Survival of patients with appendiceal neuroendocrine tumors: whether they benet from right hemicolectomy?

Background: Current guidelines recommend right hemicolectomy for appendiceal neuroendocrine tumors (A-NETs) patients with lymph node (LN) metastasis. However, prognosis of these patients is favorable, and the prognostic impact of metastatic LN is controversial. Objective: The study aims to evaluate the prognostic factors of A-NETs, and explore whether right hemicolectomy/ more extended procedure (RHCM) improves prognosis compared to less extended than right hemicolectomy (LRHC). Methods: Patients with A-NETs were identied from the Surveillance, Epidemiology, and End Results (SEER) database. The Fine-Gray proportional hazards model was established to calculate subdistribution hazard ratios of prognostic factors. A propensity score matching was performed to balance intergroup differences between the LRHC and RHCM groups, and survival difference between the after-matched groups was tested using the Gray test. Subgroup analyses were also conducted. Results: In the multivariate analysis, histological type and distant metastatic status were associated with prognosis, while tumor size and nodal status were not. After propensity score matching, the patients’ characteristics were well balanced. RHCM did not confer survival benets in the whole after-matched patients or any subgroup. Conclusions: Metastatic LN does not signicantly impact prognosis, and RHCM fails to improve prognosis compared to LRHC. Therefore, the current “LN-decided” surgical procedure may not be suitable for patients with A-NETs.


Introduction
Appendiceal neuroendocrine neoplasms (A-NENs) are rare, but they represent the most common tumor of the appendix, and their incidence has increased moderately in recent years [1][2][3][4]. Different from other neuroendocrine neoplasms, A-NENs are rarely accompanied with carcinoid syndrome, and their discovery is usually accidental after surgery for acute appendicitis, with around 0.5% of all appendectomies [5].
Some studies have reported lymph node (LN) metastasis is a hallmark of aggressive behavior and a predictor for poor prognosis, while others have reported favorable prognosis even in patients with LN metastasis, raising doubts about the prognostic value of metastatic LNs [1,6,7]. However, previous studies mostly included patients with both well and poor differentiated A-NENs, the former was classi ed into appendiceal neuroendocrine tumors (A-NETs) and the latter appendiceal neuroendocrine carcinomas (A-NECs) in the 2019 WHO Classi cation of Digestive System Tumors, without considering different biological characteristics between these two types [6,8]. It has been proved that the A-NETs are of indolent characteristic and favorable prognosis compared to the A-NECs [9,10].
Oncological right hemicolectomy is suggested for patients with LN metastasis or suspected LN metastasis, such as tumor size ≥ 2cm, to eliminate tumor dissemination [1,11]. However, many studies now argue that this "LN-decided" surgical procedure fails to improve outcome [12][13][14][15][16]. Previous studies are mainly retrospective studies and case series. Considering the rarity and favorable prognosis of A-NETs, a randomized trial is di cult to conduct because of the enormous sample size required and the long follow-up period needed. Clinicians sometimes face dilemma that whether or not to perform a reoperation or an extended operation after initial appendectomy [12,14].
Using a large, population-based database, our study focuses on A-NETs rather than A-NECs or the mixture of both. We aim to evaluate the prognostic factors, especially LN status, of A-NETs, and explore whether right hemicolectomy or more extended surgery improves prognosis compared to less than right hemicolectomy.

Availability of data and materials section
The Surveillance, Epidemiology, and End Results (SEER) database collects information on cancer from registries sponsored by the US National Cancer Institution. In current, the SEER database consists of population-based cancer registries that cover 34.6% US population. The database collects data (e.g., patient demographics, primary tumor location, tumor morphology, diagnosis, and rst course of treatment) and tracks the life status of patients.
The permission to access the database was obtained with the reference number 12005-Nov2019. The SEER database is an open-access cancer database that only contains deidenti ed patient data. Therefore, this study was exempted from the approval of the review board of the A liated Jinhua Hospital, Zhejiang University School of Medicine. Patients with A-NETs from 2000 to 2015 were identi ed by using the SEER Stat software. Patients diagnosed before 2000 were excluded because some variables were not collected on the database until 2000; patients diagnosed after 2015 were excluded to ensure adequate follow-up period.
Patients met the following criteria were excluded: patients with incomplete documentation, such as tumor size and LN status, were excluded; patients with multiple primary tumors were excluded to eliminate the survival impact from other tumors; patients with survival time less than 1 month were excluded because these patients are at risk of death of perioperative complications.

