Symptom associations during early recovery after pancreatic cancer surgery: an exploratory cross-sectional network analysis Authors: Ningning Xia, Neng Shi, Yuan song, Yi Miao, Han Yan Published: 2026 ABSTRACT AbstractBackgroundPatients undergoing pancreatic cancer surgery often experience multiple symptoms during early recovery, but the structural relationships among these symptoms remain unclear.MethodsA cross-sectional network analysis was conducted on 213 patients after pancreatic cancer surgery. Symptoms during the first postoperative week were retrospectively assessed on day 7 using the M.D. Anderson Symptom Inventory gastrointestinal cancer module. Symptom prevalence, severity, centrality, bridge strength, and network stability were examined.ResultsOf the 22 symptoms, 17 occurred in more than 30% of patients. The most prevalent symptoms were lack of appetite, disturbed sleep, distress, pain, and fatigue. Lack of appetite, vomiting, and fatigue had the highest mean severity scores. Within the symptom network, vomiting, distress, and sweating showed the highest strength centrality, with vomiting occupying the most central position. Swelling, fatigue, and irritability showed relatively high bridge strength. The correlation stability coefficient for strength was 0.516.ConclusionsSymptoms were common but generally mild during the first postoperative week. Network analysis provided a structured description of symptom associations and may inform future longitudinal studies on postoperative symptom trajectories and patient-centered outcomes.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12885-026-16383-7.key words: Pancreatic cancer, Pancreatectomy, Postoperative symptoms, Symptom burden, Network analysis, Patient-reported outcomes Background Pancreatic cancer is a highly malignant tumor with poor prognosis and substantial disease burden. According to GLOBOCAN 2022, the global age-standardized incidence and mortality rates of pancreatic cancer are 4.7 and 4.2 per 100,000, respectively, with mortality approaching incidence [1]. In China, the age-standardized incidence and mortality rates are 4.4 and 3.9 per 100,000, respectively, indicating a high disease burden [2]. For patients with resectable pancreatic cancer, surgery remains the main potentially curative treatment [3]. However, pancreatic cancer surgery is associated with extensive surgical trauma, complex perioperative management, and multiple postoperative symptoms [4]. Previous studies have shown that patients may experience fatigue, impaired perceived health, anorexia, pain, and gastrointestinal discomfort during postoperative recovery [5]. A greater number of symptoms in patients with cancer is often related to higher symptom frequency, greater severity and distress, and poorer quality of life [6]. Postoperative recovery after pancreatic cancer surgery is not limited to the management of surgical complications. Patients must also gradually adapt to changes in oral intake, digestive function, pain, sleep, physical strength, and emotional state. Lee et al. emphasized that symptom management in patients with pancreatic cancer requires a multidimensional approach, including nutritional assessment, gastrointestinal symptom control, pain management, activity support, and psychosocial care [7]. From the patient perspective, Ristau et al. developed a pancreatic cancer-specific disease trajectory model and showed that patients experienced a dynamic process of coping and adaptation across diagnosis, treatment, recovery, and life reconstruction [8]. These studies indicate that postoperative symptoms in cancer patients should not be viewed as isolated clinical events but rather understood within the broader context of holistic rehabilitation and continuous care. In recent years, there has been growing attention on symptom clusters in cancer patients. A symptom cluster generally refers to a group of two or more concurrent symptoms that are related to one another and relatively independent from other symptom clusters [9]. Compared with single symptoms, symptom clusters may have a greater impact on patients’ physical function, emotional status, quality of life, and survival outcomes [10]. Expert consensus has also emphasized that understanding and managing cancer symptom clusters remains an important goal of symptom science [11]. However, existing symptom cluster studies mainly identify symptom combinations at the cluster level. While these methods can be used to describe symptom patterns, they are less able to further demonstrate the relative position of individual symptoms within the overall symptom structure or to indicate which symptoms may have stronger connections across different symptom domains. Network analysis provides a complementary perspective for understanding complex symptom relationships. In symptom network, symptoms are represented as nodes, and edges between symptoms reflect conditional associations within the network. Compared with traditional analyses of symptom prevalence, severity, and symptom clusters, network analysis can further