A Ten Year Overview of Adrenal Vein Sampling in Croatia: Insights and Outcomes

ABSTRACT

Background

Adrenal vein sampling (AVS) is the gold standard for subtype differentiation in primary aldosteronism (PA), yet it remains technically demanding and resource intensive. However, data on long‐term performance, learning curves, and concordance with imaging in real‐world clinical practice are limited.

Methods

We conducted a retrospective single‐centre study including patients with PA who underwent AVS between January 2016 and June 2025. AVS was performed under continuous adrenocorticotropic hormone stimulation using a sequential sampling protocol. Procedural success, learning curve, and diagnostic concordance with computed tomography (CT) were evaluated.

Results

A total of 251 patients underwent 268 AVS procedures. Overall procedural success was achieved in 79% of cases. When stratified by radiologist experience, success rates increased significantly from 70% during the first 3 years to 87.5% in the subsequent 7 years (p = 0.003), demonstrating a clear learning curve. Repeat procedures were required in 7% of patients and were successful in 82% of cases. Among 240 patients with conclusive AVS results, concordance between CT and AVS was 61%. CT demonstrated a sensitivity of 69% and specificity of 55% for correct PA subtype classification, with substantial false‐positive and false‐negative rates.

Conclusions

CT alone is an unreliable surrogate for subtype classification in PA and may lead to inappropriate management decisions. In contrast, AVS achieves high and progressively improving success rates, largely driven by radiologist experience, while also providing reliable diagnostic information for subtype classification. These findings support centralising AVS procedures in institutions with extensive experience, emphasising its crucial role in the optimal management of PA.

1. Introduction

Primary aldosteronism (PA), the most common cause of secondary hypertension, is characterised by autonomous aldosterone secretion. It comprises two principal subtypes: unilateral aldosterone‐producing adenoma and bilateral adrenal hyperplasia, whose distinction is crucial for clinical management. While unilateral disease can often be cured by adrenalectomy, bilateral disease requires long‐term medical therapy [1, 2, 3].

In the diagnostic evaluation of PA, cross‐sectional imaging, most commonly computed tomography (CT), is routinely performed. It plays an essential role in excluding adrenocortical carcinoma and delineating adrenal and vascular anatomy to facilitate subsequent procedures [4, 5]. However, CT alone is not reliable for accurate subtype classification. Therefore, adrenal vein sampling (AVS) is recommended for definitive subtype determination. The concordance between CT‐based assessment and AVS results is only about 50%, highlighting the limited predictive value of imaging and establishing AVS as the gold standard for subtype differentiation [4, 6, 7].

Despite its central role in accurate subtype differentiation in PA, AVS remains a technically demanding, invasive, and resource‐intensive procedure, with variable success rates and substantial inter‐centre heterogeneity [4]. Variation in patient preparation, catheterisation techniques, and analytical criteria can significantly influence diagnostic yield [8]. Accordingly, AVS is generally confined to tertiary referral centres with experienced interventional radiologists [9, 10].

This study provides a comprehensive overview of our 10‐year experience with AVS in patients with PA, evaluating procedural success, learning curve, and concordance with CT to inform optimal clinical decision‐making.

2. Methods

2.1. Study Design and Patient Selection

This retrospective single‐centre study was conducted at University Hospital Centre Zagreb between January 2016 and June 2025. All patients were diagnosed with PA according to the established guideline criteria [11], and postoperative outcomes were assessed using the PASO criteria [12]. The study population consisted of 251 PA patients who underwent 268 AVS procedures, with 17 patients having repeat AVS. The most common indications for PA screening were arterial hypertension in 43% of patients, hypokalaemia in 35%, and adrenal incidentaloma in 22%; 45% of patients had both arterial hypertension and hypokalemia. Two patients who met the criteria for autonomous cortisol secretion at inclusion (serum cortisol > 138 nmol/L after a 1 mg dexamethasone suppression test) were excluded from the analysis [11]. Figure 1 shows the flowchart of all PA patients who underwent AVS during the study period.

Figure 1.

Flowchart of all patients with PA who underwent AVS.

