Development of a multivariate model predictive of post-adrenalectomy renal function decline in patients with primary aldosteronism: a large-cohort single-center study.

Abstract

Purpose: To develop a multivariate liniear model for predicting long-term (> 3 months) post-adrenalectomy renal function decline in patients with primary aldosteronism (PA). The model aims to help identify patients who may experience a significant decline in renal function after surgery.

Methods: We retrospectively analyzed the clinical data of 357 patients who were diagnosed with PA and underwent adrenalectomy between September 2012 and February 2023. LASSO and multivariate linear regression analyses were used to identify significant risk factors for model construction. The models were further internally validated using bootstrap method.

Results: Age (P < 0.001), plasma aldosterone concentration (PAC) measured in the upright-position (PAC_U, P = 0.066), PAC measured after saline infusion (PAC_afterNS, P = 0.010), preoperative blood adrenocorticotropic-hormone level (ACTH, P = 0.048), preoperative estimated glomerular filtration rate (eGFR, P < 0.001) and immediate postoperative eGFR (P < 0.001) were finally included in a multivariate model predictive of post-adrenalectomy renal function decline and the coefficients were adjusted by internal validation. The final model is: predicted postoperative long-term (> 3 months) eGFR decline =-70.010 + 0.416*age + 6.343*lg PAC_U+4.802*lg ACTH + 7.424*lg PAC_afterNS+0.637*preoperative eGFR-0.438*immediate postoperative eGFR. The predicted values are highly related to the observed values (adjusted R = 0.63).

Conclusion: The linear model incorporating perioperative clinical variables can accurately predict long-term (> 3 months) post-adrenalectomy renal function decline.