Histopathological Landscape of Precursor Lesions of Gastro-Entero-Pancreatic Neuroendocrine Neoplasms

Recently, attention on neuroendocrine neoplasms of the gastro-entero-pancreatic tract (GEP-NENs) has grown, but despite efforts to better characterize these tumors, little is, as yet, understood about their precursor lesions. In fact, the well-established relationship between dysplasia and neoplasia described for epithelial gastro-entero-pancreatic tumors cannot be demonstrated in the neuroendocrine field.

In order to start from a common step, a brief reminder is required concerning the current WHO classification of neuroendocrine neoplasms (NENs) of the digestive system. In general, regardless of the site of occurrence, these lesions can be categorized within the morphological tiers of well-differentiated neuroendocrine tumors (NETs), poorly differentiated neuroendocrine carcinomas (NECs), and mixed neuroendocrine-non-NENs (MiNENs). NETs are further graded as G1, G2, or G3 on the basis of proliferative activity, assessed in terms of mitotic rate and Ki-67 labeling, while NECs are regarded as high-grade cancers by definition and are distinguished in small-cell NECs and large-cell NECs; In addition, the neuroendocrine component of MiNENs can be both well or poorly (more frequently) differentiated. This classification was first developed for pancreatic NENs and subsequently extended to the neuroendocrine lesions of the entire gastro-entero-pancreatic tract (GEP-NENs) (Table 1) [1]. When compared with the previous 2010 WHO classification, the main novelty of the current taxonomy relates to the introduction of the category of well-differentiated NETs graded G3: the recognition of these entities is of utmost importance both with regard to a clinical and translational point of view. In fact, from the clinical standpoint, G3 NENs represent an intermediate prognostic stage between the almost always favorable outcome of low-intermediate grade NENs (G1 and G2) and the dismal course of NECs; on the other hand, the question whether G3 NETs constitute the link between NETs and NECs is still unresolved. Indeed, they may represent the last stage of progression of a unique “neuroendocrine transformation way,” highly divergent from NEC natural history. Precisely for this last purpose, a more in-depth knowledge of neuroendocrine precursor lesions along the gastro-entero-pancreatic tract could be of great help. Since GEP-NENs are clearly a rare and heterogenous group of tumors, with different anatomic locations and clinicopathological features, the identification of general hallmarks concerning the distinctive characteristics and the natural history of precursor lesions is challenging.

Due to the relatively high incidence of gastric NENs in the setting of chronic autoimmune gastritis (AG), their precursor lesions have been extensively investigated. Some knowledge has also been acquired about neuroendocrine preneoplastic lesions in the duodenum and pancreas. On the contrary, neuroendocrine precursor lesions in the esophagus, small and large bowel, and appendix have not been fully characterized yet.

The aim of this review was to summarize the state-of-the-art in the knowledge of the preneoplastic neuroendocrine lesions of the gastro-entero-pancreatic tract, highlighting the most relevant biologic and morphologic aspects in each anatomical site: this sort of snapshot may result a useful tool in order to design future research projects aimed at reaching a fuller understanding of the underlying biology and early development of these diseases. Many questions still remain unanswered in this field, especially why some precursor lesions progress toward fully developed cancers and others seem to arrest in early stages.

Esophageal NENs (E-NENs) are extremely rare, accounting for 0.04–1% of GEP-NENs [2]. Their recently increasing incidence is probably due to the increased diagnostic awareness of foci of neuroendocrine differentiation within the framework of esophageal adenocarcinomas or squamous-cell carcinomas. Indeed, it is well known that NECs share their molecular background with AdCs and SCCs, making the finding that more than 90% of E-NENs are NECs, while less than 50 cases of esophageal NETs have been reported, significant [2]. E-NENs usually arise in the lower third of the esophagus [3], possibly because the majority of neuroendocrine cells are located in the submucosal glands of the lower esophagus or because E-NENs may develop in the context of Barrett mucosa [2, 4‒7]. So far, no precursor lesions of E-NETs have been described.