2.2. AVS Protocol and Interpretation

Before the procedure, patients underwent unenhanced and contrast‐enhanced CT to visualise the adrenal glands and localise the adrenal veins (13). AVS was performed after an overnight fast, between 8 and 9 am. Patients were positioned supine for at least 1 h before sample collection. The AVS procedure involved continuous stimulation with adrenocorticotropic hormone (ACTH) at a rate of 50 µg/h, initiated 30 min before sampling and maintained throughout the procedure [13]. Sequential blood samples were collected first from the right adrenal vein, then from the infrarenal segment of the inferior vena cava (IVC), followed by the left adrenal vein, and again from the infrarenal IVC. AVS was considered technically successful if the selectivity ratio (SR) was at least 5:1, determined by comparing cortisol levels in each adrenal vein to those in the IVC. Unilateral disease was diagnosed if the lateralisation index (LI) was ≥ 4:1, calculated as the aldosterone/cortisol (A/C) ratio between the dominant and non‐dominant adrenal veins. In cases of unilateral successful AVS or LI between 2.5 and 4, alternative indices proposed in our previous study were used for PA subtyping. Ipsilateral adrenalectomy was considered if the A/C ratio was > 3.5 or the AV/IVC ratio was > 3.4, whereas contralateral adrenalectomy was considered if the A/C ratio was ≤ 0.63 or the AV/IVC ratio was ≤ 0.37. Medical treatment with a mineralocorticoid receptor antagonist was recommended if the A/C ratio was 0.64–1.89 or the AV/IVC ratio was 0.38–0.68. If the A/C and AV/IVC ratios fell outside these ranges, repeat AVS or medical treatment was considered [14].

From 2015 to 2022, all AVS procedures were performed by a single interventional radiologist (AMA). During this time, a second interventional radiologist (EK) completed supervised training under the primary operator and subsequently began performing procedures independently. Since 2023, AVS procedures have been routinely conducted by both interventional radiologists.

2.3. Hormone Assessments

Serum aldosterone was measured using an automated chemiluminescence immunoassay (CLIA) on the IDS iSYS (Immunodiagnostic Systems Limited, Boldon, UK). Serum cortisol was measured by chemiluminescent microparticle immunoassay (CMIA) on the Alinity I (Abbott Laboratories, Diagnostics Division, Finisklin Business Park, Sligo, Ireland). Plasma renin activity was measured by radioimmunoassay (RIA) using the Angiotensin I RIA (DiaSource Immunoassays SA, Louvain‐la‐Neuve, Belgium) on the Wallac Wizard 1470 gamma counter (PerkinElmer, Shelton, USA).

2.4. Ethical Considerations

The study received approval from the local ethics committee.

2.5. Statistical Analysis

Statistical analysis was performed using SPSS version 26.0 for Windows. Descriptive statistics for numeric variables are presented as median and interquartile range (IQR), and for nominal variables as frequencies. Non‐parametric statistical tests were applied. Differences between two independent numerical variables were assessed using the Mann–Whitney test, while differences among more than two independent numerical variables were assessed using the Kruskal‐Wallis test. Differences between two or more independent nominal variables were assessed using the chi‐square test. Binary logistic regression (Hosmer–Lemeshow goodness‐of‐fit test) was used to identify independent predictors of unilateral PA; results are presented as odds ratios (ORs) with 95% confidence intervals (CIs). Statistical significance was set at p < 0.05.

3. Results

3.1. Patient Characteristics

Patient demographic and clinical characteristics are summarised in Table 1. The median age was 51 years (IQR: 43–58), 152 patients (61%) were male, and the median BMI was 30 kg/m² (IQR: 27–34). Unilateral disease was present in 118 patients (47%), bilateral disease in 122 patients (49%), while subtype classification was inconclusive in 11 patients (4%).

Table 1.

Patient characteristics.

