Perioperative Nutritional Changes Predict Adjuvant S-1 Dose Intensity in Resectable Pancreatic Cancer Authors: HIROO SUTO, AN FUJITA, HIBIKI SAKAI, MARIN MURAMATSU, RYOSUKE WAKAMATSU, AYAKA WAKAMATSU, MIKI KATSURAGAWA, MITSUGU TANAYA, KAZUYA HIURA Published: 2026 ABSTRACT Background/AimMaintaining treatment intensity is important during adjuvant S-1 chemotherapy for resectable pancreatic cancer; however, factors associated with reduced relative dose intensity (RDI) during adjuvant chemotherapy (AC) remain unclear. This study investigated the association between perioperative inflammatory and nutritional indices and the RDI of S-1 during AC.Patients and MethodsThis retrospective single-center study included 69 patients with resectable pancreatic cancer who underwent adjuvant S-1 chemotherapy following neoadjuvant chemotherapy (NAC) and pancreatectomy between November 2019 and February 2025. Patients were classified according to AC S-1 RDI (≥80% vs. <80%). Inflammatory and nutritional indices, as well as changes in prognostic nutritional index (PNI) and geriatric nutritional risk index (GNRI) from NAC initiation to AC initiation, were compared between the groups.ResultsAC S-1 RDI was significantly associated with overall survival and recurrence-free survival (both p=0.02), and patients with RDI ≥80% had significantly better overall survival than those with RDI <80% (p=0.02). At AC initiation, PNI and GNRI were significantly higher in the RDI ≥80% group than in the RDI <80% group. Changes in PNI and GNRI from NAC initiation to AC initiation were also more favorable in the RDI ≥80% group. On multivariate analysis, these nutritional indices were associated with maintenance of RDI ≥80% for S-1 as AC. Receiver operating characteristic analysis showed that GNRI at AC initiation had the highest predictive performance for reduced AC S-1 RDI (<80%) (area under the curve = 0.77; optimal cut-off of 98.7).ConclusionNutritional status at AC initiation and perioperative changes in nutritional status were associated with maintenance of AC S-1 RDI in patients with resectable pancreatic cancer. The GNRI at AC initiation may be useful for assessing the risk of reduced S-1 RDI during AC. Introduction Pancreatic cancer is a malignancy with one of the poorest prognoses, and its 5-year survival rate remains low. Surgical resection is the only potentially curative treatment for pancreatic cancer; however, only approximately 15-20% of patients are eligible for resection (1), and recurrence remains common even after curative resection. Therefore, perioperative treatment plays an important role in preventing recurrence. The Prep-02/JSAP-05 trial demonstrated the usefulness of neoadjuvant chemotherapy (NAC) with gemcitabine plus S-1 (GS therapy) for resectable pancreatic cancer (2). In addition, adjuvant chemotherapy (AC) with S-1 after pancreatic cancer resection was established as a standard treatment based on the JASPAC-01 trial (3). In clinical practice, however, some patients are unable to continue AC as planned because of adverse events or deterioration of their general health condition. Besides treatment completion, the relative dose intensity (RDI) has recently emerged as an essential factor influencing the therapeutic effect of chemotherapy, with reduced RDI associated with poorer treatment outcomes (4-7). However, the factors associated with RDI during AC in pancreatic cancer remain to be fully elucidated. Inflammatory indices, such as the neutrophil-to-lymphocyte ratio (NLR) and platelet-to-lymphocyte ratio (PLR) (8-12), and nutritional indices, such as the prognostic nutritional index (PNI) and geriatric nutritional risk index (GNRI) (13, 14), are associated with chemotherapy tolerance and prognosis in various malignancies, including pancreatic cancer. These indices reflect the systemic inflammatory and nutritional status and have attracted attention as useful markers in perioperative management. However, few studies have examined their association with the RDI of AC during perioperative treatment for pancreatic cancer. Therefore, in this retrospective study, we aimed to identify factors associated with the RDI of S-1 therapy administered as AC in patients who underwent surgery for resectable pancreatic cancer. We focused on the association between AC RDI and perioperative inflammatory and nutritional indices, especially changes in nutritional status from NAC initiation to AC initiation. Patients and Methods Patients. This retrospective single-center observational study included patients diagnosed with resectable