Gastric NENs account for 4% of all NENs, with an incidence of 0.4/100,000 persons per year [8, 9]. They are a heterogeneous group of lesions that can be further classified according to the type of neuroendocrine cell differentiation in (i) enterochromaffin-like (ECL) cell NETs and (ii) antral NETs. Antral NETs account for a minority of gastric NETs [10], while ECL NETs are further subclassified in three officially recognized groups (type 1, 2, and 3) and two provisory groups (type 4 and 5) [11]. Consequently, ECL-cell NETs are better characterized, and especially in the settings of type 1 and type 2, the precursor lesions have been properly recognized and described, although their neoplastic potential has not been fully understood [9, 12‒14].

Type 1 ECL-cell NETs (80–90% of gastric NETs) arise in the context of AG (i.e., type A chronic AG). AG is a chronic inflammatory gastric disease that electively involves the fundus and body of the stomach (type A gastritis) [15, 16] and it is distinguished from antral gastritis associated with Helicobacter pylori (type B gastritis) [17‒19]. AG consists of an immune-mediated aggression to the parietal cells resulting in hypo/achlorhydria, which leads to a compensatory hyperplasia of antral gastrin cells (G-cells) and finally to hypergastrinemia [9, 20‒23]. Pathogenesis of type 1 NETs is strongly related to oxyntic gland atrophy-induced hypo/achlorhydria since long-standing hypo/achlorhydria triggers ECL-cell hyperplasia, which may further progress to dysplasia and, ultimately, to a fully developed neoplasia [13, 23‒27]. In this context, AG is considered a “precancerous condition,” while hyperplastic/dysplastic proliferations are “preneoplastic lesions.” Coherently with the proposed pathogenetic model, preneoplastic lesions arise within the cancerization field of atrophic gastric mucosa [12, 23].

These precursors are morphologically heterogeneous, so they have been classified into hyperplastic and dysplastic lesions. The hyperplastic ones comprise diffuse, linear, micronodular, and adenomatoid hyperplasia, while enlarged micronodules, fused micronodules, micronodules with new stroma, and microinfiltrative lesions are regarded as dysplastic [28‒31] (Table 2) (shown in Fig. 1).

Assessing the risk of neoplastic evolution in each type of ECL-cell proliferation is extremely complex; however, Vanoli and colleagues [14] suggest that severe ECL-cell hyperplasia (i.e., > 6 chains of linear hyperplasia per mm) and ECL-cell dysplasia (primarily microinfiltrative lesions) are associated with a significantly increased risk of NET development. Furthermore, a positive relationship between OLGA staging and neuroendocrine lesions has been reported: the higher the stage of gastritis, the higher the prevalence of ECL hyperplastic lesions and NETs [23]. Of note, in the series reported by Rugge and colleagues [23], 89.5% of NETs were associated with a score 3 atrophy of the body but an OLGA stage of II. Therefore, OLGA tends to underestimate the risk of NET development, due to the greater importance of the atrophy of the antrum compared to the body and fundus within this scoring system.

Type 2 ECL-cell NETs account for about 5–7% of gastric ECL NETs. They arise in patients with Zollinger-Ellison syndrome (ZES) that occurs in multiple endocrine neoplasia type 1 (MEN1) [28, 32, 33]. Type 2 ECL NETs also develop in patients with hypergastrinemia, caused by a duodenal, or less frequently, a pancreatic gastrinoma in the context of MEN1. The peritumoral oxyntic mucosa is hypertrophic because of the gastrin stimulation and may harbor hyperplastic/dysplastic ECL-cell proliferations [9]. As opposed to type 1 NETs, severe linear hyperplasia alone is strongly associated with type 2 NET development. Furthermore, while ECL-cell hyperplasia can arise in patients with sporadic gastrinomas [34], ECL-cell dysplasia and ECL-cell NETs only develop in patients with MEN1 syndrome, suggesting that MEN1 mutation sensitizes ECL cells to the mitogenic effect of gastrin [30].

