Li YC, Pan D, Liu HN, Pei ZC, Li YQ, Han ZX, Liu WL. Baseline cytokines and programmed death-ligand 1 prognostic value in advanced gastric cancer with sintilimab plus chemotherapy. World J Gastroenterol 2026; 32(40): 120447 [DOI: 10.3748/wjg.120447]
Corresponding Author of This Article
Wen-Lou Liu, MD, PhD, Department of Oncology, The Affiliated Hospital of Xuzhou Medical University, No. 99 Kunpeng Road, Economic and Technological Development Zone, Xuzhou 221000, Jiangsu Province, China. liuwenlou@163.com
Research Domain of This Article
Gastroenterology & Hepatology
Article-Type of This Article
research-article
Open-Access Policy of This Article
This article is an open-access article which was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/
Baishideng Publishing Group Inc, 7041 Koll Center Parkway, Suite 160, Pleasanton, CA 94566, USA
Share the Article
Li YC, Pan D, Liu HN, Pei ZC, Li YQ, Han ZX, Liu WL. Baseline cytokines and programmed death-ligand 1 prognostic value in advanced gastric cancer with sintilimab plus chemotherapy. World J Gastroenterol 2026; 32(40): 120447 [DOI: 10.3748/wjg.120447]
Yong-Cheng Li, Department of Medical Oncology, Xuzhou Central Hospital, Southeast University, Xuzhou 221009, Jiangsu Province, China
Di Pan, Hao-Nan Liu, Zi-Cheng Pei, Yu-Qi Li, Zheng-Xiang Han, Wen-Lou Liu, Department of Oncology, The Affiliated Hospital of Xuzhou Medical University, Xuzhou 221000, Jiangsu Province, China
Co-corresponding authors: Zheng-Xiang Han and Wen-Lou Liu.
Author contributions: Liu WL and Han ZX designed the study and they contribute equally to this study as co-corresponding authors; Li YC and Pan D contribute equally to this study as co-first authors; Liu HN, Pan D and Li YQ collected the clinical data and analyzed the data; Pan D, Li YQ and Liu WL wrote the paper; Liu WL, Pan D, Li YQ and Pei ZC revised the paper; all authors contributed to the article and approved the submitted version.
Institutional review board statement: The study protocol was reviewed and approved by the Ethics Committee of the Affiliated Hospital of Xuzhou Medical University (Approval No. XYFY2024-KL408-01).
Informed consent statement: The requirement for written informed consent was waived by the Ethics Committee of the Affiliated Hospital of Xuzhou Medical University due to the retrospective nature of this study.
Conflict-of-interest statement: There is no conflict of interest associated with any of the senior author or other coauthors contributed their efforts in this manuscript.
STROBE statement: The authors have read the STROBE Statement—checklist of items, and the manuscript was prepared and revised according to the STROBE Statement—checklist of items.
Data sharing statement: The data included in this study can be obtained from the corresponding author.
Corresponding author: Wen-Lou Liu, MD, PhD, Department of Oncology, The Affiliated Hospital of Xuzhou Medical University, No. 99 Kunpeng Road, Economic and Technological Development Zone, Xuzhou 221000, Jiangsu Province, China. liuwenlou@163.com
Received: February 27, 2026 Revised: March 31, 2026 Accepted: April 21, 2026 Published online: October 28, 2026 Processing time: 199 Days and 21.6 Hours
Abstract
BACKGROUND
Gastric cancer is a major global health challenge, with sintilimab plus chemotherapy now established as a first-line treatment. However, identifying reliable biomarkers to predict individual patient responses remains a significant clinical hurdle. While cytokines are key modulators of the tumor microenvironment and immune evasion, their specific prognostic value in advanced gastric cancer (AGC) patients receiving this combination therapy has not been fully elucidated. This study investigates whether baseline peripheral cytokine levels can serve as effective biomarkers to improve risk stratification and predict long-term clinical outcomes in this patient population.
AIM
To investigate the relationship between peripheral blood cytokines and long-term prognosis in AGC patients treated with sintilimab combined with chemotherapy.