pancreatic cancer who underwent NAC with GS therapy, followed by surgical resection, at Sapporo Kosei General Hospital between November 2019 and February 2025. During the study period, 93 patients received NAC with GS therapy. Of these, 11 patients did not undergo pancreatectomy because of disease progression or non-tumor-related reasons. Among the 82 patients who underwent pancreatectomy after NAC, S-1 therapy was initiated as AC in 69 patients who were included in the main analysis (Figure 1 and Figure 2). In contrast, 13 patients did not receive AC postoperatively, owing to disease progression, concerns regarding adverse events, poor performance status, transfer to another hospital, and non-tumor-related reasons. Flow diagram of patient selection. Among 93 patients diagnosed with resectable pancreatic cancer who underwent neoadjuvant chemotherapy (NAC) with gemcitabine plus S-1 (GS) therapy, 82 patients underwent pancreatectomy after NAC. Of these, 69 patients who underwent adjuvant S-1 chemotherapy were included in the main analysis and classified according to an adjuvant chemotherapy (AC) relative dose intensity (RDI) (≥80% vs. <80%). Perioperative treatment schedule. GS therapy: Gemcitabine (GEM) on days 1 and 8 plus S-1 on days 1-14 in a 3-week cycle for 2 cycles. Adjuvant S-1 therapy: S-1 on days 1-28 followed by a 14-day rest period in a 6-week cycle for 4 cycles. BSA: Body surface area. Treatment continuity and AC treatment intensity were evaluated as indices of perioperative treatment. For treatment continuity, patients who completed two cycles of GS therapy as NAC or four cycles of S-1 therapy as AC, regardless of dose reduction, were categorized into the completion group, whereas those who discontinued treatment because of adverse events or patient condition constituted the discontinuation group. To evaluate the treatment intensity of AC, the RDI of S-1 therapy was calculated as the actual cumulative dose divided by the planned cumulative dose. Patients were classified into high-RDI (RDI ≥80%) and low-RDI (RDI <80%) groups. Patient characteristics and surgery-related variables. Patient characteristics at NAC and AC initiation were collected, including sex, age, body mass index (BMI), body surface area, PS, white blood cell count, hemoglobin, platelet count, absolute neutrophil count (ANC), total lymphocyte count, monocyte count, serum albumin, total bilirubin, aspartate aminotransferase, alanine aminotransferase, lactate dehydrogenase, blood urea nitrogen, creatinine clearance, and C-reactive protein. The following data items were also collected at AC initiation: Surgery-related variables, including blood loss, operative time, postoperative complications, and R0 resection status. Postoperative complications were evaluated according to the Clavien-Dindo classification (15), and complications as grade ≥IIIa were defined as major postoperative complications. Laboratory values obtained closest to the initiation of NAC and AC were used. Inflammatory and nutritional indices and their changes. NLR, PLR, lymphocyte-to-monocyte ratio (LMR), PNI, modified Glasgow prognostic score (mGPS), and GNRI at NAC and AC initiation were evaluated (8-13, 16-18). The definition and formula for each index are shown in Supplementary Table I. Higher NLR, PLR, and mGPS values and lower LMR, PNI, and GNRI values indicate poorer systemic inflammatory or nutritional status. Table IPatient characteristics according to the relative dose intensity (RDI) of adjuvant S-1 chemotherapy. Patient characteristics and hematological and biochemical parameters were evaluated at the initiation of adjuvant chemotherapy (AC). Surgery-related variables were evaluated at the time of surgery or during the postoperative period. Values are presented as the mean±standard deviation or n (%).ALB: Albumin; ALT: alanine aminotransferase; ANC: absolute neutrophil count; AST: aspartate aminotransferase; BMI: body mass index; BSA: Body surface area; BUN: blood urea nitrogen; Ccr: creatinine clearance; CRP: C-reactive protein; Hb: hemoglobin; LDH: lactate dehydrogenase; PLT: platelet count; PS: performance status; T-Bil: total bilirubin; TLC: total lymphocyte count; WBC: white blood cell count. aPearson’s chi-squared test; bMann-Whitney U-test. Patient characteristics according to the relative dose intensity (RDI) of adjuvant S-1 chemotherapy. Patient characteristics and hematological and biochemical parameters were evaluated at the initiation of adjuvant chemotherapy (AC). Surgery-related variables were evaluated at the time of surgery