Variable	All patients (N = 251)	Unilateral PA			Bilateral PA (n = 122)	*p value
		Total (n = 118)	Right (n = 41)	Left (n = 77)
Age at adrenal vein sampling (years)	51 (43–58)	51 (45–56)	49 (46–57)	51(45–56)	53 (42–63)	0.137
Gender (m/f)	152/99	72/46	26/15	46/31	77/45	0.738
BMI (kg/m2)	30 (27–34)	29 (26–33)	29 (26–32)	29 (24–33)	31 (28–34)	0.051
Indication for screening						0.21
1 = hypokalemia	89	48	13	35	39
2 = adrenal incidentaloma	54	27	10	17	26
3 = arterial hypertension	107	42	17	25	57
Potassium (mmol/L)	3.3 (2.9–3.8)	3.1 (2.6–3.5)	2.9 (2.6–3.8)	3.0 (2.3–3.4)	3.6 (3.2–4.0)	< 0.001
Aldosterone to potassium ratio (ng/dL/mmol/L)	8 (6–14)	10 (8–18)	13.5 (8.3–19.3)	9.9 (7–14)	7 (5–10)	< 0.001
Imaging						< 0.001
0 = normal	81	19	7	12	57
1 = right adrenal lesion	40	26	23	3	12
2 = left adrenal lesion	100	58	5	53	38
3 = bilateral lesions or hyperplasia	30	15	6	9	15
Maximal tumor diameter (mm)	15 (11–19)	16 (11–19)	18.5 (13–22.8)	16 (11–20)	14 (10–18)	0.208
Glucose (mmol/L)	5.4 (4.8–6.1)	5.4 (4.9–6.2)	4.9 (4.6–6.5)	5.3 (4.8–5.9)	5.4 (4.8–6.1)	0.972
Diabetes mellitus (yes/no)	52/197	27/90	9/32	18/58	24/97	0.542
Dyslipidemia (yes/no)	63/186	26/91	6/35	20/56	35/86	0.236
Hypertension duration (years)	10 (4–15)	10 (5–15)	10 (3.3–13.8)	10 (7–15)	10 (3–15)	0.948
Antihypertensive DDD (defined daily dose)	4.5 (2.3–6.5)	5.3 (2.7–7.0)	4.2 (1.9–7.5)	5.8 (3–6.8)	4.2 (2.0–6.2)	0.037
Systolic blood pressure (mmHg)	150 (135–160)	150 (140–160)	150 (141–159)	150 (135–165)	150 (135–160)	0.714
Diastolic blood pressure (mmHg)	95 (85–100)	95 (85–100)	100 (95–109)	100 (80–100)	90 (85–100)	0.113
Aldosterone (pmol/L)	758.5 (552.5–1100)	911 (609–1299)	1159 (724–1606)	737 (548–1290)	680 (498–964)	< 0.001
Plasma renin activity (µg/L/h)	0.2 (0.1–0.4)	0.1 (0.1–0.2)	0.1 (0.1–0.2)	0.1 (0.1–0.4)	0.3 (0.1–0.5)	< 0.001
Aldosterone to renin ratio (pmol/L)/(µg/L/h)	4680 (2045–7840)	6300 (3363–10109)	7708 (5793–12325)	5060 (3375–7980)	2717 (1295–5510)	< 0.001
Aldosterone in saline supression test (pmol/L)	483 (370–724)	557 (423–889)	733 (521–1223)	639 (420–896)	393 (315–579)	< 0.001
Cortisol after a 1 mg overnight dexamethasone suppression test (nmol/L)	30 (28–43)	31 (28–48)	30 (26–53)	35 (27–67)	30 (29–41)	0.934

Note: *p value for the difference between patients with unilateral and bilateral primary aldosteronism (PA).

Overall, hypokalaemia (≤ 3.6 mmol/L) was present in 35% of patients and occurred more frequently in those with unilateral disease than in those with bilateral disease (61% vs. 39%; p < 0.001). Compared with patients with bilateral PA, those with unilateral disease had lower potassium concentrations (3.1 vs 3.6 mmol/L; p < 0.001) and a higher aldosterone to potassium ratio (10 vs 7 ng/dL/mmol/L; p < 0.001). Imaging findings suggestive of unilateral adrenal hyperplasia or adenoma were also more common in unilateral disease than in bilateral disease (63% vs 37% of patients; p < 0.001). In addition, patients with unilateral disease required a higher defined daily dose (DDD) of antihypertensive medication (5.3 vs 4.2; p = 0.037), had higher basal aldosterone concentrations (911 vs 680 pmol/L; p < 0.001), lower plasma renin activity (PRA) (0.1 vs 0.3 µg/L/h; p < 0.001), a higher aldosterone to renin ratio (ARR) (6300 vs 2717 pmol/L/µg/L/h; p < 0.001), and higher aldosterone concentrations following the saline infusion test (557 vs 393 pmol/L; p < 0.001).

In multivariate analysis, including the aldosterone to potassium ratio, radiological findings, DDD, PRA, and aldosterone concentration following saline infusion, radiological findings, DDD, and PRA emerged as independent predictors of unilateral PA (p = 0.004, 0.048, and 0.003, respectively).

3.2. AVS Learning Curve and Success Rate

Over a 10‐year period (2016–2025), 251 patients underwent AVS, with a total of 268 procedures performed. Annual AVS success rates were plotted against the cumulative number of procedures to illustrate the learning curve over time (Figure 2). Overall, 211 of 268 procedures (79%) were successful. Unsuccessful adrenal vein cannulation occurred in 47 right veins, seven left veins, and three cases bilaterally. Seventeen patients underwent repeat AVS procedures, which were successful in 82% of cases (14/17).