Type 3 ECL-cell NETs represent about 10–15% of gastric NETs. These tumors are not associated with hypergastrinemia and/or AG and precursor lesions have not been identified in the peritumoral mucosa since they arise in the context of normal oxyntic mucosa without hyperplastic or dysplastic ECL-cell proliferation [9].

Type 4 ECL-cell NETs arise in association with hypergastrinemia and parietal cell hyperplasia in patients without MEN1 and ZES. In these cases, they seem to be related to a defect in acid secretion from parietal cells due to a defect/lack of proton pump [9, 33, 35]. Type 5 ECL-cell NETs develop in patients with hypergastrinemia caused by long-term (at least 1 year) proton pump inhibitor use in absence of medical conditions such as AG, gastrinoma, or MEN1 [36]. The pathogenesis of gastric NETs in long-term PPI users is still not understood [36]. According to our previous statements concerning the pathogenesis of type 1 and type 2 ECL-cell NETs and the relationship with hypergastrinemia, the incidence of type 5 ECL-cell NETs among long-term PPI users should be high, but the vast majority of these patients do not develop any NETs, despite having persistent hypergastrinemia [36‒42]. This might indicate that chronic hypergastrinemia itself may not be sufficient for the development of gastric NETs and other contributing factors (environmental factors and/or genetic predisposition) are needed [22, 36, 43‒45].

In comparison with those in other organs, the peculiarity of gastric NET is their easy access and possibility of removal by endoscopy. With this approach, the surgery for these lesions has almost been completely abandoned, given the possibility of endoscopic treatment ranging from simple polypectomy to mucosal/submucosal endoscopic resection when of large size (shown in Fig. 2) [46, 47].

In contrast to the NETs developing in the oxyntic mucosa, no precursor lesions have been recognized for antral NETs. In fact, although hyperplasia of gastrin-producing cells (G-cell) has been reported in conditions that promote achlorhydria, this does not appear to increase the neoplastic transformation risk. This is true also for hyperplasia of somatostatin-producing cells (δ-cells) in the setting of duodenal ulcer disease. Moreover, neuroendocrine cell dysplasia has never been described in the antrum [48].

In conclusion, all types of ECL-cell NETs, except type 3, arise in the context of chronic hypergastrinemia (a necessary, but not sufficient condition) that leads to ECL-cell hyperplasia, then to ECL-cell dysplasia, and finally to a fully developed NET. Neither putative pathogenetic mechanisms nor precursor lesions have been found for type 3 ECL-cell NETs. Focusing on this last distinction, La Rosa and Solcia [11] further classified gastric NETs arising in the oxyntic mucosa into two categories: tumors not associated with hypergastrinemia (type 3 NETs) and tumors in which hypergastrinemia plays a crucial pathogenetic role.

Duodenal NENs represent up to 22% of NENs [49]. According to tumor-cell differentiation, duodenal NENs can be distinguished in (i) G-cell NETs, (ii) D-cell NETs, (iii) gangliocytic paraganglioma, and (iv) NECs [50, 51]. G-cell NETs are the most common (>60%), either functioning (i.e., associated with ZES) and nonfunctioning. D-cell NETs and gangliocytic paraganglioma constitute 21% and 9% of cases, respectively, while NECs are exceedingly rare. Most duodenal NETs are sporadic and nonfunctioning and are commonly diagnosed incidentally in older (usually between 50 and 70 years) women (M:F ratio 1.5:1) [52]. Functioning NETs are linked with MEN1 (15%–23% of cases) and are often multiple [53]. Until very recently, precursor lesions of duodenal NENs had been found only in patients diagnosed with MEN1 and had been associated only with gastrin or somatostatin-producing tumors [54].

