Increased Circulating/Systemic Neutrophil-Lymphocyte Ratio Predicts Recurrence After Curative Resection in Patients With Well-Differentiated Pancreatic Neuroendocrine Neoplasm Based on the 2017 World Health Organization Classication: A Single-Center Retrospective Study

Background The prognostic values of inammation-based markers in well-differentiated pancreatic neuroendocrine neoplasms, diagnosed according to the new 2017 World Health Organization classication, have remained unclear. Therefore, we assessed the recurrence predictive ability of such markers after curative resection in patients with these neoplasms. Methods Circulating/systemic neutrophil-lymphocyte, monocyte-lymphocyte, platelet-lymphocyte, and platelet-white cell ratios were evaluated in 120 patients who underwent curative resection for well-differentiated pancreatic neuroendocrine neoplasms without synchronous distant metastasis between 2001 and 2018. Recurrence-free-survival and overall survival were compared using Kaplan–Meier analysis and log-rank tests. Univariate or multivariate analyses, using a Cox proportional hazards model, were used to calculate hazard ratios with 95% condence intervals. Results Univariate analysis demonstrated that preoperative neutrophil-lymphocyte ratio, tumor size, European Neuroendocrine Tumor Society TMN classication, 2017 World Health Organization classication, and venous invasion were associated with recurrence. The optimal preoperative neutrophil-lymphocyte ratio cut-off value was 2.62, based on receiver operating characteristic curve analysis. In multivariate analysis, a higher preoperative neutrophil-lymphocyte ratio (HR=3.49 95% CI 1.05-11.7; P=0.042) and 2017 World Health Organization classication (HR=8.81, 95% CI 1.46-168.2; P=0.015) were independent recurrence predictors. preoperative


Background
Pancreatic neuroendocrine neoplasm (PanNEN) is a biologically heterogeneous and relatively rare malignancy, with an incidence rate of approximately 5 cases per 1 million person-years, which accounts for 1-2% of primary pancreatic neoplasms [1]. In recent years, the incidence of PanNEN detected clinically has signi cantly increased because of the advances in imaging modalities during the past few decades [2]. When the disease is clinically detected before it becomes symptomatic, the lesions are typically localized, increasing the possibility of curative resection, and improving prognosis [3]. Although surgical resection is currently the only curative treatment for PanNEN [4], recurrence could occur at intervals, and therefore, reoperation for recurrent lesions may occasionally be required. Reoperation for distant metastases can lead to excellent long-term survival [5]. Even if unresectable metastases occur, novel targeted drugs, such as the multiple tyrosine kinase inhibitor sunitinib and the mTOR inhibitor everolimus, have been approved and registered for antiproliferative therapy for well-differentiated PanNEN [6,7]. Therefore, it is essential to identify recurrence earlier. For this reason, indicators that could predict recurrence after surgery are required for the optimal management of PanNEN.
Several studies, however, have demonstrated that tumorigenesis and clinical manifestations of welldifferentiated PanNEN are distinctively different from poorly differentiated PanNEN (neuroendocrine carcinoma; NEC), and thus, the determinants of treatment should be considered separately [8,9]. The prognosis in NEC is poor [10], and the National Comprehensive Cancer Network (NCCN) guidelines recommend platinum-based systemic chemotherapy for patients with NEC [6]. For this reason, the 2017 World Health Organization (WHO) introduced signi cant changes to the classi cation of PanNEN. Of note, a new category of well-differentiated neoplasms, neuroendocrine tumors G3 (NET-G3), was introduced, and these are distinct from poorly differentiated NEC-G3 [11,12].
Recently, systemic immune-in ammatory markers have been reported as factors that in uence the outcomes of treatments, such as surgery, and the e cacy of chemotherapy in patients with various types of malignancies [13][14][15][16][17][18]. However, the ability of systemic immune-in ammatory markers to predict prognosis in patients with sole, well-differentiated PanNEN, based on the 2017 WHO classi cation [11], other than NEC or synchronous distant metastasis, has remained unknown.
In this study, we sought to evaluate whether systemic immune-in ammatory markers can be preoperative prognostic factors for predicting recurrence and overall survival after curative resection in patients with well-differentiated PanNEN based on the new 2017 WHO classi cation.