or during the postoperative period. Values are presented as the mean±standard deviation or n (%). ALB: Albumin; ALT: alanine aminotransferase; ANC: absolute neutrophil count; AST: aspartate aminotransferase; BMI: body mass index; BSA: Body surface area; BUN: blood urea nitrogen; Ccr: creatinine clearance; CRP: C-reactive protein; Hb: hemoglobin; LDH: lactate dehydrogenase; PLT: platelet count; PS: performance status; T-Bil: total bilirubin; TLC: total lymphocyte count; WBC: white blood cell count. aPearson’s chi-squared test; bMann-Whitney U-test. To evaluate perioperative changes in nutritional status, nutritional indices at NAC and AC initiation were compared. In particular, changes in PNI and GNRI, calculated as the value at AC initiation minus the value at NAC initiation, were assessed. Statistical analysis. Continuous variables are presented as the mean ± standard deviation and categorical variables as a number and percentage. Between-group comparisons were performed using the Mann-Whitney U-test, chi-square test, or Fisher’s exact test, as appropriate. Survival analyses were performed to evaluate the association between completion of GS or S-1 therapy and prognosis. Overall survival (OS) was defined as the time from pancreatectomy to death and recurrence-free survival (RFS) as the time from pancreatectomy to recurrence or death. Survival analyses were performed using the Kaplan-Meier method, and between-group comparisons were performed using the log-rank test. The association between RDI during perioperative chemotherapy and survival outcomes was evaluated using univariable Cox proportional hazards models, with RDI entered as a continuous variable per 10% increase. Patients were classified high-RDI and low-RDI groups according to the RDI of S-1 therapy during AC. Logistic regression analysis was performed to explore factors associated with maintenance of high RDI (RDI ≥80%). Separate multivariable logistic regression models were constructed for each nutritional index and adjusted for BMI, creatinine clearance, and ANC. Creatinine clearance was entered into the model per 10 ml/min increase, and ANC was entered as ×10³/μl. Receiver operating characteristic (ROC) curves were generated to evaluate the predictive ability of each nutritional index for reduced AC RDI (<80%). The area under the curve (AUC) was calculated, and optimal cut-off values were determined using the Youden index. Statistical analyses were performed using JMP® Student Edition 18 (SAS Institute Inc., Cary, NC, USA). A value of p<0.05 was considered statistically significant. No missing values were observed for the variables used in the analyses, thus precluding missing-value imputation. Ethical considerations. This study was conducted in accordance with the Ethical Guidelines for Life Science and Medical Research Involving Human Subjects and was approved by the Ethics Committee of Sapporo Kosei General Hospital, Hokkaido P.W.F.A.C. (approval number: 775). Because this was a retrospective observational study using medical records data, informed consent was obtained through an opt-out approach. Results Patient cohort and association of NAC and AC completion with survival. Among the 69 patients who underwent S-1 therapy as AC, the association between NAC and AC completion and survival outcomes was evaluated. The NAC completion and discontinuation groups comprised 36 and 33 patients, respectively, whereas the AC completion and discontinuation groups numbered 44 and 25 patients, respectively. OS and RFS showed no significant differences according to the NAC completion status (OS: p=0.63; RFS: p=0.83) (Figure 3A). Similarly, survival analysis according to the AC completion status showed no significant difference in OS or RFS (OS: p=0.23; RFS: p=0.20) (Figure 3B). Overall survival (OS; left panels) and recurrence-free survival (RFS; right panels) according to completion of (A) neoadjuvant chemotherapy (NAC) and (B) adjuvant S-1 chemotherapy (AC). Completion was defined as completion of two cycles of gemcitabine-S-1 therapy as NAC or four cycles of S-1 therapy as AC, regardless of dose reduction. Among the 82 patients who underwent pancreatectomy after NAC, 69 (84.1%) underwent AC, whereas 13 (15.9%) did not. Baseline characteristics of patients at NAC initiation according to AC initiation are shown in Supplementary Table II. The association between NAC completion status and survival outcomes for the entire cohort (n=82) is shown in Supplementary Figure 1. Table IIComparison of inflammatory and nutritional indices at neoadjuvant