Figure 2.

AVS success over time.

When the study period was divided into the first 3 years (2016–2018) and the subsequent 7 years (2019–2025), the AVS success rate improved significantly from 70% to 87.5% (χ ² = 9.115, p = 0.003), demonstrating a clear learning curve.

A second interventional radiologist (EK), who joined the AVS programme in 2023, achieved a procedural success rate of 79% (75/95), comparable to that of the primary operator (AMA).

No serious complications were reported. Minor adverse events included one case of iodine allergy and one case of inguinal haematoma (Table 2).

Table 2.

The concordance between CT and AVS in the subtyping of PA.

Variable	N	%
Concordance	147	61
Discordance	7	3
Normal CT/Unilateral disease AVS	19	8
Bilateral adenoma CT/Unilateral disease AVS	15	6
Unilateral adenoma CT/Bilateral disease AVS	52	22

3.3. Diagnostic Concordance Between CT and AVS in the Subtyping of PA

Data from 240 patients were included in the analysis, as AVS results were inconclusive in 11 patients. Overall, concordance between AVS and CT was observed in 147 patients (61%). Discrepancies between CT and AVS included: in 52 patients (22%), CT indicated unilateral disease while AVS revealed bilateral disease; in 15 patients (6%), CT suggested bilateral disease whereas AVS demonstrated unilateral disease; in seven patients (3%), both procedures indicated unilateral disease but on opposite sides; and in 19 patients (8%), CT appeared normal while AVS detected unilateral disease.

There were 76 true positive cases, 59 false positive cases, 71 true negative cases, and 34 false negative cases. The sensitivity and specificity of CT for correctly identifying PA subtypes were 69% and 55%, respectively.

When patients were stratified by age, concordance between AVS and CT in patients aged ≤ 40 years was observed in 28 cases (76%). Discrepancies between CT and AVS were observed in nine patients (24%): in seven patients (19%), CT indicated unilateral disease while AVS revealed bilateral disease; in one patient (3%), CT suggested bilateral disease whereas AVS demonstrated unilateral disease; and in one patient (3%), CT appeared normal while AVS detected unilateral disease.

3.4. Surgical Outcomes According to the PASO Criteria

Of the 118 patients with unilateral PA, 95 underwent adrenalectomy (right‐sided adrenalectomy was performed in 42% of patients); some patients declined surgery, while others are still awaiting the procedure. Postoperative evaluation at 6–12 months was available for 77 patients (81%). Complete clinical remission was achieved in 31 patients (40%), partial clinical remission in 42 patients (55%), and no clinical remission was observed in four patients (5%). Complete biochemical remission was achieved in 62 patients (80%), partial biochemical remission in 13 patients (17%), while two patients (3%) showed no biochemical remission.

Among the four patients without clinical remission, three achieved complete biochemical remission, while one achieved partial biochemical remission. Regarding the two patients without biochemical remission, both achieved partial clinical remission postoperatively.

4. Discussion

AVS is the gold standard for subtyping PA, yet it remains a technically demanding procedure, limited by low standardisation, high cost, and the need for experienced interventional radiologists. Despite these challenges, acceptable success rates can be achieved in experienced centres, where well‐trained interventional radiologists and standardised protocols help optimise procedural outcomes. In our cohort, AVS was successful in 79% of procedures, which is consistent with outcomes reported in other high‐volume centres worldwide. A large multicentre study of 1625 consecutive patients in 19 tertiary centres across Asia, Australia, Europe, and North America documented an 80% AVS success rate [15].

Nonetheless, reported AVS success rates in the literature vary widely. Notably, a nationwide survey in Germany reported a success rate of 40% [16], while Ota et al. documented a rate as high as 99% [17]. Some of this variability may arise from differences in sampling techniques and selectivity thresholds, which complicate comparisons across studies [4]. In line with this procedural heterogeneity our centre applied a SR threshold of ≥ 5 for ACTH stimulated sampling and used a sequential rather than simultaneous sampling strategy. Sequential sampling has generally been associated with lower bilateral success rates [5] and may therefore have influenced procedural outcomes. In contrast, other centres employ different methodological approach. For example, the SPARTACUS trial used a lower SR threshold of ≥ 3 [18]. In addition, a substantial proportion of centres perform AVS without ACTH stimulation [19]. Taken together, these inconsistencies highlight the need for greater standardisation of AVS protocols to improve diagnostic consistency and support clinical decision making.

