Development and validation of model for sparing adrenal venous sampling in diagnosing unilateral primary aldosteronism

Context: Current guidelines recommend adrenal venous sampling (AVS) to identify unilateral primary aldosteronism (UPA) before offering adrenalectomy. However, AVS is costly and technically challenging, limiting its use to expert centres. Objective: To establish a model to predict UPA, and therefore, bypass the need for AVS prior to surgery. Design and setting: The model was developed in a Chinese cohort and validated in an Australian cohort. Previously published prediction models of UPA were also tested. Participants: primary aldosteronism patients with a definite subtyping diagnosis based on AVS and/or surgery. Main outcome measure: Diagnostic value of the model. Results: In the development cohort (268 UPA and 88 bilateral primary aldosteronism), combinations of different levels of low serum potassium (<= 3.0 or 3.5 mmol/l), high PAC (>= 15-30 ng/dl), low PRC (<= 2.5-10 mu IU/ml) and presence of unilateral nodule on adrenal CT (>8-15mm in diameter) showed specificity of 1.00 and sensitivity of 0.16-0.52. The model of serum potassium 3.5 mmol/l or less, PAC at least 20 ng/dl, PRC 5 mu IU/ml or less plus a unilateral nodule at least 10mm had the highest sensitivity of 0.52 (0.45-0.58) and specificity of 1.00 (0.96-1.00). In the validation cohort (84 UPA and 117 bilateral primary aldosteronism), the sensitivity and specificity of the model were 0.13 (0.07-0.22) and 1.00 (0.97-1.00), respectively. Ten previous models were tested, and only one had a specificity of 1.00 in our cohorts but with a very low sensitivity [0.07 (0.04-0.10) and 0.01 (0.00-0.06) in our development and validation cohorts, respectively]. Conclusion: A combination of high PAC, low PRC, low serum potassium and unilateral adrenal nodule could accurately determine primary aldosteronism subtype in 13-52% of patients with UPA and obviate the need for AVS before surgery.