PA Using Different ARR Cut‐Off and Confirmatory Testing at Different Centers

Site/Years of Study	Patients, No.	Diagnostic Approach	AVS Performed/AVS Success, %	CT Abnormality, %	Hypokalemia, %	AVS Lateralization Index to Define Unilateral	Cases deemed to be BAH (or not definitively known), %
Torino, IT24/1994–2002	587	ARR >1111 + aldosterone >416, IVSS >139 pmol/L, CT, AVS	60/success not stated	Not stated	24.9	>4:1	70
Rochester241994–2000	120	ARR >555 + aldosterone >416, oral NaCl suppressed urinary aldosterone >33 nmol/d, CT, AVS	50/success not stated	Not stated	37.0	>4:1	72
Brisbane241992–2002	631	ARR >833, FST aldosterone >166, CT, AVS	100/success not stated	Not stated	21.7	>2:1 with nondominant suppression	65
Singapore24/1995–2001	177	ARR >555 + aldosterone >416, IVSS >277 pmol/L, CT, AVS	Not stated	Not stated	37.3	>5:1	50
Belfast25/1983–2009	100	IVSS >137 pmol/L, CT, AVS	93/54	48	91.7	>2.2:1 and suppression of nondominant	48
Germany26/2006–2007 (multicenter)	553	ARR + high aldosterone + confirmatory test, CT	32/success not stated	Not stated	56.1	Not stated	83
Italy‐227/2001–2004	126	ARR >1111 or >833 post‐captopril, CT, AVS	34/93 success	Not stated	30.5	Not stated	57
Tokyo28/1995–2005	93	PRA <1.0 ng/dL + aldosterone >332 + furosemide or captopril stimulation, CT, AVS	100/success not stated	47	Not stated	Adrenal vein aldosterone >6900	26
Sendai29/4.5 y	87	ARR + captopril test, CT, AVS	100/98 success	81.7	56.3	>3:1	24
Philadelphia30/2001–2007	97	ARR >550 + aldosterone >415, CT, AVS	100/92 success	89	Not stated	>4:1	46
Bethesda31/To 2010	114	ARR, saline suppression, captopril test, CT, AVS	100/100 success	96	Not stated	>4:1	25
Paris32/2009–2010	101	ARR >64 pmol/mU + aldosterone >500 or urine aldosterone >63 nmol/L d, saline suppression >139, CT, AVS	100/86 success	44	55	>4:1	51
Calgary 2005–2011	83	ARR >550, CT, AVS	100/96 success	66.2	67.5	>3:1	42

Abbreviations: ARR, aldosterone‐renin ratio; AVS, adrenal vein sampling; BAH, bilateral adrenal hyperplasia; CT, computed tomography; FST, fludrocortisone suppression test; IVSS, intravenous saline suppression. Aldosterone is reported in pmol/L and renin in ng/mL/h.

Discussion

These data confirm, in a specialized endocrine hypertension unit with experience in PA diagnosis, that despite omitting a confirmatory test for PA, we do not have an inordinately higher rate of bilateral AVS results that might otherwise suggest a high proportion of non‐PA patients. Therefore, in our select population, it is unlikely that we are making a false diagnosis of PA at a higher rate or, at least, not submitting an excess number of patients to invasive localizing tests. In fact, in <3% of surgical candidate patients sent for AVS with a diagnosis of PA, it could be strongly argued that they have false‐positive PA as a result of their lack of hypokalemia, absent CT masses, and bilateral AVS with marginally high ARR to start. A 3% false‐positive rate may well be acceptable when screening for a potentially curable, serious disease with a diagnostic algorithm designed for maximum sensitivity.

Among those who went to surgery, 95.5% had either a meaningful improvement in BP or outright normalization of BP without medications. This figure is comparable to other centers9, 25, 26, 27, 28, 29, 30, 31, 32 and confirms the high value of approaching this problem with a test characteristic set for sensitive detection of cases. Although not formally evaluated in this study, there would likely be significant time and cost savings for the workup of PA when one bypasses the need for a specialized testing unit, nurse, and laboratory costs for any suppression test. This modification to the diagnostic pathway for PA may improve any estimates of the cost‐benefit analysis for PA diagnosis that includes AVS and surgical approaches.