chemotherapy (NAC) and adjuvant chemotherapy (AC) initiation according to the relative dose intensity (RDI) of adjuvant S-1 chemotherapy. Values are presented as the mean±standard deviation or n (%).GNRI: Geriatric nutritional risk index; LMR: lymphocyte-to-monocyte ratio; mGPS: modified Glasgow Prognostic Score; NAC: neoadjuvant chemotherapy; NLR: neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio; PNI: prognostic nutritional index. aMann-Whitney U-test; bFisher’s exact test; cPearson’s chi-squared test. mGPS was analyzed as 0 vs. 1-2. Comparison of inflammatory and nutritional indices at neoadjuvant chemotherapy (NAC) and adjuvant chemotherapy (AC) initiation according to the relative dose intensity (RDI) of adjuvant S-1 chemotherapy. Values are presented as the mean±standard deviation or n (%). GNRI: Geriatric nutritional risk index; LMR: lymphocyte-to-monocyte ratio; mGPS: modified Glasgow Prognostic Score; NAC: neoadjuvant chemotherapy; NLR: neutrophil-to-lymphocyte ratio; PLR: platelet-to-lymphocyte ratio; PNI: prognostic nutritional index. aMann-Whitney U-test; bFisher’s exact test; cPearson’s chi-squared test. mGPS was analyzed as 0 vs. 1-2. Association between perioperative chemotherapy RDI and survival outcomes. The association between RDI during each perioperative chemotherapy and survival outcomes was evaluated using univariable Cox proportional hazards models per 10% increase in RDI (Figure 4). Higher S-1 RDI in AC was significantly associated with better OS and RFS (both p=0.02). In contrast, RDI of GS during NAC was not significantly associated with OS or RFS. Given the primary aim of this study to explore factors associated with treatment intensity during adjuvant S-1 chemotherapy, and the observed association between AC S-1 RDI and survival outcomes in univariate analyses, subsequent analyses focused on factors associated with S-1 RDI in AC. Association of relative dose intensity (RDI) of S-1 in adjuvant chemotherapy (AC) and of gemcitabine (GEM) and S-1 in neoadjuvant chemotherapy (NAC) with survival outcomes. Forest plots showing hazard ratios for association of RDI during NAC and AC with overall (A) and recurrence-free (B) survival according to RDI. Univariate Cox proportional hazards models were used to evaluate the association between each RDI, entered as a continuous variable per 10% increase, and survival outcomes. CI: Confidence interval. The distribution of S-1 RDI in AC is shown in Figure 5A. Among the 69 patients who underwent AC, 34 were classified into the RDI ≥80% group and 35 into the RDI <80% group. Survival analysis according to the 80% cutoff showed that OS was significantly better for the RDI ≥80% group than the RDI <80% group (p=0.02) (Figure 5B). In contrast, RFS did not significantly differ between the groups (p=0.09) (Figure 5C). Relative dose intensity (RDI) of S-1 in adjuvant chemotherapy (AC) and its association with survival outcomes. Patients were classified into two groups according to AC S-1 RDI (≥80% vs. <80%) for survival analyses. (A) Distribution of S-1 RDI during AC. Kaplan-Meier curves for (B) overall (OS) and (C) recurrence-free survival (RFS) according to AC S-1 RDI. The patients’ characteristics and perioperative factors according to AC S-1 RDI are shown in Table I. No significant differences in age, sex, BMI, performance status, and surgery-related variables were observed between the groups. Among the laboratory parameters, only serum albumin was significantly higher for the RDI ≥80% group (p<0.01). Baseline characteristics at NAC initiation according to the S-1 RDI in AC are shown in Supplementary Table III. Table IIIPredictive performance of nutritional indices for relative dose intensity (RDI) of adjuvant S-1 chemotherapy (AC) <80%.AUC: Area under the curve; (Δ)GNRI: (change in) geriatric nutritional risk index; NAC: neoadjuvant chemotherapy; NPV: negative predictive value; (Δ)PNI: (change in) prognostic nutritional index; PPV: positive predictive value. RDI <80% was defined as the positive event in the receiver operating characteristic analysis. PNI AC and GNRI AC indicate values at the initiation of adjuvant chemotherapy. ΔPNI and ΔGNRI were calculated as follows: value at AC initiation minus value at NAC initiation. Values lower than the cut-off indicate a higher risk of reduced S-1 RDI in AC. Predictive performance of nutritional indices for relative dose intensity (RDI) of adjuvant S-1 chemotherapy (AC) <80%. AUC: Area under the curve; (Δ)GNRI: (change in) geriatric nutritional risk index; NAC: neoadjuvant chemotherapy; NPV: