Development and validation of a prognostic nomogram based on inflammation and Nutrition-Related Indexes for predicting postoperative survival in patients with pancreatic neuroendocrine tumors

Abstract

Background

Pancreatic neuroendocrine tumors (pNETs) are rare malignant tumors. While surgical resection is the primary therapeutic approach, postoperative survival outcomes remain suboptimal. This study aims to investigate the prognostic value of inflammation and nutrition-related indicators, such as Prognostic nutritional index (PNI), Neutrophil-lymphocyte ratio (NLR), Platelet-lymphocyte ratio (PLR), Systemic immune-inflammation index (SII), and Lymphocyte-monocyte ratio (LMR), in patients with pNETs undergoing surgery. Furthermore, we sought to develop and validate a prognostic nomogram incorporating independent predictors of postoperative overall survival (OS).

Methods

A retrospective analysis was conducted on the clinical data from 102 patients with pathologically confirmed pNETs. Univariate and multivariate Cox regression analyses were performed to identify prognostic factors. A nomogram was developed based on independent risk factors, and its predictive performance was internally validated.

Results

The analysis revealed that low preoperative levels of PNI, low NLR, low LMR, low SII, and low PLR were significantly associated with reduced postoperative survival. Multivariate Cox regression analysis identified PNI (HR = 0.153; 95% CI, 0.033–0.720), NLR (HR = 0.153; 95% CI, 0.037–0.633), and AJCC stage (HR = 20.126; 95% CI, 5.036–80.422) as independent prognostic factors. The developed nomogram demonstrated excellent discriminative ability, with time-dependent AUC values of 0.878 (3-year), 0.860 (5-year), and 0.830 (10-year), along with a C-index of 0.824 (95% CI 0.748, 0.901) upon internal validation. Risk stratification using the nomogram effectively distinguished low- and high-risk groups, and its validity was confirmed using Kaplan-Meier survival curves.

Conclusion

Despite limitations inherent to its retrospective, single-center design and small sample size, this study establishes a novel nomogram integrating inflammatory-nutritional biomarkers with clinicopathological staging. This tool provides clinicians with a personalized prognostic assessment framework to guide postoperative management strategies for pNETs patients.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-025-14864-9.

Background

Pancreatic neuroendocrine tumors (pNETs) are a heterogeneous group of tumors that arise from neuroendocrine cells [1]. Accounting for 1–3% of all pancreatic malignancies, pNETs constitute one of the most prevalent pancreatic cancer subtypes, second only to pancreatic ductal adenocarcinoma [2–4]. Epidemiologic surveillance data suggest that the incidence of pNETs is approximately 1 per 100,000 people [5, 6]. With the recent advancements in diagnostic techniques, the detection and diagnosis rates of pNETs have been steadily increasing. However, survival rates of pNETs have not kept pace with the rising incidence, and the prognosis for patients remains unsatisfactory [7, 8]. While surgical resection remains the cornerstone of curative-intent therapy [9], substantial prognostic heterogeneity persists even among histologically comparable cases.

Disease grading and staging are often regarded as the most valuable predictors of pNETs survival [8, 10]. Nevertheless, significant interindividual variability in tumor biological behavior—manifested as divergent metastatic patterns and treatment responses—challenges the precision of stage-based prognostication [11–13]. Therefore, identifying additional effective biomarkers is essential for refining treatment strategies and improving long-term survival outcomes for patients [14].

Emerging evidence underscores the dual prognostic significance of systemic inflammation and nutritional status in oncological outcomes [15–17]. Interactions between inflammation and the tumor microenvironment regulate tumor cell proliferation and metastasis. The levels of systemic inflammatory responses such as lymphocytes, neutrophils, monocytes, and platelets, can provide insights into the prognosis of cancer patients. Additionally, nutritional status affects a patient’s therapeutic response, inflammatory response, and treatment toxicity [14, 18]. Albumin, a classical biomarker of nutritional status, has been shown to predict the prognosis of multiple cancers [14, 19]. Several studies have demonstrated that preoperative hypoproteinemia is strongly correlated with postoperative complications in cancer patients [20, 21].

