A combined systemic immune-inflammation index and prognostic nutritional index score for predicting overall survival after resection of pancreatic ductal adenocarcinoma: a retrospective cohort study Authors: Ruihan Hou, Aorui Wang, Chengxu Du, Xinda Yang, Weihong Zhao, Haitao Lv, Dongrui Li Published: 2026 ABSTRACT AbstractBackgroundThe systemic immune-inflammation index (SII) and prognostic nutritional index (PNI) have been individually associated with prognosis in various cancers. This study aimed to develop and validate a combined SII-PNI-based risk score for predicting overall survival (OS) in patients with resectable pancreatic ductal adenocarcinoma (PDAC).MethodsThis retrospective study enrolled 375 consecutive patients with PDAC who underwent curative-intent resection at The Second Hospital of Hebei Medical University between January 2014 and July 2025. Using X-tile software, optimal cut-off values for SII and PNI were determined based on 1-year OS in a training cohort (constituting 70% of the patients). A combined risk score was constructed using Cox regression coefficients and dichotomized at the optimal cut-off (− 0.231). The model’s performance was assessed using receiver operating characteristic (ROC) curve analysis, Kaplan–Meier survival curves, and multivariate Cox regression.ResultsThe combined SII-PNI model achieved AUCs of 0.758 in the training cohort and 0.750 in the validation cohort. Kaplan–Meier analysis showed significantly worse OS in the high-risk group (p < 0.001). In univariate analysis, elevated SII (p = 0.044) and reduced PNI (p < 0.001) were both significantly associated with poorer OS. Multivariate Cox regression analysis confirmed PNI (HR = 0.891, 95% CI: 0.863–0.919, p < 0.001) and age (HR = 1.029, 95% CI: 1.013–1.046, p < 0.001) as independent prognostic factors, whereas SII and sex were not.ConclusionThe combined SII-PNI score is a simple, cost-effective, and powerful predictor of prognosis in resectable PDAC, enabling reliable postoperative risk stratification.Keywords: Pancreatic ductal adenocarcinoma, Systemic immune-inflammation index, Prognostic nutritional index, Overall survival, Prognosis Introduction Pancreatic ductal adenocarcinoma (PDAC) is a highly aggressive malignancy with a persistently poor prognosis, ranking among the leading causes of cancer-related deaths worldwide [1, 2] For the 15–20% of patients who present with resectable disease, surgical resection remains the cornerstone of potential cure [3, 4]. However, the long-term outcomes post-surgery are heterogeneous, with a high rate of recurrence and a five-year overall survival rate that rarely exceeds 30%. This variability underscores the critical need for reliable and accessible prognostic tools to identify high-risk patients who might benefit from more aggressive adjuvant therapy and tailored surveillance.Given the substantial global burden of pancreatic and pancreatitis-related diseases, accurate postoperative risk stratification remains of high clinical relevance [5]. Beyond traditional tumor-node-metastasis (TNM) staging, the host’s systemic condition is increasingly recognized as a pivotal determinant of cancer outcomes [6–9]. Systemic inflammation and nutritional status are two key host-derived factors that play a crucial role in the pathogenesis and progression of PDAC [10]. The systemic immune-inflammation index (SII), calculated as (platelet count × neutrophil count)/lymphocyte count, integrates these cellular components to quantify the balance between pro-tumor inflammation and anti-tumor immunity [11–13]. It has emerged as a promising prognostic marker in various cancers. Similarly, the prognostic nutritional index (PNI), defined as serum albumin (g/L) + 5 × lymphocyte count (×10⁹/L), reflects the patient’s nutritional status and immune competence [13, 14]. Low PNI is a well-documented indicator of cancer cachexia and is associated with poor survival in gastrointestinal malignancies [10, 15, 16].Composite immune-nutritional markers integrating platelet, albumin, and inflammatory parameters have demonstrated robust prognostic value in gastrointestinal malignancies, including colorectal cancer [17, 18]. Moreover, inflammation-based hematologic markers have been consistently associated with long-term outcomes across stage I–III solid tumors [19], underscoring the broad applicability of such biomarkers in oncological prognostication. Individually, both high SII and low PNI have been significantly associated with worse survival in PDAC [20, 21]. Serum nutritional biomarkers have been shown to critically influence postoperative risk and recovery, highlighting the biological and clinical importance of host nutritional status [22].However, inflammation and nutrition are biologically interconnected pathways that collectively shape the host-tumor interaction [23, 24]. Relying on a single index may provide an incomplete prognostic picture. We hypothesize that a composite model integrating both indices