negative predictive value; (Δ)PNI: (change in) prognostic nutritional index; PPV: positive predictive value. RDI <80% was defined as the positive event in the receiver operating characteristic analysis. PNI AC and GNRI AC indicate values at the initiation of adjuvant chemotherapy. ΔPNI and ΔGNRI were calculated as follows: value at AC initiation minus value at NAC initiation. Values lower than the cut-off indicate a higher risk of reduced S-1 RDI in AC. Association of inflammatory and nutritional indices with S-1 RDI in AC. Perioperative inflammatory and nutritional indices were compared according to AC RDI (Table II). At NAC initiation, no significant differences in NLR, PLR, LMR, PNI, mGPS, and GNRI were observed between the RDI ≥80% and RDI <80% groups. In contrast, at AC initiation, PNI and GNRI were significantly higher in the RDI ≥80% group (both p<0.01). Changes in nutritional indices from NAC initiation to AC initiation are shown in Figure 6. PNI was significantly higher for the RDI ≥80% group (Figure 6A), whereas no significant difference was observed for the RDI <80% group (Figure 6B). The change in PNI was significantly higher for the RDI ≥80% group (Figure 6C). The GNRI did not significantly change for the RDI ≥80% group (Figure 6D), whereas it was significantly lower for the RDI <80% group (Figure 6E). The change in GNRI was also significantly greater for the RDI ≥80% group (Figure 6F). Changes in nutritional indices from neoadjuvant chemotherapy (NAC) to adjuvant chemotherapy (AC) and their association with the relative dose intensity (RDI) of S-1 in AC. (A) Prognostic nutritional index (PNI) during NAC and AC in patients with RDI ≥80%. (B) PNI during NAC and AC in patients with RDI <80%. (C) Change (Δ) in PNI according to RDI. (D) Geriatric nutritional risk index (GNRI) during NAC and AC in patients with RDI ≥80%. (E) GNRI during NAC and AC in patients with RDI <80%. (F) Change (Δ) in GNRI according to RDI. ΔPNI and ΔGNRI were calculated as follows: value at AC initiation minus value at NAC initiation. Significantly different at: *p<0.05, **p<0.01. Multivariable logistic regression analysis was performed to identify factors associated with maintaining AC S-1 at an RDI ≥80% (Figure 7). Each nutritional index was entered separately into the model, with adjustment for BMI, creatinine clearance, and ANC. BMI, creatinine clearance and ANC were selected as clinical factors reflecting body size, renal function, and bone marrow function, respectively. PNI at AC initiation, change in PNI, GNRI at AC initiation, and change in GNRI were each significantly associated with maintaining an AC S-1 RDI ≥80% after adjustment. Forest plots showing the results of multivariate logistic regression analyses of nutritional indices associated with maintaining a relative dose intensity (RDI) ≥80% for S-1 in adjuvant chemotherapy (AC). (A) Prognostic nutritional index (PNI) at AC initiation. (B) Change (Δ) in PNI from neoadjuvant chemotherapy (NAC) to AC. (C) Geriatric nutritional risk index (GNRI) at AC initiation. (D) Change (Δ) in GNRI from NAC to AC. Odds ratios and 95% confidence intervals (CIs) are shown after adjustment for body mass index (BMI), creatinine clearance (Ccr), and absolute neutrophil count (ANC), entered as continuous variables; Ccr was scaled per 10 ml/min increase, and ANC was expressed as ×10³/μl. ΔPNI and ΔGNRI were calculated as the value at AC initiation minus the value at NAC initiation. Furthermore, ROC analysis was performed to evaluate the predictive performance of these nutritional indices for an AC S-1 RDI <80% (Table III). The AUC values were 0.73, 0.65, 0.77, and 0.70 for PNI at AC initiation, change in PNI, GNRI at AC initiation, and change in GNRI, respectively. The GNRI at AC initiation showed the best predictive performance, with an AUC of 0.77 and accuracy of 78.3%. The optimal cut-offs determined by ROC analysis were 48.1 for PNI at AC initiation, 4.2 for change in PNI, 98.7 for GNRI at AC initiation, and 3.7 for change in GNRI. Therefore having values below these cut-offs may indicate an increased risk of an S-1 RDI <80% in AC. Discussion This study investigated the association between AC S-1 RDI and survival outcomes, as well as the factors associated with AC S-1 RDI, in patients with resectable pancreatic cancer who underwent perioperative treatment. A high S-1 RDI was significantly associated with better OS, suggesting the importance of maintaining treatment intensity during AC. In addition, PNI and GNRI at AC initiation, as well as perioperative changes