Study Limitations

Our study has several important limitations. First, AVS itself may not be foolproof for defining PA subtypes; it may be possible to get false “negative” in the setting of unilateral disease. Similarly, bilateral findings on AVS does not prove bilateral hyperplasia since, theoretically, non‐PA patients would be expected to have symmetric adrenal aldosterone secretion and AVS is not a PA confirmatory test unto itself. By design, this is a retrospective study and is therefore prone to selection bias in the performance of AVS, which could not be avoided without performing a prospective, randomized trial of confirmatory testing vs no testing using 100% AVS as a gold standard in all patients with abnormal ARR. Our clinic aggressively screens for PA by measuring ARR in all referred patients, but AVS is not offered to patients who have complex medical comorbidities or in whom surgery is refused a priori. Thus, patients proceeding to AVS are already a select group in which to study outcomes. By virtue of being a referral center for resistant hypertension and endocrine hypertension, there is likely a significant referral bias in patients coming for assessment and our population may be particularly enriched in patients with a high probability of having primary aldosteronism. Since low renin can cause a high ARR in the absence of high aldosterone, omission of a confirmatory step may result in a much higher false‐positive diagnosis of PA among less select populations with milder forms of disease or simple low‐renin hypertension. AVS would not be indicated in such patients without more compelling evidence of a true PA diagnosis, which emphasizes the importance of the entire clinical context. Finally, we cannot exclude the possibility that the clinic physicians are using a nonrandom, context‐informed approach to generate a pre‐test probability that directs selection of PA patients for AVS. This is reflected in the relatively high prevalence of hypokalemia in our population, which may predict APA.33 However, this practice is not out of keeping with prior studies showing that more restrictive use of AVS may be possible by using clinical variables in a prediction model.32, 34 Our study adds to this rationale by affirming that a dedicated PA clinic can use clinical context to carefully select suitable patients to bypass confirmatory testing and permits offering of AVS to all potential surgical candidates with even borderline ARR screening.

Second, since these observations have not been prospectively validated, they are not sufficient to justify a whole scale change in practice and cannot be transformed into a rigid protocol for PA diagnosis. Our definitions of high‐probability PA are admittedly arbitrary but not unreasonable since such patients usually had either hypokalemia or adrenal masses accompanying their high ARR. Yet, the conclusions do question the need for routine, mandatory confirmatory testing for PA in experienced hypertension units where AVS is systematically performed before surgery in patients with high‐probability PA. Prior reports of confirmatory testing characteristics have been limited by lack of prospective, blinded comparisons to a uniform gold standard35, 36; therefore, uncertainty remains as to the necessity and accuracy of such testing. Additionally, where an accurate confirmatory test is lacking, one must question the accuracy of various cut‐off values for first‐line screening tests. The results presented here highlight the feasibility of using a relatively low cut‐off value for definition of abnormal ARR without generating an inordinate amount of further testing for confirmation. Indeed, using a cut‐off of 550 (well below that used by many other centers), we found surgical PA in 6 of 11 patients who barely qualified as having abnormal findings, possibly reflecting the full spectrum of the biochemistry of PA. Future prospective investigations should be designed to examine the differences in resource utilization and diagnostic outcomes between matched patients who undergo routine confirmatory PA testing vs those who do not.

Comparison of our results with other published series is also subject to limitations in validity. Several of the other large series comprise time periods that are now several years past. Clinical practices at different centers may certainly have evolved over time with respect to AVS utilization, which could quite significantly change the proportions of lateralized or nonlateralized disease that is found in the most up‐to‐date series. Similarly, several published series include patients who did not undergo AVS and therefore residual diagnostic uncertainty could relegate a larger proportion of patients to a nonoperative/nonlateralized category.

Conclusions

We have demonstrated that with aggressive screening for PA in a select population with a high prevalence of PA risk factors such as hypokalemia and adrenal masses, patients who are surgical candidates have a high proportion of lateralizing disease when subjected to AVS. Omission of a confirmatory test such as saline suppression does not appear to result in inappropriate AVS utilization, especially when the population is enriched in patients with high‐probability PA features. Nevertheless, even patients who lack classic clinical features of PA will often prove to have surgically amenable disease. Clinicians should be cautious not to exclude such patients from further investigation on the basis of a negative confirmatory PA test, where pre‐test probability is high or false‐negative rates of the confirmatory test are unknown. Further prospective studies are needed to determine whether newly described prediction rules can accurately direct PA investigation prior to the stage of confirmatory testing.

Author Contributions

Gregory Kline conceived of the study, assisted in conduct of the study and analysis of results, cowrote the paper, and approves of the final version; Dr Kline takes primary responsibility for the entire study from inception to publication. Janice Pasieka and Adrian Harvey assisted in the recruitment of patients and the clinical care of patients. Benny So performed all of the AVS procedures. Valerian Dias coordinated all of the laboratory handling and quality control for the AVS procedures.

J Clin Hypertens (Greenwich). 2014;16:488–496. ©2014 Wiley Periodicals, Inc.