would provide a more comprehensive assessment of the host’s systemic status, potentially yielding superior prognostic accuracy compared to either marker alone [25, 26]. Therefore, the primary objective of this study was to develop and validate a novel, combined risk score based on pre-operative SII and PNI for predicting overall survival in a large, homogeneous cohort of patients with resectable PDAC undergoing curative-intent surgery [27, 28]. Methods Study population In this single-center retrospective study, we retrospectively reviewed 375 consecutive patients who underwent curative-intent resection for PDAC at The Second Hospital of Hebei Medical University between January 2014 and July 2025. Follow-up for all patients began on the date of surgery and continued until death or the last follow-up visit, with overall survival as the endpoint event. The final follow-up cutoff date was September 30, 2025. The median follow-up time, calculated using the reverse Kaplan-Meier method, was 33.2 months (95% confidence interval: 31.5–35.8 months). Exclusion criteria included incomplete laboratory data, perioperative mortality (within 30 days), or prior malignancy. Clinicopathological data were collected from the electronic medical record system. The study was approved by the Institutional Review Board of The Second Hospital of Hebei Medical University (Approval No. 2025-Y028). Given the retrospective nature of the study, the requirement for informed consent was waived. Laboratory parameters Preoperative blood samples collected within 7 days before surgery were used for analysis. Risk score calculation The prognostic risk score was calculated for each patient using the following formula derived from the Cox regression model: where: Age is the patient’s age in years. Sex is assigned as 1 for Male and 0 for Female. Patients were classified into high-risk (Risk Score > − 0.231) or low-risk (Risk Score ≤ − 0.231) groups based on the optimal cut-off value determined by X-tile software. Cut-off determination Optimal cut-offs for SII and PNI were determined using X-tile software (version 3.6.1, Yale University) based on 1-year OS in the training cohort (random 70% of the total cohort). A combined risk score was constructed using Cox regression coefficients and dichotomized at the optimal cut-off (− 0.231). Statistical analysis ROC curve analysis, Kaplan–Meier survival analysis (log-rank test), and multivariate Cox proportional hazards models were performed using R software (version 4.3.0). A two-sided p-value < 0.05 was considered statistically significant.To enhance clinical interpretability, continuous variables (such as SII, PNI, and age) were standardized in the multivariate Cox regression analysis, with hazard ratios (HR) and 95% confidence intervals (CI) reported per one standard deviation increase (or per 100-unit increase, depending on the variable distribution). To evaluate the predictive performance of different models, we compared the discriminative ability of three multivariate Cox proportional hazards models: Model 1 (base model) included age and sex; Model 2 (PNI model) included age, sex, and PNI; Model 3 (combined model) included age, sex, PNI, and SII. The C-index was used to assess the predictive accuracy of each model, and its 95% confidence interval (CI) was calculated. Optimal cut-off values for SII, PNI, and the combined risk score were determined using X-tile software based on patient 1-year overall survival status. The choice of 1-year OS as the time point was primarily based on the following considerations: Pancreatic ductal adenocarcinoma is characterized by highly aggressive biological behavior, with extremely high risks of recurrence and death in the early postoperative period (especially within the first year). Therefore, identifying patients at high risk for early mortality has more direct clinical relevance for guiding aggressive adjuvant treatment strategies and intensive follow-up protocols. Cut-off values based on short-term endpoints are more aligned with clinical decision-making needs for early postoperative intervention. Results Patient characteristics A total of 375 patients with resectable pancreatic ductal adenocarcinoma who underwent curative-intent resection between January 2014 and July 2025 were included in this study. The cohort was randomly divided into training (70%, n = 262) and validation (30%, n = 113) sets. The baseline characteristics were well-balanced between the two cohorts (Table 1). The median age was 64 years (range: 38–82), and 56.3% of patients were male.The median follow-up time for the entire cohort was 33.2 months (calculated using the reverse Kaplan-Meier method, 95% CI: 31.5–35.8 months; follow-up cutoff: September 30, 2025). Table 1Baseline characteristics of the study cohortCharacteristicValueAge, median (range), years64 (38–82)Sex, n (%) Male211(56.3) Female164(43.7)BMI, median (IQR), kg m⁻²23.1(21.2–25.0)Under-weight (< 18.5), n (%)43(11.5)Pre-operative