Several peripheral blood biomarkers related to inflammation and nutrition have been investigated for their prognostic value in patients with tumors [15, 22]. These indexes include the neutrophil-lymphocyte ratio (NLR), lymphocyte-monocyte ratio (LMR), systemic immune-inflammatory index (SII), platelet-lymphocyte ratio (PLR), and prognostic nutritional index (PNI). These indexes are easily obtainable through simple, non-invasive, and cost-effective tests. In recent years, the clinical relevance of these biomarkers has been increasingly recognized, and they have emerged as important prognostic tools for various malignancies, including bladder cancer [20], gastric cancer [23], nasopharyngeal carcinoma [24], and pancreatic cancer [14]. Jun-Xi Xiang et al. demonstrated that integrating PNI with tumor grade and metastasis effectively predicts overall survival in patients with gastro-entero-pancreatic neuroendocrine tumors [14]. In pNETs, a recent study shows that the preoperative PNI was associated with postoperative progression [21]. However, the combined effect of inflammatory response and nutritional state on predicting the prognosis of pNETs patients has not been thoroughly investigated, and there is a lack of predictive modeling based on PNI and NLR for predicting OS of pNETs. This study therefore aims to: (1) systematically evaluate the prognostic significance of PNI, NLR, PLR, SII, and LMR in surgical pNETs; (2) develop and internally validate a novel nomogram integrating significant biomarkers with established clinicopathological factors.

Materials and methods

Study design and patient population

The data of patients with pNETs were collected from The Second Affiliated Hospital Zhejiang University School of Medicine between 2013 and 2020. A total of 102 patients whose surgical specimens were confirmed to be pNETs were enrolled in this retrospective study. The inclusion criteria were as follows: (I) Patients with pNETs who received surgical treatment and whose diagnosis was confirmed via pathology; (II) No history of major organic diseases such as heart, lung, or kidney disease; (III) No history of chemotherapy, radiotherapy, prior upper abdominal surgery, or other malignancies; (IV) Successful completion of surgery.

The exclusion criteria were: (I) A history of other malignancies or prior upper abdominal surgery; (II) Presence of multiple lesions or distant metastases that cannot be radically resected; (III) Comorbidities involving critical organs such as the heart, lungs, or kidneys, preventing surgery; (IV) Age over 75 years or under 18 years; (Ⅴ) Incomplete clinical data or lack of follow-up records. The study was approved by the Ethics Committee of The Second Affiliated Hospital Zhejiang University School of Medicine, and conducted in accordance with the guidelines of the Declaration of Helsinki.

Ourinstitutional protocol for pancreatic neuroendocrine tumors (pNETs) strictly adheres to the 2021 Chinese Guidelines for the Diagnosis and Treatment of Pancreatic Neuroendocrine Tumors, implemented through a multidisciplinary team (MDT) approach (Additional file 2). Our preoperative evaluation protocol aligns with the 2021 Chinese Guidelines (Additional file 2).

Data collection

We collected demographic information, baseline clinical measurements, and laboratory results from 102 patients. The patient data included age, sex, body mass index (BMI), surgical approach, tumor location, tumor size, tumor stage and grade, Ki-67 index, lymph node metastasis, and hematological parameters. Additionally, we calculated the PNI, NLR, PLR, LMR, and SII using the following formulas: PNI = albumin (g/L) + 5 × lymphocyte count (× 10⁹/L); NLR = neutrophils / lymphocytes (× 10⁹/L); PLR = platelet / lymphocyte (× 10⁹/L); LMR = monocyte / lymphocyte (× 10⁹/L); SII = neutrophil × platelet/lymphocyte (× 10⁹/L).

Patients were followed up after treatment through outpatient examinations or telephone check-ins until July 2024. The follow-up continued until the patient passed away. Overall Survival (OS) was defined as the period from diagnosis to death from any cause or the last follow-up.