Data collection
The age and year at diagnosis, gender, race, histological type, differentiated grade, tumor size, depth of invasion, status of LNs and distant metastasis, surgical procedure, causes of death and survival months were retrieved from the SEER database. Race was classi ed into white, black and other. Histological type was classi ed into pure and mixed, and the former included malignant carcinoid tumor, enterochroma n cell carcinoid, neuroendocrine cell carcinoma and the latter included composite carcinoid, adenocarcinoid tumor, atypical carcinoid tumor. Differentiated grade was classi ed into well and moderate differentiation. As previous studies reported, depth of invasion was obtained by combining the data of the collaborative stage and the extent of disease, resulting in three categories: invasion of the lamina propria (LP), invasion or through the muscularis propria (MP/TMP), or invasion through the serosa and adjacent structures (TS). [18]. LN status was categorized into no lymph nodes examined (NLNE), lymph nodes examined with negative lymph nodes (LNN) and lymph nodes examined with positive lymph nodes (LNP). According to "RX Summ -Surg Prim Site" values in the database, surgical procedure was divided into two categories: right hemicolectomy/ more extended procedure (RHCM) and less extended than right hemicolectomy (LRHC). The RHCM involves removing the cecum, the total ascending colon, the hepatic exure, a portion of the transverse colon, and part of the terminal ileum, along with fat and lymph nodes (the exact number of retrieved LNs is not de ned because of the characteristic of our population-based study and non-consensus of speci c LN number in current guidelines), or more extended surgery; the extent of the LRHC is lesser than the RHCM, including appendectomy, ileocolectomy, and partial and segmental colon resection.

4.Data analysis
Demographic and clinical characteristics of the cohort were reported as medians with interquartile ranges (IQR) for continuous variables, and frequencies for categorical variables. Continuous variables were compared using Student t test or Mann-Whitney U test. Categorical variables were compared using chisquared test or Fisher exact test.
The prognosis of patients with well and moderately differentiated A-NETs is favorable, and the incidence of non-cancer speci c death (non-CSD) cannot be simply ignored when conducting a survival analysis [19]. Traditional Cox proportional hazard models only considered two statuses of outcome (e.g. alive and death), while competing risk models considered the presence of competing events (e.g. non-CSD). Thus, the competing risk model, rather than the Cox proportional hazard model, was applied in the study.
The time to cancer speci c death (CSD) was calculated from the date of diagnosis to the date of death of cancer; the time to non-CSD was calculated from the date of diagnosis to the date of death of other causes. CSD was regarded as the outcome event, and non-CSD was regarded as the competing event.
Univariate and multivariate analyses were conducted. In the univariate analyses, the cumulative incidences of CSD were calculated, and the differences were tested using the Gray tests. Variables signi cant in the univariable analyses then entered into the multivariable analysis. In the multivariate analysis, subdistribution hazard ratios (SHRs) were calculated to predict the association of variables with CSD (patients with tumor invasion of LP were excluded in both univariate and multivariate analyses because none of them dead of CSD).
To reduce biases from confounders and achieve balance between the RHCM and LRHC groups, a propensity score matching was performed. Based on demographic and clinical characteristics (i.e., age, gender, race, histological type, differentiation, tumor size, depth of invasion and status of LNs), patients were matched with a 1:1 ratio using the nearest neighbor method (caliper set to 0.05) [20]. Absolute standardized differences were calculated to evaluate pre-and after-matched balance, and a "love plot" was plotted to present them [21]. The differences less than 10% support intergroup balance, and 0% is considered no bias. In the after-matched patients, the univariate analysis between the RHCM and LRHC groups was conducted to identify whether RHCM rendered a survival bene t compared to LRHC.
In addition, subgroup analyses of the before-matched patients strati ed by age, gender, race, histological type, differentiation, tumor size, nodal status, depth of invasion and metastatic status were conducted to explore whether RHCM improved outcome in a certain group, and a "forest plot" were plotted to present the results. Considering some patients did not undergo lymphadenectomy or had no LNs retrieved, another subgroup analysis (univariate and multivariate competing risk analyses) was conducted in patients with LNs retrieved.
The difference was considered to be statistically signi cant when the 2-side P-value was less than 0.05. R software (version 3.6) was applied for data analysis, and R packages survival, survminer, forestplot and cobalt were used.

