Introduction
The 2017 WHO classification of pancreatic neuroendocrine neoplasms (PanNENs) [
1] and subsequently the 2019 WHO classification of digestive system tumors [
2] stratified high-grade neuroendocrine neoplasms (NENs) into neuroendocrine tumor (NET) G3 and neuroendocrine carcinoma (NEC), mainly based on histology and genetic profile [
3,
4]. While NETs G3 generally have retained the organoid pattern and the wild-type
TP53 and
Rb1 gene profile of the well-differentiated NENs, NECs display usually a disorganized pattern, significant cellular atypia, and genetic abnormalities in the
TP53 and
Rb1 genes. Since this stratification of high-grade digestive system NENs has clinical implications and seems also to be pertain to pulmonary NENs [
5,
6] and other non-digestive organs [
7‐
10], investigations on high-grade NENs, in particular the NETs G3, have received great attention. Despite these studies, data on NET G3, especially in non-pancreatic organs, are still scarce [
11,
12]. Moreover, despite the WHO definition of criteria distinguishing NET G3 from NEC, diagnostic difficulties can occur in each case, notably in liver metastases.
We studied NETs G3 in a consultation series of 1513 NENs. The NETs G3 were identified using the criteria of the WHO classification 2019. As many NETs G3 presented as metastasis with a clinically unclear origin, the tumors were tested by a panel of markers with known site specificity, e.g., ISLET-1, CDX2, serotonin, and TTF-1. The special aims of this study were to investigate (1) the prevalence of NET G3 regarding origin and metastasis; (2) the characterization of NET G3 regarding Ki67 index and p53, Rb1, and SST2 expression; (3) the value of immunohistochemistry in the differential diagnosis of NET G3 versus NEC; and (4) the agreement between referral diagnosis and final diagnosis.
Discussion
This study evaluated 1513 NENs and identified 130 NETs G3 of various origin, applying the criteria of histological differentiation and proliferative activity as defined in the 2019 WHO classification of digestive system tumors [
16].
Previous studies on well-differentiated NENs with a Ki67 index above 20% and deriving from various organs identified the pancreas as the most frequent site of origin in 46–65% of the cases [
9‐
11]. We confirm this data on NET G3 in the pancreas with a prevalence rate of 42%. Next in frequency in our cohort was the lung with 20%, a figure that is considerably higher than the 8% reported by Velayoudom-Cephise et al. [
9]. The discrepancy is probably due to that our study included much more metastases than the other investigation. Other sites of origin that were included in previous studies [
10,
11] as well as in our study were the stomach, the ileum, and the rectum. The available data (although difficult to extract from the literature [
9,
10]) and our data indicate that gastric NETs G3 account for approximately 11%, which is higher than that of the ileum and rectum. In our cohort, NETs G3 from the ileum and the rectum accounted for about 5% and 3%, respectively. These low rates are in line with the rarity of respective cases reported [
10,
18‐
20]. Among the very rare sites of NET G3 were the duodenum and papilla of Vater, ascending the colon, presacral region, breast, and prostate, with a rate of 1% each. Taken together, these prevalence data clearly indicate that the NETs G3 are unevenly distributed in the body. They parallel the prevalence of NETs G1/G2 in the pancreas, lung, stomach, and colon, but not in the ileum, appendix, and rectum. Noteworthy is this discrepancy in the ileum, as it again points to the special role that ileal NETs play among the remaining NETs in terms of histological, hormonal, and genetic features [
21].
Among all NETs of the pancreas and the lung, the NETs G3 of this series were found in a percentage of 13% and 15%, respectively, and accounted for 9% of all NENs. These numbers are high compared with the data presented by Rindi (2.1%) and by Kim (9.7%) in pancreatic NENs [
22,
23] and our own study on pulmonary NENs (12%) [
5]. The reason for these discrepancies may be that, first, it must be assumed that NETs G3 are overrepresented in this study because of the case selection bias inherent in all consultation series. Second, included in the number of our NETs G3 are not only primaries but also the respective metastases which significantly exceeded (in a ratio of 1: 2–3) the number of primaries in the pancreas and the lung. The finding that metastases of pancreatic and pulmonary NETs G3 were more frequent than primaries suggests that some primary NETs G1/G2 progress to NETs G3 during metastatic spread. Similar conclusions may be drawn from other studies, which showed that there is often an evolution of the proliferative activity in NETs of the pancreas and the lung during the metastatic process [
24,
25]. An exception from these observations in pancreas and lung NETs G3 seems to be the stomach NET G3, since the percentage of primary NETs G3 (mostly type 3 gastric NET) was higher (70%) than that of metastatic NETs G3. This is probably due to their early detection and removal by endoscopy [
8].
The reason for the assumed progression of NET G1/G2 to NET G3 from the pancreas and the lung may be a genetic change that occurs during the metastasizing process, particularly to the liver. We found that the immunohistochemical expression of p53 and Rb1 was normal in NETs G1/G2 and commonly present in NECs. In NETs G3, abnormal expressions of p53 and Rb1 were found in 8% and 7%, respectively. The small difference between NETs G3 and NETs G1/G2 could be an indication that few NETs G3 probably have or develop
TP53 and
RB1 gene abnormalities that relate them to NECs. This hypothesis was also recently discussed by Pelosi et al. based on genetic studies on subsets of NETs and NECs in the gastrointestinal tract and the lung [
26,
27]. Since we only discovered p53 and Rb1 abnormalities in NET G3 metastases and not in primaries, we suspect that these genetic abnormalities mainly develop during the metastasizing process, particularly to the liver.
