A retrospective review of the medical records of pediatric patients (age 0-18 years) diagnosed with insulinoma was performed. Insulinoma was diagnosed biochemically (serum insulin >2.0 U/l during the hypoglycemia <3.0 mmol/l) and by imaging (US, CT, MRI, endoscopic US), and verified by histopathology.

Collected data for the analysis included family history (parents were interviewed on known malignancies, benign lesions, ulcer, cholelithiasis or hypoglycemia in relatives), clinical symptoms prior to diagnosis, its onset and severity, results of the biochemical, hormonal, genetic and histopathological investigations. Whenever possible, clinical data during follow-up period were analyzed.

Screening for the signs of MEN1 syndrome included hormonal analysis (parathyroid hormone (PTH), cortisol, adrenocorticotropic hormone (ACTH), gastrin, prolactin, insulin growth factor 1 (IGF1)), imaging (brain MRI, abdomen US/CT, thyroid US), blood biochemical analysis (Ca, Ca++, glucose, ALT, AST), NET markers (serum chromogranin A and serotonin, urinal 5-Hydroxyindolacetic acid (5-HIAA)) and was performed at the first visit and during follow-up.

Fasting test was performed according to the local protocol and required capillary glucose measurements using an automatic blood glucose meter for professional use every 3 hours if blood glucose (BG) was ≥4 mmol/l, every hour if BG was 3.5-3.9, and every 30 min. if BG was ≤3.4 mmol/l. A critical sample was obtained when BG was less than 3 mmol/L and included serum glucose, insulin and cortisol in all cases. Additionally, serum 3-hydroxybutyrate (BHB), and C-peptide, were measured in 15, and 13 cases, respectively.

Blood biochemistry was performed using Hitachi 912 Analyzer with standard reagents. Glucose was measured on plasma from venous blood samples with Cobas 8000 hexokinase assay analyzer (Roche^®) with normal values 3.3 to 6.1 mmol/l. Bedside glucose values, or continuous glucose monitoring values, were not used for diagnostic fasting measurements. Serum BHB measurements were performed using a precision Xtra meter (Abbott Pharmaceuticals), with a reported assay range of 0 to more than 8 mmol/L. Urine ketone bodies were measured on an automated iChemVELOCITY analyser (Beckman Coulter Life Sciences, Krefeld, Germany) with urine test strips. Levels of insulin, C-peptide, cortisol, ACTH and PTH were measured using Cobas 6000 analyzer (Roche Diagnostic, Switzerland). Prolactin and IGF1 levels were measured using Vitros 3600 (Johnson & Johnson) and a Liason (DiaSorin) analyzer, respectively. Serum NET markers were evaluated using standard immunoassay method, urine 5-HIAA — using liquid chromatography method.

Histological and immunohistochemical (IHC) studies were performed on sections 3–5 µm thick prepared from paraffin blocks. For IHC studies, an Autostainer (Autostainer, LabVision, type 480s, UK) was used. Sections were deparaffinized and antigenicity was restored in buffer pH 9.0 in a PT Module (Thermo Scientific, UK). To determine the type of tumor, a spectrum of antibodies was used: Chromogranin A (clone LK2H10), synaptophysin (clone MRQ-40), CD56 (clone MRG42), insulin (polyclone, RTU), glucagon (polyclone, RTU), somatostatin (polyclone, RTU, Cell Marque, USA), and gastrin (all from Cell Marque, USA), pancreatic polypeptide (clone EPR2330-10, Abcam, USA), Ki67 (clone MIB1, DAKO), somatostatin receptors type 2 (rabbit monoclonal EP149, Epitomix, USA) and type 5 (rabbit monoclonal UMB4, Epitomix, USA).

Somatostatin receptor (SSTR) expression analysis was performed according to the method of Volante M. et al.(20). Membrane expression of SSTR2, or membrane-cytoplasmic expression of SSTR5, was considered as positive if found in more than 30% of tumor cells.

The expression of other cytoplasmic markers was estimated according to standard semi-quantitative method for cytoplasmic markers. Tumor grade was assessed using Ki67-index according to the World Health Organization guideline (21).

Genomic DNA from peripheral blood leukocytes was extracted using standard methods (22). Molecular genetic analysis of MEN1 was performed using bidirectional direct sequencing (n=21). Samples were sequenced in both directions using the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems, FosterCity, CA, USA) and analyzed on a ABI3730XL DNA Analyzer (AppliedBiosystems, Naerum, Denmark). Sequence analysis was performed using SeqMan Software (DNASTAR, Madison, WI, USA). To detect larger deletions Multiplex Ligation-dependent Probe Amplification (MLPA) analysis was performed (n=4) according to the manufacturer’s recommendations (Salsa MLPA, P017 MEN1, MRC-Holland, Amsterdam, the Netherlands). Data were analyzed using GeneMarker (Softgenetics, Pennsylvania, USA). MEN1 DNA variant nomenclatures were given according to GenBank accession no. NM_000244.4.

Follow-up investigations included screening for MEN1 components as described for the first visit, including detection of any distant metastases, using abdominal ultrasound or MRI. In patients with genetically verified MEN1 syndrome first grade relatives (parents and siblings) underwent genetic testing for point mutation in MEN1 gene and in case of positive results, — screening for MEN1 components as mentioned above.

Demographic and clinical data were presented as median (interquartile range (IQR)). Statistical analysis was performed using StatSoft Inc., USA, version 10.0. Non-binary data were analyzed with the help of Mann-Whitney U test and chi-square test with p value < 0.05 considered as significant.

We analyzed 465 medical records of children aged less than 18 years admitted to the Endocrine Research Center with HH from 1995 to 2021. A total of 22 patients (13 females) were diagnosed with primary pancreatic insulinoma, accounting for 4.7% of all pediatric patients with HH.

