Site‐specific trends in gastroenteropancreatic neuroendocrine neoplasms in Bavaria, Germany

Abstract Introduction Neuroendocrine neoplasms (NEN) are rare and heterogeneous epithelial tumors, occurring throughout the body. For gastroenteropancreatic (GEP)‐NEN, rising incidence rates were reported for the last decades, with underlying causes remaining largely unexplained. We evaluated NEN trends stratifying by their histologic subtypes. Methods Incident cases of GEP‐NEN diagnosed between 2005 and 2019 were retrieved from the prospective, population‐based Bavarian Cancer Registry. GEP‐NEN were divided in their histologic subtypes, that is, neuroendocrine tumors (NET) G1, NET G2/G3, other NET versus small‐cell neuroendocrine carcinoma (NEC), large‐cell NEC, and other NEC. We calculated annual age‐standardized incidence rates (ASIRs) per 100,000 persons for the total of GEP‐NEN, NEN histologic subtypes, and tumor sites. We used an annual percentage change (APC) approach including a joinpoint analysis to investigate NEN incidence trends. Results ASIR of GEP‐NEN rose from 2.2 in 2005 to 4.8 in 2019, characterized by a significant increase until 2012 (APC 2005–2012: 10.1%), followed by modest rise (APC 2012–2019: 1.5%). In the last decade, this increase was mainly driven by the rise of NET G1 and G2/G3, while incidence for NEC declined. Over the study period, ASIR increased significantly for all GEP‐sites except the colon. APCs were largest for the stomach, the appendix, the pancreas, and the rectum. Conclusions This study found a significant increase in the incidence of GEP‐NET. Though this development may partially be attributable to the increased use of advanced detection techniques and changes in NEN classification, further research should also focus on the identification of NEN risk factors.


| INTRODUCTION
Neuroendocrine neoplasms (NEN) are a rare and strongly heterogeneous group of tumors showing variable clinicopathologic and biologic behavior. 1They originate from neuroendocrine cells, 2 which express the vesiclecharacteristic proteins synaptophysin and chromogranin A, 3 hallmark neuroendocrine markers used for entityspecific immunohistochemical diagnosis as well as for the monitoring of disease progression.The majority of NEN is slow growing and indolent for a long period.This may lead to a delay in diagnosis and presentation of patients in a metastatic stage of the tumor, when a curative treatment is not possible anymore. 4,5Symptoms vary according to the tumor site. 1 Localization, histology, and tumor size are the best predictors for disease development and serve as the basis for treatment decision. 6Following the distribution of neuroendocrine cells, NEN can occur throughout the body.Approximately 65% of all NEN are localized in the gastroenteropancreatic (GEP) system, about 25% in the lung and the bronchi (excluding small cell neuroendocrine carcinoma), and about 10% in other sites, for example, the thymus or the ovaries. 1 In this study, we focus on NEN located in the GEP system.
The European Neuroendocrine Tumour Society (ENETS) issued standard of care guidelines for pathology in 2009.Since then, profound changes in the classification have been made.8][9] After initial validation studies, NEN were divided for the first time into neuroendocrine tumors (NET) and neuroendocrine carcinoma (NEC) in the 4th edition of the World Health Organization (WHO) classification of tumors 2010. 10The latest classification of GEP-NEN was published in the 2019 WHO classification of tumors (5th edition). 11It differentiated GEP-NEN according to their proliferation characteristics into well-differentiated NET, graded Grade 1 (G1)-Grade 3 (G3), poorly differentiated NEC, and mixed neuroendocrine-non-neuroendocrine tumors (MiNEN).The 7th edition of the TNM classification of malignant tumors of 2009 12 included NEN for the first time, widely adapting ENETS proposals, though some differences regarding NEC as well as pancreatic and appendiceal NET persisted. 13The growing debate on NEN gave rise to further, continuous changes in the classification of NEN.Beyond other, these consisted of a change in consent on malignancy of specific histologic codes ('behavior code' change) and the inclusion of new histologic codes, for example, the functioning pancreatic neuroendocrine tumors (PanNETs), finally implemented in the first revision of the International Classification of Diseases for Oncology (ICD-O3). 14evious epidemiological research on GEP-NEN reported continuously rising incidence rates for the last decades, 6,[15][16][17][18] predominantly for countries of Western Europe, North America, and from Japan.The United States, to give just one example, experienced a strong increase of GEP-NEN rates, which rose 3.6-fold between 1973 and 2007. 15Uncertainty exists whether the observed increase in NEN incidence is due to improved detection strategies of these tumors, altering risk factors, or changes in tumor biology.Published data from Germany and other European countries have often focused on retrospective analysis of oncological studies, 19 mostly obtained from referral centers, or the analysis of specific subgroups. 20,21Due to national restrictions and data protection rules, large population-based data such as provided by the Surveillance, Epidemiology, and End Results (SEER) Program in the United States 5,22 are often not available for Europe.In particular, regarding the more recent changes in the WHO and TNM classifications, population-based incidence data are needed.
This study aims to analyze time trends in the incidence of GEP-NEN in a population-based cancer registry between 2005 and 2019 in Bavaria, Germany, stratified by age, NEN subtype, tumor site, and stage.Analyzing time trends in subgroups can contribute to the exploration of the increase in NEN incidence observed in many countries.