METHODS
Clinical and pathological information, along with pretreatment peripheral blood cytokine levels, were collected from 101 patients who received sintilimab in conjunction with chemotherapy at the Affiliated Hospital of Xuzhou Medical University (Xuzhou, China) between January 2021 and January 2023. To determine optimal cutoff values for baseline cytokines, receiver operating characteristic (ROC) curves were generated using cytokine levels measured prior to immunotherapy, subsequently classifying patients into high and low cytokine level groups. The correlations between cytokines and clinicopathological factors were assessed using both the χ2 test and t-test. To evaluate the dynamic predictive value of continuous cytokine levels at 9-, 12-, and 18-month, time-dependent ROC curves were utilized. The Kaplan-Meier method and log-rank tests facilitated the comparison of survival curves. Variables showing P < 0.05 in univariate Cox regression analysis were chosen for the least absolute shrinkage and selection operator-Cox multivariate regression analysis to pinpoint independent prognostic factors for progression-free survival (PFS) and overall survival (OS). A nomogram for predicting OS was constructed using the entire cohort and was internally validated through 1000 bootstrap resamples. To evaluate the robustness of the model, ROC curves, calibration curves, and concordance indices (C-indices) were calculated at 9-, 12-, and 18-month. The primary endpoints of the study were OS and PFS. All statistical tests were two-tailed, considering significance at α = 0.05 (P < 0.05).
RESULTS
Optimal cytokine cutoff values were determined using ROC curves: Interleukin (IL)-4 0.34 pg/mL, IL-6 2.12 pg/mL, IL-8 9.09 pg/mL, IL-10 4.10 pg/mL, and IL-17 3.03 pg/mL. Groups with low IL-6, IL-10, and IL-17 had significantly better PFS and OS (P < 0.05), while IL-4 and IL-8 had no impact. After adjusting for the specific chemotherapy backbone using Firth’s penalized Cox regression, high levels of IL-10 and IL-17 remained highly significant independent risk predictors for both PFS and OS, while IL-6 was an independent predictor exclusively for PFS. Furthermore, programmed death-ligand 1 (PD-L1) positivity and positive Epstein-Barr virus status were also identified as significant independent prognostic factors for OS.
CONCLUSION
PD-L1 expression and peripheral blood cytokine levels of IL-10 and IL-17 are independent prognostic factors for OS, while IL-6 is an independent prognostic factor exclusively for PFS in AGC patients treated with sintilimab combined with chemotherapy, potentially serving as biomarkers to identify patients who benefit from immunotherapy.
Core Tip: This study explores advanced gastric cancer (AGC) patients receiving first-line sintilimab plus chemotherapy, identifying baseline interleukin (IL)-6, IL-10, IL-17 and programmed death-ligand 1 as independent overall survival prognostic factors. A well-validated nomogram model is established, and these peripheral blood indices serve as novel biomarkers for screening AGC patients who benefit from immunotherapy.
Citation: Li YC, Pan D, Liu HN, Pei ZC, Li YQ, Han ZX, Liu WL. Baseline cytokines and programmed death-ligand 1 prognostic value in advanced gastric cancer with sintilimab plus chemotherapy. World J Gastroenterol 2026; 32(40): 120447
Gastric cancer ranks among the most prevalent cancers globally, especially in Asian nations such as China[1]. The main treatment approaches for early-stage gastric cancer involve surgical removal and chemotherapy. Nevertheless, the majority of patients receive a diagnosis at more advanced stages, which obstructs the possibility of curative treatment and results in a substantial unmet clinical need for this patient population[2,3].
Sintilimab is a monoclonal antibody targeting programmed death-ligand 1 (PD-L1). By binding to programmed death 1 (PD-1) and blocking its interaction with PD-L1 and programmed death-ligand 2, sintilimab inhibits PD-1-mediated immunosuppression, activates T-cell function, enhances tumor immunosurveillance and cytotoxicity, and induces antitumor immune responses[4]. Multiple studies have demonstrated the efficacy of sintilimab in treating malignancies, such as gastric, esophageal, liver, and lung cancers[5-9]. Results from the ORIENT-16 trial[10], presented at the 2023 American Association for Cancer Research Annual Meeting, showed that sintilimab plus chemotherapy significantly reduced mortality risk in both the overall population and the PD-L1 Combined Positive Score (CPS) ≥ 5 subgroup. Median overall survival (OS) was prolonged by 2.9 months in the overall population and 6.3 months in the CPS ≥ 5 subgroup, with consistent beneficial trends across all prespecified subgroups. These findings established sintilimab plus chemotherapy as a first-line treatment for gastric cancer with survival benefits across all populations. Based on these results, the China National Medical Products Administration approved sintilimab combined with chemotherapy for first-line treatment of gastric and gastroesophageal junction adenocarcinoma, making it China’s first PD-1 inhibitor reimbursed by national medical insurance for all first-line gastric cancer patients. However, not all patients benefit from immunotherapy[11]. Given the high cost and potentially severe toxicities of immunotherapy, identifying predictive biomarkers to select patients who would benefit from treatment remains a major challenge[12].
