Clinical significance of the preoperative main pancreatic duct dilation and neutrophil-to-lymphocyte ratio in pancreatic neuroendocrine tumors (PNETs) of the head after curative resection

Background The present study aimed to investigate the prognostic significance of preoperative main pancreatic duct dilation and the neutrophil-to-lymphocyte ratio (PD-NLR) in pancreatic neuroendocrine tumors (PNETs) of the head after curative resection. Methods Sixty-four consecutive PNETs of the head that underwent curative resection were included in the study. Preoperative main pancreatic duct dilation (PD) was defined as a pancreatic duct dilation greater than 3 mm before surgery. Patients with both PD and an elevated NLR (> 3.13), with PD or elevated NLR, or neither of these characteristics were allocated a PD-NLR score of 2, 1, or 0, respectively. Univariate, multivariate and Kaplan-Meier analyses were used to calculate overall survival (OS) and disease-free survival (DFS). Results Preoperative PD-NLR score was correlated with tumor size (P = 0.005), T-stage (P = 0.016), lymph node metastasis (P <  0.001), distant metastasis (P = 0.005), type of hormone production (P = 0.006), perineural invasion (P = 0.014), and WHO classification (P <  0.001). Patients with a high PD-NLR score had a significantly poor OS and DFS relative to those with a low PD-NLR score (P <  0.001). In the multivariate analysis, PD-NLR score was an independent predictor of OS and DFS for PNET of the head (both P <  0.05). In the analyses of the various subgroups, preoperative PD-NLR score was also a predictor of OS and DFS. Additionally, the survival predictive capability of PD-NLR score was superior to that of WHO classification. Conclusions Despite the retrospective nature and small sample size of the present study, the results suggest that preoperative PD-NLR score can serve as an independent prognostic marker of early survival in patients with PNETs of the head undergoing curative resection. Further large prospective studies are necessary to validate our findings.


Background
Pancreatic neuroendocrine tumors (PNETs) are a heterogeneous group of neoplasms and account for approximately 1-2% of all pancreatic neoplasms and 7.0% of all neuroendocrine tumors [1]. PNETs can be classified as either functional or nonfunctional. The majority of PNETs, ranging from 60 to 90% of all PNETs, are nonfunctional. Complete surgical resection of a PNET has been suggested to be the only potentially curative treatment. In particular, for PNETs localized to the head of the pancreas, resectability criteria are stricter than are those for the body or tail, indicating that an inoperable tumor invades the celiac axis, superior mesenteric artery or the retroperitoneum extensively [2]. However, PNETs are considered more indolent tumors than are tumors of the exocrine pancreas and are associated with better long-term survival rates [3][4][5]. The clinical course of the disease is usually characterized by an indolent history with a 5-year survival rate exceeding 60% [6].
Several studies have found that multiple host-related factors are associated with survival in PNETs. Intrinsic tumor characteristics used to predict disease progression include tumor size, stage and grade, Ki-67 indices, and lymph node involvement [7][8][9]. Information regarding these factors is generally useful, but most of these factors are determined only after surgery. Therefore, it is necessary to identify potential prognostic indicators available before surgery. Increasing amounts of evidence suggest that inflammatory cells are an essential component of the tumor microenvironment and play a role in tumor progression [10,11]. An elevated neutrophil-to-lymphocyte ratio (NLR) has been shown to be correlated with advanced stages and poor prognoses in a variety of human tumors, including PNET [12], hepatocellular carcinoma [13], breast cancer [14], and pancreatic cancer [15]. Moreover, recent studies have reported that the earliest consistent imaging finding of pancreatic cancer is main pancreatic duct dilation [16,17]. In addition, Gupta N reported an association between final histologic diagnosis of pancreatic malignancy and main pancreatic duct diameter as determined by endoscopic ultrasound [18]. Nanno Y et al. observed that patients with PNET who presented with main pancreatic duct dilation (PD) showed an aggressive clinical course [19].
When considered together, main pancreatic duct dilation and the neutrophil-to-lymphocyte ratio (PD-NLR) may represent a potential predictor of clinical survival. The aim of the present study was to investigate the prognostic significance of PD-NLR score in PNETs of the head after curative resection. Furthermore, we aimed to compare the predictive capability of PD-NLR score for survival with that of other predictive models.

