Risk Factors of Lymph Node Metastasis in Patients with Pancreatic Neuroendocrine Tumors (PNETs)

Background The prognostic value of lymph node metastasis in patients with PNETs is controversial. Understanding the effect of lymph node metastasis on prognosis in pancreatic neuroendocrine tumors is helpful for surgery and follow-up. The purposes of this study are to identify predictors of lymph node metastasis among patients with PNETs and determine its prognostic associations. Methods A retrospective analysis of the surveillance, epidemiology, and end results (SEER) database was performed. Patients with PNETs that underwent surgery and pathologic nodal staging were identified. Logistic regression and Cox regression were performed to identify independent predictors and prognostic factors, respectively. Results Of 1956 patients (age: 56.8 ± 13.4 years, 53.3% males), 748 (38.2%) had lymph node metastasis. On multivariable analysis, tumor located in pancreas head, distant metastasis, and poorly differentiated, undifferentiated, and unknown differentiated histology grades were three independent risk factors of lymph node metastasis. In the entire cohort, lymph node metastasis indicated a worse overall survival (HR: 1.48, 95% CI: 1.17-1.88, p < 0.001) and disease-specific survival (HR: 1.87, 95% CI: 1.41-2.48, p < 0.001) on multivariable analysis. Lymph node metastasis was associated with worse overall (HR: 1.45, 95% CI: 1.08-1.93, p = 0.012) and disease-specific survival (HR: 2.13, 95% CI: 1.48-3.05, p < 0.001) in patients without distant metastasis on multivariate analysis. Lymph node metastasis was also independently associated with worse disease-specific survival among patients in well differentiation (HR: 2.16, 95% CI: 1.35-3.46, p = 0.001) and moderately differentiation (HR: 2.67, 95% CI: 1.28-5.56, p = 0.009) groups on multivariate analysis. Conclusions Tumor located in pancreas head, distant metastasis, and poorly differentiated, undifferentiated, and unknown differentiated histology grades were three independent risk factors for lymph node metastasis. Lymph node metastasis was an independent prognostic factor of worse OS and DSS in patients with tumor located in pancreas head. Lymph node metastasis was an independent prognostic factor of worse OS and DSS in patients without distant metastasis. Lymph node metastasis was an independent prognostic factor of worse DSS in well differentiation and moderately differentiation groups.


Introduction
The neuroendocrine tumors (NETs) originate from neuroendocrine cells and may occur in many organs, including the lung, gastrointestinal tract, and pancreas. NETs occur in approximately 6/100,000 [1]. Among them, gastroenteropancreatic NETs (GEP-NETs) account for 65-75% of the whole body NETs. Pancreatic NETs (PNETs) comprise approximately half of the GEP-NETs [2] and account for approximately <3% of all pancreatic malignancies. As a rare pancreatic neoplasm with an annual incidence of 0.19/100,000-0.32/100,000 [3], the incidence of PNETs has been rising in the United States over the past several decades [4].
The natural history of PNETs is highly variable, with some tumors showing indolent behavior but others displaying an aggressive course, with local invasion and distant metastasis [5]. Several staging systems have been developed to better stratify prognosis in patients with PNETs. The World Health Organization staging system incorporates mitotic count and Ki-67 index to separate patients into three categories [6]. The eighth edition of the American Joint Committee on Cancer (AJCC) tumor-node-metastasis staging system includes primary tumor size, presence of lymph node metastasis on pathologic examination, and presence of distant metastases [7]. Lymph node status is particularly important in PNETs staging systems, because the presence of lymph node metastasis would render a patient stage III according to both the AJCC and European Neuroendocrine Tumor Society staging (ENETS) systems [8]. The prognostic significance of lymph node metastasis in PNETs is controversial. Some researchers have reported a significant association between lymph node metastasis and survival [9][10][11][12][13][14][15], while others showed no association [16][17][18][19]. Lymph node metastasis is an important marker of malignancy and, as such, may influence the type and extent of PNETs surgical management. The purposes of this study are to identify predictors of lymph node metastasis among patients with PNETs and determine its prognostic associations.