biliary drainage, n (%)117(31.2)Neoadjuvant therapy, n (%)69(18.4)Follow-up, median (range), months33.2 months Baseline characteristics of the study cohort Development of the combined SII-PNI risk score Using Cox regression analysis in the training cohort, we developed a prognostic risk score based on preoperative systemic immune-inflammation index (SII) and prognostic nutritional index (PNI) values, along with age and sex. The final risk score formula was derived as follows: where Sex was coded as 1 for male and 0 for female patients. Model performance and discrimination The combined SII-PNI model demonstrated excellent discriminative ability for predicting 1-year overall survival. In the training cohort, the area under the ROC curve (AUC) was 0.758 (95% CI: 0.702–0.814), while in the validation cohort, the AUC was 0.750 (95% CI: 0.671–0.829), indicating consistent performance across both datasets (Figs. 1 and 2). Receiver operating characteristic (ROC) curve of the combined systemic immune-inflammation index (SII) and prognostic nutritional index (PNI) model for predicting 1-year overall survival in the training cohort Receiver operating characteristic (ROC) curve of the combined SII-PNI model for predicting 1-year overall survival in the validation cohort The concordance index (C-index) for the model was 0.724 (95% CI: 0.683–0.765) in the training set and 0.718 (95% CI: 0.662–0.774) in the validation set, further confirming the model’s robust prognostic capability. Risk stratification and survival analysis Using X-tile software, we identified − 0.231 as the optimal cut-off value for risk stratification based on 1-year overall survival. Applying this cut-off, 142 patients (37.9%) were classified as high-risk and 233 patients (62.1%) as low-risk. Kaplan-Meier survival analysis revealed significantly worse overall survival in the high-risk group compared to the low-risk group (log-rank test, p < 0.001) (Fig. 3). The median overall survival was 18.3 months (95% CI: 15.2–21.4) in the high-risk group versus 32.7 months (95% CI: 28.9–36.5) in the low-risk group. Similar significant differences were observed when using median risk score stratification (log-rank test, p < 0.001) (Fig. 4). Kaplan-Meier survival curves stratified by the optimal prognostic risk score cut-off value. Patients were classified into high-risk (risk score > -0.231) and low-risk (risk score ≤ -0.231) groups based on the combined SII-PNI score Kaplan-Meier survival curves stratified by the median prognostic risk score. Patients were classified into high-risk (risk score > median) and low-risk (risk score ≤ median) groups Individual biomarker analysis When analyzed separately, both SII and PNI showed significant associations with overall survival in univariate analysis. Patients with high PNI (above median) demonstrated significantly better survival compared to those with low PNI (log-rank test, p < 0.001) (Fig. 5). Similarly, patients with high SII (above median) showed a trend toward worse survival, though with borderline significance (log-rank test, p = 0.044) (Fig. 6). Kaplan-Meier survival curves stratified by the prognostic nutritional index (PNI). Patients were dichotomized into high-PNI and low-PNI groups based on the median PNI value Kaplan-Meier survival curves stratified by the systemic immune-inflammation index (SII). Patients were classified into high-SII and low-SII groups based on the median SII value Univariate and multivariate cox regression analysis In univariate analysis, PNI (HR = 0.891, 95% CI: 0.863–0.919, p < 0.001), SII (HR = 1.000, 95% CI: 1.000-1.001, p = 0.044), age (HR = 1.029, 95% CI: 1.013–1.046, p < 0.001), and sex (HR = 1.172, 95% CI: 0.872–1.576, p = 0.292) were evaluated for their association with overall survival. Multivariate Cox regression analysis, adjusting for all variables, identified PNI (HR = 0.891, 95% CI: 0.863–0.919, p < 0.001) and age (HR = 1.029, 95% CI: 1.013–1.046, p < 0.001) as independent prognostic factors.SII, reported per 100-unit increase, yielded a hazard ratio of 1.02 (95% CI: 0.98–1.06, p = 0.312) and sex (HR = 1.158, 95% CI: 0.854–1.570, p = 0.348) did not retain independent prognostic significance in the multivariate model (Fig. 7).To assess the incremental predictive value of SII, we compared the discriminative ability of three multivariate Cox proportional hazards models. As shown in Table 2, the base model (age and sex) yielded a C-index of 0.65 (95% CI: 0.62–0.68). The addition of PNI improved the C-index to 0.68 (95% CI: 0.65–0.71). Furthermore, incorporating SII into the model (age, sex, PNI, and SII) resulted in a C-index of 0.70 (95% CI: 0.67–0.73). This stepwise increase suggests that although SII was not independently significant in the multivariate analysis, its combination with PNI provides a modest improvement in risk stratification. Forest plot of multivariate Cox regression analysis for overall survival. Hazard ratios and 95\% confidence intervals are shown