We conducted a retrospective cohort study of adult patients with PA who underwent AVS between May 2014 and July 2022 at the Leiden University Medical Center (LUMC), a Dutch tertiary referral center for adrenal disease. Patients with confirmed PA and successful bilateral AVS were included.

Patients were excluded if they had other adrenal diseases like Cushing’s disease or if there were major deviations in the pre-protocol work-up, potentially leading to unreliable AVS outcomes. We quantified the total amount of antihypertensive drugs, including MRA, as the daily defined dose (DDD), ATC/DDD WHO Index 2023. All patients underwent computed tomography (CT) or magnetic resonance imaging (MRI) of the adrenal glands prior to sampling. This retrospective study was approved by the scientific board of the LUMC, code W2020.058, and patients were given the opportunity to object to the use of their coded clinical data.

The diagnostic work-up for PA starts with the aldosterone-to-renin ratio (ARR) (12). The ARR was calculated as the plasma aldosterone concentration (PAC) in pmol/L divided by the plasma renin concentration (PRC) in mU/L. An ARR >100 pmol/mU along with spontaneous hypokalemia confirmed the diagnosis PA (19). Before May 2015, PAC was measured in nmol/L and the plasma renin activity (PRA) in µg/L/hour; in this scenario an ARR >0.85 nmol/µg/hour and hypokalemia confirmed the diagnosis. If patients did not meet these criteria with an ARR >30 pmol/mU, an additional salt loading test (SLT) was performed, during which two liters of 0.9% sodium chloride (NaCl) were administered in a seated position (20). Plasma aldosterone measurements were taken before and immediately after saline infusion. A consistently elevated PAC >179 pmol/L after SLT confirmed PA (21). During both the ARR and SLT, antihypertensive medication known to interfere with renin and/or aldosterone was substituted with non-interfering alternatives ( Appendix S1 ) (12). Potassium was supplemented to maintain normal range (3.5-5.0 mmol/L).

Serum aldosterone and renin were determined using chemiluminescence technology (ImmunoDiagnostic Systems GMBH, Germany), while cortisol levels were measured through electro-chemiluminescence immunoassay (Elecsys Cortisol gen2 ECLIA Roche Diagnostic, Germany). The analytical variation was 4.0%-6.3% for aldosterone, 3.4%-4.3% for renin and 2.5%-4.1% for cortisol. Before May 2015, a DiaSorin Plasma Renin activity RIA (Gammacoat Plasma Renin Activity RIA, CA1553, DiaSorin, Italy) was utilized. The ARR was clinically validated for both aldosterone and renin assays, establishing a cut-off of 31 pmol/mU [equivalent to 1.12 (ng/dl)/(µU/ml)], with a sensitivity of 99% and a specificity of 79% (20).

AVS was performed under continuous intravenous stimulation of synthetic adrenocorticotropic hormone (ACTH) Synacthen^® at 50 µg/hour. This ACTH infusion enhances the specificity of AVS by minimizing fluctuations in cortisol levels, assuming symmetrical cortisol secretion, as cortisol functions as an internal control (22). Autonomous cortisol secretion was ruled out with a dexamethasone suppression test in case of clinical suspicion for Cushing’s syndrome or in case of incidental detected adenomas. The right common femoral vein was punctured and a 5F sheath was inserted. Selective catheterization of the left and right adrenal vein was performed under fluoroscopic guidance. At least two blood samples were obtained from each adrenal vein, along with two peripheral samples from the sheath with the tip in the inferior vena cava. Aldosterone and cortisol levels were measured in these blood samples, and the A/C ratios of the LAV, RAV and IVC were compared to determine lateralization and/or suppression. The selectivity index (SI), defined as the cortisol ratio between the adrenal veins and the IVC, was used to assess for sampling adequacy. A SI index of 3-fold greater in both adrenal veins indicated successful bilateral sampling ( Supplementary Table S1 ) (23, 24).

The A/C ratio was determined by calculating the average of the two samples from each site. To distinguish patients with unilateral disease, the study utilized the lateralization index (LI), (A/C ratio of the dominant vein)/(A/C ratio of the non-dominant vein), and contralateral suppression index (CSI), (A/C ratio non-dominant vein)/(A/C ratio IVC) (25). In our center, a LI ≥ 4 was considered indicative of unilateral disease. However, adrenalectomy was offered from LI ≥3 onwards, given the high likelihood of biochemical cure and clinical improvement. Additionally, a CSI <1 was considered consistent with unilateral disease ( Supplementary Table S1 ). Patients failing to meet the criteria for both LI and CSI were categorized as having bilateral disease. The LAV/IVC index was defined as the A/C ratio between the left adrenal vein and the inferior vena cava.