Statistical analysis

Statistical analysis was performed using SPSS (version 23) and R software (version 4.3.3). Normally distributed data were presented as mean ± SD, while non-normally distributed data were presented as median and interquartile range, based on the results of the Shapiro-Wilk test. Comparisons between two groups were performed using Student’s t-test for normally distributed data and the Mann-Whitney U test for non-normally distributed data. Receiver operating characteristic (ROC) curves were generated to determine the optimal cut-off values for NLR, PLR, LMR, and SII. The Kaplan-Meier (KM) curve and log-rank test were used to estimate and compare survival differences.

Variable selection for the multivariable Cox regression model followed a two-stage process: (1) Univariate screening: All candidate variables underwent preliminary univariate Cox regression analysis. Factors demonstrating significant association with overall survival (P < 0.05) were retained for further modeling. (2) Multivariable selection: Statistically significant predictors from univariate analysis were subsequently entered into a backward stepwise multivariate Cox regression model (P < 0.05). This rigorous approach ensured inclusion of only robust independent prognosticators in the final nomogram. The nomogram underwent internal validation using bootstrap resampling, and a calibration curve was plotted to assess model accuracy. To further evaluate the discriminatory efficacy of the nomogram, time-dependent ROC curve analysis was conducted, and metrics such as the area under the curve (AUC) and concordance index (C-index) were calculated. Additionally, a clinical decision curve analysis (DCA) was performed to assess the clinical applicability of the model and quantify the net benefit within different threshold probabilities. The nomogram score for each patient was calculated using the constructed model. The optimal cut-off value for the nomogram score was determined using the surv_cutpoint function in the survminer package of R, and patients were divided into high- and low-risk groups based on this cut-off. The prognostic performance of this classification was further validated by plotting Kaplan-Meier survival curves and conducting log-rank tests.

In this analysis, the rms package was used for nomogram construction and calibration curve plotting, the riskRegression and ggplot2 packages for time-dependent ROC curve analysis, and the ggscidca package for clinical decision curve analysis. The survminer package was used for survival analysis, and the nomogramFormula package was used for calculating nomogram scores. All tests were two-sided, and a P < 0.05 was considered statistically significant.

Results

Patient characteristics

A total of 102 patients with pNETs were enrolled in the final analysis between 2013 and 2020 based on the inclusion criteria. The patient characteristics are listed in Table 1. The mean age of patients was 53.29 ± 13.39, and 50% of them were women. The mean BMI of the patients was 23.48 ± 3.43 kg/m2. Abdominal pain was reported in 27 patients (26.5%), hypertension in 21 patients (20.6%), and diabetes in 8 patients (7.8%). The mean value of PNI was 49.28 ± 4.80, and the median values of NLR, PLR, SII, and LMR were 2.16 (1.55, 3.14), 123.70 (99.12, 153.96), 411.95 (299.54, 632.50), and 3.91 (3.05, 5.09), respectively.

Table 1.

Demographics and clinical characteristics of pancreatic neuroendocrine tumor patients

Characteristics	Total (102)
Age (years)	53.29 ± 13.39
Gender (M/F)	51/51
BMI (kg/m2)	23.48 ± 3.43
PNI	49.28 ± 4.80
NLR	2.16 (1.55, 3.14)
PLR	123.70 (99.12, 153.96)
SII	411.95 (299.54, 632.50)
LMR	3.91 (3.05, 5.09)
Hypertension	21 (20.6%)
Diabetes	8 (7.8%)
Symptomatic diagnosis
Abdominal pain	27 (26.5%)
Dizziness	11 (10.8%)
Fever	1 (1.0%)
Jaundice	2 (2.0%)
Back pain	3 (2.9%)
Fatigue	2 (2.0%)
Diarrhea	1 (1.0%)
Abdominal distension	8 (7.8%)
Back distension	2 (2.0%)
Coma	2 (2.0%)
Numbness in hands	1 (1.0%)
Tumor location
Head	36 (35.3%)
Neck	11 (10.8%)
Body or tail	52 (51.0%)
Multiple sites	3 (2.9%)
Grade
G1	53 (52.0%)
G2	40 (39.2%)
G3	9 (8.8%)
AJCC stage
I-II	84 (82.4%)
III-IV	18 (17.6%)

