Agreement Between Molecular Imaging and Adrenal Vein Sampling for Localization in Primary Aldosteronism: A Systematic Review

BACKGROUND: Primary aldosteronism is a common cause of hypertension that is curable via surgery when localized to a single adrenal gland. Localization testing has traditionally relied upon adrenal vein sampling (AVS); however, molecular imaging has emerged as a promising noninvasive alternative. METHODS: We performed a systematic synthesis of studies comparing molecular imaging with AVS for localization in primary aldosteronism. Studies directly comparing both modalities with prespecified AVS localization criteria were eligible. Agreement was evaluated using positive percent agreement, negative percent agreement, and, where available, Cohen's κ. Because molecular imaging tracers differ in biological targets and interpretation criteria, agreement metrics were examined by tracer rather than pooled. Sensitivity analyses repeated positive percent agreement and negative percent agreement calculations using molecular imaging as the reference. RESULTS: Twenty studies were included. Across tracers, agreement between molecular imaging and AVS was moderate, with positive percent agreement values ranging from 0.6 to 0.9 and negative percent agreement from 0.5 to 0.9. Cohen's κ values ranged from 0.33 to 0.93, indicating moderate-to-strong agreement beyond chance. Each tracer was assessed separately; direct comparisons between tracers were not performed, and quantitative pooling was not undertaken because of heterogeneity in tracer protocols, patient preparation, and test interpretation. Reciprocal analyses yielded similar positive percent agreement and negative percent agreement distributions, with no consistent directional bias between modalities. CONCLUSIONS: Molecular imaging shows moderate and variable agreement with AVS for localization testing in primary aldosteronism, suggesting complementary rather than competing roles. Molecular imaging may serve as an adjunct or alternative within diagnostic algorithms; however, patient outcome-based studies are still needed to define its clinical utility.