Methods
A retrospective analysis of the National Cancer Institute's SEER database was performed for patient diagnosed between 2004 and 2016. The SEER database is a comprehensive database that collects information on several clinical and pathologic aspects of multiple cancers and is approximated to encompass 28% of the US population. Pancreatic neuroendocrine tumors were identified using their ICD-O-3 codes (8013/3, 8150/3, 8151/3, 8152/3, 8153/3, 8155/3, 8156/3, 8240/3, 8241/3, 8242/3, 8243/3, 8245/3, 8246/3, 8247/3, 8248/3, and 8249/3). Patients without positive histological diagnosis and those that did not undergo surgical resection were excluded. Patients without available information regarding their lymph node status were also excluded. Only patients that met criteria for pathologic nodal staging (PN staging) were included based on the SEER variable CS lymph Nodes Eval (CS Lymph Nodes Eval code 3 and code 6). In this way, patients without pathological nodal staging were excluded from the analysis. A flowchart demonstrating patient selection can be viewed in Figure 1. All tumors were primary tumors. The study was exempted from Institutional Review Board approval, due to SEER's use of unidentifiable patient information.

Statistical Analysis
Univariate and multivariable logistic regressions were performed to ascertain the possible factors associated with the presence of lymph node metastasis. Univariate and multivariable analyses using the Cox proportional hazards model were performed to ascertain the prognostic role of nodal metastasis, with both overall survival (OS) and disease-specific survival (DSS) as endpoints. All statistical tests used two-tailed p values, and 0.05 was set as the threshold for significance. Statistical analysis was performed using the IBM SPSS Statistics v.22.0.0 (IBM Corp, Armonk, NY).

Results
Overall, 1956 patients were identified, of which 748 patients (38.2%) had lymph node metastasis. The median number of lymph node examined was 10 (mean 12:8 ± 8:5), and the median number of metastatic lymph nodes was 2 (mean 3:5 ± 3:8). Most tumors were located in the body and tail of pancreas (1016 (51.9%)), followed by the head of pancreas (632 (32.3%)) and the overlapping lesion and other specified parts of pancreas (308 (15.7%)). Most patients were males (1043 (53.3%)) and 301 (15.4%) patients had synchronous distant metastasis at diagnosis. Patients were followed for a median period of 49.1 months (range 0-155 months). Detailed demographic, clinical, and pathologic features of the study cohort are listed in Table 1.

Subgroup Analysis of Tumor Location, Distant
Metastasis, and Histology Grade. Subgroup analysis of tumor location, distant metastasis, and histology grade were executed to ascertain whether the diagnosis of lymph node metastasis was independent prognostic factor in these groups of patients. According