All patients underwent surgery. Of these, 36 (35.3%) had tumors located in the pancreatic head, 11 (10.8%) in the neck, 52 (51.0%) in the body or tail, and 3 patients had tumors located in multiple sites. Based on the World Health Organization (WHO) pathological grading standard, 53 (52.0%) cases were classified as G1, 40 (39.2%) as G2, and 9 (8.8%) as G3. In addition, according to the 8th edition of the AJCC TNM staging criteria, the majority of patients were classified as stage I-II (n = 84, 82.4%), while 18 patients (17.6%) were classified as stage III-IV.

Survival analysis for PNI- SFA and PNI- SMI

ROC analysis was performed using tumor recurrence as the endpoint to determine the optimal cut-off points. The optimal cutoff values for the PNI, NLR, LMR, SII and PLR were 49.675 (AUC: 0.692, 95% CI: 0.573, 0.812), 3.413 (AUC: 0.497, 95% CI: 0.335, 0.660), 2.467 (AUC: 0.490, 95% CI: 0.331, 0.649), 256.155 (AUC: 0.580, 95% CI: 0.409, 0.750), and 150.319 (AUC: 0.530, 95% CI: 0.409, 0.750), respectively (Fig. 1A). Based on the cutoff values above, patients were divided into high- and low-value groups (Additional file 1: Supplementary Tables 1–5).

Fig. 1.

ROC curves and Kaplan-Meier curves of PNI, NLR, LMR, SII, and PLR for OS in pNETs patients. (A) ROC curves; (B-F) Kaplan-Meier curves of PNI, NLR, LMR, SII, and PLR

Kaplan-Meier curves were utilized to investigate the association between the above indexes and OS in patients with pNETs. As shown in Fig. 1B-F, survival was significantly lower in patients with low PNI (P = 0.009), low NLR (P = 0.022), low LMR (P = 0.027), SII (P = 0.006) and low PLR (P < 0.001).

Univariate and multivariate cox’s regression analysis

Univariate and multivariate Cox’s regression analysis were used to identify independent prognostic factors (Table 2). According to univariate analysis, PNI, NLR, PLR, SII, LMR, tumor grade (ref: 1–2), and AJCC stage (ref: I-II) were significantly associated with OS. Prognostic factors (P < 0.05) identified in univariate analysis were included in multivariate analyses. Multivariate Cox analysis identified PNI (HR = 0.153; 95% CI, 0.033–0.720; P = 0.017), NLR (HR = 0.153; 95% CI, 0.037–0.633; P = 0.010), and AJCC stage (HR = 20.126; 95% CI, 5.036–80.422; P < 0.001) as independent prognostic factors for OS. Interaction analysis revealed no significant effect modification between PNI and AJCC stage (HR = 1.04; 95% CI, 0.88–1.21; P = 0.649) or NLR and AJCC stage (HR = 0.95; 95% CI, 0.75–1.38; P = 0.725) (Additional file 1: Supplementary Table 6).

Table 2.

Univariable and multivariable analysis for overall survival in patients with pancreatic neuroendocrine tumor patients

