Background: Recent biochemical diagnostic guidelines for insulinomas require demonstration of hypoglycemia with inappropriately elevated (nonsuppressed) insulin, C-peptide, or proinsulin, but these criteria may overlap with those in patients without insulinomas. Use of an "amended" insulin-glucose ratio that accounts for the normal variation in insulin secretion according to prevailing glycemia may improve diagnostic accuracy.
Objective: To compare the diagnostic accuracy of current diagnostic guideline criteria with the amended insulin-glucose ratio in patients with a suspected insulinoma.
Design: Retrospective cohort study.
Setting: 2 specialized university departments in Germany.
Patients: 114 patients with suspected hypoglycemia over 10 years having diagnostic prolonged fasts.
Measurements: Glucose, insulin, C-peptide, and the amended insulin-glucose ratio were measured during and at discontinuation of prolonged fasts.
Results: Of 114 patients who were evaluated, 49 had surgical resection of histologically confirmed insulinomas. Insulinoma was excluded in 65 patients; follow-up for a mean of 10 years (range, 0 to 16 years) showed no progressively severe hypoglycemic events or diagnoses of insulinoma. Patients with insulinoma had lower glucose levels and higher insulin and C-peptide levels overall than did control patients at the end of prolonged fasts, but there was considerable overlap. The amended insulin-glucose ratio correctly identified 48 of 49 patients with insulinoma and excluded the diagnosis in 64 of 65 control patients, resulting in positive and negative predictive values of 0.98 (95% CI, 0.89 to 1.00) and 0.99 (CI, 0.92 to 1.00), respectively, compared with 0.75 (CI, 0.63 to 0.85) and 0.98 (CI, 0.89 to 1.00), respectively, for glucose, insulin, and C-peptide concentration criteria.
Limitation: The study had a retrospective design, no proinsulin concentrations were available, and a nonspecific insulin immunoassay (crossreactive with proinsulin) was used.
Conclusion: The amended insulin-glucose ratio showed improved diagnostic accuracy over established criteria that use glucose, insulin, and C-peptide concentrations.
Primary funding source: None.