Beyond differences in procedural protocols, operator expertise represents a key determinant of AVS success. Previous reports suggest that procedural proficiency is achieved after approximately 30 cases, with sustained competence requiring at least 15 procedures per year [9]. In our centre, annual AVS volumes exceeded 20 cases following the initial 3‐year period, peaking at 42 procedures in 2022. When the study period was stratified accordingly, a marked improvement in procedural success rates was observed. These findings highlight the importance of cumulative operator experience and procedural volume in optimising AVS outcomes. However, in contrast to the primary operator, the second radiologist achieved high procedural success rates from the outset of independent practice, likely reflecting the benefit of prolonged supervised training and prior experience gained alongside the primary operator. This suggests that structured mentorship and cumulative institutional expertise may attenuate the individual learning curve associated with AVS.

Despite its widespread availability and routine use in the diagnostic work‐up of PA, CT remains an unreliable surrogate for AVS. In our cohort, concordance between CT and AVS was only 61%, closely mirroring the 60% concordance reported in a large meta‐analysis of 38 studies [20]. Moreover, the sensitivity and specificity of CT‐based subtyping (69% and 55%, respectively) in our study indicate substantial rates of both false‐positive and false‐negative classifications. From a clinical perspective, reliance on CT alone would have had significant consequences: one in four patients would have been exposed to unnecessary or inappropriate adrenalectomy, while 14% of patients with unilateral disease confirmed by AVS would have been incorrectly classified as non‐surgical candidates and thus denied potentially curative adrenalectomy. Additionally, our results showed concordance between CT and AVS in only 76% of patients with PA aged ≤ 40 years. From a clinical perspective, relying on CT alone would have exposed 19% of young patients with PA to unnecessary adrenalectomy. These findings are consistent with previous reports demonstrating the limited ability of imaging to predict functional lateralisation [21, 22, 23, 24]. Notably, the SPARTACUS trial reported similar short‐term clinical outcomes between CT‐guided and AVS‐guided management strategies [18]. However, a growing body of evidence supports the superiority of AVS. A systematic review of 11 studies, including 1325 AVS‐guided and 907 CT‐guided patients, demonstrated significantly higher rates of complete biochemical success following AVS‐guided adrenalectomy [25]. In addition, the AVIS‐2 study highlighted the limitation of imaging showing that 34% of patients with PA had normal adrenal imaging, while 7% exhibited bilateral nodules [15]. Despite this evidence, adrenalectomy is still sometimes performed without clear biochemical confirmation of unilateral aldosterone excess, potentially exposing up to one third of patients to unnecessary surgery [26].

Although AVS generally achieves high success rates, technical challenges occasionally require repeat procedures when initial cannulation is incomplete. Decisions about repeat AVS remain difficult, as formal guidelines for a second attempt are lacking. Available evidence indicates that a substantial proportion of repeat procedures result in successful cannulation [27]. In our cohort, repeat AVS was required in 7% of cases; the first attempt achieved a technical success rate of 79%, while repeat procedures were successful in 82% of patients. These outcomes compare favourably with previous reports indicating approximately 75% success after an initial failure [27]. Collectively, these findings support the use of repeat AVS when clinically indicated.

The primary limitation of this study is its retrospective design. However, several important strengths should be noted. Our analysis offers a detailed 10‐year evaluation of AVS performance, capturing procedural success rates, the learning curve, and concordance with CT imaging. All procedures were performed in a high‐volume tertiary referral centre by two interventional radiologists with extensive experience. A standardised AVS protocol was applied consistently throughout the study period, ensuring uniform procedural conduct and enhancing the comparability of our findings with previous studies. These factors strengthen the validity of our observations and provide a real‐world perspective on AVS implementation in experienced centres.

In conclusion, our findings reinforce the central role of AVS as the gold standard for subtyping PA, substantially outperforming CT, which often misclassifies lateralisation and may lead to inappropriate management decisions. The high procedural success in our cohort highlights the importance of radiologist expertise and experience, supporting the centralisation of AVS to specialised, high‐volume centres. When performed with standardised protocols by skilled interventional radiologists, AVS is a reliable, reproducible, and indispensable tool for guiding optimal management in patients with PA.

Data Availability Statement

The data that support the findings of this study are available from the corresponding author upon reasonable request.