in nutritional status from NAC initiation to AC initiation (changes in PNI and GNRI), were independently associated with maintenance of an AC S-1 RDI ≥80%. The GNRI at AC initiation showed the highest predictive performance for an S-1 RDI <80% in AC, suggesting its potential usefulness as an indicator of risk of RDI reduction. In this study, neither NAC nor AC completion were significantly associated with OS or RFS. In contrast, with RDI as an indicator of treatment intensity during perioperative chemotherapy, univariable Cox proportional hazards analysis showed that the S-1 RDI in AC was significantly associated with both OS and RFS, whereas gemcitabine and S-1 RDIs in NAC were not. The different associations of RDI with prognosis between NAC and AC may be explained by differences in their treatment objectives. NAC is administered to reduce tumor burden and improve resectability, and its effect on prognosis is ultimately mediated through surgical resection (2). Therefore, differences in dose intensity during NAC may be influenced by multiple factors, including surgical outcomes and postoperative course, and may not be directly reflected in survival outcomes. In addition, because this study included only patients who underwent surgical resection, patients unable to proceed to surgery because of disease progression or a deterioration in their general health condition during NAC were not included. This selection criterion may have attenuated the apparent association between NAC RDI and prognosis. In contrast, AC is administered postoperatively to control micrometastatic disease, and differences in treatment intensity may be more directly reflected in prognosis. In the present study, the choice of an RDI cut-off of 80% was based on a previous report, which observed an association between an RDI <80% during AC and poor prognosis of patients with pancreatic cancer (19). In addition, AC RDI ≥80% has been reported to be associated with better OS in patients with pancreatic ductal adenocarcinoma, including those treated with NAC (20). Furthermore, the RDI distribution of our cohort tended to separate at approximately 80%, supporting the appropriateness of this cut-off value. Patients with an RDI ≥80% had significantly better OS than those with an RDI <80%, consistent with previous findings (19, 20). Although RFS did not differ significantly between the groups, it tended to be better in patients with an RDI ≥80%. The limited sample size may have contributed to the lack of statistical significance. Higher S-1 RDI during AC has also been reported by others to be associated with better OS and RFS (21), whereas delayed initiation of S-1 AC has been associated with worse survival and delayed postoperative nutritional recovery (22). Together, these findings support the clinical importance of maintaining S-1 AC intensity and postoperative nutritional recovery. PNI and GNRI at AC initiation were significantly associated with maintenance of AC RDI ≥80%. In pancreatic cancer, preoperative PNI and postoperative GNRI have been reported to be associated with reduced continuation of S-1 as AC and increased adverse events (23, 24). The present findings extend these observations by showing their association with the S-1 RDI in AC. A distinctive feature of this study is its focus on perioperative changes in nutritional status. At NAC initiation, no significant difference in PNI or GNRI was observed between the groups; however, values at AC initiation and changes in these indices were associated with maintenance of AC S-1 RDI ≥80%. Although the patterns of change differed between PNI and GNRI, these findings suggest that maintaining or improving nutritional status throughout the perioperative period, rather than the preoperative nutritional status alone, may influence the treatment intensity of postoperative AC. This interpretation is also suggested by reports on other cancer types. In patients with colorectal cancer, recovery from low PNI before AC initiation was associated with favorable outcomes (25), while severe malnutrition defined by the Global Leadership Initiative on Malnutrition criteria was an independent poor prognostic factor in patients with biliary tract cancer undergoing adjuvant S-1 chemotherapy (26). These findings support the clinical relevance of nutritional status around AC initiation. Multivariable logistic regression analysis was performed to identify the factors independently associated with maintenance of an AC S-1 RDI ≥80%. In addition to BMI, which is known to influence the continuation of