Currently, there are a limited number of biomarkers that can predict the response to immunotherapy, especially in peripheral blood. Among the existing biomarkers are PD-L1 expression, the ratio of neutrophils to lymphocytes, the ratio of platelets to lymphocytes, and various subsets of lymphocytes in peripheral blood[13,14], yet their application in clinical settings is constrained[15]. Consequently, it is crucial to develop biomarkers that are reliable, sensitive, and specific for forecasting outcomes of immunotherapy. Cytokines are integral to the processes of cancer development, progression, and regulation. For instance, interleukin (IL)-4 facilitates interactions within the tumor microenvironment (TME) by activating tumor-associated macrophages and myeloid-derived suppressor cells[16,17]; IL-6 and IL-8 trigger angiogenesis mediated by vascular endothelial growth factor, promoting tumor vascularization[18]; IL-10 boosts antibody production, aids humoral immunity, and inhibits antitumor immune responses[19]; and IL-17 fosters an epithelial-mesenchymal transition-like transformation in gastric cancer stem cells, which increases their invasiveness[20]. Although cytokines in peripheral blood have been confirmed as predictive biomarkers for immunotherapy across melanoma, lung, and esophageal cancers, their prognostic significance in patients with advanced gastric cancer (AGC) undergoing treatment with sintilimab in combination with chemotherapy is still unclear. This study aims to explore the correlations between the levels of IL-4, IL-6, IL-8, IL-10, and IL-17 and treatment outcomes in AGC patients receiving sintilimab plus chemotherapy.
MATERIALS AND METHODS
Clinical data of patients
A retrospective analysis was conducted on the clinical data of patients with AGC who were treated at the Affiliated Hospital of Xuzhou Medical University (Xuzhou, China) between January 2021 and January 2023.
The inclusion criteria: (1) Histologically verified gastric or gastroesophageal junction adenocarcinoma; (2) Imaging evidence supporting unresectable locally advanced or metastatic disease; (3) Initiation of first-line treatment with the combination of sintilimab and chemotherapy; (4) Availability of peripheral blood cytokine measurements collected within one week prior to the initial chemotherapy with sintilimab; (5) Having undergone a minimum of 2 cycles of sintilimab; (6) Being at least 12 months post-curative surgery; and (7) HER2 status negative or unknown.
The exclusion criteria: (1) Previous antitumor therapies; (2) Recent use of steroids, immunosuppressants, or anti-inflammatory medications; (3) A history of autoimmune disorders; (4) Presence of active severe infections or systemic inflammation; (5) Significant dysfunctions related to cardiac, hepatic, or renal systems; (6) Diagnosis of concurrent cancers; and (7) Incomplete clinical information.
Ethical approval for this study was granted by the Ethics Committee of the Affiliated Hospital of Xuzhou Medical University, and informed consent was acquired from all participants. All research procedures were conducted in accordance with the ethical principles of the Declaration of Helsinki. The process for patient selection and exclusion is depicted in the patient flow diagram (Supplementary Figure 1).
Clinical data collection
Clinical information, such as age, gender, clinical stage, and laboratory test findings, was obtained from the hospital’s electronic medical record system. Cytokine levels (IL-4, IL-6, IL-8, IL-10, and IL-17) were documented prior to treatment (within one week before immunotherapy). The assessment of tumor response adhered to the Response Evaluation Criteria in Solid Tumors (RECIST 1.1) guidelines, along with baseline computed tomography (CT) scans conducted prior to treatment initiation. Follow-up CT scans were carried out every three months, commencing from the third cycle.
