Abstract
Precise anatomic localization of insulinomas is crucial for surgical treatment. Current routine noninvasive imaging techniques, including CT, MRI, and (68)Ga-DOTA-somatostatin analog (DOTA-SSA) PET/CT, have limited sensitivity. Endoscopic ultrasound is highly sensitive but invasive. In this prospective multicenter study, we compared the diagnostic accuracy of (68)Ga-NODAGA-exendin-4 (exendin) PET/CT with all routine imaging procedures for the localization of insulinomas. Methods: Sixty-nine adults with biochemically proven adult endogenous hyperinsulinemic hypoglycemia underwent exendin PET/CT and current routine imaging. Images were evaluated in a clinical reading and in an expert reading. Image quality was determined by quantitative analysis. Results: Based on clinical readings, the accuracy of exendin PET/CT (94.4%; 95% CI, 84.6%-98.8%) was greater than that of DOTA-SSA PET/CT (64.8%; 95% CI, 50.6%-77.3%), contrast-enhanced CT/contrast-enhanced diffusion-weighted imaging-MRI (83.3%; 95% CI, 70.7%-92.1%), and endoscopic ultrasound (82.8%; 95% CI, 64.1%-94.1%). In 13% of patients, a correct diagnosis was only reached after exendin PET/CT. Interobserver agreement between readings was higher for exendin PET/CT than for DOTA-SSA PET/CT and contrast-enhanced CT/contrast-enhanced diffusion-weighted imaging-MRI (Cohen κ, 1.0 vs. 0.5 and 0.55). Exendin PET/CT provided a higher insulinoma-to-background ratio (15.3 ± 6.7 vs. 5.2 ± 3.0) and contrast-to-noise ratio (22.6 ± 11.1 vs. 5.1 ± 3.7) than did DOTA-SSA PET/CT. Conclusion: This study demonstrates the superiority of exendin PET/CT in a unique prospective comparison to all current routine imaging modalities for preoperative localization of benign insulinomas, providing the level of evidence needed for clinical implementation.