Post-operative cure was assessed within the first-year post-adrenalectomy. The definition of biochemical cure, according to the Post-Adrenalectomy Surgical Outcomes (PASO) criteria, was used (5). Biochemical cure was defined as the correction of hypokalemia (if present pre-surgery) and normalization of the ARR post-operatively; when ARR was not normalized, the salt loading test was repeated. In addition, we assessed clinical improvement, defined as improved control of hypertension, a reduction in antihypertensive medication use, and symptom resolution.

A search strategy ( Appendix S2 ) was developed using different variations of the keywords ‘primary aldosteronism, ‘adrenal vein sampling’ and ‘subtyping’. The PubMed database was explored based on titles; 22 abstracts were screened. Full texts of the eligible studies were evaluated, and a total of 8 studies were included for the literature overview, focusing on unsuccessful sampling of the RAV and using the LAV/IVC index to predict lateralization.

Baseline characteristics were reported as mean ± standard deviation (SD) or median and interquartile range (IQR) if not normally distributed. Categorical variables were expressed as absolute numbers and percentages. Differences between unilateral and bilateral PA patients were tested using the independent T-test and Mann-Whitney U-test. Additionally, the Kruskall Wallis test was used to compare multiple groups for numerical values, and the Chi-squared test for categorical values (26). Differences between pre- and post-adrenalectomy outcomes in the unsuccessfully sampled group were tested using the Wilcoxon signed-rank test and McNemar test for categorical data. Receiver operating characteristics (ROC) analysis was used to calculate cut-off values for both the LAV/IVC index and A/C ratio, aiming to predict the disease subtype. The predictive value of these indices was measured by the area under the curve (AUC). The ratio with the highest AUC, representing the highest predictive value, was selected for further analyses. The Youden index was used to select the optimal cut-off values, ensuring a specificity over 90%, to limit false positive errors and their clinical consequences associated with misclassifying patients (27). Other optimal cut-off values with different desired specificities (>85% and >95%) were identified and presented in the supplementary. A post-hoc power calculation was conducted. Assuming an α of 0.05 and a power of 0.80, calculations showed that a sample size of at least 26 patients in each group would be necessary to detect a difference between a 90% cure rate in the intervention group and a 50% in the reference group (assuming that half of the patients has unilateral disease). A p-value of ≤0.05 was considered statistically significant. All statistical analyses were performed using IBM SPSS Statistics version 29.0.

Between 2014 and 2022, 92 patients with PA were identified, of whom 82 underwent AVS. Among these, 54 (66%) samplings were bilaterally successful and included in the study. Of these, 28 patients had bilateral disease, while 26 had unilateral disease. The unilateral disease group exhibited a more severe phenotype of PA with a higher incidence of hypokalemia (85% vs. 32%, p<0.001) and a trend towards a higher ARR ratio than those in the bilateral group (353 vs. 215, p=0.21). Antihypertensive medication use was higher in the unilateral disease group (4.6 vs. 3.6). Both groups showed a substantial prevalence of cardiovascular comorbidities (e.g. chronic kidney disease - defined as reduced eGFR or presence of albuminuria - and left ventricular hypertrophy (LVH) - defined as meeting ECG or ultrasonographic criteria for LVH) at baseline ( Table 1A ).

Adrenal imaging revealed abnormalities in 58% of the unilateral disease group and 32% of the bilateral disease group (p = 0.01). In patients with unilateral disease, imaging showed both ipsilateral and contralateral abnormalities, including adenomas and hyperplastic adrenal glands. In the bilateral disease group, no abnormalities in both adrenal glands were seen, but unilateral abnormalities were observed in 10 patients on either the left or right side. Full details are provided in Supplementary Table S2 .