All sporadic neuroendocrine duodenal neoplasms are traditionally thought to arise in absence of any forerunner lesion, despite the abundance and the variety of neuroendocrine cells resident in the duodenum [2, 12, 49, 54‒58]. Additionally, no duodenal precursor lesion has been described in patients with neurofibromatosis type 1 (NF1, von Recklinghausen’s disease) nor Von Hippel-Lindau (VHL) disease, who bear an inherited predisposition for the development of ampullary or periampullary NETs [59].

Recently, Merchant and colleagues [60] reported an association between G-cell hyperplasia and sporadic gastrinomas in a small series of 18 patients with H. pylori gastritis treated with PPIs. This observation suggests that, as in the gastric mucosa, a proliferation of G-cells may occur in the duodenum [61].

Within the spectrum of MEN1-related G- and D-cells preneoplastic lesions, hyperplasia and dysplasia can be distinguished on the basis of specific morphometric criteria (Table 3) [48, 62]. These precursor lesions are typically multiple and can occur either within the crypts or within Brunner’s glands [55].

Although duodenal neuroendocrine preneoplastic lesions are morphologically similar to those of ECL cells in type A AG, gastric and duodenal neuroendocrine precursors have a different biological substrate [12, 50, 63]. In fact, 50% of MEN1 microinvasive lesions in the duodenum show allelic deletion of the MEN1 gene, while hyperplastic endocrine cell precursor lesions retain heterozygosity [50, 64]. This is consistent with the hypothesis that allelic deletion of the second MEN1 gene is a pivotal event in the development of gastrin and somatostatin-producing neoplasms [50, 55]. Moreover, the observation of chronologically distinct second MEN1 allele deletion patterns in synchronous tumors of MEN1 patients supports the concept that each gastrin/somatostatin-producing duodenal tumor arises from independent cell clones, justifying the typical multifocality of the duodenal disease [55].

NENs are the most common malignancies of the jejuno-ileal tract [65], with an estimated incidence of 0.28–0.89/100,000 per year [66]. To date, most small intestine NETs (SI-NETs) are thought not to be preceded by precursor lesions [2] and, indeed, endocrine cell dysplasia has never been described [50]. Moreover, there is no known genetic defect that associates with ileal NENs, even though jejunal NETs have been reported in patients with MEN1 syndrome [67‒69]. Recently, Sei et al. [70] have suggested the existence of a multiphasic carcinogenetic process in SI-NETs, studying a family with a particular genetic alteration. According to their results, SI-NETs originate from a peculiar type of serotonin-producing enterochromaffin cell (EC) that retains the capability to dedifferentiate and acquire stem properties [71]. These cells have been named “active reserve intestinal stem cells” (ARISCs), since they physiologically participate in the constant self-renewal of the intestinal mucosa, but their activity increases in the context of chronic damage. ARISCs are characterized by a unique molecular signature, with expression of BMI1, HOPX, TERT, and Lrig1 [72‒74]. Starting from these assumptions, Sei et al. [75] studied ileal resections from patients with personal and/or familial history of multiple SI-NETs. Abnormal crypts with micronodules of chromogranin A-positive ECs (i.e., aberrant crypt containing endocrine cell clusters [ACECs]) widely distributed in the macroscopic tumor-free mucosa were detected. Moreover, clusters of EECs in ACECs seem to express typical ARISC marker genes, such as BMI1 and HOPX [70]. Globally, these results suggest that ARISCs and ACECs may be precursors of SI-NETs. Hyperplastic and dysplastic proliferative changes of the distal small intestine are far from being systematically defined and further studies are needed to better understand the pathogenesis of SI-NETs, especially considering their prognostic impact in advanced stages (5-year mortality of 21% in metastatic settings) [49].

