The overriding role of surgery and tumor grade for long‐term survival in patients with gastroenteropancreatic neuroendocrine neoplasms: A population‐based cohort study

Abstract Background Gastroenteropancreatic neuroendocrine neoplasms (GEP‐NENs) comprise a heterogeneous disease group. Factors that affect long‐term survival remain uncertain. Complete population‐representative cohorts with long‐term follow‐up are scarce. Aim To evaluate factors of importance for the long‐term survival. Methods and results An Observational population‐based study on consecutive GEP‐NEN patients diagnosed from 2003 to 2013, managed according to national guidelines. Univariable and multivariable survival analyses were performed to evaluate overall survival (OS) and to identify independent prognostic factors. One hundred ninety eligible patients (males, 58.9%) (median age, 60.0 years; range, 10.0–94.2 years) were included. The small bowel, appendix, and pancreas were the most common tumor locations. The World Health Organization (WHO) tumor grade 1–3 distributions varied according to the primary location and disease stage. Primary surgery with curative intent was performed in 66% of the patients. The median OS of the study population was 183 months with 5‐ and 10‐year OS rates of 66% and 57%, respectively. Only age, WHO tumor grade, and primary surgical treatment were independent prognostic factors for OS. Conclusion The outcomes of GEP‐NEN patients are related to several factors including age and primary surgical treatment. WHO tumor grading, based on the established criteria, should be routine in clinical practice. This may improve clinical decision‐making and allow the comparison of outcomes among different centers.

survive for several years. 5,[8][9][10] In terms of the prevalence in this group of patients, GEP-NENs are the most common gastrointestinal malignancy after colorectal cancer. 1 This translates into a considerable number of patients requiring long-term surveillance and treatment.
An enhanced understanding of the biology of this disease and the development of novel diagnostic approaches (i.e., molecular detection, receptor-based approaches, and metabolic positron emission tomography [PET]) and treatment options (i.e., biological or targeted treatments and improved surgical approaches) have increased the complexity of the clinical management of neuroendocrine tumors. 3,11 Some of these efforts have likely contributed to an improved survival rate of subgroups of GEP-NEN patients. 5,8 This study aimed to evaluate the long-term survival in a population-based cohort of consecutive GEP-NEN patients treated in routine practice who were classified according to current grading and staging criteria of the given time period. 12,13

| MATERIALS AND METHODS
This observational study included all GEP-NEN patients treated at a single hospital that covers a geographically well-defined area. The manuscript follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement. 14

| Study population
The study population comprised unselected consecutive patients with GEP-NENs diagnosed at the Stavanger University Hospital between 2003 and 2013, as reported previously in greater detail. 4 This is the only hospital that serves a regional area of approximately 380 000 inhabitants; thus, patients consisted of a population-representative cohort from the Norwegian southwest coast and a mixed urban-rural area.
We excluded nine patients with an unknown primary tumor location. Moreover, three patients with esophageal and two patients with bile duct primary tumors were excluded due to small sample sizes. This study does not include mixed neuroendocrine neoplasm (MINEN). Also, Goblet cell carcinoma as a specifically defined entity was not included.
Thus, the final study population included 190 consecutive patients.

| Clinical workup for primary treatment decision-making
Routine evaluation of patients encompassed clinical examination, necessary blood tests including tumor marker detection (i.e., chromogranin A [CgA]), and standard oncologic imaging (i.e., multiphase computed tomography [CT], magnetic resonance imaging [MRI], and somatostatin receptor-scintigraphy [SRS]), as described in available guidelines. [15][16][17][18] Positron emission tomography (PET) imaging ( 68 Ga-DOTA-somatostatin analog-PET/CT) and metabolic PET-imaging were not routinely performed. Transthoracic echocardiography was used for suspected carcinoid heart disease. Endoscopy, including endoscopic ultrasound (EUS) and video capsule endoscopy, were available if indicated. Diagnostic step sequences and appropriate adjustments were made according to the clinical presentation, for example, in symptomatic patients, adequate diagnostic steps were performed to locate the primary tumor and to evaluate the disease stage. In contrast, if a gastroenteropancreatic neuroendocrine tumor (GEP-NET) was incidentally discovered during surgery for a tentative diagnosis (i.e., indication for surgery) other than GEP-NET, a postoperative evaluation was performed.
2.3 | Classification and staging according to morphology In the present study, the most current criteria 19 (Table 1) were applied by two pathologists (D.L. and E.G.) during the independent re-grading of the tumors and included the novel distinction between grade 3 neuroendocrine tumors (NET) and grade 3 neuroendocrine carcinomas (NEC).