S-1 as AC (27, 28), creatinine clearance and ANC were included as covariates because renal and bone marrow function are related to dose reduction and treatment discontinuation. PNI at AC initiation, change in PNI, GNRI at AC initiation, and change in GNRI were all independently associated with maintenance of an AC S-1 RDI ≥80%. In the ROC analysis, GNRI at AC initiation showed the best predictive performance for an S-1 RDI <80% in AC, with an AUC of 0.77, accuracy of 78.3%, and optimal cut-off of 98.7. This cut-off was practically consistent with the commonly used GNRI criterion of >98, indicating no nutritional risk (18). Therefore, patients with a GNRI below approximately 98 at AC initiation may require careful attention regarding the risk of reduced RDI during postoperative S-1 AC. In addition, previous studies have reported that postoperative body weight loss is associated with reduced treatment intensity and continuity of AC in pancreatic and gastric cancer (19, 29). The present finding that the GNRI, which incorporates body weight, showed the best predictive performance is consistent with these reports. This study has several limitations. Firstly, being a retrospective single-center study with a limited sample size, its statistical power may have been insufficient. Although age did not significantly differ between the two RDI groups, patients in the group with RDI <80% tended to be older (p=0.08), suggesting residual confounding. Secondly, this study only included patients who underwent AC, excluding those who did not. Therefore, the effects of a poor nutritional status or deterioration in their general health condition may have been underestimated. Thirdly, detailed factors that may influence nutritional status, such as nutritional intervention, energy intake, skeletal muscle mass, and body composition, could not be fully evaluated. Fourthly, causal relationships could not be established because this was an observational study. In particular, RDI was determined after AC initiation; therefore, the association between RDI and prognosis should be interpreted as exploratory, considering the possibility of time-dependent bias. Future prospective studies are needed to determine whether improvement in the GNRI directly contributes to maintaining the RDI or improving prognosis. Despite these limitations, this study suggests that a patient’s nutritional status at AC initiation and perioperative changes in nutritional status are associated with maintenance of treatment intensity of postoperative S-1 AC. Further multicenter prospective studies are warranted to validate the cut-off and predictive performance of the GNRI and evaluate comprehensive perioperative management strategies, including nutritional intervention. Conclusion In patients with resectable pancreatic cancer, maintenance of RDI during postoperative S-1 AC was associated with better survival outcomes. The GNRI at AC initiation and perioperative changes in GNRI were independently associated with maintenance of an AC S-1 RDI ≥80% and may be useful for predicting the risk of a reduced AC RDI. These findings suggest that evaluating and managing perioperative nutritional status and its changes may help maintain treatment intensity during AC and support the development of individualized treatment strategies. Further prospective studies are needed to validate the cut-off and predictive performance of the GNRI in larger cohorts. Supplementary Material Available at: https://doi.org/10.6084/m9.figshare.32947901 Conflicts of Interest The Authors have no conflicts of interest to declare. Authors’ Contributions Conceptualization: HSu, RW, and KH. Data curation: AF, HSa, and MM. Formal analysis: AF, HSa, MM, and KH. Investigation: AF, HSa, MM, AW, MK, and MT. Methodology: HSu, RW, and KH. Project administration: KH. Resources: AW, MK, and MT. Software: AF, HSa, MM, and KH. Supervision: KH. Validation: HSu, RW, AW, MK, MT, and KH. Visualization: AF, HSa, MM, and KH. Writing - original draft: HSu and KH. Writing – review and editing: all Authors. All Authors read and approved the final manuscript. Acknowledgements The Authors would like to thank Editage (www.editage.jp) for English language editing. Artificial Intelligence (AI) Disclosure An artificial intelligence-based language model (ChatGPT, OpenAI, San Francisco, CA, USA) was used to assist with language refinement and clarity. All AI-assisted content was carefully reviewed, verified, and revised by the Authors, who take full responsibility for the integrity and accuracy of the final manuscript.