Based on the obtained diagnostic data, the multidisciplinary board was held, implying the participation of surgical oncologist, medical oncologist, pathologist and radiation oncologist. Taking into account tumor size more than 2 cm and hypoglycemia syndrome radical pancreatic resection was planned. A distal spleen-preserving pancreatectomy was performed.

Taking into account multiple microadenomas in the pancreatic parenchyma the recurrence rate of the hyperinsulinemic hypoglycemia is relatively high. Thus, patient is examined every 3-6 month. In our case there are no signs of hypoglycemia and no pancreatic tumors on imaging during 30 months of follow-up by CT and PET-CT scanning.

Microscopic examination showed well differentiated neuroendocrine tumor with 2 mitoses and no necrosis in 10 HPF. There are multiple large microadenomas in the surrounding pancreatic tissue. Tumor cells are positive for synaptophysin and chromogranin A, as well as for insulin ( Figures 1A, B ). There is no expression of glucagon, somatostatin and pancreatic polypeptide. Ki-67 is less than 1.5%, Grade 1. The cells of microadenomas express insulin and are negative for glucagon and somatostatin ( Figures 1C, D ).

Macrodissection of tumors was performed under guidance of pathologist. DNA was extracted from four formalin-fixed, paraffin-embedded (FFPE) tumor tissues by using GeneRead DNA FFPE kit (Qiagen, Germany), and from whole peripheral blood by using standard phenol-chloroform extraction protocol. One tumor sample was insulinoma, three others were small proliferative insulin-expressing monohormonal endocrine cell clusters (IMECCs).

Next generation sequencing (NGS) was performed by using Ion AmpliSeq targeted amplification technology with AmpliSeq Comprehensive Cancer Panel (Thermo Fisher Scientific, United States) with exon coverage of 409 oncogenes and tumor suppressor genes.

The bioinformatic workflow for sequencing data analysis was based on Torrent Suite software (version 5.10.1). Annotation was performed by ANNOVAR (4). CNVs were called using CNVpanelizer R package (version 1.22.0) ran with default parameters (5). Putative CNV status is assigned when there is a significant difference between observed copy number and the one expected from the bootstrapped distribution. A reliable CNV is called when the upper bound of the copy number ratio is below the lower bound of background (noise) copy number for deletion or when the lower bound of the copy number ratio is above the upper bound of background (noise) copy number for amplification.

Sanger sequencing was performed for detection of the Trp372Arg mutation in YY1 gene. PCR was performed in 25 µl reactions. The reaction mixture for PCR consisted of the following reagents: 8% glycerol, 68 mM Tris-HCl with pH 8.3, 17 mM (NH₄)₂SO₄, 0.01% Tween-20, 0.1 mg/ml BSA, 0.2 mM each dNTP, 1.5 units Taq polymerase, 0.12 pM each primer, 2.5 mM MgCl₂. On top, 40-60 μl of mineral oil was layered. PCR reaction was performed with the following parameters: 95°C for 5 min, followed by 30 cycles of 95°C for 40 seconds, 61°C for 40 seconds and 72°C for 40 seconds. Final elongation was at 72°C for 5 minutes. Sanger sequencing was performed on ABI3500 according to the manufacturer’s protocol (Thermo Fisher, USA). Sequencing results were analyzed in Chromas software and compared to GenBank database using the BLAST algorithm. Primer sequences for Trp372Arg are listed in Table 1 .

Quantitative PCR (qPCR) is one of the methods that are widely used to detect copy number changes. This method is preferable due to its low consumable and instrumentation costs, high sensitivity and fast development time of assay (6–9). qPCR was performed to confirm the results of CNVPanelizer packages for CEBPA and FOXL2 gene, the first and the second exons of IRS2 gene and the first and the ninth exons of ATRX gene.

The qPCR mixes were prepared according to the GenPak PCR Core protocol (Isogene Lab ltd, Russia) in 20 µl reactions containing 1 ng genomic DNA and using SYBR green as a reporter and ROX as a reference dyes. All qPCR reactions were run in triplicate on a QuantStudio 5 Real-Time PCR System for Human Identification (Thermo Fisher Scientific, USA). The PCR conditions were as follows: 95°C for 5 min, followed by 40 cycles of 95°C for 20 seconds, 60°C for 20 seconds and 72°C for 20 seconds. Data were analyzed by the ΔΔCt method (10). The relative copy number was estimated by comparison with a normal blood control DNA sample. We used the B2M housekeeping gene as an endogenous control. Primer sequences for qPCR are shown in Table 1 .

Point mutation Trp372Arg (Chr14:100,743,807, hg19) in YY1 gene was not found in the tested samples.

Median NGS read coverage was 486x, 461x, 471x, 454x for four tumor samples and 217x for a blood sample.

Gain status was confirmed in all samples for FOXL2 gene. For CEBPA, only in insulinoma and IMECC #1 gain status was confirmed (p-value <0.001 and <0.01 respectively).

Amplification of the first exon of the IRS2 gene was also confirmed in all samples (p-value <0.001 for all samples). However, the data on the second exon was various. Results qPCR with CNVPanelizer matched only in insulinoma sample (p-value <0.01).