In the colon and rectum, NETs represent a relatively tiny fraction of the whole neoplastic burden accounting for the 0.4% of all types of colorectal malignancies. In contrast, primary adenocarcinomas with less than 30% of areas of neuroendocrine differentiation are quite common: these neoplasms, despite the expression of neuroendocrine markers, should not be classified as MiNENs according to the WHO 2019 classification [76]. NETs are mostly found in the rectum (54%), followed by cecum (20%) [77, 78]. NETs represent 50% to 70% of all appendiceal neoplasms, accounting for 19% of all gastrointestinal NETs [79].

In the colon and rectum, multifocal small NETs have been described mainly in association with long-standing inflammatory bowel disease (IBD) [80‒83]; these neoplasms are diagnosed by chance in specimens resected because of IBD complications, but they are rarely grossly visible [54].

There is still very little knowledge on the natural history of colorectal NETs and their biologic and morphologic correlates: to date, neither a genetic nor a definite hormonal background has been recognized [54] and the reports about the presence of precursor lesions are conflicting. Some authors [68, 84] report EC hyperplasia in the mucosa surrounding IBD-related NETs, but this finding has not been confirmed by others [80]. EC dysplasia has never been reported or defined in distal GI locations [50]. The association between colorectal NETs and long-standing IBDs suggests that chronic inflammation may represent a stimulus for endocrine cell growth [85]. Moreover, colorectal NET multifocality may be explained by the field effect of chronic inflammation [43]. On the other hand, the mucosal damage induced by long-standing IBDs makes the recognition of precursor lesions in the cancerization field more difficult.

Appendiceal NETs tend to be single and confined to the tip (75% of cases) of the organ. Compared to other sporadic NENs, they occur in younger patients (between 30 and 40 years old [86]) and there is no significant association with chronic IBD. Little is known about precursor lesions of appendicular NETs and most authors [50, 87, 88] agree that appendiceal NENs do not show evidence of EC hyperplasia in the surrounding mucosa. Ultrastructural and immunohistochemical studies [89‒91] recognized an association between appendiceal NETs and Schwann cell processes. On this basis, Moyana and colleagues [68] described discrete subepithelial aggregates composed of neurons, Schwann cells, and neuroendocrine cells, named “subepithelial neuroendocrine complex” (SNC), within the macroscopically healthy mucosa surrounding appendiceal NETs. This finding suggests that appendiceal NETs arise from discrete units, possibly explaining the fact that those neoplasms are usually isolated [68]. The SNCs may sometimes be symptomatic by themselves, possibly representing the pathophysiologic substrate for the historical reports of so-called neurogenic appendicopathy. This last entity consists of the intermingling of a proliferation of pale spindle cells and Schwann cells in the mucosal lamina propria of the appendix, leading to an increased release of neuropeptides such as substance P and vasoactive intestinal peptide, with the consequent onset of acute abdominal pain mimicking acute appendicitis [92]. Since the descriptions of SNCs and neurogenic appendicopathy overlap, it is reasonable that they are part of the spectrum of a same disease.

Pancreatic NETs are generally rather indolent and slow growing and can be associated with hormonal syndromes, whereas NECs are aggressive, fast-growing, usually nonfunctioning neoplasms. PanNECs are sporadic, while about 10–20% of PanNETs are syndromic [2, 93, 94].

The vast majority of inherited PanNETs can be ascribed to four syndromes. The most frequent is MEN1, where multiple nonfunctioning NETs arise virtually in all affected patients. In Von Hippel-Lindau disease, PanNETs are diagnosed in about 5–17% of affected patients [95]. More rarely, PanNETs have been described in patients with NF1 [95] and tuberous sclerosis [95]. Recently, other two hereditary conditions linked to the development of PanNETs have been proposed: glucagon cell hyperplasia and neoplasia [96‒98] and familial insulinomatosis [98] (Table 4).

Many studies investigating possible precursor lesions of PanNETs focused on syndromic tumors, especially MEN1-related [55, 59, 95, 102, 103]. So far, no precursor lesions have been described in sporadic PanNETs [59, 104].