Regarding long-term outcomes, OS was calculated from the date of the first treatment administration until death from any etiology. We determined progression-free survival (PFS) as the duration between the onset of therapy and either the first documented evidence of disease advancement or patient mortality.
Laboratory methods
Cytokine concentrations (IL-4, IL-6, IL-8, IL-10, and IL-17) were evaluated using enzyme-linked immunosorbent assay. Peripheral blood specimens were obtained, centrifuged to separate serum, diluted to ensure they fell within standardized concentration ranges, and subsequently analyzed using colorimetric methods. The assessment of PD-L1 expression was performed through immunohistochemistry (IHC) with PD-L1-specific antibodies, and the results were expressed as CPS. Serum tumor markers, including carcinoembryonic antigen (CEA), alpha-fetoprotein (AFP), carbohydrate antigen (CA) 199, CA724, and CA125, were quantified via chemiluminescent immunoassays. The status of microsatellite instability (MSI) was evaluated using IHC techniques and MSI testing. The normal ranges for the tumor markers are as follows: AFP ≤ 20 ng/mL[21], CEA ≤ 5.8 ng/mL[22], CA199 ≤ 37 U/mL[23], CA125 ≤ 37 U/mL[24], CA724 ≤ 6.9 U/mL[25]. Baseline peripheral cytokine concentrations were retrospectively retrieved from the clinical laboratory database at the Affiliated Hospital of Xuzhou Medical University.
Blood samples (5 mL) were obtained from fasting peripheral venous sources in the morning and placed into standard tubes that did not contain anticoagulants. These samples were left to coagulate naturally at room temperature for at least 30 minutes before being centrifuged at 1000 × g for 15 minutes. The serum that separated was then collected and stored at -20 °C for future analysis. Before proceeding with the analysis, the frozen samples were entirely thawed, mixed thoroughly, and centrifuged to eliminate any particulates, with the number of freeze-thaw cycles limited to a maximum of two. Levels of IL-4, IL-6, IL-8, IL-10, and IL-17 in the serum were measured using a multiplex bead-based flow fluorescent immunoassay. These quantitative analyses were conducted on a BD FACSCantTM II flow cytometer (Becton, Dickinson and Company, NJ, United States) employing specific assay kits (12-Cytokine Detection Kit, Catalog Number: RS-12-01, Registration No. 20222401165) from Qingdao Raisecare Biotechnology Co., Ltd. (Qingdao, China). All methodologies were performed in strict compliance with the manufacturer’s guidelines and established clinical laboratory protocols, accompanied by ongoing internal quality control. In alignment with clinical laboratory performance validation standards, the coefficients of variation for both intra-assay and inter-assay were confirmed to be ≤ 15%. PD-L1 expression was evaluated using IHC through the 22C3 pharmDx assay. The CPS was determined by pathologists and was calculated by taking the total count of PD-L1-stained cells (inclusive of tumor cells, lymphocytes, and macrophages), dividing this by the total number of viable tumor cells, and then multiplying by 100.
Treatment protocols
Sintilimab was administered intravenously at 200 mg every 3 weeks. Chemotherapy regimens included SOX (oxaliplatin + tegafur/gimeracil/oteracil), XELOX (oxaliplatin + capecitabine), docetaxel, or albumin-bound paclitaxel, dosed individually and infused intravenously every 3 weeks.