present the connectivity characteristics of different symptoms within the overall structure and identify symptoms with higher connectivity or bridging roles in the network. At present, evidence on the network structure of early postoperative symptoms in patients with pancreatic cancer remains limited, particularly during the first week after surgery, which is a critical phase of recovery. Therefore, network analysis was performed to examine the association structure among patient-reported symptoms during the first postoperative week in patients undergoing surgery for pancreatic cancer. This study aimed to describe the characteristics of early postoperative symptom burden and to identify symptoms with higher connectivity and bridging roles within the symptom network, thereby providing a basis for further understanding early postoperative symptom experiences and for future research on postoperative symptom trajectories in patients with pancreatic cancer. Methods Study design and participants A cross-sectional design was utilized to assess the symptoms experienced by patients undergoing pancreatectomy for pancreatic cancer. Participants were recruited from the Department of Pancreatic Surgery at BenQ Hospital, affiliated with Nanjing Medical University, between January and June 2025. Consecutive sampling was employed, and all eligible patients undergoing pancreatectomy for pancreatic cancer during the study period were invited to participate. Inclusion criteria were as follows: (1) diagnosis of pancreatic cancer confirmed by clinical and pathological examination in accordance with the National Comprehensive Cancer Network (NCCN) guidelines[3]; (2) scheduled to undergo pancreatectomy for pancreatic cancer; (3) aged 18 years or older, with adequate communication ability; and (4) provision of written informed consent and willingness to complete study assessments. Patients were excluded if they were scheduled for palliative surgery or exploratory laparotomy, or if they had severe comorbid conditions, such as other malignancies, severe infection, or heart failure. The sample size for this study was calculated using the standardized effect size approach for symptom network analysis. The formula for calculating sample size was: n=[(Zα/2+Zβ)/f]2[12], where Zα/2 = 1.96 corresponds to a significance level of α = 0.05, Zβ=0.84 corresponds to a statistical power of 0.80, and f = 0.22 represents the standardized effect size. Based on this formula, the estimated minimum required sample size was 162 participants. Taking into account a 10% attrition rate, the target sample size was set at no fewer than 180 participants. A total of 213 patients completed the assessment and were included in the final cross-sectional network analysis. This study was approved by the hospital’s ethics committee (Ethics No. 2025-KL046). Measures Demographic and clinical characteristics Demographic and clinical data were collected using a structured questionnaire developed by the team. Demographic variables included age, sex, religious belief, occupation, marital status, smoking history, place of residence, living arrangement, alcohol consumption, and medical payment method. Clinical variables included tumor location, hypertension, diabetes mellitus, coronary heart disease, previous cerebral infarction, and previous hospitalization history. Information was obtained through patient interviews and verified against electronic medical records to ensure accuracy. M.D. Anderson symptom inventory -gastrointestinal cancer module (MDASI-GI) Symptom burden was assessed using the M.D. Anderson Symptom Inventory (MDASI) [13]. MDASI is a validated instrument widely used to evaluate symptom severity in cancer patients, including core symptom items as well as disease-specific modules. Given its relevance to gastrointestinal malignancies, the MDASI gastrointestinal cancer module (MDASI-GI) [14] was used as the framework for perioperative symptom assessment. Considering the specific clinical characteristics of pancreatic cancer patients in the perioperative period-particularly during the early postoperative phase-the original symptom items were contextually expanded based on a literature review and clinical practice experience. The expanded items were reviewed and confirmed by pancreatic surgeons and oncology nursing experts to ensure content validity. A total of 22 symptom items were included, covering both physical and psychological symptoms. Each symptom was rated on a 0–10 numeric scale (0 = not present, 10 = as severe as imaginable). The total symptom score was calculated as the sum of all symptom ratings; a higher score indicates a greater overall symptom burden. Data collection Eligible patients were approached by trained research staff during hospitalization and were informed about the study objectives and procedures. Written informed consent was obtained prior to participation. Demographic and clinical data were extracted from the electronic medical record system by authorized research personnel. Symptom assessments were conducted on postoperative day 7 (POD7). At that time point, participants were asked to retrospectively rate the severity of symptoms experienced during the first postoperative week using the MDASI-GI module. This approach was adopted to capture the overall symptom burden experienced by patients during the first postoperative week after surgery for pancreatic cancer. All assessments were administered face-to-face in a private hospital setting. Research staff provided standardized instructions and clarification when necessary while ensuring that responses were self-reported and free from external influence. Completed questionnaires were reviewed on site to ensure completeness and data accuracy. Statistical analysis All data organization, descriptive statistics, and symptom network modeling in this study were conducted using R software (version 4.4.1). In the descriptive analysis, categorical data were presented as frequency and percentage (n, %). For skewed distribution of continuous data, the median and interquartile range (M [P25, P75]) were used, with the mean as a supplementary evaluation metric to reflect the overall severity of symptoms. The construction and analysis of the symptom network were primarily carried out using the bootnet and qgraph packages. The cor_auto method was first used to compute the correlation matrix among symptom variables, followed by network structure estimation using the Extended Bayesian Information Criterion graphical lasso algorithm (EBICglasso) [15, 16]. Network visualization was performed using the qgraph package, with the Fruchterman-Reingold algorithm applied for optimal spring layout. In the network graph, nodes represent different symptoms, and edges represent the partial correlations between symptoms. The thickness of the lines reflects the strength of the associations [15]. Community detection of the symptom network was performed based on the random walk algorithm (Walktrap) from the igraph package. To identify key symptoms within the network, three core centrality measures were computed: strength (Strength, the total weight of a node’s connections to other nodes, reflecting its local importance in the network), closeness centrality (Closeness, the inverse of the shortest path length from a node to all other nodes), and betweenness centrality (Betweenness, the number of shortest paths that pass through a node). Additionally, the networktools package was used to compute bridge strength, which is the absolute sum of the edge weights between a node and other nodes in different communities, in order to identify core bridging symptoms that connect different symptom clusters. To assess the stability and reliability of the network model, 1000 bootstrap resampling iterations were performed using the bootnet package. Non-parametric bootstrap method was used to evaluate the accuracy of edge weights, while case-dropping bootstrap method was employed to assess the stability of centrality indices and calculate the related stability coefficient (CS coefficient). According to conventional recommendations, a CS coefficient > 0.25 indicates acceptable stability, and a CS coefficient > 0.50 indicates good stability [17]. Results General data analysis A total of 213 patients undergoing surgery for pancreatic cancer were included in this study. Among them, 121 were male (56.8%) and 92 were female (43.2%). The mean age of the participants was 61.28 ± 13.45 years. Most participants were married (93.4%) and lived with family members (97.2%). Regarding occupation, retired individuals accounted for the largest proportion (46.5%), followed by employees (21.6%) and farmers (21.1%). In terms of place of residence, 105 patients (49.3%) lived in urban areas, 65 (30.5%) in rural areas, and 43 (20.2%) in towns. Most patients were non-smokers or had quit smoking for more than 12 months (77.5%), and most reported no alcohol consumption (81.7%). Regarding tumor location, 138 patients (64.8%) had tumors located in the pancreatic head or neck, while 75 patients (35.2%) had tumors located in the pancreatic body or tail. The most common comorbidities were hypertension (39.4%) and diabetes mellitus (29.1%), while coronary heart disease (5.6%) and previous cerebral infarction (3.8%) were less frequent. In addition, 130 patients (61.0%) had a history of previous hospitalization. Detailed demographic and clinical characteristics are presented in (Table 1). Table 1Demographic and clinical characteristics of patients undergoing pancreatic cancer surgery (n = 213)VariableCategoryn (%)VariableCategoryn (%)SexMale121 (56.8)Smoking statusNon-smoker or quit > 12 months165 (77.5)Female92 (43.2)Quit < 12 months11 (5.2)Religious beliefNo204 (95.8)Current smoker37 (17.4)Yes9 (4.2)Alcohol consumptionNone174 (81.7)OccupationFarmer45 (21.1)Occasional36 (16.9)Employee46 (21.6)Heavy drinking3 (1.4)Self-employed5 (2.3)Tumor locationPancreatic head/neck138 (64.8)Retired99 (46.5)Pancreatic body/tail75 (35.2)Others18 (8.5)HypertensionNo129 (60.6)Marital statusMarried199 (93.4)Yes84 (39.4)Unmarried7 (3.3)Diabetes mellitusNo151 (70.9)Widowed7 (3.3)Yes62 (29.1)Place of residenceRural65 (30.5)Coronary heart diseaseNo201 (94.4)Town43 (20.2)Yes12 (5.6)Urban105 (49.3)Previous cerebral infarctionNo205 (96.2)Living statusLiving alone6 (2.8)Yes8 (3.8)Living with family207 (97.2)Previous hospitalizationNo83 (39.0)Medical payment methodSelf-pay39 (18.3)Yes130 (61.0)Medical insurance144 (67.6)New Rural Cooperative Medical Scheme30 (14.1)Symptom Incidence and Severity Analysis Demographic and clinical characteristics of patients undergoing pancreatic cancer surgery (n = 213) Symptom Incidence and Severity Analysis Among the 22 symptoms included in this study, 17 symptoms had an incidence rate exceeding 30%. The five most prevalent symptoms were lack of appetite (97.2%), disturbed sleep (80.8%), distress (77.9%), pain (77.5%), and fatigue (75.1%). Regarding symptom severity, the symptoms with the highest mean scores were lack of appetite (mean = 1.34), vomiting (mean = 1.33), fatigue (mean = 1.21), diarrhea (mean = 1.14), and disturbed sleep (mean = 1.12). Notably, although vomiting was not among the five most prevalent symptoms, it showed the second highest mean severity score. The symptoms with the lowest incidence were difficulties in remembering (24.4%), numbness (24.9%), and both irritability and swelling (26.3%). Sweating had the lowest mean severity score (mean = 0.78). Detailed information is presented in (Table 2). Table 2Prevalence and severity of symptoms among patients (n = 213)ItemSymptomOccurrence [n(%)]Symptom score [median (P25, P75)]Mean scoreA4Lack of appetite207 (97.2)1 (0.5, 2)1.34A21Disturbed sleep172 (80.8)0.75 (0.25, 2)1.12A6Distress166 (77.9)0.5 (0.25, 1.25)0.91A17Pain165 (77.5)0.5 (0.25, 1.5)0.92A13Fatigue160 (75.1)1 (0.25, 2)1.21A1Dry mouth159 (74.6)0.5 (0.12, 1.5)0.90A11Bloating158 (74.2)0.75 (0, 1.75)1.04A16Sadness157 (73.7)0.75 (0, 1.75)1.05A22Taste alteration145 (68.1)1 (0.5, 1.25)1.11A10Constipation133 (62.4)0.5 (0.25, 1.5)0.95A18Sweating131 (61.5)0.5 (0, 1.25)0.78A7Nausea121 (56.8)0.75 (0.06, 2)1.09A9Vomiting96 (45.1)1.25 (0.25, 2.25)1.33A19Itching96 (45.1)0.5 (0.25, 1.25)0.92A12Drowsiness85 (39.9)0.5 (0, 1.25)0.83A2Shortness of breath79 (37.1)0.5 (0, 1.5)0.92A15Diarrhea73 (34.3)0.75 (0.25, 2)1.14A20Difficulties in swallowing60 (28.2)0.75 (0, 1.75)1.07A8Irritability56 (26.3)0.5 (0, 1.25)0.80A5Swelling56 (26.3)0 (0, 2)0.94A3Difficulties in remembering52 (24.4)0 (0, 1)0.80A14Numbness53 (24.9)0 (0, 1)0.80Symptom Network Analysis Prevalence and severity of symptoms among patients (n = 213) Symptom Network Analysis The symptom network was constructed using the EBICglasso method (Fig. 1). The resulting network contained 90 non-zero edges, of which 84 (93.3%) were positive and 6 were negative. Among the positive associations, the strongest symptom pairs were vomiting and difficulties in swallowing, dry mouth and distress, and drowsiness and sadness, suggesting close conditional associations among these symptoms in the estimated network. Symptom network during the first postoperative week Based on the centrality indices shown in Fig. 2, the top three symptoms in terms of strength centrality were vomiting (rS = 1.110), distress (rS = 0.950), and sweating (rS = 0.883). Vomiting showed the highest strength centrality, indicating that it had the strongest direct connections with other symptoms in the network. The symptom with the lowest strength centrality was numbness (rS = 0.243). Centrality indices of each symptom in the network For betweenness centrality, the top three symptoms were lack of appetite (rB = 83), vomiting (rB = 62), and fatigue (rB = 62), suggesting that these symptoms were more frequently located on the shortest paths between other symptoms in the network. For closeness centrality, the top three symptoms were lack of appetite (rC = 0.001965), vomiting (rC = 0.001926), and constipation (rC = 0.001844), indicating that these symptoms were relatively close to other symptoms within the overall network structure. The bridge symptom analysis showed that swelling, fatigue, and irritability had the highest bridge strength values across symptom clusters (Figs. 3 and 4). Specifically, swelling had the highest bridge strength (0.270), followed by fatigue (0.166) and irritability (0.132). Sadness (0.126) and constipation (0.102) also showed relatively high bridge strength values. These findings suggest that swelling, fatigue, and irritability may function as cross-cluster symptoms in the postoperative symptom network. Bridge symptoms in the network Bridge strength of each symptom in the network Network stability and accuracy were evaluated using non-parametric bootstrap procedures. The correlation stability coefficient for strength was 0.516, indicating good stability of the strength centrality index. Detailed bootstrap results are provided in Supplementary Figures S1 and S2. Nodes represent individual symptoms assessed by the MDASI-GI. Edges represent regularized partial correlations estimated using the EBICglasso method. Green edges indicate positive associations, while red edges indicate negative associations. Thicker edges represent stronger