for age, sex, SII, and PNI Table 2Comparison of discrimination ability among different prognostic modelsModel C-index95% Confidence IntervalBase model (age + sex)0.650.62–0.68PNI model (age + sex + PNI)0.68Combined model (age + sex + PNI + SII)0.700.67–0.73 Comparison of discrimination ability among different prognostic models Risk score distribution The distribution of prognostic risk scores across the study population showed a bimodal pattern, with clear separation between the identified risk groups (Fig. 8). The optimal cut-off value of -0.231 effectively distinguished patients with divergent survival outcomes. Distribution of prognostic risk scores. The optimal cut-off value of -0.231 (dashed line) was determined using X-tile software The concordance index for the combined SII-PNI model was 0.724 (95% CI 0.683–0.765) in the training set and 0.718 (95% CI 0.662–0.774) in the validation set. Post-operative SII and PNI were available in 198 patients at POD7; however, their inclusion did not improve the C-index (0.724 vs. 0.726, P = 0.81) and are therefore not reported further. Discussion This study developed and validated a novel, combined risk score based on the Systemic Immune-Inflammation Index (SII) and Prognostic Nutritional Index (PNI) for predicting overall survival in patients with resectable pancreatic ductal adenocarcinoma (PDAC). The principal finding of our research is that this integrated SII-PNI model demonstrates robust and consistent discriminative ability, with AUCs approximating 0.75 in both training and validation cohorts. More importantly, it effectively stratified patients into distinct high-risk and low-risk groups with significantly different survival outcomes, underscoring its potential utility in postoperative risk assessment. The core of our findings reveals a fascinating hierarchy between the two biomarkers. While both SII and PNI were significant in univariate analysis, only PNI emerged as an independent prognostic factor in the multivariate Cox model [15, 29]. This result strongly suggests that nutritional and immune competence, as captured by PNI, may be a more fundamental driver of prognosis in resectable PDAC than the systemic inflammation reflected by SII [30, 31]. This observation can be interpreted through the lens of cancer cachexia, a hallmark of PDAC. PNI, which incorporates serum albumin and lymphocyte count, is a direct gauge of the patient’s nutritional reserve and adaptive immune capacity [25]. A low PNI signifies a state of catabolism and impaired immune surveillance, creating a permissive environment for tumor recurrence and metastasis. On the other hand, SII, a ratio of platelets and neutrophils to lymphocytes, primarily reflects a pro-inflammatory and pro-thrombotic state. Our results imply that in the context of resected PDAC, the detrimental effects of malnutrition and lymphocytopenia may subsume or precede the prognostic information offered by systemic inflammation alone. Our findings regarding the supremacy of PNI align with a growing body of evidence. For instance, a study by also identified PNI, but not other inflammatory indices, as an independent predictor for survival in biliary tract cancers, highlighting the cross-cancer importance of nutritional status. Conversely, our result that SII lost its independent significance contrasts with studies in, where SII remained a powerful prognosticator. In the multivariate analysis of this study, after adjusting for PNI, age, and sex, SII was no longer an independent prognostic factor for overall survival (HR = 1.000, 95% CI: 0.999–1.001, p = 0.312). The hazard ratio was infinitely close to 1, and the confidence interval included 1, clearly indicating that after accounting for the stronger influence of nutritional status (PNI), the direct impact of systemic inflammatory burden (as represented by SII) on patient prognosis was minimal to negligible.This discrepancy underscores the unique tumor microenvironment and pathophysiology of PDAC, where the profound metabolic dysregulation and cachexia might render nutritional status a more dominant prognostic axis than in other malignancies. Our study thereby contributes to a more nuanced understanding of which biomarker is most relevant in specific clinical contexts. Despite PNI’s dominance, the combined SII-PNI model did not diminish in value. The integration of SII, even as a non-independent factor, likely captures complementary aspects of the host-tumor interaction, leading to a model with superior discriminative power (AUC 0.75) compared to using either marker alone. Although SII and sex did not reach conventional statistical significance in the multivariate Cox regression analysis, It is noteworthy that, despite SII and sex losing statistical significance in the multivariate analysis, the combined model achieved a higher C-index (0.70) compared to the model containing only PNI (0.68). This highlights the core value of the SII-PNI combined score: by integrating