| Statistics
The statistical calculations were performed using SPSS version 25 for Mac (IBM, Armonk, NY) and R 3.6.3. 26 The R-package "relsurv" version 2.2-3 was used for the relative survival calculation. 27 In the descriptive analyses, categorical data are reported as numbers and percentages, and continuous data are reported as medians and ranges or interquartile ranges (IQRs). Non-parametric tests were used for comparisons between subgroups. OS was estimated by the Kaplan-Meier method, and the log-rank test evaluated differences between subgroups. Cox proportional hazard analyses were performed to assess independent predictive factors of OS and DSS. Factors with a p-value < .2 in the univariable analysis were included in the multivariable models, and these selections were run with a backward stepwise model. The results of the Cox regression analyses are expressed as hazard ratios (HRs) with 95% confidence intervals (CIs).
All tests were two-sided, and a p-value < .050 was considered statistically significant.

| RESULTS
Of  Table 2. Small bowel tumors were primarily localized in the ileum. Among the 33 patients with pancreatic NEN, four patients had clinically functional insulinomas.
While males showed a slight predominance of 58.9%, no significant differences were observed between sexes in the distribution of the primary tumor location. However, in patients whose primary tumors were in the appendix, the median age of 30.4 (range, 10.0-84.9) years was significantly lower than the median age of 62.5 (range, 19.5-90.5) years in patients whose primary tumors were in other locations (p < .001).
Disease characteristics are displayed as WHO tumor grades 1-3 according to the primary tumor location in Figure 1 and UICC disease stage in Figure 2 Surgical treatment according to the UICC disease stage is shown in Figure 2(B), and primary surgery was performed significantly more often in patients with stage I-III disease than in patients with stage IV disease (p < .001).
The 30-days mortality varied among treatment groups, with the lowest mortality of 2.4% observed in patients who underwent surgery with curative intent (Table 2). Although the early mortality was higher after palliative surgery and highest when no primary surgery could be performed, these differences did not reach statistical significance.
In contrast, the 90-days mortality was significantly different among groups (p = .031), with threefold increased mortality (16.7%) in patients who underwent palliative surgery. Patients who did not undergo surgery had a fourfold higher mortality rate (22.5%) than patients treated with curative intent (a 90-day mortality rate of 6.4%) ( Table 2). Survival was significantly better in patients who underwent primary surgery at the time of diagnosis (p < .001) (Figure 3(B)). Moreover, tumor grade was significantly associated with OS, and a poor prognosis was seen in those with WHO grade 3 tumors (p < .001) (Figure 3(C)). In the univariable survival analysis, primary tumor location was statistically significant (p < .001), as better survival was observed in patients with tumors in the appendix, duodenum, or rectum, while worse survival was observed in patients with primary colon tumors (Figure 3(D)).
As shown in Table 2, survival was similar in both sexes (p = .381). However, younger patients (i.e., median age ≤ 60 years) had significantly better OS (p < .001) than those above 60 years of age, and patients with incidentally discovered tumors had better survival (p = .003) than symptomatic patients. In addition, survival differed significantly (p < .001) among UICC stages, and a poor prognosis was observed in stage IV patients.
In the Cox multivariable survival analysis, only age, WHO tumor grade, and primary surgical treatment were independent predictors of survival (Table 3). Regarding tumor grade as a prognostic factor, a  (Table 3).
DSS rates of 73.1% and 68.1% were observed at 5 and 10 years, respectively. A median DSS was not calculated for the whole cohort.
During the follow-up, 57 patients (30.0%) died from proven advanced GEP-NEN disease, and the best DSS was observed in patients with appendiceal and duodenal NENs; this was in contrast to a relatively dismal prognosis in patients with colon-NENs, as indicated by the OS in Figure 3(D). Intermediate survival was achieved in patients whose primary tumors were located in the rectum, small intestine, stomach, and pancreas, and a better DSS was observed in patients with tumors of rectal and small intestinal origins. Notably, even patients with pancreatic-NENs had at least a 50% median survival at 10 years.
In Figure 4, the estimated OS (Kaplan-Meier plot) is depicted alongside the calculated relative survival curve for age and sexadjusted populations without a diagnosis of GEP-NENs. The curves are relatively comparable during the first 2 years. After that, the relative survival curve plateaued, which shows that the excess deaths attributed to the disease mainly occurred during the first 2 years.