Study population
A total of 64 patients who underwent curative resection for PNET of the head were retrospectively reviewed from September 2002 to July 2016 at the First Affiliated Hospital, Zhejiang University School of Medicine. The diagnosis of PNET was made based on standard histologic criteria. Information on the following characteristics was collected for each patient: pathologic features including tumor size, lymph nodes and stage, gender, surgical approach, symptoms, and patient age. Tumor stage was classified according to the 8th edition of the American Joint Committee on Cancer (AJCC) staging system, and the grade of each PNET was determined based on the 2010 WHO classification of NETs of the GEP system. The retrospective measurement data on the main pancreatic duct were obtained through magnetic resonance cholangiopancreatography (MRCP) or enhanced CT by at least two experienced radiologists. According to the literature, main pancreatic duct dilatation (PD) was defined as a main pancreatic duct with a maximal diameter greater than 3 mm [18,20]. PD was observed in 19 of 64 patients (29.7%). The NLR was calculated by dividing the absolute neutrophil count by the absolute lymphocyte count. Receiver operating characteristic (ROC) curve analysis showed that the area under the curve (AUC) of NLR was 0.776 and that the optimal cut-off value was 3.13. These values indicated that NLR showed high sensitivity and high specificity in predicting overall survival. As a result, the NLR values were categorized into two groups: ≤ 3.13 and > 3.13. High NLR (> 3.13) was observed in 16 of 64 patients (25%). Patients with both PD and an elevated NLR (> 3.13), PD or an elevated NLR, or neither were allocated a PD-NLR score of 2, 1, or 0, respectively. The study was approved by the Ethics Committee of First Affiliated Hospital of Zhejiang University School of Medicine.

Follow-up
The patients were monitored after surgery by outpatient visits or telephone calls. Overall survival (OS) was calculated as the time from the date of surgery to the date of death from any cause or the date of last known contact. Disease-free survival (DFS) was calculated as the time from initial diagnosis until relapse. Follow-up was routinely carried out every 6 months for the first 5 years and yearly thereafter. Follow-up examinations included laboratory tests and imaging techniques [12].

Statistical analysis
All of the statistical analyses were performed using SPSS software, version 16.0 for Windows (SPSS, Chicago, IL, USA). Area under the curve values, obtained from ROC curve analysis, were used to compare the predictive efficacies of PD-NLR score and those of other predictive models. Differences between groups were analysed using Pearson's chi-square test, Fisher's exact test or the Mann-Whitney U test as appropriate. Cox proportional hazard models were used to estimate hazard ratios for OS and DFS and to determine independent risk factors. All p values were two sided and considered significant when less than 0.05.
Associations of PD, NLR, and PD-NLR score with survival PD was predictive of inferior OS for PNETs (HR = 9.229, 95% CI 2.879-29.585, P < 0.001), whereas patients with a high NLR had shorter OS than patients with a low NLR (HR = 8.837, 95% CI 2.990-26.117, P < 0.001). Given the predictive ability of PD and NLR for survival, we analysed the prognostic significance of PD-NLR score. Patients with a PD-NLR score of 1 had shorter OS (HR = 20.055, 95% CI 2.444-164.556, P = 0.005) and DFS (HR =14.132, 95% CI 3.083-64.774, P = 0.001) than patients with a PD-NLR score of 0 (Fig. 1). Furthermore, a high tumor grade, the presence of lymph node (LN) metastasis and perineural invasion, a large tumor and a high T stage tumor were prognostic factors for poor OS (P < 0.05 for all), whereas a symptomatic tumor; a high tumor grade; the presence of LN metastasis, distant metastasis or perineural invasion; a large tumor and a high T stage tumor were associated with poor DFS (P < 0.05 for all).

Prognostic value of preoperative PD-NLR in different PNET subgroups
As WHO classification, tumor size, type of hormone production and perineural invasion have been identified as prognostic factors for PNET, we next investigated the prognostic value of preoperative PD-NLR score in different subgroups of PNET patients to exclude these factors. Patients with a low PD-NLR (score 0) showed better OS and DFS than those in the high PD-NLR group (score 1 or 2) among patients with grade 1 or 2 tumor (both P < 0.001) ( Fig. 2a and b). Furthermore, in the patients with tumor size > 2.5 cm, high PD-NLR score was significantly associated with decreased OS (P = 0.001) and DFS (P < 0.001) (Fig. 2c and d), and the prognostic value of OS (P < 0.001) and DFS (P < 0.001) was maintained in patients with nonfunctional PNETs (Fig. 3a and b). In addition, high PD-NLR score was a prognostic factor for poor OS and DFS in those patients without perineural invasion (both P < 0.001) ( Fig. 3c and d).