Discussion
This study demonstrated that tumor located in pancreas head, distant metastasis, and poorly differentiated, undifferentiated, and unknown differentiated histology grades were three independent risk factors associated with the diagnosis of lymph node metastasis in patients with PNETs, suggesting that a high degree of suspicion for lymph node metastasis should be present in these patients. Lymph node metastasis was also identified as an independent predictor of worse overall and disease-specific survival and in patients with PNETs. And lymph node metastasis was an independent prognostic factor of worse overall survival and diseasespecific survival in patients without distant metastasis. Furthermore, lymph node metastasis was an independent prognostic factor of worse disease-specific survival in patients with the well-differentiated and moderately differentiated groups.
Data concerning the incidence of lymph node metastasis in patients with PNETs is multifold. Hill and colleagues retrospectively analyzed 728 patients with PNETs and identified lymph node metastasis in only 24.9% using the SEER database [20]. The rate of lymph node metastasis that was identified in our study was 38.7%, which is in line with previously published data by Hashim and colleagues, who reported a rate of 37.6% [5]. Moreover, a 33.3% rate of lymph node metastasis was reported by the PET/CT screening imaging [21]. However, our rate of 38.7% may be an overestimation of the true incidence of lymph node metastasis due to the exclusion of low-risk patients that did not undergo nodal sampling due to a low degree of suspicion.
This study identified tumor located in pancreas head, distant metastasis, and poorly differentiated, undifferentiated, and unknown differentiated histology grades as three independent predictors of lymph node metastasis in patients with PNETs. As is well known, pancreatic resections are associated with significant morbidity, and there is interest in 6 Gastroenterology Research and Practice   Gastroenterology Research and Practice minimizing the impact of surgery [22]. Hence, enucleation and central pancreatectomy in selected patients, especially in those with small tumor size and tumor not close to the main pancreatic duct, could be considered as alternative management strategies to radical surgery. For instance, Tsutsumi and colleagues suggested that regional lymphadenectomy should be performed for all PNETs ≥ 15 mm [23], while Jiang and colleagues suggested that regional lymphadenectomy in PNETs ≤ 25 mm is not necessary [24]. Curran and colleagues found there was no lymph node metastasis in patients with PNETs ≤ 1 cm in low tumor grade, so they suggested regional lymphadenectomy is unnecessary in these patients [25]. However, assessing lymph node status preoperatively by radiological examination or endoscopy cannot be accurate and facile; preoperative determination of the risk of lymph node metastases attains critical importance. Therefore, patients without known lymph node status but fulfilling these factors should be considered for pancreatectomy with regional lymphadenectomy instead of tumor enucleation or central pancreatectomy without nodal staging.
The results of this study also demonstrated that lymph node metastasis was correlated with worse OS and DSS on multivariate analysis. Moreover, the number of more lymph nodes metastasis has a certain influence on OS and DSS, which suggests that standard regional lymphadenectomy may guide accurate staging, thus having a good influence on the prognosis.
The subgroup analysis results of multivariate analysis by tumor location grouping were also worse OS and DSS. The OS and DSS were statistically significant only in PNETs patients without distant metastasis, in either univariate or multivariate analyses. This suggested that lymphadenectomy is not necessary in PNETs patients with distant metastasis. On multivariate analysis based on the grouping of tumor differentiation grade, there was no significant difference in OS among each subgroup. However, OSS only showed significant difference between the well differentiation group and the moderately differentiation group. The effect of lymph node metastasis to prognosis is controversial in patients with PNETs. Some reports have associated lymph node metastasis with a shorter OS and DSS [9][10][11][12][13][14][15]. However, several other studies have reported that lymph node metastasis has no effect on survival [16][17][18][19]. It is difficult to interpret the real reasons of the contradictory results reported by different institutions. The inconsistency may be in part due to inadequate lymph node sampling rates during pancreatic resections and various measured outcome (OS, 5-year OS, DFS, DSS, etc.). Therefore, we interpreted this data to recommend that regional lymphadenectomy may be a safer choice for even low-grade patients. Meanwhile, lymph node metastasis is related to pathological features such as lymphatic, vascular, and neural invasion [26]. So, it is possible that with the extension of follow-up time, patients with high-grade PNETs will recurrence and die, so OS and DSS will become statistically significant.
There were certain several limitations for this study. The design of the study was retrospective. Large databases contain patients from different institutions and time periods.
This can lead to patient selection bias, coding ,errors and missing information. Other variables, such as performance status, comorbidities, mitotic count, Ki-67 index, and surgical information (duration, blood loss, and postoperative complication), were not captured in the SEER database. In addition, there was no information on whether adjuvant therapies (chemotherapy, targeted therapy, and endocrine therapy) were given to patients with PNETs, which may have contributed to a better analysis. Despite these limitations, large database like SEER is the best database available today. We are in agreement of prospective study to better understand the risk factors of lymph node metastasis in PNETs and their effects on prognosis through the inclusion of more impact factors.

Conclusion
In conclusion, this study was performed on PNETs patients who underwent surgical treatment in the SEER database. The research results confirmed that tumor located in pancreas head, distant metastasis, and poorly differentiated, undifferentiated, and unknown differentiated histology grades were three independent risk factors for lymph node metastasis. This suggested that regional lymphadenectomy should be carefully considered when choosing surgical treatment for these patients. In addition, lymph node metastasis is an independent prognostic factor of worse OS and DSS in patients with tumor located in the pancreas head. Lymph node metastasis was an independent prognostic factor of worse OS and DSS in patients without distant metastasis. However, lymph node metastasis was an independent prognostic factor for worse DSS in well differentiation and moderately differentiation groups. Prospective studies are required to more comprehensively understand the risk factors of lymph node metastasis and determine criteria in performing regional lymphadenectomy in patients with PNETs.

Data Availability
The data used to support the findings of the study "Risk factors of lymph node metastasis in patients with pancreatic neuroendocrine tumors (PNETs)" have been deposited in the SEER database. The method of obtaining data has been described in detail in the manuscript.