Pancreatic neuroendocrine neoplasms are histopathologically classified as well differentiated pancreatic neuroendocrine tumors (WD-PanNETs) or poorly differentiated pancreatic neuroendocrine carcinomas (PD-PanNECs) according to the 2010 World Health Organization (WHO) classification system. PanNECs are further subclassified into small cell and large cell carcinomas. Grading of PanNET is based solely on the mitotic rate, assessed as the number of mitotic figures (MF) per 10 high power fields (hpf) or the Ki67 labeling index. Staining for Ki67 is a useful surrogate in cytology as the recommendation is to count at least 40 hpfs (for a mitotic count), which is more than what can be found in most cell blocks from fine needle aspirations. Grading follows a three-tiered system: low grade (grade 1 [G1]) have less than 2 MF per 10 hpf (Ki67 index < 3%), intermediate grade (grade 2 [G2]) have 2 to 20 MF per 10 hpf (Ki67 index 3-20%), and high grade (grade 3 [G3]) have greater than 20 MF per 10 hpf (Ki67 index >20%). (2)

PanNETs are uncommon tumors, representing only 1-2% of all pancreatic neoplasms. These tumors tend to be small (1-3 cm), circumscribed masses that can be partially cystic. They most commonly occur in adults (peak incidence between 30-60 years). The majority of these tumors are asymptomatic and incidentally detected on imaging; however, a subset can be functional and present with clinical symptoms (sometimes life-threatening) related to excessive production of pancreatic hormones (insulin, glucagon, somatostatin, vasoactive intestinal polypeptide, pancreatic polypeptide, serotonin, adrenocorticotropic hormone, or calcitonin). A minority (10-20%) of PanNETs are syndromic with multiple endocrine neoplasia type 1 or von Hippel-Lindau syndrome (VHL), tuberous sclerosis, neurofibromatosis type 1, among others.

Histologically, these tumors show three main growth patterns including trabecular, glandular, and solid. Endocrine cells are typically round and monomorphic. Their nuclei are also round with fine chromatin that is uniformly dispersed. Nucleoli can vary from inconspicuous to prominent. The cytoplasm is pale to eosinophilic and usually granular. Rarely, vacuolated or clear cytoplasm is seen, the latter is more common in VHL. The large, pleomorphic cells of endocrine atypia can be present, however, cytologic atypia does not predict malignant behavior. Amyloid can be seen most commonly in association with insulinomas. Somatostatinomas on the other hand are associated with glandular architecture and psammomatous calcifications. While the stroma in PanNETs is often scant, it is highly vascular and hyalinization is common.

PanNEC must also be considered when evaluating neuroendocrine neoplasms of the pancreas. Classification of pancreatic neuroendocrine neoplasms has evolved over the past several decades to incorporate genetic, immunophenotypic, and molecular characteristics that better align with the behavior and treatment response of these tumors. According to the 2010 WHO classification, G1 and G2 lesions represented PanNETs while all high grade, G3, lesions were grouped together as PD-PanNECs. (2) In 2012, Yachida, et al. demonstrated that G3 pancreatic neuroendocrine lesions are in fact heterogeneous and that there is a molecular genetic distinction between PD-PanNECs and G3 WD-PanNETs by immunohistochemical studies and sequencing analysis of KRAS, CDKN2A/P16, TP53, SMAD4/DPC4, DAXX, ATRX, PTEN, Bcl2, and RB1. In their study, Yachida, et al. found that aberrant p53 expression (either null pattern or robust nuclear pattern) and loss of Rb protein were associated with PanNECs. On the other hand, all the WD PanNETs in their study showed normal (wild type) p53 expression and nuclear positivity for Rb. (3)

Early molecular studies identified mutations in MEN1, ATRX, and DAXX to be the most commonly found in PanNETs. (4, 5, 6) While loss of DAXX and/or ATRX protein expression defines WD-PanNET and abnormal p53 and/or Rb expression defines PD-NEC, there is some debate on the significance of SMAD4/DPC4 loss in differentiating G3 WD-PanNET from PD-PanNEC. SMAD4/DPC4 is a tumor suppressor in the TGF-beta signaling pathway and is inactivated in 55% of pancreatic adenocarcinomas. (7) Kim, et al. found loss of SMAD4 expression in three of seven PanNECs in their study cohort while none of their G3 PanNET cases showed loss of SMAD4/DPC4, leading to their conclusion that SMAD4/DPC4, along with p53, p16, and MUC1, may be helpful in differentiating G3 WD-PanNETs from PD-PanNEC. (8) However, other studies have found limited utility in evaluating for SMAD4/DPC4 loss when differentiating G3 WD-PanNET from PD-PanNEC. In the study by Yachida et al., SMAD4/DPC4 was lost in only one case of PanNEC. All other PanNECs and WD-PanNETs in their study cohort did not show loss of SMAD4/DPC4. (3) Similarly, Tang, et al. found that loss of SMAD4/DPC4 expression was only present in one of their PD-PanNECs and this case also had a concurrent p53 mutation. Importantly no concurrent abnormalities in DAXX/ATRX along with p53, Rb, or SMAD4 were found in the study by Tang et al. demonstrating that mutations in DAXX and ATRX support a diagnosis of PanNET while p53 and Rb mutations support PanNEC. (9) Thus, molecular studies can assist with morphologically ambiguous cases. (10, 11)

The 2010 WHO classification was too broad and did not capture the heterogeneous nature of high-grade neuroendocrine lesions. In 2017, a new diagnostic category was added, G3 WD-PanNET. These lesions have a Ki67 index >20% and resemble lower grade PanNET at the genomic level (ATRX, DAXX, MEN1). Thus, high grade neuroendocrine neoplasms of the pancreas include both well differentiated neuroendocrine tumors and poorly differentiated neuroendocrine carcinomas. (12, 13, 14) Distinction between PD-PanNEC and G3 WD-PanNET is clinically important as PD-PanNEC shows response to chemotherapy while G3 WD-PanNET shows limited response, however it is associated with better prognosis that PanNEC. (15)

The patient in this case had a laboratory workup that was negative for functional PanNET and she did not display any symptoms for the same (denied flushing/diarrhea, nausea, early satiety, hypoglycemia, or peptic ulcers). A diagnosis of WD- PanNET was confirmed by immunohistochemical studies (positive synaptophysin, chromogranin, and INSM1, while BCL10 was negative in tumor cells). The Ki67 index was 1% consistent with a Grade 1 WD-PanNET. Because of the proximity to the pancreatic duct, her tumor was not amenable to enucleation and any resection would have required consideration of a Whipple procedure. Given the size (<2 cm) and pathology (well differentiated and low Ki-67), resection was not indicated, and she was followed with serial surveillance with CT imaging.