Amplification of the first exon of ATRX in insulinoma sample that we noticed on CNVPanelizer was confirmed with a statistical significance of p-value < 0.01. The downstream loss of ATRX which we checked in the ninth exon was also confirmed with statistical significance of p-value < 0.05. These results show that CNVPanelizer can reliably detect CNVs in separate exons.

Pancreatic neuroendocrine neoplasms (panNENs or pNENs) are rare tumors of the pancreas that account for up to 2% of all pancreatic neoplasms. However, based on autopsy studies, prevalence of panNENs has been reported to be up to 10% (13). The 2017 World Health Organization classification divided panNENs into two categories, well-differentiated pancreatic neuroendocrine tumors (panNETs or PNETs) and poorly differentiated pancreatic neuroendocrine carcinomas (panNECs) (14). In the vast majority PNETs occur sporadically (~90%), but up to 5-10% are associated with genetic syndromes like multiple endocrine neoplasia type 1, neurofibromatosis type I, Von Hippel–Lindau syndrome and tuberous sclerosis complex (15). The most common functioning PNETs are insulinomas with an incidence of 4-7 per 100,000 persons per year (16–18). Most of insulinomas, more than 90%, are benign. Insulinomas are composed of producing beta cells that are actively secreting a large amount of insulin that results in episodic hyperinsulinemia and is the most frequent cause of persistent hyperinsulinemic hypoglycemia (1). In addition to Ki-67, which differentiates malignant and benign nature, tumor size is critical for survival rate prognosis. Thus, insulinomas >2 cm in diameter have a 10-year survival rate nearer to 30%, whereas for those <2 cm, the survival rate is close to 100% (19).

The molecular genetic of insulinomatosis is yet to be understood. This phenomenon is not related to mutation in MEN1 gene and is more similar to sporadic benign insulinomas (1). In insulinomas, mutations in MEN1, ATRX and DAXX that are often mutated in all PNETs occur in no more than 10 percent: 3%, 8% and 3%, respectively (20). In addition many other genes were seen to be mutated in insulinomas (20, 21) and several regions with amplification (7p, 3p, 5q and 13q) were identified as early events and may be involved in tumorogenesis (22). However, the most common mutation in insulinoma is a gain of function mutation Trp372Arg in YY1 gene that occurs in 30% in Asian population and in 13% in Caucasian (German) population (23, 24). Mutation Trp372Arg increases the activity of YY1 as transcription factor, which results in greater transcription of IDH3A, UCP2 and increases the expression of ADCY1 and CACNA2D2 which regulate the insulin secretion (22). It is assumed that mutations in YY1 gene are driver mutations for insulinomas (25). In our case of insulinoma, we did not find Trp372Arg mutation in the YY1 gene nor mutations in MEN1 or DAXX, however, loss of ATRX was found and confirmed by qPCR ( Figures 2 – 4 ).

As for other CNVs, alterations were found in 13q34 (amplification of IRS2), 13q14.2 (loss of RB1) and 5q22.2 (loss of APC) ( Figures 2 , 3 ). With regards to the precursor lesions of PNETs, the information about their molecular genetic profile is very limited. Microadenomas are shown to harbor mutations in MEN1 (13, 14, 26, 27). Hadano et al. in their work showed that sporadic microadenomas have a significantly lower expression of ATRX and overexpression of cytokeratin-19 (CK19) compared to hyperplasia of pancreatic islet cells (13). In our studied samples of micro-tumors, we have not found alteration in MEN1, but gain status in IRS2, FOXL2 and CEBPA genes was found, suggesting that they can play a role in the development of micro- to macro-tumors, to wit insulinomas, and, accordingly, insulinomatosis.

Besides the identity in monohormonality, our macro and micro-tumors samples share CNVs that harbor the same genes, CEBPA, FOXL2 and IRS2. CCAAT enhancer-binding protein alpha (CEBPA) and Forkhead Box L2 (FOXL2) are known for being able to arrest or suppress cell proliferation respectively (28–30). The IRS2 (Insulin receptor substrate 2) protein plays an important role in the response to stimuli for cytokines and growth factors, like insulin and insulin-like growth factor 1, and promotes proliferation and survival of normal and cancer cells by mediating signaling from INSR, IGF1R, EPOR, MPL, VEGFR, LEP, GH and IFNB1/IFNG proteins. The stimulation of insulin receptor results in IRS2 association with the p85 subunit of PI3K and GRB2, activating the PI3K/AKT/MTOR and MAPK pathways and leading to proliferation and differentiation (31, 32).

It was shown that increased expression of CEBPA is involved in apoptosis of pancreatic β-cells exposed to proinflammatory cytokines IL1β, IFNγ, and TNFα (33). Also it was shown that CEBPA induce the transdifferentiation of B cells into macrophages, and in co-expression with the transcription factors Oct4 (Pou5f1), Sox2, Klf4 and Myc enhances reprogramming into induced pluripotent stem cells (34).

Mohanty in his work showed that overexpression of IRS2 stimulates proliferation of β-cells and increases insulin secretion. It also protects β-cells from d-glucose-induced apoptosis. On the contrary, the repression of IRS2 in INS-1 cells leads to downregulation of proliferation (35).

As insulinomatosis seems to be a disease that affects the entire population of β-cells, we can assume that this triad of genes, CEBPA, FOXL2 and IRS2, participates in the disturbance of morphogenesis of β-cells, although of course, this statement requires more detailed investigations.