Homeostasis Model Assessment of β-Cell Function for Diagnosis of Insulinoma

Abstract

Abstract Context Diagnosis of insulinoma is based on different criteria from the 72-hour fasting test according to current guidelines (Endocrine Society [ES], European [ENETS], and North American [NANETS] Neuroendocrine Tumor Societies), including assessment of β-cell function by glucagon stimulation test. Objective This study tested whether the homeostasis model assessment of insulin secretion, including assessment of β-cell function, (HOMA-B) at the end of the fasting test provides comparable efficacy for insulinoma diagnosis. Methods In 104 patients with suspected insulinoma, 72-hour fasting tests were performed with frequent assessment of glucose, insulin, and C-peptide in venous blood. HOMA-B values using insulin and C-peptide were calculated at the end of the fasting test, as defined by the lowest glucose concentration from each participant. Results HOMA-B was more than 6.5-fold higher in patients with (n = 23) than in those without (n = 81) insulinoma (insulin and C-peptide; both P < .001). HOMA-B (cutoff using insulin >253 a.u. and C-peptide >270 a.u.) had a sensitivity of 0.96, 0.78 to 1.00, and a specificity of 0.96 or greater (≥0.89-0.99) for insulinoma diagnosis. ES and ENETS/NANETS criteria reached a diagnostic sensitivity of less than or equal to 0.96 (≤0.78-1.00) and ≤0.83 (≤0.61-0.95) as well as specificity of ≤0.85 (≤0.76-0.92) and less than or equal to 1.00 (≤0.96-1.00) for insulin, and C-peptide, respectively. Using insulin for HOMA-B, sensitivity tended to be higher compared to ENETS/NANETS criteria (P = .063) and specificity was higher compared to ES criteria using insulin and C-peptide (both P < .005). Conclusion HOMA-B, as calculated at the end of the fasting test employing defined cutoffs for insulin and C-peptide, provides excellent diagnostic efficacy, suggesting that it might represent an alternative and precise tool to diagnose insulinoma.