Statistical analysis
Qualitative data were summarized by means of n (%), while quantitative data were presented using median values alongside their respective interquartile ranges (IQRs). To establish the optimal cut-off points for baseline peripheral cytokines, receiver operating characteristic (ROC) curve analysis was employed, focusing on maximizing the Youden index; this approach transformed continuous data into dichotomous variables (classified as high or low) for clinical relevance. In assessing survival outcomes, both PFS and OS were evaluated. The Kaplan-Meier method helped estimate survival probabilities, and group differences were assessed using the log-rank test. A univariate Cox proportional hazards regression analysis was conducted on the complete cohort (n = 101) to identify potential prognostic indicators, calculating hazard ratios (HRs) and 95%CIs. To identify the most significant prognostic features while reducing the risk of overfitting, a least absolute shrinkage and selection operator (LASSO)-Cox regression model with 10-fold cross-validation was implemented. Features chosen by the LASSO model were then included in the multivariate analysis. Notably, to tackle the challenges posed by monotone likelihood and the extreme HRs recorded in the univariate analysis (which often leads to inflated maximum likelihood estimates), Firth’s penalized Cox proportional hazards regression was applied to all multivariate models, utilizing the coxphf package. Additionally, to mitigate potential confounding bias stemming from treatment variability, the specific chemotherapy regimen (paclitaxel-based vs platinum doublets) was included as a covariate in all multivariate Firth Cox models, irrespective of its significance in the univariate analysis. To examine the evolving prognostic significance of cytokines treated as continuous variables, time-dependent ROC curve analysis was carried out utilizing the timeROC package to compute the area under the curve (AUC) at 9-, 12-, and 18-month. Ultimately, a prognostic nomogram for OS was developed based on the independent variables identified in the multivariate Firth Cox model. Rather than employing a split-sample technique, internal validation was meticulously conducted through 1000 bootstrap resamples of the entire cohort. This approach yielded an optimism-adjusted concordance index (C-index) and calibration curves (for 9-, 12-, and 18-month) to assess the model’s ability to discriminate and its predictive precision. All statistical evaluations were executed using R software (version 4.4.0), with a two-sided P value of less than 0.05 regarded as statistically significant.
RESULTS
Baseline characteristics of the study population
A total of 101 patients with AGC who received sintilimab combined with chemotherapy were included in this study. The baseline characteristics are detailed in Table 1. The median age was 63 years (IQR 54-71 years), and 54 patients (53.5%) were male. The majority of patients had an Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 1 (62.4%) and metastatic disease (86.1%). Regarding the chemotherapy regimen, 48 patients (47.5%) received platinum doublets (SOX/XELOX), while 53 patients (52.5%) received paclitaxel-based regimens. The PD-L1 positive rate was 79.2%. Based on the optimal cut-off values determined by ROC curves, patients were categorized into high-expression groups for IL-6 (32.7%), IL-10 (37.6%), and IL-17 (50.5%). At the data cutoff, 82 PFS events (81.2%) and 60 OS events (59.4%) were observed. The median PFS for the entire cohort was 8.36 months (95%CI: 8.14-8.58), and the median OS was 14.94 months (95%CI: 11.2-18.9).
Table 1 Comparison of clinical data of patients, n (%).
Determination of optimal cut-off values for cytokines
The analysis of the ROC curve related to predicting mortality based on cytokine levels in patients prior to immunotherapy is illustrated in Table 2. Figure 1 depicts the ROC curve, revealing the AUCs associated with diagnosing death using baseline levels of IL-4, IL-6, IL-8, IL-10, and IL-17 prior to treatment, which were measured at 0.678, 0.651, 0.667, 0.779, and 0.654, respectively. From the ROC curve analysis, the baseline cytokine levels of IL-4, IL-6, IL-8, IL-10, and IL-17 were categorized into two groups: High cytokine level (high-level) and low cytokine level (low-level). IL-4 was defined as ≤ 0.34 pg/mL and > 0.34 pg/mL, demonstrating a sensitivity of 95.0% alongside a specificity of 53.7%; IL-6 was categorized into ≤ 2.12 pg/mL and > 2.12 pg/mL, exhibiting a sensitivity of 73.3% and a specificity of 65.9%; IL-8 was classified as ≤ 9.09 pg/mL and > 9.09 pg/mL, showing a sensitivity of 94.2% and a specificity of 51.3%; IL-10 was segmented into ≤ 4.10 pg/mL and > 4.10 pg/mL, resulting in a sensitivity of 91.7% and a specificity of 75.6%; finally, IL-17 was classified as ≤ 3.03 pg/mL and > 3.03 pg/mL, achieving a sensitivity of 97.6% and a specificity of 39.0%.
Figure 1 Receiver operating characteristic curves for prediction of death by pre-immunotherapy cytokine levels in gastric cancer patients.
ROC: Receiver operating characteristic; AUC: Area under the curve; IL: Interleukin; OS: Overall survival.
Table 2 Receiver operating characteristic area under the curve for prediction of death in patients before immunotherapy by of cell factors.