Patients
We analyzed 132 consecutive cases who underwent surgery for primary, histologically con rmed PanNEN at the Department of Surgery, Tohoku University Hospital, between 2001 and 2018. Eight patients with synchronous hepatic metastasis durig the surgery, two patients with NEC, one patient not suitable for curative resection, and one patient with an active infection at blood sampling were excluded from the study. Finally, 120 patients with well-differentiated PanNEN were enrolled in this study. Patient characteristics (age, sex), perioperative factors (serum albumin, hormonal secretion, tumor location, clinical stage), pathological ndings (2017 WHO classi cation, tumor size, lymph node metastasis, lymphovascular invasion), and prognosis were investigated retrospectively. Histopathological ndings were assessed by experienced pathologists (FF, and HS). For all the patients, visual assessment ''eyeballing calculation'' was performed to assess Ki-67 index. TNM staging was adopted according to the European Neuroendocrine Tumor Society (ENETS) classi cation [19], and the new 2017 WHO classi cation of NET by the gastro-entero-pancreatic (GEP) system was used for histopathological classi cation [11]. Peripheral blood routine tests were performed within 14 days before surgery, according to our internal institutional policy. The serum neutrophil-lymphocyte ratio (NLR) was calculated as the number of neutrophils divided by the number of lymphocytes. The monocyte-lymphocyte ratio (MLR), platelet-lymphocyte ratio (PLR), and platelet-white cell ratio (PWR) were calculated in the same manner.
Clinical follow-up Postoperative follow-up evaluation included physical examinations, laboratory tests, and enhanced computed tomography (chest and abdominal cavity), once every six months. Treatment after recurrence was determined by the available evidence at the time of surgery and based on the patient's condition.

Statistical analysis and software
Recurrence-free-survival (RFS) and overall survival (OS) were calculated from the date of surgery to the date of recurrence, the date of death from any cause, or the date of last follow-up. To determine the appropriate cutoff values, we used receiver operating characteristic (ROC) curves and determined the area under the curve (AUC). Differences between groups were determined using t-tests in the case of normally distributed variables or by the Wilcoxon rank-sum test in the case of abnormally distributed variables for examining differences in continuous variable distributions, and Pearson's chi-square tests for categorical variables. RFS probabilities were compared for various categories of interest using the Kaplan-Meier method with the log-rank test.
Prognostic factors were assessed with univariate and multivariate analyses, using Cox's proportional hazards model. Hazard ratios (HR) with 95% con dence intervals (CIs) were calculated. P< 0.05 was considered to indicate statistical signi cance.

Ethics approval
This study was approved by the ethics committee of the Tohoku University Hospital (Approval No. 2020-1-322). It was performed in adherence to the tenets of the Declaration of Helsinki and its later amendments. The need to obtain written informed consent was waived due to the retrospective nature of the study.

Characteristics of patients with resected well-differentiated PanNEN
The demographic and clinicopathological features of the 120 patients who underwent curative resection of PanNEN are shown in Table 1. The median age was 60 years (range 12-88 years), and the median follow-up period in all patients was 64 months (range, 6-185 months). There were no perioperative deaths. The pathological ndings (based on the 2017 WHO classi cation) were NET-G1 in 73 patients, NET-G2 in 45, and NET-G3 in 2. The median tumor size was 14.5 mm (range 4-168 mm). Pathology investigations con rmed lymph node metastasis in 18 patients (15.0%). Postoperative recurrences were observed in 12 cases (10%). The sites of recurrence were in the liver in 10 patients, the para-aortic lymph node in 1, and the lung in 1. The 5-and 10-year RFS rates for the entire cohort were 92.0% and 78.7%, respectively. Three patients died due to PanNEN, 10 patients died due to other diseases, and the remaining 107 patients were alive at the end of the surveillance period. Thus, the 5-and 10-year disease-speci c survival rates were 100% and 92.5%, respectively. The NLR was signi cantly higher in patients with recurrence than in those without recurrence (median NLR: 2.40 vs 1.90, p = 0.001), while the MLR, PLR, and PWR were not statistically signi cantly different between those with and those without recurrence (Fig. 1).
An ROC curve was used to determine the cutoff value associated with postoperative recurrence. Each cutoff value of NLR and tumor size was de ned as the highest log-rank statistic of any threshold. The optimal cutoff values for preoperative NLR and tumor size were 2.62 mm and 25 mm, respectively (Additional le1: Fig. S1.)  Results are expressed as mean ± SD or as absolute number