associations. The layout was generated using the Fruchterman–Reingold algorithm. Discussion Network analysis was carried out to examine the association structure among symptoms during the first postoperative week in patients following pancreatectomy for pancreatic cancer. Previous studies on postoperative symptoms in pancreatic cancer have mainly focused on symptom prevalence, severity, and their impact on quality of life, whereas less attention has been paid to the interconnections among symptoms and their relative positions within the overall symptom structure. In current work, symptoms were common during the first postoperative week, with 17 of the 22 symptoms occurring in more than 30% of patients, although the overall symptom severity was relatively low. Meanwhile, symptom prevalence, severity, and network connectivity were not fully consistent. These findings suggest that postoperative symptom burden cannot be fully captured by identifying only the most frequent or most severe symptoms. Network analysis provides additional information by presenting the conditional associations among symptoms and their relative positions within the estimated symptom network. Consistent with previous symptom network studies in cancer populations [5, 6], the present findings suggest that some symptoms may have greater connectivity within the early postoperative symptom structure and warrant further investigation in future studies. The most prevalent symptoms in this study were lack of appetite, disturbed sleep, distress, pain, and fatigue. These symptoms reflect several major aspects of early recovery after pancreatic cancer surgery, including oral intake, sleep, pain control, emotional adjustment, and physical recovery. In terms of severity, lack of appetite, vomiting, fatigue, diarrhea, and disturbed sleep had relatively higher mean scores. Previous studies have shown that patients after pancreatic cancer surgery commonly experience fatigue, anorexia, pain, and gastrointestinal symptoms, which may affect both physical recovery and the treatment experience [5]. Lee et al. emphasized that symptom management in pancreatic cancer should include nutritional assessment, gastrointestinal symptom control, pain management, activity support, and psychosocial care [7]. The present findings further indicate that, within the specific context of the first postoperative week, early postoperative symptom management after pancreatic cancer surgery should not be reduced to the separate management of isolated symptoms, but rather understood within the broader context of rehabilitation and continuous care. In the strength centrality analysis, vomiting, distress, and sweating showed relatively high connectivity in the symptom network. Although vomiting is not among the most common symptoms, its severity is relatively high, and it occupies the most central position in the network. The first week postoperatively is a period of gastrointestinal recovery and gradual restoration of oral intake. Vomiting may occur concurrently with reduced feeding tolerance, abdominal distension, fatigue, fluid replacement needs, and emotional distress. Bromley-Dulfano et al. reported that gastrointestinal symptoms, including nausea and vomiting, are common after pancreatic resection and may persist during postoperative recovery [18]. Here, the close connection between vomiting and other symptoms suggests that early postoperative vomiting may be related not only to gastrointestinal response, but also to nutritional intake, physical recovery, and overall postoperative recovery status. These findings suggest that vomiting may be considered as a candidate symptom for closer observation in future longitudinal studies, together with oral intake, abdominal bloating, hydration status, analgesic use, and accompanying symptoms. Distress also occupied a relatively central position in the symptom network. Patients recovering from pancreatic cancer surgery may experience not only physical problems such as pain, fatigue, and eating difficulties, but also uncertainty about prognosis, subsequent treatment, and functional recovery. Previous studies have shown that psychological distress in patients with pancreatic cancer is closely associated with fatigue, sleep disturbance, pain, and quality of life [19, 20]. From the patient perspective, postoperative symptoms are not simply the accumulation of physiological problems, but are embedded in a broader experience of bodily discomfort, emotional response, and expectations for recovery. Ristau et al. proposed a pancreatic cancer trajectory model from the patient perspective, emphasizing that patients continuously experience adaptation and coping throughout diagnosis, treatment, recovery, and life reconstruction [8]. In this study, the close association between distress and other symptoms suggests that early postoperative symptom assessment should incorporate patients’ subjective experiences and emotional responses, rather than focusing solely on surgical complications or physiological indicators. In this study, sweating also