	Univariable analysis				Multivariable analysis
	OR	95%CI	P -value		OR	95%CI	P -value
Age (years) (ref:≤60)	1.120	0.403–3.114	0.828
Female (ref: male)	0.887	0.339–2.318	0.807
BMI (kg/m2) (ref:≤18)	0.218	0.029–1.658	0.141
PNI	0.189	0.052–0.691	0.012		0.153	0.033–0.720	0.017
NLR	0.340	0.119–0.973	0.044		0.153	0.037–0.633	0.010
PLR	0.130	0.033–0.517	0.004		0.070	0.005–1.014	0.051
SII	0.229	0.076–0.689	0.009		2.118	0.292–15.356	0.458
LMR	0.274	0.083–0.906	0.034		0.218	0.041–1.168	0.075
Hypertension	0.886	0.263–2.981	0.845
Diabetes	0.529	0.061–4.550	0.562
Symptomatic diagnosis	2.691	0.900-8.045	0.076
Tumor location (ref: head and neck)	0.939	0.684–1.289	0.696
Grade (ref: 1–2)	49.231	5.678-426.843	< 0.001		15.012	0.726-310.298	0.080
AJCC Stage (ref: I-II)	16.667	5.023–55.299	< 0.001		20.126	5.036–80.422	< 0.001
Preoperative leukocyte (109/L) (ref:≤9.5)	0.625	0.071–5.494	0.672
Preoperative neutrophil (109/L) (ref:≤6.3)	0.760	0.084–6.878	0.807
Preoperative lymphocyte (109/L) (ref:≤3.2)	0.000	0.000–0.000	1.000
Preoperative monocyte (109/L) (ref:≤0.6)	0.670	0.137–3.284	0.621
Preoperative platelet (109/L) (ref:≤350)	0.000	0.000–0.000	1.000
Preoperative albumin (g/L) (ref:≤55)	0.000	0.000–0.000	1.000
Preoperative globulin (g/L) (ref:≤40)	0.000	0.000–0.000	1.000
Preoperative triglyceride (mmol/L) (ref:≤1.71)	0.831	0.289–2.388	0.731
Preoperative cholesterol (mmol/L) (ref:≤5.69)	1.960	0.597–6.436	0.267
Preoperative CA199 (U/mL) (ref:≤37)	3.244	0.815–12.919	0.095
Preoperative CEA (ng/mL) (ref:≤5)	0.000	0.000–0.000	1.000
Preoperative CA125 (U/mL) (ref:≤24)	2.281	0.515–10.100	0.277

Bold values are achieved statistical significance (P < 0.05) in univariate or multivariate analysis

Nomogram establishment and validation

To predict the 3-year, 5-year, and 10-year OS of pNETs, a nomogram was developed based on the results of multivariate Cox analysis. Three independent prognostic variables-PNI, NLR, and AJCC stage—were incorporated into the nomogram (Fig. 2). An integrated score was calculated for each patient, with higher total scores correlating with lower survival rates.

Fig. 2.

The nomogram for OS of pNETs patients

The nomogram was validated using the bootstrap technique. The calibration curve showed that the predictions for 3-year, 5-year, and 10-year OS closely matched the actual observations (Fig. 3A). Time-dependent ROC curve analysis demonstrated that the AUC values of the nomogram at 3 years, 5 years, and 10 years were 0.878, 0.860, and 0.830, respectively (Fig. 4A). The C-index and 95% CI of the nomogram were 0.824 (95% CI 0.748, 0.901). These results suggest that the nomogram is well-calibrated and possesses strong predictive ability.

Fig. 3.

Calibration curves and DCA of the nomogram in pNETs patients. (A) Calibration curves of the nomogram at 3-year, 5-year and 10-year; (B-D) DCA of the nomogram at 3-year, 5-year and 10-year

Fig. 4.

ROC curves and Kaplan-Meier curves of the nomogram in pNETs patients. (A-B) ROC curves; (C) Kaplan-Meier curves

Additionally, the nomogram scores for all patients were calculated. The optimal cut-off value for the nomogram scores was determined using the surv_cutpoint function in the survminer package of R, with the optimal cut-off value set at 122.07. Based on this value, patients were classified into high-risk and low-risk groups (Fig. 4C). Kaplan-Meier analysis showed that the high-risk group had significantly worse prognoses than the low-risk group at different time points (P < 0.05).

To estimate the clinical benefit of the nomogram, DCA was performed. The DCA curves demonstrated that for the 3-year, 5-year, and 10-year predictions, the nomogram had a net benefit for risk thresholds ranging from 0.05 to 0.9, 0.07–0.96, and 0.09–0.99, respectively (Fig. 3B-D). This suggests that the nomogram has good clinical applicability and can be useful in clinical decision-making. These results further confirm the nomogram’s robust clinical relevance.