Microadenomas (less than 5 mm) are typical, but not pathognomonic findings in MEN1 [95, 105], and frequently progress into single macrotumors (more than 5 mm) [55, 62, 103, 106]. Concerning the pathogenesis of MEN1-related PanNETs, Hermann and colleagues [107] studied the expression of four transcription factors that govern the development and the differentiation of the pancreas and duodenum in PanNETs with production of specific hormones.

Vortmeyer and colleagues [108] attempted to elucidate the histogenesis of PanNETs by studying typical and atypical structures located in unaffected pancreatic tissue of MEN1 patients. Using a combination of genetic and morphological analysis, the authors provided evidence of a nonislet cell origin of PanNETs. They showed that MEN1 bilallelic inactivation, which is typical of fully developed MEN1-related PanNETs, can be detected in atypical cell proliferation originating in the ductal/acinar system but not in islet tissue. Furthermore, cells of the ductal/acinar system retain embryonic pluripotency in adult pancreatic tissue and they are able to induce the formation of functional islet-like structures [108]. On these bases, the neuroendocrine differentiation process in the context of ductal/acinar system could be altered by early loss of wild-type MEN1 allele in the setting of MEN1. According to this theory, the cells of the ductular/acinar system could be the real target of the “second hit” that leads to the development of MEN1-related PanNETs.

Starting from the observations of Vortmeyer, Anlauf, Klöppel, Perren and colleagues [93, 102, 103, 105], the morphological features and hormone expression pattern of MEN1-associated pancreatic microadenomas have been reappraised. Using a combination of fluorescence in situ hybridization and hormone immunostaining technique, they investigated site of origin of endocrine tumors in the pancreas of MEN1 patients. Since all the cells of these patients carry heterozygous MEN1 mutations, they used LOH as a molecular marker for neoplastic growth and investigated this allelic loss/retention in microadenomas and macrotumors, monohormonal endocrine cells clusters (MECCs), endocrine and exocrine structures entrapped in microadenomas, and in normal and enlarged islets.

Allelic loss of one MEN1 allele was found in 86% of macrotumors and in all microadenomas. Interestingly, n 95% of MECCs [103] (islet-like endocrine cell clusters, usually enriched in glucagon-positive cells) also show allelic loss of one MEN1 allele, thus suggesting that these lesions may represent a minute form of microadenomas [54]. The cells of the glucagon-cell-rich islets show retention of heterozygosity of the MEN1 gene and only have a hyperplastic nature. Moreover, all morphologically normal islets, even if hyperplastic, invariably retained both MEN1 alleles. The retention of heterozygosity was also observed in exocrine tissue entrapped in microadenomas and in the endocrine cells located in the context of microadenomas [103]. Interestingly, when cells aggregate in microadenomas (monoclonal and monohormonal clusters) LOH was detected. This pointed out that hormonally programmed neuroendocrine cells are able to transform into monoclonal NET cells by only one genetic event (i.e., the activation of the germline MEN1 mutation by loss of MEN1 gene locus on the second allele) [102].

Taken together, these findings suggest that MECCs are the forerunners of pancreatic microadenomas in MEN1 [103]. Hence, Anlauf, Klöppel, Perren and colleagues have proposed that MEN1-associated endocrine tumors most frequently arise from the endocrine cell compartment in normal islets. Nevertheless, more rarely, MECCs have also been observed in ducts and hyperplastic islets, where they can progress to microadenomas by 11q13 LOH [103]. This is consistent with Vortmeyer’s results [108].

In conclusion, even if the cell of origin of PanNETs remains controversial, there is evidence that pluripotent stem cells give origin to MECCs within the islet and the ductal epithelium [59, 108]. It is not clear if the sequence of morphologic and genetic changes observed in MEN1-related PanNETs also applies to sporadic PanNETs, but, to date, the aforementioned precursor lesions have not been observed in solitary nonhereditary NETs [62, 95].