exhibited relatively high central intensity, despite its lower average severity score. In the early postoperative period after pancreatic cancer surgery, sweating may be reported in the context of pain, disturbed sleep, temperature fluctuation, nausea or vomiting, anxiety, fluid shifts, and medication-related effects. These symptoms and clinical conditions may influence patients’ comfort, oral intake, mobilization, and perceived recovery. Physiologically, sweating is influenced by thermoregulatory and autonomic mechanisms and may be affected by pain, emotional stress, anxiety, and changes in autonomic nervous activity [21, 22]. Pancreatic cancer surgery involves substantial surgical trauma and perioperative stress; pain, volume changes, hypothermia or temperature instability, inflammatory responses, medication use, and psychological stress may all influence autonomic regulation during recovery [23–25]. Therefore, the network position of sweating should be interpreted within the broader context of early postoperative recovery, particularly in relation to pain control, temperature status, fluid balance, medication effects, and emotional stress. Bridge strength analysis showed that swelling, fatigue, and irritability had relatively high cross-community connectivity. Among them, swelling had the highest bridge strength, although its prevalence and severity were not prominent. Because swelling was assessed as a patient-reported symptom in this study, its clinical interpretation should be linked to the early postoperative context after pancreatic resection. During this period, perceived swelling may be related to perioperative fluid retention, tissue edema, reduced mobility, inflammatory responses, and nutritional status. Previous research has shown that early postoperative fluid retention after pancreatic resection is common and is associated with increased postoperative complications [26]. In the present network, the bridge position of swelling indicates that patient-reported swelling was connected with symptoms from different domains. Future longitudinal studies should combine patient-reported swelling with objective indicators, such as fluid balance, body weight change, serum albumin, inflammatory markers, mobility, and postoperative complications, to clarify its clinical significance during recovery. Fatigue also showed a prominent bridge role in the network. It was one of the most common symptoms and had relatively high severity, while also showing clear connectivity across symptom communities. Postoperative fatigue may arise from surgical trauma, reduced oral intake, disturbed sleep, and pain, and may further affect ambulation, willingness to eat, confidence in recovery, and daily functioning [5, 27, 28]. Studies based on patient-reported outcomes have also shown that fatigue and loss of appetite are prominent symptoms after pancreatic cancer surgery and may change across the recovery trajectory [28]. Irritability showed lower prevalence and severity, but ranked among the symptoms with relatively high bridge strength. This symptom may reflect the outward expression of physical discomfort, poor sleep, and emotional stress. The bridge symptoms identified in this study suggest that postoperative symptom experience is not confined to a single physical, psychological, or functional domain, but occurs across overlapping domains. The present findings also show that symptom prevalence, severity, and network position provide different types of information. Lack of appetite was the most prevalent symptom, reflecting the common difficulty of oral intake recovery during the early postoperative period. Vomiting was less prevalent but more severe and occupied the most central position in the network. Fatigue was both highly prevalent and prominent in the bridge strength analysis. These differences indicate that traditional descriptive indicators and network indicators reflect different dimensions of symptom burden. Prevalence identifies symptoms that patients most commonly experience, severity reflects symptoms that are more burdensome to patients, and network indicators suggest which symptoms are more closely connected with others. Network analysis should therefore be viewed as a complement to, rather than a replacement for, routine symptom assessment. In this study, its main role was to provide additional information for understanding symptom combinations and the broader recovery context. From the patient perspective, recovery after pancreatic cancer surgery is not a linear accumulation of single symptoms, but a process involving simultaneous changes in oral intake, sleep, physical strength, emotional state, and daily functioning. Patients’ perceptions of recovery often arise from the combined effects of multiple symptoms and their impact on daily activities, treatment confidence, and restoration of normal life. The disease trajectory perspective proposed by Ristau et al. highlights that patients with pancreatic cancer experience an ongoing transition from survival-oriented coping to life reconstruction during illness