| DISCUSSION
GEP-NENs comprise a heterogeneous group of tumors. 1,3 Many factors including age, symptomatic disease, primary tumor location, disease stage, tumor grade, and primary surgical treatment are relevant for prognostic prediction. 6,8,10,28,29 In this study, the most common tumor origins were the small intestine (31.6%), appendix (25.3%), and pancreas (17.4%). This is in agreement with a Canadian population-based study, which indicated the small bowel as the predominant location, although pancreas NENs were not included in that report. 30 In a recent national survey from Iceland that studied a national population comparable to our hospital's  (Continues) regional population (i.e., ≈380 000), the tumor origin distribution was slightly different, with 23.1% occurring in the small bowel, 30.8% in the appendix, and 9.6% in the pancreas. 6 These authors reported an incidence of 3.85/100 000 from 2000 to 2014, which is lower than the incidence of 5.83/100 000 reported in our previous study 4 and by others 5 during a similar time period.
The relative indolent nature of this group of tumors is reflected in a promising long-term prognosis, with 5-and 10-year DSS rates of 73.1% and 68.1%, respectively, and 5-and 10-year OS rates of 66% and 57%, respectively. However, as shown in this study, survival is related to several factors many of which cannot be moderated. Moreover, the increasing gap (see Figure 4) between the OS curve and the calculated relative survival curve, particularly during the first 2 years, may indicate that an excess mortality risk in GEP-NEN patients compared with the age-and gender-adjusted general population, is mainly attributed to the early time period after a GEP-NEN diagnosis is confirmed.
Decreased survival has been associated with older age, the presence of symptoms, primary tumor location, tumor grade, and disease stage. 6,8,10 This study confirms the prognostic value of age for both OS and DSS, with a 4-to 5-fold increased risk of death in patients older than 60 years (i.e., median age). While survival also varied according to primary tumor location, only the appendix (with an excellent prognosis) and colon (with a more dismal prognosis) sites were significantly associated with prognosis. However, the primary tumor location did not retain its independent association with prognosis in the multivariable analysis, which was also the case for the disease stage. This is partly in contrast to recent population-based studies, 5,8 which reported that sex, tumor differentiation, stage, and primary site were independent predictors of OS. However, attention should be paid to the differences among these studies, as they included patient populations with tumor locations outside the gastrointestinalpancreatic sites, contained reporting bias due to a national registry, and considered different definitions to describe tumor differentiation.
Thus, caution is warranted when making direct comparisons.
The prognostic relevance of tumor grading, as first proposed in Europe more than a decade ago 31 and eventually embraced by clinicians worldwide, [32][33][34] is also supported by observations in this study.
As shown in our research, the distribution of grades varied according  F I G U R E 4 Overall survival (OS) and relative survival of the patient cohort. Notably, the two curves have a similar pattern, particularly during the early time period of 2-3 years Tumor grade not only serves as a prognostic factor but may also serve as a predictive factor for the selection of GEP-NEN patients for chemotherapy treatment, 34 particularly in cases with a welldifferentiated morphology despite a high Ki67 index (>55%). 38 In line with long-standing treatment principles, surgical treatment was used whenever feasible. 10,21,22,39 The high proportion of patients (78.9%) who were surgically treated, including 66.3% treated with curative intent in the total cohort, is partially explained by the assumption that removal of the primary tumor even in stage IV patients may be beneficial. 40 53 Moreover, the application of observed epigenetic modifications to serve as potential prognostic biomarkers, or even as therapeutic targets, may present novel opportunities in the future. 54 In addition, metabolic grading using FDG-PET 55 and the development of artificial intelligence (AI) imaging approaches (i.e., tumor heterogeneity) seem to be promising prognostic stratification tools. 56 Finally, knowledge-based supportive care for GEP-NEN patients, including those with a lengthy course trajectory, and even those with advanced disease, should not be overlooked. 57

ACKNOWLEDGMENTS
The authors want to thank all the health professionals who contributed to the care of these patients. This study was financially supported in part by an unrestricted research grant (#2016) provided by the CarciNor, which is the Norwegian patient advocacy association for patients with neuroendocrine tumors.

CONFLICT OF INTEREST
The authors have stated explicitly that there are no conflicts of interest in connection with this article. All authors have reviewed and consented to the submitted final version of this manuscript.

AUTHORS' CONTRIBUTIONS
All authors had full access to the data in the study and take responsibility for the integrity of the data and the accuracy of the data analy-

ETHICAL STATEMENTS
This study was approved by the Institutional Review Board (Study # 2011/1659-16) according to the general guidelines provided by the Regional Committees for Medical and Health Research Ethics. This approval includes a consent to publish the results .

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.