| METHODS
Data were retrieved from the population-based Bavarian Cancer Registry, which at present covers a population of more than 13 million residents.Being established in 1998, it records all cancer cases in Bavaria reported by inpatient and outpatient facilities. 23We included all incident cases of malignant NEN, diagnosed between 2005 and 2019, as for most GEP entities, completeness of coverage was achieved for these years.Completeness of coverage, which is estimated by the ratio of mortality to incidence, is a data quality measure of cancer registries. 24Internationally, it serves as an indicator of data reliability in cancer registrybased research.The percentage of death certificate only (DCO) cases continuously declined, from 2009 onwards falling below 15%.
Annual age-standardized incidence rates (ASIRs) per 100,000 persons with corresponding 95% confidence intervals (CI) 25 were calculated for the total of GEP-NEN and its histologic subtype (NET G1, NET G2/G3, other NET, SCNEC, LCNEC, and other NEC), as well as for each GEP site, using the European standard population 26 and the average Bavarian population of the respective year.To evaluate GEP-NEN time trends, we applied an annual percent change approach (APC) including joinpoint analysis and also calculated average APCs (AAPCs) for the whole study period 2005-2019. 27APCs as well as AAPCs were estimated with corresponding 95% CI for all GEP-NEN, their histologic subtypes, GEP sites, and stages, according to Union for International Cancer Control (UICC).We used a tiebreak method 28 by adding the minimum value for the ASIR in the time series to each component to approximate the APCs and AAPCs of NET G2/G3.Specific sites were selected for further analysis, including stratification by age group, pre-defined in 0-39, 40-59, 60-79, 80+ years of age, and UICC stage.The statistical analyses were performed using SAS (SAS Institute, Inc.), version 9.4 and R (R Foundation for Statistical Computing), version 4.1.1.

| RESULTS
Between 2005 and 2019, a total of 9236 cases of GEP-NEN were identified from the registry data (Table 1).The most frequent GEP-NEN site was the small intestine (30.8%), followed by the pancreas (  2B).Incidence of SCNEC, LCNEC, and other NET remained rather stable.The inverse trends of NET and NEC were reflected in the analyses of GEP sites, stratifying NEN cases by histologic subtype (Figure S1).
T A B L E 1 GEP-NEN by site: Descriptive analysis of cases in Bavaria.Over the study period, NEN incidence increased for all UICC-stages (Table 3).The increase declined stepwise from UICC Stage I to UICC Stage IV.The strongest rise (18.7% [95% CI 13.0%; 24.8%]) was found for incidence of early stage NEN (UICC I), the lowest for NEN in an already metastasized stage (UICC IV, 3.6% [95% CI 2.2%; 5.0%]).Figure S2 shows the distribution of stages according to GEP sites.

