Functioning human insulinomas

Sixty-seven insulinomas were investigated by immunohistochemistry using site-directed antibodies against insulin, proinsulin, chromogranin A, HISL-19, and four proteins directly or indirectly involved in the proteolytic processing of proinsulin: the prohormone convertases PC2 and PC3, carboxypeptidase H (CPH) and 7B2. Results were expressed in a six-grade score according to the frequency of immunoreactive tumour cells. Insulin was expressed by all tumours, appearing in either a diffuse or a polarized pattern and being detected in more than 30% of tumour cells in all cases but three. Proinsulin was also expressed in all tumours, with more than 50% of tumour cells immunoreactive in all cases but 5. It was consistently localized in the Golgi apparatus. In about half the cases, moreover, it also showed diffuse cytoplasmic staining, usually with a very sparse distribution. Trabecular and solid insulinomas did not present specific, homogeneous patterns of insulin immunostaining. However, insulin immunoreactivity was much more abundant in trabecular than in solid neoplasms, being present in virtually all tumour cells (score 6) in 50% and 8% of cases, respectively. Virtually all insulinomas expressed PC2, PC3, CPH and 7B2, usually in 30–100% of tumour cells, with a frequency significantly related to that of insulin. However, detection of PC2 and 7B2 was slightly less frequent than that of PC3 and CPH. In consecutive sections these proteins were found to be mostly co-localized with insulin and chromogranin A but not with proinsulin. They were heavily expressed in all 10 tumours with more than 10% of cells showing cytoplasmic proinsulin immunoreactivity, indicating that the leakage of proinsulin from the Golgi compartment is not associated with faulty expression of converting enzymes and possibly reflects a saturated processing capacity. HISL-19 immunoreactivity was found in both Golgi apparatus and insulin stores, indicating that the relevant antigen is different from all other proteins investigated. These results do not support a defect in expression or localization of proinsulin-processing enzymes in most insulinomas.