Correlation between clinicopathological features and pretreatment levels of cytokines
In AGC patients, Spearman rank correlation analysis revealed that PD-L1 expression was significantly and positively correlated with both IL-6 (rho = 0.300, P = 0.002) and IL-10 (rho = 0.231, P = 0.020) levels. However, while group-wise comparisons (Table 3) showed distribution differences, the continuous rank correlation did not reach statistical significance for age with IL-6 (rho = -0.185, P = 0.064) or IL-10 (rho = 0.040, P = 0.689). Similarly, no significant correlation was observed between CA19-9 and IL-10 (rho = -0.029, P = 0.771). No other clinicopathological features [body mass index, ECOG PS, Sex, stage, Epstein-Barr virus (EBV), and MSI status] showed significant associations with cytokine levels (P > 0.05; Table 3).
Table 3 Comparison of cytokine distribution with different clinical characteristics, n (%).
The PFS of the IL-4 low-level group was 8.46 months (95%CI: 8.30-8.73), while the PFS of the high-level group was 7.91 months (95%CI: 7.70-8.52), with no significant difference (P = 0.174, Figure 2A). The PFS of the IL-6 low-level group was 8.58 months (95%CI: 8.42-8.75), which was significantly better than that of the high-level group (7.57 months, 95%CI: 7.10-7.84, P < 0.001, Figure 2B). However, there was no significant difference in PFS between the IL-8 low-level group and the high-level group (8.58 months, 95%CI: 8.35-8.75 vs 8.22 months, 95%CI: 7.91-8.52, P = 0.554, Figure 2C). The PFS of the IL-10 low-level group was significantly prolonged compared to the high-level group (8.66 months, 95%CI: 8.52-9.00 vs 7.59 months, 95%CI: 7.30-7.84, P < 0.001, Figure 2D). The PFS of the IL-17 low-level group was 8.70 months (95%CI: 8.52-9.20), which was significantly better than that of the high-level group (7.84 months, 95%CI: 7.69-8.22, P < 0.001, Figure 2E).
Figure 2 Survival curves of progression free survival for two groups of patients.
A: Interleukin (IL)-4; B: IL-6; C: IL-8; D: IL-10; E: IL-17. The shaded areas represent the 95%CIs. IL: Interleukin.
The OS of the IL-4 low-level group was 15.03 months (95%CI: 11.34-19.07), which was similar to that of the high-level group [11.68 months, 95%CI: 9.00-not available (NA), P = 0.831, Figure 3A]. The OS of the IL-6 low-level group was significantly better than that of the high-level group (18.80 months, 95%CI: 13.90-19.19 vs 8.03 months, 95%CI: 7.57-NA, P < 0.001, Figure 3B). There was no statistically significant difference in OS between the IL-8 low-level group and the high-level group (18.42 months, 95%CI: 13.90-NA vs 10.96 months, 95%CI: 9.66-18.96, P = 0.098, Figure 3C). The OS of the IL-10 low-level group was significantly prolonged compared to the high-level group (18.94 months, 95%CI: 17.09-19.20 vs 8.03 months, 95%CI: 7.62-11.08, P < 0.001, Figure 3D). The OS of the IL-17 low-level group was 19.15 months (95%CI: 18.80-19.26), which was also significantly better than that of the high-level group (9.66 months, 95%CI: 8.03-11.08, P < 0.001, Figure 3E).
Figure 3 Survival curves of overall survival for two groups of patients.
A: Interleukin (IL)-4; B: IL-6; C: IL-8; D: IL-10; E: IL-17. The shaded areas represent the 95%CIs. IL: Interleukin.
Univariate survival analysis
Univariate Cox regression analysis was performed on the entire cohort to identify prognostic factors (Table 4). For PFS, high expressions of all three cytokines were significantly associated with an increased risk of progression: IL-10 (HR = 10.00, P < 0.001), IL-6 (HR = 8.12, P < 0.001), and IL-17 (HR = 3.72, P < 0.001). Conversely, PD-L1 positivity was a strong protective factor (HR = 0.10, P < 0.001). Notably, the chemotherapy regimen (paclitaxel-based vs platinum doublets) showed no significant association with either PFS (HR = 1.02, P = 0.945) or OS (HR = 0.85, P = 0.538), confirming baseline comparability between treatment groups. The prognostic patterns for OS were consistent with PFS, with high expressions of IL-10, IL-6, and IL-17 all demonstrating significant adverse associations (all P < 0.001).