Regarding VHL disease, neuroendocrine precursor lesions have not been identified so far [59, 95]. Some studies have however proposed several putative precursor lesions, such as microadenomas, pancreatic ductular proliferation (the so-called nesidioblastosis or ductulo-insular complexes), islet hyperplasia, islet dysplasia, and peliotic change of islets even though there is no consensus among authors [59, 104, 105, 109‒112].

As previously mentioned, it is unclear if the sequence of morphological and genetic changes observed in MEN1-related PanNETs also applies to sporadic PanNETs, and, to date, no precursor lesions have been described in solitary nonhereditary PanNETs [62, 95]. However, given that hormone-producing nonhereditary PanNETs are definitely the most frequent neuroendocrine pancreatic lesions in the general population (50–60% of total of PanNETs [113]), ongoing study is still required.

Indeed, recent studies of gene expression and master regulator analysis, alongside investigation of super-enhancer signatures, have been performed in order to investigate the cell of origin for nonfunctioning PanNETs [114‒118]. In this setting, DNA-methylation profiling has been used [119‒121]. With this type of analysis, Di Domenico and colleagues [118] identified at least 2 cells of origin for PanNETs, alpha-like and beta-like, with two possible evolutionary pathways for PanNET development. Based on epigenetic similarities, PanNETs cluster in alpha-like, beta-like, and intermediate tumors, which show different mutational spectra, stage of disease, and prognosis [122]. Further work is needed to better understand the origin and the therapeutic strategies in intermediate tumors.

Poorly differentiated NECs (gastrointestinal poorly differentiated NECs [GI-NECs]) are high-grade gastrointestinal malignancies characterized by the expression of immunohistochemical markers of neuroendocrine differentiation together with poorly differentiated morphology and a higher proliferation rate than well-differentiated NETs. This category includes both small-cell carcinoma and large-cell NEC [123‒125]. There is a substantial lack of information regarding the potential precursor lesions of GI-NECs. This is probably due to the exceptional rarity of these malignancies, despite the number of new diagnoses is increasing largely after the recent improvement to the classification system. These neoplasms can be most frequently found in the colorectal tract, where the vast majority of poorly differentiated tumors immunoreactive for neuroendocrine markers are “common” adenocarcinomas with areas of neuroendocrine differentiation. Nevertheless, according to the WHO 2019 classification, the diagnosis of NEC can be made when the neuroendocrine component exceeds 70% of the whole tumor mass, while if the neuroendocrine component is between 30 and 70% the malignancy should be classified as MiNEN [124].

To date, the few studies that have explored the parthenogenesis of GI-NECs have suggested that these lesions originate from the same multipotent stem cells, which also may give rise to non-NECs. This hypothesis is supported by the frequent observation of a well-differentiated adenocarcinoma/adenoma sited superficially in the mucosal layer overlying the aggressive invasive neuroendocrine component of colorectal MiNENs [125]. From the molecular standpoint, several authors have reported that colonic NECs share with non-NENs alterations in TP53, CDKN2A, KRAS, and ERBB2 loci [124]. More recently, Karkouche et al. [125] reported 20% BRAF mutation rate in a retrospective 10 patient series with colorectal NECs. Of great interest, as in the case of colorectal adenocarcinomas RAF and RAS mutations have been shown to be mutually exclusive, but interestingly all BRAF-mutated NECs reported in a paper by Kamran and coworkers were microsatellite stable [123]. Hence, it is entirely reasonable to admit that NETs and NECs present different pathogenetic pathways and therefore originate from different types of precursor cells [93, 126‒128]. Due to the rarity of GEP-NENs and the difficulty of analyzing the nonneoplastic counterpart of the neuroendocrine cells of the different organs in which they develop, a carcinogenic cascade for GEP-NENs has not yet been well elucidated.