SII, which reflects systemic inflammation, with PNI, which indicates nutritional status, this composite marker provides a more comprehensive depiction of the patient’s host immune-nutritional state, thereby achieving improved risk stratification beyond that possible with a single parameter.The principal advantage of our risk score is its pragmatism. Derived from routine, low-cost blood tests, it can be seamlessly integrated into clinical workflows without additional burden. This tool allows for the identification of high-risk patients who may benefit from more aggressive adjuvant therapy, intensified nutritional support, and closer post-operative surveillance. We acknowledge several limitations of our study. Its single-center, retrospective design inherently introduces potential selection bias and limits the generalizability of our findings and Its study population was exclusively Chinese, and the applicability of our risk score to other ethnic groups requires verification. Although we constructed the SII-PNI combined score, the multivariate Cox regression model primarily adjusted for basic demographic characteristics such as age and sex, and did not adequately adjust for key tumor-related pathological factors, including AJCC TNM stage, lymph node metastasis status, tumor differentiation, margin status, and whether adjuvant chemotherapy was administered. Tumor burden and treatment differences are core determinants of patient prognosis, and the absence of these variables may lead to residual confounding, potentially affecting the accuracy of our estimates of the associations between SII, PNI, and prognosis. Therefore, the conclusions of this study need to be further validated and confirmed in large prospective cohorts or multi-center databases containing detailed clinicopathological information and standardized treatment protocols. These limitations directly inform avenues for future research. Priority should be given to external validation in large, multi-center, and prospective cohorts. Furthermore, investigating the dynamic changes of the SII-PNI score during adjuvant therapy could reveal its value in monitoring treatment response.From a biological mechanism perspective, the prognostic value of the SII-PNI score may reflect the complex tumor-host interactions characteristic of pancreatic ductal adenocarcinoma. The profound metabolic reprogramming within the PDAC tumor microenvironment (such as aerobic glycolysis, i.e., the Warburg effect) not only drives tumor growth but also extensively consumes host nutrients, leading to cancer cachexia manifesting as hypoalbuminemia (reflected by low PNI). Concurrently, tumor necrosis and hypoxia can induce systemic inflammatory responses, resulting in elevated neutrophil and platelet counts (reflected by high SII). This chronic state of inflammation-nutrition imbalance can further induce immune dysfunction; for example, the expansion of circulating myeloid-derived suppressor cells and regulatory T cells can suppress the anti-tumor function of effector T cells, thereby creating a “fertile ground” for tumor growth and metastasis. Therefore, the SII-PNI score may serve not merely as a statistical prognostic indicator but also as a window into the state of metabolic, inflammatory, and immune network dysregulation in patients with PDAC. Conclusion The combined SII-PNI score is a reproducible, inexpensive, and powerful predictor of postoperative survival in PDAC. Prospective multicenter studies are needed to validate its clinical utility for individualized management. Acknowledgements Not applicable. Abbreviations Systemic immune-inflammation index Prognostic nutritional index Pancreatic ductal adenocarcinoma Overall survival Area under the curve Confidence interval Hazard ratio Receiver operating characteristic Tumor-node-metastasis Authors’ contributions Ruihan Hou contributed to data curation, formal analysis, and writing – original draft; Aorui Wang, Xinda Yang, Chengxu Du, Weihong Zhao, and Haitao Lv contributed to investigation and resources; Dongrui Li contributed to conceptualization, supervision, funding acquisition, writing – review & editing, and serves as the corresponding author. Funding This work was supported by the Hengrui-Hebei Innovation and Development Medical Cooperation Program (Grant No. HR202502088). The funder had no role in study design, data collection, analysis, interpretation, or writing of the manuscript. Data availability The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request. Declarations Ethics approval and consent to participate This study was approved by the Institutional Review Board of the Second Hospital of Hebei Medical University (Approval No. 2025-Y028). Written informed consent was waived due to the retrospective design. Consent for publication Not applicable. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. References Associated Data Data Availability Statement The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.