Adrenalectomy was performed in 22 patients with unilateral disease, surgery was recommended for an additional 3 patients but not (yet) performed. Post-adrenalectomy, 93% of the unilateral left group and 100% of the unilateral right group achieved biochemical cure of PA ( Table 1B ). Symptom improvement, particularly better cognitive functioning such as improved concentration, was observed in 53% of the patients. Antihypertensive medication use could be reduced in nearly all patients (95%) and fully eliminated in 13% ( Table 1B ). Follow-up data was not available for three patients: one was lost to follow-up, and two patients no longer had hypertension, antihypertensive medication use, or symptoms, but their clinicians did not verify biochemical cure ( Supplementary Figure S1 ). Regarding histology, the pathology of the adrenal glands revealed adenoma in 14 patients and hyperplasia was found in 14 patients. The observed pathological variants of aldosterone-producing adenomas in our cohort were KCNJ5 (N=5), ATP1A1 (N=5), and CACNA1D (N=4) mutations.

The A/C ratio differed between groups (p<0.001): unilateral right-sided (median: 0.9, IQR [0.4-2.0]), bilateral (median: 3.8, IQR [2.8-9.3]) and left-sided disease (median: 10.6, IQR [5.5-15.3]) ( Table 1B ). Similar differences were observed in the LAV/IVC index (p<0.001): unilateral right-sided (median: 0.6, IQR [0.1-1.0]), bilateral (median: 2.1, IQR [1.7-3.2]) and unilateral left-sided disease (median: 4.8, IQR [3.2-5.9]). ROC analysis evaluated the accuracy of the A/C ratio and LAV/IVC index in predicting disease subtype ( Figure 1 ). Both ratios showed good predictive ability; however, the A/C ratio had a lower area under the curve. Therefore, this study focused on the LAV/IVC values. For cut-off values off the A/C ratio, see Appendix S3 .

Using the LAV/IVC index, a cut-off value of ≥4.4 predicted unilateral left-sided disease with a sensitivity of 60% and specificity of 93% (AUC 0.85, 95% confidence interval (CI) 0.73–0.96), Figure 2A . In other words, when roughly half of the patients have unilateral disease, the LAV/IVC index of ≥4.4 has a positive predictive value of 93% for left-sided disease. However, this cut-off value missed 40% of patients with left-sided disease. Additional cut-off values derived from the same ROC-curve of >4.1 and >5.9 showed a sensitivity of 60% and 27%, with a specificity of 89% and 96%, respectively ( Figure 2A ).

The optimal cut-off value of the LAV/IVC index for the left adrenal vein was found at <1.2 for predicting right-sided disease with a sensitivity of 96% and specificity of 91% (AUC 0.98, CI 0.95-1.00), Figure 2C . Other cut-off values of <1.1 and <1.5 derived from the same ROC-curve, resulted in a sensitivity of 91% and 100% with a specificity of 100% and 82%, respectively.

To predict bilateral disease, a cut-off of value <2.4 was found to have a sensitivity of 64% and a specificity of 93% (AUC 0.85, CI 0.73–0.96). Alternative cut-off values of ≤1.8 and <2.5 derived from the same ROC-curve provided a sensitivity of 39% and 64% with a specificity of 100% and 87%, respectively ( Figure 2B ).

Combining the calculated (optimal) cut-off values, we developed a clinical decision model ( Figure 3 ). Additional cut-off values for different specificities (>85% and >95%) for the clinical decision model are provided in Supplementary Figure S2 . Validating this model on our own study cohort of bilaterally successful sampled patients (N=54), we found that the disease subtype could have been predicted for 74% of the patients. Of whom, 20% were diagnosed with right-sided disease, 35% with bilateral disease, and 19% with left-sided disease. To further validate the tool, we extended the analysis with patients who had unsuccessfully right-sided sampling followed by an adrenalectomy ( Table 2 ). During the study period, right-sided sampling failed in 24 patients, of whom 17 were predicted to have unilateral disease by our clinical model. Among those who opted for surgery according to our model’s recommendation, 14/14 (100%) achieved biochemical cure post-adrenalectomy. In this group (male 79%, age 49 [43-60] years, BMI 30.1 ± 6.1 kg/m²), differences were observed postoperatively in both the number of antihypertensive medications (in DDD) and the prevalence of uncontrolled hypertension. Additionally, 36% reported a reduction in symptoms, particularly cognitive improvements ( Table 2 ).

Various studies have investigated the use of the LAV/IVC index to predict disease subtype. Using the previously described search strategy, 8 articles were identified comparing the LAV/IVC index (28–35). Their cut-off values, including sensitivity, specificity, LI, and SI together with our own data is presented in Table 3 .