Patient characteristics
In total, 901 patients were identi ed from the SEER database, with 658 patients undergoing LRHC and 243 patients undergoing RHCM. In the LRHC group, over a half of (68.39%) A-NETs were less than 10mm, while only 9.57% were more than 20mm; in the RHCM group, nearly a half (46.09%) of A-NETs were more than 20mm, while only 22.22% were less than 10mm. Only 20.97% of the patients underwent lymphadenectomy when they were performed LRHC, and most of the retrieved LNs were negative. In comparison, most patients (91.77%) underwent lymphadenectomy in the RHCM group, and a third of the group (33.33%) had metastatic nodes (Table 1). In patients proved to have negative LNs, 56.57% (142/251) underwent RHCM, while in patients proved to have metastatic LNs, 73.64% (81/110) underwent it.
Intergroup differences indicated clinicopathological characteristics were not balanced between the two groups. In the RHCM group, A-NETs were poorer differentiated grade, deerper invasion, and had higher proportions of LN and distant metastasis.

Survival analysis
The endpoint date of follow-up was November 2019, and the median follow-up duration was 148 months (IQR: 138-167). In total, 14 patients (1.55%) and 18 patients (2.00%) suffered from CSD and non-CSD, respectively ( Figure 1). Note that more patients suffered from non-CSD than CSD (13 patients suffered from non-CSD versus 6 patients suffered from CSD) in the LRHC group, while there were more CSD than non-CSD in the RHCM group (8 patients suffered from CSD versus 5 patients suffered from non-CSD).
In univariate analyses, age, histological type, differentiated grade, tumor size, depth of invasion, status of LNs and distant metastasis were all signi cantly associated with CSD ( Table 2). In patients undergoing lymphadenectomy (361 patients), the patients with LNP had worse prognosis than those with LNN, and the 10-year CSD cumulative incidences was 0.96% in former while there was no CSD happened in latter during the 10-year period (P=0.54).
Surgical procedure, which did not reach statistical signi cance in the univariate analysis, also entered into the next multivariate analysis to explore the surgical impact on survival when controlling confounders. The surgical procedure and the signi cant variables in the univariate analyses then entered into multivariate analysis. Only histological type and status of distant metastasis kept statistically signi cant in the multivariate analysis (P all < 0.01), while surgical procedure, LN status and tumor size were not (Table 2).
Results of subgroup analyses of the patients with LNs retrieved (361 patients) were shown in Appendix Table 1. Age, race, histological type and distant metastasis were signi cantly associated with CSD in the univariate analyses, and entered into the multivariable analysis (Surgical procedure did not reach statistical signi cance in the univariate analysis, but it also entered into the multivariate analysis). The results of the entire cohort and the subgroup of patients with LNs retrieved are consistent. Histological type and status of distant metastasis kept statistically signi cant in the multivariate analysis (P = 0.01 and 0.02, respectively), while surgical procedure, LN status and tumor size were not.

Patient characteristics and survival analyses after propensity score matching
Patients in the LRHC and RHCM groups were matched 1:1, and there were 136 patients in each aftermatched group. Table 3 showed the demographic and clinical characteristics after matching, and P values of each variable were more than 0.05. Appendix Figure 1 showed the total absolute standardized difference was less than 10%, although the absolute standardized differences of some variables were more than 10%. In the after-matched patients, there was no CSD happened in both LRHC and RHCM groups in the 5-year follow-up period, and the 10-year CSD cumulative incidences were 0.00% in the LRHC group and 0.75% in the RHCM group, and there was no signi cant difference of CSD between the two groups (P = 0.67) (Appendix Figure 2).