It has been shown that SST2 is expressed in most digestive organ NETs including NETs G3 [
15] while only in about 50% in lung NETs [
5,
6,
15,
21,
28]. Here, we found that NETs G3 of digestive organ tumors express SST2 somewhat less frequent than the respective NETs G1/G2, but still distinctly more than NECs. In the lung, however, NETs G3 turned out to be the NENs with the lowest SST2 expression rate (13%), since not only NET G1/G2 (44%) but also NECs (32%) express SST2 at a distinctly higher rate [
29].
Infrequent sites of NET G3 recorded in our series and/or reported in the literature included Vater papilla, ascending colon, and presacral region. No NETs G3 have so far been observed by us and others in the esophagus, jejunum, appendix, and hepatobiliary tract [
10,
18]. We also diagnosed a NET G3 in a liver metastasis from primary tumors of the prostate and breast, respectively. In both cases, the metastases presented as solid tumor tissue with endocrine appearance. In the case of the breast metastasis, its origin was revealed by the positivity for the markers ER and PgR. In the case of the other metastasis, an origin from the prostate was confirmed by the subsequent resection of a prostate carcinoma which had a solid neuroendocrine component negative for the prostate marker NKX3.1. Since in the prostate and the breast the classification of NENs does not include a NET G3 category [
30,
31], the discussion of the presented neuroendocrine breast and prostate liver metastases with NET G3 morphology serves only to point out that metastases with NET G3 features can occur in the liver, that derive from primaries in organs, in which NETs G3 have so far not been observed and discussed.
In the consultation requests, the issue of tumor origin played an important role in the tumors identified as NET G3, since most NET G3 samples came from metastatic sites. Helpful in solving this issue was the use of a panel of transcription factors known as markers for gastroenteropancreatic or pulmonary NETs such as CDX2 [
32], ISLET-1 [
33,
34], TTF-1 [
35], and serotonin [
36‐
38]. This differential testing also revealed that NETs G3 are similar, or identical, in their expression rates to NET G1/G2. Noteworthy is that serotonin was a very indicative marker for ileal NETs, when it was diffusely positive and accompanied by a nested histology with peripheral palisading of the tumor cells.
Another frequent request was the differentiation of NET G3 from NEC, MiNEN, and occasionally NET and other non-neuroendocrine carcinomas. Well- or poorly differentiated histology, i.e., organoid versus diffuse architecture, rare and small necrosis versus common geographic necrosis and moderate versus high-grade cellular atypia are the main defining criteria for NET versus NEC and most MiNEN in our study, as well as in other recent studies [
10,
11,
23]. These criteria found a high degree of agreement (84%) by pathology experts, who re-evaluated 196 high-grade NENs with regard to their distinction in NET G3 and NEC [
11]. However, assessment of histological differentiation may be difficult in biopsies [
39] and occasionally also in resection specimens [
40]. This is also true for the Ki67 index, which as a median value is a guiding parameter and significantly separates NET G3 from NEC (30% versus 60% in our study). The individual values, however, may overlap with the values of the other category. In these difficult cases, as could be demonstrated in our series, p53 and Rb1 immunohistochemistry was helpful for distinguishing NETs G3 from NECs. These data that are in line with previous studies in the pancreas [
15,
40] and the lung [
5,
6] also include NETs G3 from other organs (see Table
3) which, except for a case report on an ileal NET G3 [
19], have so far not been examined.
Matching the referral diagnoses with the final diagnoses revealed that the distinction of NET G3 from NEC achieved the lowest concordance. This concerned particularly the consultation cases, received till 2017, and may be explained by the definition of NEC provided in the 2010 WHO classification of digestive tumors, in which NECs were distinguished from NETs by a Ki67 index above 20%. Thus, all NENs with a Ki67 index > 20% were classified as NEC. This problem was only solved with the release of the 2017 WHO classification of endocrine tumors, in which the NET G3 category was defined for the pancreatic NENs. Since then, the NET-NEC issue is slowly losing its importance as a diagnostic problem, as we have noticed a slight decrease in the number of inconsistent NET-NEC diagnoses in the years 2018 to 2021 compared to the period of 2009 to 2017 (5.8 versus 6.7 cases per year).
In conclusion, the 130 NETs G3 identified in our consultation series mostly originated from the pancreas and lung and showed a high metastatic potential that often manifested itself as metastasis to the liver. It seems that a low number of NET G3 acquire an abnormal expression of p53 and RB1 during the metastatic process, indicating mutations in the respective genes. Our case review also showed that the potential of NETs to develop into NETs G3 is site-specific, with very few NETs G3 in the ileum where NETs G1/G2 are otherwise frequent. Primaries or metastases from NENs with the features of NET G3 can occasionally emerge from organs that are known to develop neoplasms with neuroendocrine differentiation, which however are difficult to classify and compare to the NENs of the digestive system or lung [
30,
31]. The main diagnostic problem encountered in our consultation series, particularly in biopsies, is the distinction of NET G3 from NEC, for which the application of the markers p53 and Rb1 proved to be very helpful.
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.