Discussion

pNETs are a rare group of tumors characterized by neuroendocrine differentiation, and their prevalence is increasing [1, 3]. Surgery remains the primary treatment for pNETs patients; however, postoperative prognosis is generally poor, underscoring the need for further improvement in survival outcomes [22, 25]. Current guidelines recommend using validated tools that integrate nutritional intake and inflammatory indicators for screening, as malnourished neuroendocrine tumors patients are at a higher risk of complications and mortality [14, 26]. Accurate prognostic assessment helps clinicians optimize treatment strategies and improve survival rates. This study investigates the prognostic value of inflammatory and nutritional indices, including PNI, NLR, LMR, SII and PLR, for pNETs, and constructs a nomogram based on independent risk factors to predict postoperative OS in pNETs patients. Model validation demonstrated excellent predictive efficiency and clinical utility.

In this study, we included a total of 102 patients with pNETs and analyzed their demographic and clinical characteristics. The mean age of the patients was 53.29 years, with an equal gender distribution (51 males and 51 females). The mean BMI was 23.48 kg/m². Approximately 26.5% of patients presented with abdominal pain, 20.6% had hypertension, and 7.8% had diabetes, which is consistent with other reports on pNETs [8, 27, 28]. These clinical symptoms may be related to the tumor’s anatomical location, hormone secretion, or disease progression [27]. Additionally, the majority of tumors were located in the body or tail of the pancreas (51.0%), followed by the head of the pancreas (35.3%). This distribution may be related to the richer vascular supply of the pancreatic body and tail. Notably, most patients had low-grade tumors (52.0% were G1 and 39.2% were G2), and 82.4% of patients were at AJCC stage I-II, indicating that early-stage pNETs were more frequently diagnosed in this cohort. Overall, these patient characteristics provide a foundation for further prognostic analysis and the development of a risk prediction model.

This study explored the relationship between nutritional and inflammatory markers, such as PNI, NLR, LMR, SII, and PLR, and the prognosis of patients with pNETs. The results indicated that lower PNI, NLR, LMR, SII, and PLR were all associated with poorer OS, suggesting that inflammatory and nutritional statuses significantly impact the prognosis of pNETs patients. Through univariate and multivariate Cox regression analyses, we identified PNI, NLR, and AJCC staging as independent prognostic factors for OS in pNETs patients. AJCC staging, as the gold standard for clinical cancer management, is widely used for prognostic evaluation in various tumor types [29]. However, despite providing comprehensive tumor information, AJCC staging alone has certain limitations in predicting patient prognosis [30–32]. On one hand, AJCC staging mainly relies on tumor anatomical features, size, and metastasis, neglecting biological factors such as the overall immune status and inflammatory response of the body. On the other hand, some pNETs patients with early AJCC stages may still experience poor prognosis due to immune suppression or latent inflammatory responses [27, 33]. Therefore, relying solely on AJCC staging may fail to capture all factors that influence patient prognosis. We further introduced PNI and NLR as independent prognostic factors. PNI integrates the patient’s nutritional status and immune function and has been shown to be closely associated with the prognosis of various cancers [20, 24, 34]. In pNETs patients, a low PNI value may indicate a higher malignancy and more pronounced immune suppression, leading to poor prognosis [21]. NLR, as a marker of inflammation, reflects the immune status of the tumor microenvironment. An elevated NLR typically indicates a chronic inflammatory response within the tumor microenvironment, which not only contributes to tumor initiation and progression but also promotes metastasis [8, 15, 20]. Previous studies have demonstrated the strong prognostic value of NLR in pancreatic cancer [14, 22]. By combining these markers with AJCC staging, we can more comprehensively assess the prognostic risk in pNETs patients. Notably, interaction analyses demonstrated consistent prognostic effects of PNI across AJCC stages and NLR across tumor grades, supporting the stability of these biomarkers in diverse clinical contexts. PNI and NLR, as supplementary indicators, provide insights into the patient’s immune function and inflammatory response, thereby compensating for the limitations of AJCC staging in terms of immune and nutritional factors.