Comparative performance of PD-NLR and other predictive models
To further evaluate the prognostic value of PD-NLR, PD, NLR and WHO classification, ROC analysis was performed, and AUC values were compared. PD-NLR had a higher AUC value for OS and DFS (0.886 and 0.889) than PD or NLR (Fig. 4). In addition, the discriminatory capability of PD-NLR was superior to that of the  (Table 4).

Discussion
The present study showed that preoperative PD-NLR score, a novel and easily accessible inflammation-based score derived from main pancreatic duct dilation and the neutrophil-to-lymphocyte ratio, was an independent prognostic factor for PNET of the head undergoing curative resection. Furthermore, high preoperative PD-NLR score was associated with high tumor grade, large tumor size, LN metastasis, distant metastasis and perineural invasion. In addition, we found that the predictive capability of PD-NLR score was superior to that of other predictive models, including PD, NLR and WHO classification.
Recently, there has been accumulating evidence that PD is associated with a high risk for pancreatic tumors, including pancreatic cancer or PNETs [17][18][19][20]. Concerning the mechanism of PD in pancreatic tumors, mechanical compression by the tumor or cancer cell   [19,21,22]. In addition, the pancreas with PD may be the site where pancreatic cancer originates. Tanaka S et al. reported that main pancreatic duct dilatation (> 2.5 mm) and presence of a pancreatic cyst (> 5 mm) were both strong independent predictors of the subsequent development of pancreatic cancer [23]. Nanno Y et al. found that PNET patients with main pancreatic duct involvement and dilation showed significantly worse recurrence-free survival than did those without ductal involvement (P < 0.001), with a 5-year recurrence-free rate of 41% [19]. Therefore, PD can act as a significant prognostic biomarker in pancreatic tumors. Systemic inflammation is correlated with worse survival in cancer patients in a variety of cancer types [24,25]. The NLR, which can reflect the systemic inflammation status, has been shown to be a reliable predictive marker for different types of cancer, such as PNET [12], pancreatic ductal adenocarcinoma [15], renal cell carcinoma [26], breast cancer [14] and liver cancer [13]. Salman T et al. revealed that an advanced stage was accompanied by significantly higher NLR and platelet-to-lymphocyte ratio (PLR) in patients with NETs [27]. Recently, Luo G et al. conducted a retrospective analysis of 165 PNETs and reported that NLR was an independent predictor of overall survival for patients with PNETs [28]. Tong Z et al. demonstrated that increased NLR was related with advanced T stage, LN metastasis, tumor thrombus formation, and advanced grade in patients with PNETs [29]. Additionally, our previous study found that NLR and PLR were significantly higher in patients with PNETs than in matched healthy volunteers. Furthermore, NLR, but not PLR, is an independent prognostic factor of both OS and DFS in patients with PNET [30]. In general, preoperative NLR, a prognosis-related serum biomarker, can be useful for predicting disease prognosis in patients with surgically resected PNETs.
In our study, we found that high PD-NLR score was correlated with tumor size, T-stage, lymph node metastasis, type of hormone production, perineural invasion, and WHO classification. Moreover, patients with high PD-NLR score were more likely to have distant metastasis than patients with low PD-NLR score. All of these findings indicate that PD-NLR can reflect the tumor progression and tumor burden. Further analysis revealed that PD-NLR score was an independent predictive factor for patients with surgically resected PNETs of the head. Patients with a high PD-NLR score had significantly poorer OS and DFS than did those with a low PD-NLR score. For the subgroups of patients with grade 1 or 2 tumor or tumor size > 2.5 cm, we similarly found that high PD-NLR score was significantly associated with decreased OS and DFS. Furthermore, the predictive capability of PD-NLR score for survival was superior to that of WHO classification. All of these findings confirm that preoperative PD-NLR score can predict the outcomes of patients with PNET of the head undergoing curative resection.
The major limitations of the present study are its retrospective nature and single-center design. A further limitation is the small sample size. Therefore, large prospective studies are necessary to validate our findings. Another limitation was the short study duration of 46 months, as PNETs generally have an indolent disease course. In addition, only patients who underwent curative resection were included in the study. Despite these limitations, the present study is valuable as it identifies PD-NLR score as a potential prognostic marker to predict survival in patients with PNET of the head undergoing curative resection.