Discussions
In the population-based study, we proved that the prognosis of A-NETs was extremely favorable, and the impact of non-CSD should be emphasized. In survival analysis, histological type and status of distant metastasis were associated with prognosis, while tumor size and LN status were not. Right hemicolectomy or extended surgery failed to render survival bene ts compared to less than right hemicolectomy, even in the subgroups of patients with metastatic LN and with ≥ 2.0 cm tumor size.
A-NENs are tumors of heterogeneous entities, and prognoses are various among different types [22]. Previous studies have shown that well and moderate A-NENs, namely A-NETs, are of indolent clinical course and the prognosis of them is favorable, which was proved in our study with only 1.55% patients suffered from CSD in the entire cohort. In addition, our study found non-CSD made up a similar proportion (2.00%) as CSD. To our knowledge, there are only one study reported the impact of non-CSD in patients with appendiceal cancer [23]. The study of Jingjing Wu et al. focused on the entire appendiceal cancer and variables associated with non-CSD, while our study paid attention to the A-NETs and variables associated with CSD. Our results suggested non-CSD should be considered in clinical situations. For example, when making clinical decisions for patients who are old or accompanied with severe underlying diseases, appendectomy, rather than right hemicolectomy, seems to be more suitable. It was also the reason why our study applied the competing risk model, rather than the Cox proportional hazard model, because the latter underestimates the cumulative incidences of CSD [24,25].
Histological type is associated with prognosis and serves as an important predictor for survival [16]. Hsu C et al. compared characteristics and outcomes of malignant carcinoid tumor, goblet cell carcinoid and composite goblet cell carcinoid-adenocarcinoma from the American National Cancer Database, and found malignant carcinoid tumor had the best prognosis and composite goblet cell carcinoidadenocarcinoma had the worst [22]. However, with deeper understanding of goblet cell carcinoid, it was classi ed to goblet cell adenocarcinoma in the 2019 WHO classi cation of tumors because of its minor neuroendocrine component [17]. Therefore, we excluded goblet cell adenocarcinoma in our study, and also con rmed the prognosis of pure A-NETs was much better than the mixed type. Distant metastasis is another well-established predictor for poor prognosis, and our study found MRHC only rendered survival bene ts in the subgroup of distant metastasis.
Tumor size and LN status used to be regarded as main prognostic features of A-NETs. Tumor size ≥ 2 cm is the indicator for right hemicolectomy in both the National Comprehensive Cancer Network (NCCN) and the ENETS guidelines because the risk of LN metastasis increases dramatically in ≥ 2 cm A-NETs [1,11]. However, neither the tumor size nor the nodal status was signi cantly associated with CSD when controlling confounders in our study, and there are some studies supporting our ndings. Rault-Petit et al. conducted a national study including 403 patients with nonmetastatic A-NETs, 26 of whom was surgically proved to have metastatic nodes, and only 1 patient had recurrence during the follow-up [6]. Recently, Daskalakis et al. conducted a meta-analysis including A-NENs patients, nding there was no signi cant difference in disease-speci c survival between patients with and without lymph node metastasis (10-year disease-speci c survival: 99.2% in patients without LN metastasis versus 95.6% in patients with LN metastasis) [26].
On the other hand, results of some previous studies are different from our ndings, and suggest metastatic LN is the predictor for poor outcome. Note that these previous studies included patients with both A-NETs and A-NECs [27]. When focusing on patients with well differentiated A-NETs, some previous studies reported favorable prognosis even in LN metastasis patients. Lambert et al. reported 114 children and adolescents with well differentiated A-NETs, some of who had metastatic nodes (the exact number was not provided), and all patients were alive and disease free at the last follow up [7]. Combining results of ours and previous studies, we concluded that metastatic LNs should not be regarded as a signi cant predictor for poor prognosis in patients with A-NETs.
In 1987, Moertel et al. studied 150 patients with pure appendiceal carcinoid tumors, and found recurrences occurred very late and only happened to patients with ≥ 2 cm tumors [28]. Therefore, they suggested right hemicolectomy was appropriate only for young patients with ≥ 2 cm tumors. As mentioned in the Background section, current guidelines for A-NETs were "LN-guided". According to current guidelines, RHCM is routinely suggested for patients with LN metastasis or high probabilities of LN metastasis, such as vascular or lymph vessel invasion and mesoappendiceal in ltration > 3 mm, to eliminate metastatic lesions and achieve long-term survival. However, our study suggested RHCM failed to render survival advantage compared to LRHC in both univariate and multivariate analyses. Our nding is supported by results of some previous studies. Crown et al. analyzed a multi-center cohort (61 patients) and found more radical surgical procedures, such as colon resection, did not improve long-term outcomes [12]. Bamboat et al. conducted a retrospective review of 48 appendiceal carcinoid patients, and they failed to nd right hemicolectomy confer a prognostic bene t over appendectomy for tumors greater than 2.0 cm [29]. But, as far as we know, there has been no high-quality evidence comparing the prognostic bene ts between LRHC and RHCM.
Considering the low incidence and long-term survival of A-NENs, it is di cult to conduct a randomized controlled trial to compare survival advantage between LRHC and RHCM. In our study, a propensity score matched analysis was performed to achieve intergroup balance, and surprisingly, we failed to observe survival advantage of RHCM over LRHC in the after-matched patients. Lukas Marti et al. hypothesized the reason for non-bene t of RHCM in appendiceal carcinoma was that LNs metastasis was limited, and ileocaecal resection was su cient for removing metastatic LNs [13]. But his hypothesis was based on LN metastatic patterns in colon cancer patients and needs further con rmation, and we hypothesized that this non-bene t is "blamed" for the current "LN-decided" surgical strategy. Results of the subgroup analyses that patients with positive LNs failed to bene t from the RHCM further strengthened this hypothesis.
On the other hand, colon resection has been reported to increase operation time, blood loss, postoperative complication rates and length of stay, and the postoperative complications such as anastomotic leakage will have a negative impact on both short-term and long-term outcomes [12,30]. Reoperation also serves as an option for patients with local or distant recurrence during the follow-up period, and previous studies reported patients underwent reoperation could achieve long-term survival [7]. Thus, we concluded that LRHC, such as ileocaecal resection or appendectomy, was curative for A-NETs with LN metastasis, and MRHC did not improve prognosis. In addition, long follow-up term is vital for A-NET patients because we observed happenings of CSD in both RHCM and LRHC groups even after 10 years.
Although our study is based on large population in the USA, it is also characterized by the following limitations. First, the SEER database does not contain records of vascular or lymph vessel invasion, margin states, Ki-67 index and mitotic rate, which may be associated with prognosis. Also, information about post-operative complications, which has been reported to have an in uence on survival, is not available in the database. Moreover, the treatment records are incomplete and there is no detailed information regarding chemotherapy and radiotherapy. Finally, this study is a retrospective study and biases such as selection bias and treatment bias are inevitable in our study.