Comparison of clinical variables in relationship to RFS after curative resection
The results of the univariate and multivariate analyses for each of the clinicopathological variables are shown in Table 3. According to univariate analysis, the recurrence risk was about six times higher in patients with a high NLR than in those with a low NLR (95% CI 1.81-18.5, P = 0.004). Additionally, the TMN clinical-stage, 2017 WHO classi cation G2/3, tumor size, and venous invasion were also signi cantly predictive factors for recurrence (P < 0.05 for all). In contrast, age, sex, albumin, hormonal function, tumor location, lymph node metastasis, and lymphatic invasion were not signi cant predictors of recurrence. Moreover, in multivariate analysis, higher NLR (HR = 3.49, 95% CI 1.05-11.7, P = 0.042) and 2017 WHO classi cation G2/3 (HR = 8.81, 95% CI 1.46-168.2, P = 0.015) were independent predictive factors for recurrence. A higher NLR showed a signi cant correlation with shorter RFS (median RFS duration, 117.8 months, P < 0.001) (Fig. 2a) and poor OS (median OS duration, 95.2 months, P = 0.032) after curative resection (Fig. 2b).  (Table 4). Additionally, a higher NLR was signi cantly associated with a shorter RFS in nonfunctional PanNEN (P = 0.009) (Fig. 3a). Contrary to nonfunctional PanNEN, NLR was not associated with RFS in functional PanNEN (P = 0.094) (Fig. 3b).