Although PLR, LMR, and SII showed significant prognostic value in univariate analysis, they did not emerge as independent prognostic factors in multivariate Cox regression analysis. While these markers were associated with OS, they may have been influenced by other clinical factors such as AJCC staging, tumor size, and pathological grading in the complex multivariable context. Additionally, individual differences and immune status could also impact the performance of these markers [22, 35, 36]. Therefore, while these inflammation-related markers show potential in univariate analysis, their lack of independence in multivariate analysis suggests that combining them with additional clinical and molecular biomarkers may further improve the accuracy of prognostic predictions.

Nomograms, as personalized survival prediction tools, are widely used in cancer prognosis evaluation. Compared to the traditional TNM staging system, nomograms can incorporate multiple clinical variables to provide more precise risk assessments [34, 37, 38]. YM Jiang et al. showed that, compared with the TNM staging system, the nomogram has a higher net benefit and predictive accuracy in patients with nasopharyngeal carcinoma [24].In this study, we developed a pNETs nomogram model based on three easily accessible prognostic factors: PNI, NLR, and AJCC stage. This model successfully provides a simple and clinically applicable tool. Internal validation was performed using the bootstrap method, and calibration curves along with time-dependent ROC curve analysis demonstrated that the model’s predicted results were highly consistent with actual observations. The AUC values were 0.878, 0.860, and 0.830, and the C-index was 0.824, confirming the accuracy, stability, and clinical value of the model. Furthermore, the Kaplan-Meier survival curves also validated the effectiveness of the model.

Admittedly, the present study has several limitations. Firstly, the uncontrolled biases could not be completely eliminated owing to the nature of single-center retrospective studies; Secondly, while the prevalence of pNETs is increasing, their overall percentage remains low, leading to a small sample size in this study. In addition, the absence of external validation of the nomogram means that prospective studies with larger sample sizes are needed to further validate the general applicability of the predictive model; Thirdly, the standardized cut-off value of PNI, NLR, LMR, SII and PLR remain controversial, as different studies have applied various statistical methods to determine these cut-off values. Therefore, large-scale multicenter randomized controlled trials, along with parameters such as molecular diagnostics and targeted therapies, will be incorporated in future prognostic studies for pNETs.

Conclusions

In conclusion, we developed a nomogram using PNI, NLR, and AJCC staging to predict overall survival in pNETs patients, with strong predictive accuracy confirmed by internal validation. The model effectively stratifies patients into risk groups and demonstrated clinical utility across varied risk thresholds. This tool offers a simple, individualized approach for prognosis in pNETs, aiding in clinical decision-making. Future studies with larger, diverse cohorts are needed for external validation to further confirm its applicability.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviation

AJCC

American Joint Committee on Cancer

AUC

Area under the curve

BMI

Body mass index

C-index

Concordance Index

DCA

Decision curve analysis

KM

Kaplan-Meier

LMR

Lymphocyte-monocyte ratio

NLR

Neutrophil-lymphocyte ratio

OS

Overall survival

PLR

Platelet-lymphocyte ratio

pNETs

Pancreatic neuroendocrine tumors

PNI

Prognostic Nutritional Index

ROC

Receiver operating characteristic

SD

Standard deviation

SII

Systemic immune-inflammation index

WHO

World Health Organization

Author contributions

BZ and SY conceived and designed the hypotheses of this study, performed the surgery and revising the final manuscript. NY and MJ performed the data collect and data analysis, drafting the article and revising the final manuscript. YJ, JQ and ZG performed data collect, data analysis and organizing the data and revising the final manuscript. All authors have approved the final manuscript to be submitted.

Funding

This project was supported by the National Natural Science Foundation of China (No. 82270684).

Data availability

Data is provided within the manuscript or supplementary information files.

Declarations

Ethics approval and consent to participate

The study involving humans were approved by The Second Affiliated Hospital Zhejiang University School of Medicine. The study was reviewed and approved by the Research Ethics Committee of the Second Affiliated Hospital Zhejiang University School of Medicine (approval number: 2020-047) and was conducted in accordance with the guidelines of the Declaration of Helsinki. The participants provided their written informed consent to be involved in this study.

Consent for publication

All authors approved the publication.

Competing interests

The authors declare no competing interests.