The median age at the time of first symptoms was 10.45 years. Median age at the time of diagnosis was 11.5 years, giving the median delay in diagnosis of 10 months (Table 1). Twelve children (55%) were misdiagnosed to have epilepsy and were treated with anticonvulsants before the hypoglycemia was revealed. All patients had typical clinical features of hypoglycemia, including drowsiness (73%), seizures (73%), syncope (68%), progressive weight gain (45%), learning and behavioral difficulties (45%). A third of the patients experienced hypoglycemic coma prior to diagnosis (7/22; 32%).

Median duration of diagnostic fasting test was eight hours (Table 1). In all cases it resulted in laboratory hypoglycemia with a mean serum glucose level of 1.76 ± 0.63 mmol/L. Median serum insulin level taken during hypoglycemia was 20.9 U/L, median C-peptide level 3.4 ng/mL (n=13). Serum hydroxybutate was less than 0.5 mmol/l in all cases (n=15). Urine ketone bodies were undetectable in all measurements in cases when serum ketones were not available (n=7).

Localization of the tumor only by transabdominal ultrasound was possible in seven patients. Contrast enhanced CT and/or MRI were used in 18 cases. Additional endoscopic ultrasound was needed in four cases with inconclusive results of MRI and CT. Figure 1 represents results of different imaging technics used for the evaluation of pancreatic lesions in our cohort of patients.

Five children (23%) had multiple pancreatic lesions. The 30 lesions were found equally in all parts of the pancreas (tail; n=11, head; n=10, body; n=9).

Prior to surgery, 12 children required hyperglycemic medication. Diazoxide and octreotide alone were used in nine and two cases, respectively. One patient (case#12) had a combination of both drugs. Doses of diazoxide ranged from 100 to 300 mg/day. In 7 out of 10 patients, diazoxide was sufficient to maintain normoglycemia. Octreotide was given by subcutaneous injections every 8 hours in doses of 400 mcg/day. Other patients were managed with frequent feeds and/or continuous dextrose infusion.

Surgical treatment was performed in all children: insulinoma enucleation in 11 (50%), partial pancreatic resection in 8 (36%), subtotal pancreatectomy in 2 (9%). In one patient (5%) with multiple lesions, repeated surgeries ended up with pancreatogastroduodenal resection (Table 2). Removed tumors varied in size from 0.3 to 6 cm. Six tumors were less than 1 cm, 14 ranged from 1 to 2 cm, and 10 were larger than 2 cm.

Histopathological studies confirmed the presence of insulinoma in all cases. Lesions showed polymorphic histology: most of the tumors had trabecular architecture (Figures 2D–O), rarely solid (Figures 2A–C) or mixed. Majority of lesions were encapsulated and well circumscribed from the surrounding pancreatic tissue. One patient (case #12) had multifocal insulinoma without capsule and invasion to the surrounding tissue (Figures 2M–O). This patient had liver metastases at the time of diagnosis.

Immunohistochemical studies were performed in 17 cases. All lesions demonstrated the expression of Synaptophysin and Chromogranin A. Among the children with multiple pancreatic lesions (n=5), some of the NETs were negative for insulin staining. Of these, glucagon expressing tumors were found in two cases (Figures 2H–L). Others were classified as non-functioning. Compound expression of gastrin and insulin was found in 1 case. SSTR2 and SSTR5 expression was analyzed in 13 lesions and was positively expressed in 6 and 5 cases, respectively (Figures 2G, K, L). Ki67 index and/or mitotic indexes were measured in a total of 19 lesions. In ten cases we found G2 differentiation grade (Table 3).

We performed biochemical screening for the MEN1 syndrome components at the time of insulinoma diagnosis. This revealed mild normocalcemic hyperparathyroidism in two patients (cases #4 and #5), mildly elevated serotonin in three patients (cases #4,5,6), and high levels of Chromogranin A and 5-HIAA in one (case #6). There were no biochemical or radiological signs of pituitary adenoma. Analysis of the family history was possible in 20 cases and revealed malignancies in relatives (n=6), ulcer (n=2) and hypoglycemia (n=1). None of the interviewed relatives had a history of hyperparathyroidism or insulinoma (Table 2).

MEN1 gene sequencing revealed pathogenic variants in 8 out of 21 children (38%). Of them, 2 missense, 1 nonsense, 1 splicing and 4 frameshift mutations were identified (Table 2). Five variants were previously described in MEN1 patients (23–27). Mutations p.P277R (case #1), p.Tyr313Ilefs*55 (case #3), and p.Leu48SerfsTer69 (case #8) are novel. With regards to a latter variant, a different substitution in the same codon was previously described in the literature as disease causing (28).

We compared main clinical characteristics in children with sporadic insulinomas (n=13) and those with pathogenic variants in the MEN1 gene (n=8) (Table 4). Children with MEN1 syndrome tended to be younger at the time of insulinoma presentation, with a higher number of lesions and more often G2 grade. Significant difference between the two groups was, however, only found in the number of lesions (Table 4).

Genetic testing of parents was performed in six families and revealed MEN1 carriers in five. All relatives with MEN1 mutations were investigated. Despite the absence of suspicious family history and minimal clinical presentations, four of out the five MEN1 parental carriers and two additional relatives were found to have components of MEN1 syndrome, but none compatible with insulinoma (Table 2).

Follow-up data were available in 17 patients. Median [25-75%] follow-up duration was 4.2 [1.3-5.2] years. There was no recurrence of insulinoma during follow-up. All children with genetically confirmed MEN1 syndrome developed various MEN1 components during next 2-13 years with hyperparathyroidism and hyperprolactinemia being the most common findings (n=5 and 3 resp.), (Table 2). In one patient, liver metastases were found eight years after the pancreatectomy (case #20).