Wanting to summarize the information known to date and presented in this review in the various sites of the gastrointestinal tract, it emerges how, while the origin of GEP-NEC remains elusive, the existence of MiNEN suggests that GEP-NECs and non-NEN components develop from a common precursor lesion [129]. The whole-genome sequencing study of Kawasaki et al. [122] revealed comparable mutational signatures and driver gene mutation patterns in GEP-NEC and adenocarcinoma organoids, supporting this hypothesis. RB1 mutations were common in GEP-NEC organoids, as were severe chromosomal abnormalities such as whole-chromosome-level LOH and extensive chromosomal deletions and translocations. Interestingly, structural variation was the primary driver of RB1 changes in GEP-NECs, implying that chromosomal abnormalities came first. TP53 and RB1 alterations are prevalent in NECs, and genetically engineered mouse models showed their contribution to NE tumorigenesis, in which loss of Trp53 and Rb1 induced lineage reprogramming to generate NEC [130, 131].

When we focus on the molecular background of precursor lesions of GEP-NENs, the few available data relate exclusively to PanNETs. Therefore, PanNETs, unlike GEP-NECs, lack TP53 and RB1 mutations, as well as other driver gene mutations common in adenocarcinomas arising in the same site, and instead gain DAXX, MEN1, and ATRX mutations [128]. Furthermore, in hereditary forms and their precursors, some germline mutations, such as those in MEN1, CDKN1B, VHL, NF1, and TSC2, confer multifocal PanNET predisposition [132]. If confirmed in the other sites of gastrointestinal tract, these genetic backgrounds might be the source of separate molecular pathways that contribute to the pathobiology differences between GEP-NECs and GEP-NETs.

As far as the metastatic and progression potential of GEP-NENs is concerned, very few studies are available; worthy of note is the work of Azzoni et al. [133] which reports that in ileal NEN, HER-2/neu overexpression plays a role in the carcinogenetic process and by triggering the altered expression of c-Met and MTA-1 may activate the molecular pathways promoting tumor progression and metastasis development.

This review highlights the differences in precursor lesions between different sites within the gastro-entero-pancreatic system and in different scenarios. For some anatomical sites (i.e., esophagus) and in some specific settings (i.e., sporadic vs. inherited neoplasms in duodenum and pancreas), little is still known, possibly due to the rarity of these neoplasms. In other cases (i.e., stomach type 1 and type 2 ECL NETs, sporadic duodenal gastrinoma, MEN1-related duodenal NETs, colon-rectum, VHL-related pancreatic NETs), some evidence of a hyperplasia-dysplasia sequence (or at least hyperplasia or dysplasia as precursor lesions) has been documented. Furthermore, some specific lesions have also been described as putative precursors, namely, ACECs in the jejunum and ileum and MECC for MEN1-related pancreatic NETs, both of which seem to derive from stem cells (Table 5). Notwithstanding this, there are no conclusive studies to date, and these are eagerly awaited in the future.

The authors declare no conflict of interest regarding the present work.

This research was partially funded by a grant from the Italian Health Ministry’s research program NET-2016–02363853. The funding agency had no role in the design and performance of the study.

Conceptualization: Francesca Galuppini, Gianmaria Pennelli, Claudio Luchini, Luca Mastracci, Alessandro Vanoli, Massimo Milione, Federica Grillo, and Matteo Fassan. Methodology and writing – original draft preparation: Rachele Biancotti, Carlo Alberto Dal Pozzo, Paola Parente, Gianluca Businello, Valentina Angerilli, and Francesca Galuppini. Data curation: Rachele Biancotti, Carlo Alberto Dal Pozzo, and Francesca Galuppini. Writing – review and editing: Stefano Realdon, Edoardo Savarino, Fabio Farinati, Anna Caterina Milanetto, Claudio Pasquali, and Roberto Vettor. Supervision: Matteo Fassan. All the authors have read and agreed to the published version of the manuscript.