Evaluation of a euglycaemic clamp procedure as a diagnostic test in insulinoma patients

Abstract

In 15 patients with insulinoma, six patients after successful removal of this tumour, two patients with previous pancreas resection because of hypoglycaemia elsewhere, and 10 control subjects, the diagnostic usefulness of euglycaemic clamp procedures (without exogenous insulin) was assessed in comparison with prolonged starvation. Only insulinoma patients developed sustained hypoglycaemia (less than or equal to 2.3 mmol l-1) within 2-44 h without caloric intake, because of inappropriately elevated immunoreactive insulin (IR-insulin) concentrations. IR-proinsulin values were elevated in most (7 out of 10), but not in all insulinoma patients. The steady-state glucose infusion rate necessary to maintain a stable plasma glucose concentration of 4.4-5.0 mmol l-1 was significantly (P less than or equal to 0.001) higher in insulinoma patients (2.5 +/- 0.6 mg kg-1 min-1) than in pancreas resected patients (0.6 +/- 0.2 mg kg-1 min-1), or in control subjects (0.5 +/- 0.1 mg kg-1 min-1). Due to a considerable degree of overlap, sensitivity (0.44) and specificity (0.95) were too low for such a procedure to qualify as a diagnostic test. There was no correlation of glucose infusion rates to IR-insulin values (r = 0.024, P = 0.461). One reason for this was the development of insulin resistance in some, but not in all insulinoma patients. When, in analogy to insulin/glucose ratios, a diagnostic index was derived by multiplying the steady state glucose infusion rate by the steady state IR-insulin concentration, the diagnostic accuracy was greatly increased (sensitivity and specificity 0.94, respectively), but still lower than that of 'amended' insulin/glucose ratios in fasting plasma or at the time of discontinuation of prolonged fasts (1.00). Somatostatin infusions inhibited insulin secretion (IR-C-peptide plasma concentrations) by 52-88% in subjects without insulinoma and in those insulinoma patients whose tumour cells ultrastructurally contained plenty of normal secretory granules, and to a lesser degree when only abnormal or virtually no secretory granules were present, i.e. in more de-differentiated tumours. In contrast to this significant (P = 0.036) association, malignancy, i.e. the presence of metastases, could not be predicted from whether or not insulin secretion was resistant to the inhibitory action of somatostatin. In conclusion, euglycaemic clamp experiments are less reliable for detecting or excluding a functioning insulinoma than the relation of glucose and insulin values during starvation. The inhibition of insulin secretion by somatostatin depends on the presence of normal beta-granules, and does not distinguish adenomas from carcinomas.