and treatment [8]. The present findings are consistent with this perspective: postoperative symptom management should not only identify and treat specific symptoms, but also consider how symptom combinations influence recovery experience and functional rebuilding. Future studies may further integrate patient-reported outcomes, qualitative interviews, and longitudinal symptom monitoring to identify symptom patterns most relevant to recovery quality and quality of life. Overall, this study provides a structured perspective for understanding postoperative symptoms in pancreatic cancer. Network analysis can complement traditional descriptive analysis by presenting conditional associations among symptoms and their relative positions within the overall symptom structure. Based on the present findings, future research may focus on whether symptom combinations involving vomiting, distress, lack of appetite, and fatigue are associated with delayed oral intake recovery, prolonged hospitalization, postoperative complications, unplanned healthcare use, or reduced quality of life. Longitudinal and interventional studies are still needed to determine whether network-informed symptom assessment can be translated into clinical strategies with practical value. The findings should be interpreted as exploratory and hypothesis-generating rather than as evidence for prioritizing specific clinical interventions. Centrality and bridge strength indicate patterns of symptom co-occurrence within the estimated network, but they do not establish causal relationships or prove that targeting these symptoms would improve postoperative outcomes. Therefore, the main contribution of this study is to provide a structured description of symptom associations during early recovery and to identify symptom combinations that warrant further evaluation in longitudinal and patient-centered studies.Several limitations should be acknowledged. First, this was a single-center cross-sectional study; therefore, the temporal sequence and causal relationships among symptoms cannot be determined, and the generalizability of the findings requires further validation. Second, symptom assessment was conducted on postoperative day 7 by retrospectively rating symptoms experienced during the first postoperative week, which may have introduced recall bias and could not capture daily symptom fluctuations. Third, the network was constructed mainly using patient-reported symptoms and did not fully incorporate laboratory indicators, perioperative treatment factors, postoperative complications, or recovery outcomes. As a result, the interpretation of some nonspecific symptoms remains limited. Future multicenter, longitudinal, and repeated-measures studies are needed to examine changes in symptom networks during postoperative recovery and to evaluate whether early symptom structures can predict patient-centered outcomes. Supplementary Information Supplementary Material 1. Acknowledgements In addition to the funding agencies, we would also like to extend our thanks for the invaluable contributions by the study participants. Abbreviations Correlation stability Extended Bayesian Information Criterion graphical lasso Global Cancer Observatory M.D. Anderson Symptom Inventory M.D. Anderson Symptom Inventory gastrointestinal cancer module National Comprehensive Cancer Network Postoperative day 7 Strengthening the Reporting of Observational Studies in Epidemiology Authors’ contributions N.X. and N.S. contributed equally to this work.N.X. and N.S. wrote the main manuscript text and performed data analysis.Y.S. collected the data.H.Y. prepared the figures.Y.M. conceived and designed the study.All authors reviewed and approved the manuscript. Funding The research was supported by Nanjing medical science and technology development fund numbered YKK 23258 and Science and Technology Development Project of Nanjing Medical University (Grant No. NMUB20240150). Data availability The original contributions presented in the study are included in the article and data analyzed are available from the corresponding author on reasonable request. Declarations Ethical approval and consent to participate This study complies with the Declaration of Helsinki. This study was approved by the Ethics Committee of the Affiliated BenQ Hospital of Nanjing Medical University (number 2025-KL046). All participants in this study provided written informed consent.The study was reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) statement.Study registration: This observational cross-sectional study was not prospectively registered. 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. Ningning Xia and Neng Shi contributed equally to this work. Contributor Information Ningning Xia, Email: 781901805@qq.com. Yi Miao, Email: miaoyi@njmu.edu.cn. References Associated Data Supplementary Materials Supplementary Material 1. Data Availability Statement The original contributions presented in the study are included in the article and data analyzed are available from the corresponding author on reasonable request.