Discussion
The current study demonstrated that elevated preoperative NLR and 2017 WHO classi cation independently predicted recurrence in patients with well-differentiated PanNEN after curative surgery. No previous studies have demonstrated that increased NLR serves as an independent prognostic factor in patients with well-differentiated PanNEN, as de ned by the 2017 WHO classi cation. This may be of potential clinical bene t in these patients. Furthermore, we observed that elevated preoperative NLR was predictive of a signi cantly shorter RFS in nonfunctional PanNEN patients.
Previously, PanNEN with lymph node metastasis, a higher Ki-67 index, and a higher 2010 WHO grade were reported to be associated with a signi cantly higher risk of recurrence [20,21]. In contrast, a large international cohort study showed that the ENETs TNM classi cation was superior to the Union for International Cancer Control/American Joint Committee on Cancer/WHO staging system and could more accurately predict the clinical outcome of patients [22]. We revealed that the ENETs TNM classi cation was related to RFS in univariate analysis, but not in multivariate analysis. A possible explanation for this nding is that we assessed the patients with curatively resected PanNEN and excluded metastatic stage patients in our present study. Clinically, the preoperative Ki-67 index obtained by ne-needle aspiration biopsy is less accurate due to intra-tumoral heterogeneity [23], which highlights the requirement of preoperative non-invasive prognostic indicators, such as in ammation-based markers. Preoperative precise assessment of recurrence risk of the patients allows clinically more relevant selection of an optimal surgical strategy, such as enucleation or additional lymph node dissection.
In terms of systemic in ammation-based markers in PanNEN, preoperative NLR and PLR have been reported to be useful for predicting lymph node metastasis or recurrence [24][25][26][27][28]. However, these studies included a moderate number of patients with distant metastatic stage IV or poorly differentiated PanNEN (NEC), as de ned as 2010 WHO NET-G3. Generally, poorly differentiated PanNEN has substantial distant metastases and a distinctly poor prognosis [29]. The 2010 WHO classi cation of NET-G3 included both well-differentiated and poorly differentiated PanNEN, resulting in a morphologically and biologically heterogenous population [30]. Consequently, the 2017 WHO classi cation of NET-G3 was recategorized as only well-differentiated PanNEN, distinctively different from NEC. Indeed, the median RFS (6.7 months) and median OS (15.3 months) of surgically resected NEC were markedly shorter than in welldifferentiated PanNEN in our institute (Additional le2: Fig. S2.). Furthermore, the value of NLR for NEC was signi cantly higher than that in patients with well-differentiated PanNEN (Additional le3: Fig. S3.).
In our present study, we assessed the e cacy of NLR to predict recurrence in well-differentiated PanNEN, other than NEC or distant metastasis, based on the 2017 WHO classi cation. Hence, more prolonged RFS and OS were detected in our present study than in the previously reported ones [24][25][26][27][28].
NLR was recently reported to be associated with tumor progression in several human malignancies [14][15][16][17][18]. In addition, NLR could serve as a predictive marker in patients with not only PanNEN but also gastrointestinal NEN [27]. We previously reported that the NLR was a useful diagnostic marker for predicting intraductal papillary mucinous neoplasm with high-grade dysplasia/invasive carcinoma to differentiate low-grade dysplasia [31]. Previous studies reported that a high NLR was signi cantly consistent with accumulation of tumor in ltrating CD66b neutrophils or CD163 + macrophages in patients with PanNEN and pancreatic cancer, which results in poor RFS and OS [28,32]. In general, neutrophils are markers of acute in ammation and could possibly promote tumor development and progression by providing an adequate tumor microenvironment via the production of cytokines and chemokines [33]. In addition, an increased number of lymphocytes play a crucial role in the host's anticancer immune response; thus, lymphocytosis is generally associated with better prognosis and a more favorable response to chemotherapy or radiation therapy in a variety of cancers [34]. Therefore, in cancer patients, peripheral blood neutrophilia and lymphopenia may re ect a weak anticancer reaction and worse clinical outcomes [35].
Regardless of the histological ndings, there are hormonally functional and non-functional phenotypes in PanNEN. According to an epidemiological survey, the number of PanNEN patients has increased rapidly. In particular, hormonally non-functional PanNEN was most prevalent and increased signi cantly [36,37].
To the best of our knowledge, no previous studies have demonstrated possible roles of NLR as a prognostic factor for RFS in distinct categories of nonfunctional and functional PanNEN. A high NLR has also been proven to be a risk factor of recurrence in nonfunctional PanNEN. In contrast, the NLR was statistically unrelated to RFS in functional PanNEN. One reason for this might be that we analyzed a relatively small number of the patients with hormonally functional PanNEN; only three patients in this subgroup had recurrence during the follow-up period.
Surveillance at shorter intervals might be required in patients with nonfunctional well-differentiated PanNEN with a high NLR and 2017 WHO G2/G3 classi cation to detect recurrence earlier after surgery.
Furthermore, almost all well-differentiated PanNENs express somatostatin receptors; hence, somatostatin receptor scintigraphy should be considered in PanNEN patients with a high NLR to assist identi cation of distant metastases that could be missed by computed tomography or positron emission tomography before and after surgery [38][39][40]. Well-differentiated PanNEN with the risk factors described above may receive clinical bene ts by adjuvant treatments such as somatostatin analogs after surgery. However, there is no current evidence or clinical indication for adjuvant therapy and further studies that focus on these high-risk groups are required.
This study had some limitations. First, it was a retrospective review of a single, high-volume institution in the eld of pancreatic tumors. Second, although consecutive patients were enrolled, they were collected over a relatively long period, during which treatment strategies changed reasonably. It awaits further investigations to verify these ndings and to identify the optimal cutoff values and the usefulness of NLR in the clinical management of PanNEN patients.

Conclusion
In summary, results of our present study clearly demonstrated that the NLR could serve as a useful preoperative marker of clinical recurrence risks after the surgery. It is considered a convenient screening