Esophagus
Over the study period, NEN ASIR increased for the age groups (40-59, 60-79 and 80+ years) for all sites (Figure S3).For the appendix, an increase in NEN incidence was also found for the youngest age group (0-39 years)  F I G U R E 1 Age-standardized incidence of GEP-NEN, stratified by type of NEN.ASIR, age-standardized incidence rate (European standard population); GEP-NEN, gastroenteropancreatic neuroendocrine neoplasm; NEC, neuroendocrine carcinoma; NET, neuroendocrine tumor.For relevant data see Table S2.
from 0.1 to 0.8 per 100,000 residents (Figure S3).Of these, 123 were below the age of 18. Except for the appendix, incidence of GEP-NEN increased with age.

| DISCUSSION
In this analysis of population-based data from the Bavarian Cancer Registry, an increase in ASIR of NEN was observed for most GEP sites during the study period.This increase was highest for NEN of the stomach, followed by those of the appendix, pancreas, and rectum.The increase in ASIR was observed for all stages; however, it declined stepwise from Stage I to Stage IV NEN.Rising trends of ASIR were observed for age groups 40 years and older.For NEN of the appendix, an increase of NEN incidence was additionally found in the youngest age group (0-39 years).8,22 NEN incidence rates varied according to gender and ethnicity, geographic region, and tumor site. 1,6,22Few studies investigated NEN in Germany 29 focusing either on NETs in general or on one selected GEP site. 30,31One study from the Netherlands investigated NEN incidence trends by histologic grade using the 2010 WHO classification.Korse et al. investigated ASIR between 1990 and 2010 for the total of NEN and for different sites. 32To our knowledge, hitherto no study has investigated ASIR of GEP-NEN stratified by histologic subtype according to the latest WHO classification of 2019.
The underlying reasons for the increase in GEP-NEN incidence still remain contentious.Some studies suggest that the increase in incidence could be explained by the incidental detection of asymptomatic NEN during routine check-up or screening. 5,33The increasing use of advanced and more sensitive diagnostic imaging techniques in clinical practice might have enhanced detection, 4,33 which would likely lead to an increase especially of early stage tumors. 18Our findings support this hypothesis, as the highest increase in ASIR was found for early stage NEN (UICC I), which are characterized by a small size.Especially small and well-differentiated rectal NEN were often found during screening colonoscopy. 33Similarly, we found a rise for NEN of the rectum but not the colon (excluding appendix, F I G U R E 2 Age-standardized incidence of (A) NET and (B) NEC, both stratified by histologic subtype.ASIR, agestandardized incidence rate (European standard population); G, Grade; LC, large cell; NEC, neuroendocrine carcinoma; NET, neuroendocrine tumor; SC, small cell.For relevant data see Table S3.
F I G U R E 3 Age-standardized incidence of GEP-NEN, stratified by tumor site: stomach, small intestine, colon, appendix, rectum, and pancreas.ASIR, age-standardized incidence rate (European standard population); GEP-NEN, gastroenteropancreatic neuroendocrine neoplasm.For relevant data see Table S4.which usually cannot be detected by colonoscopy).In Bavaria, case numbers of screening colonoscopies increased between 2010 and 2019, simultaneously to the increase in colorectal NEN in the same period.Appendiceal NEN are often incidentally diagnosed during surgical interventions, accounting for 0.2-2.3% of appendectomies. 34Our data indicate that ASIR of appendiceal NEN have increased over the study period as incidence rates of appendicitis have been stable in Western countries since 1990.Appendicectomy rates even decreased. 35The rise in appendiceal NEN may thus be a real increase.Also, the appendix was the only GEP site, in which almost 50% of the patients were aged below 40 years.This fact is striking, as in general, the age group under 40 years is characterized by far less doctor's visits compared to persons of higher age.Some studies conclude that the risk of developing NEN may be primarily attributable to genetic factors, however, of a hitherto unknown gene, since no correlation with the menin-gene or any other gene is known for appendiceal NEN. 36A gene-environment-interaction might also provide the explanation for the varying NEN incidence in different geographic regions. 37Findings of Hassan and colleagues on higher prevalence of NEN in patients whose first-degree relatives developed any cancer 38 suggest a possible hereditary basis.A link between NEN and cancer in general was also suggested by other studies, 39,40 supposing a shared cancer predisposition.However, the concurrence with other cancers might also be due to age, as NEN cases predominantly occur in higher age groups. 41Traditional risk factors such as obesity, alcohol consumption, smoking, and diabetes are assumed to increase the risk for NEN and/or GEP tumors. 41,42Continuously rising diabetes rates 43 and increasing trends of obese and overweight individuals 44 could be observed for the last three decades in Bavaria and thus might have influenced the development of NEN.In contrast, a decline in prevalence rates of smokers occurred in Bavaria for the last decade and the consumption of alcohol decreased. 45Thus, these risk factors alone may not explain the increase of incidence in GEP-NEN.

