Differentiating pituitary metastasis from hypophysitis: A two-center study using conventional MRI and clinical features.

OBJECTIVE

Pituitary metastasis (PM) and hypophysitis frequently present with overlapping clinical and radiologic features, yet their management strategies differ substantially. This study aimed to identify clinically applicable predictors using conventional MRI and basic clinical variables.

METHODS

In this two-center retrospective cohort of cases collected between January 2016 and January 2026, 64 patients (34 PM, 30 hypophysitis) were included. Diagnosis was based on histopathology or response on follow-up imaging. Two blinded neuroradiologists independently evaluated 13 conventional MRI features and maximum lesion size. Clinical variables, including age, gender, history of cancer, and presence of diabetes insipidus, were documented. Univariable logistic regression with Holm-Bonferroni correction and subsequent multivariable modeling were performed. Model performance was assessed using ROC analysis. Interobserver agreement was assessed using Cohen's κ or ICC.

RESULTS

Patients with PM were older than those with hypophysitis (mean, 56.9 vs 42.9 years; p = 0.0004), while gender distribution did not differ between groups. Age, degree of enhancement, and history of cancer were significant predictors in univariable analysis. In the multivariable model, marked enhancement emerged as the only independent discriminator of hypophysitis (p = 0.003), while age and history of cancer contributed to improved overall model performance without reaching statistical independence. The final model integrating age, degree of enhancement, and cancer history yielded an AUC, accuracy, sensitivity, and specificity of 0.843, 76.6%, 70.0%, and 82.4%, respectively.

CONCLUSIONS

Marked enhancement, younger age, and absence of cancer history constitute a practical triad for differentiating hypophysitis from PM, offering a clinically useful decision-support tool when biopsy is not feasible.