Adrenal imaging features and associated pathologic diagnoses: A contemporary, longitudinal analysis

Imaging features are key in preoperative risk assessment of adrenal nodules, but current recommendations stem from low quality evidence. This study evaluates the performance of cross-sectional imaging features in predicting adrenal pathology.

All adult adrenalectomy cases (2015-2023) with preoperative cross-sectional imaging were retrospectively reviewed; pheochromocytomas and incidental adrenalectomies were excluded. Benign imaging features were defined as size ≤4.0 cm; Hounsfield units ≤10, or Hounsfield units ≤20 on unenhanced computed tomography; absolute and relative contrast washout ≥60% and ≥40%, respectively, on adrenal protocol computed tomography; and loss of signal or homogeneity on T1-weighted magnetic resonance imaging. Performance of imaging features in identifying benign and malignant adrenal pathology were analyzed.

Among the 250 adrenal specimens included, 21% were malignant, with 22 adrenocortical carcinomas and 30 other malignant lesions, whereas 79% were benign. Compared to benign, malignant lesions were larger (6.8 cm vs 4.4 cm, P = .002) and none had densities of ≤20 Hounsfield units on unenhanced computed tomography scan or loss of signal on magnetic resonance imaging. Malignancy rate for lesions >20 Hounsfield units was 31%. Notably, 13.6% (3/22) of adrenocortical carcinomas and 63.3% (19/30) of other malignant lesions were <4.0 cm. Accuracy of imaging features in correctly identifying pathology was best (65%) using a density threshold of ≤20 Hounsfield units, and 48% of pathologically benign adrenal lesions were accurately identified as being benign on imaging.

In this large, contemporary study, using adrenal lesion size >4.0 cm alone has low specificity (57%) in identifying malignant adrenal pathology and absolute contrast washout on adrenal protocol computed tomography can miss metastatic lesions. However, a density threshold of ≤20 Hounsfield units on unenhanced computed tomography reliably excludes malignancies, potentially avoiding adrenalectomy in 48% of patients.