UICC stage
Our study has some limitations.Due to changes in classification, assessing trends in NETs remains challenging. 37We included the preceding morphology codes for the histologic GEP-NEN subtypes and accounted for known behavior code changes, for example, the change of behavior code of the histology 8240 from 1 ("borderline malignancy") to 3 ("malignant") for appendiceal NEN.However, we cannot ensure that appendiceal tumors of "borderline malignancy" with the morphologic code 8240 were completely reported to the Bavarian Cancer Registry.Also, we cannot generally rule out misclassification as shifts in the classification of NEN were introduced in the recent years, especially in 2013 when the first revision of the ICD-O3 was published. 14The term mixed adenoneuroendocrine carcinoma (MANEC) was introduced at first in the 2010 WHO classification of tumors of the digestive system, 4th edition. 10Before that date, MANEC might have been classified as NEC.The change in coding practice may have additionally contributed to the decline of reported NEC.A process of transition in registration practices is likely to have occurred prior to the introduction of the latest WHO classification of NEN in 2019, 11 leading to the successive adoption of the novel subtype NET G3 and the decline of NEC in consequence.Moreover, until the late 90s, it used to be common clinical practice to categorize only metastasized NEN as NEC.Alterations in coding and the emergence of NEN classifications might have increased awareness of NEN in clinicians and pathologists, further enhanced by a broad implementation of improved immunohistochemical techniques. 32These developments may contribute to the explanation of the increase in NEN incidence. 15Strengths of the study are the populationbased analysis for a populous German federal state with more than 13 million residents-exceeding the populations of countries like Belgium, Sweden or Portugal-as well as the stratification by tumor site.
Analyzing a large, unselected population, this study found a significant increase in the incidence of NEN, which in the last decade was mainly driven by rising rates of NET.This increase may-at least partially-be attributable to the improved detection of tumors and changes in diagnostic classification.However, it may also be associated with changes in risk factors related to Western lifestyle and the increase of associated diseases.Further research should focus on identifying those risk factors with the highest attributable risk for GEP-NEN to target preventive efforts accordingly and further enhance awareness towards NEN.

T A B L E 3
Trends in incidence of GEP-NEN by UICC stage.
19.8%).NEN were slightly more prevalent in men (54.0%) than in women (46.0%), with the exception of the appendix with almost two thirds (61.0%) occurring in females.Patients presented at first diagnosis with an average age of 61 years (±16.2years) of age.Mean age was lowest for the appendix (43 ± 21.4 years).The largest proportion of GEP-NEN was diagnosed in UICC Stage IV (21.2%).A high percentage of cases had no staging (UICC Stage X: 44.5%).For most sites, NEN cases increased with age.However, regarding appendix, almost half of the patients were <40 years old (48.5%).Within this age group, 66.8% of appendiceal NEN occurred in women.Between 2005 and 2019, ASIR for all GEP-NEN significantly increased from 2.2 per 100,000 residents in 2005 to 4.8 in 2019 (Table 2, Figure1).Rates distinctly rose be- Trends in incidence of GEP-NEN by histologic subtype.
T A B L E 2