Aim in a prospective study endoscopic ultrasonic localisation of clini
cally suspected insulinomas was compared with the findings of abdomina
l ultrasound, computed tomography and magnetic resonance imaging. Meth
od From Dec. 1990 to Jan. 1995 11 patients (8 f, 3 m, median age 42 [2
7-79] years) were enrolled in the study. The preoperative endosonograp
hic findings were compared to surgery (n=7) or clinical follow-up (n=4
). Results Endoscopic ultrasound identified a solitary tumour (mean si
ze 12.4 mm) in 7 of 11 patients, which was proven at surgery in 5 pati
ents and by transhepatic portal venous sampling after negative partial
pancreas resection in 1 patient. One patient is still awaiting surger
y. Clinical follow-up (n=2) and negative intraoperative and histologic
al findings of partial pancreas resection (n=1) confirmed a true negat
ive examination in 3 patients. One patient with negative endoscopic ul
trasound is still under medication for recurrent hypoglyctemia. Abdomi
nal ultrasound(n=11), computed tomography (n=11) and magnetic resonanc
e imaging (n=5) were negative in all investigated patients. Conclusion
Endoscopic ultrasound is highly accurate for localisation of insulino
mas and should be performed early in the preoperative management of th
ese patients.