Conclusion
In conclusion, the study has demonstrated the prognosis of A-NETs is extremely favorable, and non-CSD should be taken into consideration when making clinical decisions. Current guidelines suggest performing RHCM for patients with LN metastasis or suspected LN metastasis, while in our study, LN metastasis was proved not to be associated with poor outcome in A-NETs. Therefore, it indicates the current "LN-decided" surgical strategy was inappropriate, and LRHC was su cient for patients with A-NETs, regardless of nodal status.

Clinical Practice Points
The prognosis of well and moderately differentiated A-NETs is extremely favorable, and non-cancer speci c death should be taken into consideration when making clinical decisions Age, race, histological type and status of distant metastasis were associated with prognosis in the Fine-Gray proportional hazards model, while tumor size and LN status were not.
In our study, 42.34% and 73.21% patients underwent right hemicolectomy or extended surgery in the negative lymph node group and positive lymph node group respectively, proving current guidelines are "LN-decided".
Right hemicolectomy or extended surgery failed to render survival bene ts compared to less extended than right hemicolectomy, indicating current "LN-decided" surgical strategy is inappropriate.

Availability of data and materials
The data sources of the manuscript are derived from the de-identi ed Surveillance, Epidemiology, and End Results (SEER) database (SEER ID: 12005-Nov2019).

Ethics approval and consent to participate
The SEER database is an open-access cancer database that only contains deidenti ed patient data. Therefore, this study was exempted from the approval of the review board of the A liated Jinhua Hospital, Zhejiang University School of Medicine. a. Patients with invasion of LP were excluded from univariate and multivariate analyses because none of them dead of cancer-specific death, and analyses could not be performed in such a situation.