Results of a prospective strategy to diagnose, localize, and resect insulinomas.

Since 1982, 25 consecutive patients with benign sporadic (non-multiple endocrine neoplasia type I) insulinomas have been studied. Most were referred because either the tumor was not identified at the referring institution or the diagnosis was unclear. Each patient suffered severe neuroglycopenic symptoms for a median of 24 months before diagnosis of insulinoma, and 32% had hypoglycemic seizures. Eighteen patients (72%) had a confirmed weight gain. Each patient underwent a supervised fast until 72 hours or the onset of significant neuroglycopenic symptoms (median duration 16 hours), with serum levels of glucose (median 35 mg/dl; range 24 to 46 mg/dl), insulin (median 21 microU/ml; range 11 to 230 microU/ml), C-peptide (median 2.5 ng/ml; range 1.0 to 7.2 ng/ml), and proinsulin fraction (median 55%; range 14% to 86%) measured at the termination of the fast. Preoperative imaging with ultrasonography, computed tomography, magnetic resonance, and angiography visualized tumor in a minority of patients (26%, 17%, 25%, and 35%, respectively); in 48% of patients one or more imaging study results was positive. Selective portal venous sampling for insulin was the most informative localizing test (77% positive; no false-positive results). Tumor was resected for cure in 24 of 25 patients. Intraoperative ultrasonography identified nonpalpable tumor in seven patients and was crucial to the achievement of this high rate of surgical cure. We conclude that the diagnosis of insulinoma can be made by the results of a supervised fast, portal venous sampling is the most sensitive preoperative test for localizing insulinomas, and intraoperative ultrasonography is essential for intraoperative detection of insulinomas.