Table 4 Single factor Cox proportional hazards regression analysis of progression-free survival and overall survival.
To reduce dimensionality and select the most robust features, a 10-fold cross-validated LASSO-Cox regression model was applied to the complete cohort. For PFS, the model retained IL-6, IL-10, and IL-17 at the λ1se value (Figure 4). For OS, IL-10 and IL-17 were retained at λ1se, while PD-L1, EBV status, and IL-6 were further incorporated at λmin (Figure 5). To construct a parsimonious and robust final multivariate OS model, we incorporated the core predictors retained at λ1se (IL-10 and IL-17), key clinical factors identified at λmin (PD-L1 and EBV status), and forced in the chemotherapy regimen to adjust for treatment heterogeneity. IL-6 was excluded from the final OS multivariate model as its independent prognostic value was primarily confined to PFS.
Figure 4 Least absolute shrinkage and selection operator-Cox regularization for progression-free survival.
A: Ten-fold cross-validation curve showing partial likelihood deviance as a function of log(λ); vertical dashed line indicates λ.min and dot-dashed line indicates λ.1se. Numbers along the top axis represent the count of non-zero coefficients at each λ; B: Coefficient shrinkage paths for all candidate variables; vertical lines correspond to λ.min and λ.1se. Three variables [interleukin (IL)-6 high, IL-10 high, IL-17 high] were retained at λ.1se. (n = 101). PFS: Progression-free survival; CV: Cross validation; EBV: Epstein-Barr virus; IL: Interleukin; PD-L1: Programmed death-ligand 1.
Figure 5 Least absolute shrinkage and selection operator-Cox regularization for overall survival.
A: Ten-fold cross-validation curve of partial likelihood deviance vs log(λ); vertical lines indicate λ.min and λ.1se; B: Coefficient shrinkage paths for all candidate variables. At λ.1se, interleukin (IL)-10 high and IL-17 high were retained as core predictors; IL-6, programmed death-ligand 1, and Epstein-Barr virus were additionally selected at λ.min (n = 101). OS: Overall survival; CV: Cross validation; EBV: Epstein-Barr virus; IL: Interleukin; PD-L1: Programmed death-ligand 1.
Multivariate firth penalized Cox regression
Given the extreme HRs observed in the univariate analysis (e.g., IL-10, HR = 10.00) indicative of monotone likelihood, Firth’s penalized Cox regression was utilized to obtain bias-corrected, robust estimates (Tables 5 and 6).
Table 5 Multivariable firth-penalized Cox regression for progression-free survival.
In the multivariate PFS model, high expressions of IL-6 (HR = 3.82, P < 0.001), IL-10 (HR = 4.38, P < 0.001), and IL-17 (HR = 1.94, P = 0.041) remained independent prognostic factors for progression, while PD-L1 positivity remained an independent favorable factor (HR = 0.35, P = 0.006).
In the multivariate OS model, high IL-10 (HR = 3.35, 95%CI: 1.45-7.81, P = 0.005) and high IL-17 (HR = 3.94, 95%CI: 1.93-8.11, P < 0.001) were confirmed as independent adverse prognostic factors, whereas PD-L1 positivity (HR = 0.32, 95%CI: 0.14-0.77, P = 0.012) improved OS. Additionally, positive EBV status was identified as an independent adverse prognostic factor for OS (HR = 2.90, 95%CI: 1.09-6.73, P = 0.034). Crucially, the chemotherapy regimen showed no independent prognostic value for either PFS (P = 0.605) or OS (P = 0.623) after adjusting for cytokine profiles, indicating that the prognostic value of these biomarkers is independent of the chemotherapy backbone.
Internal validation and prognostic nomogram
Based on the multivariate OS model, a prognostic nomogram integrating IL-10, IL-17, PD-L1, EBV status, and chemotherapy regimen was constructed to predict survival probabilities at 9-, 12-, and 18-month (Figure 6). Internal validation via 1000 bootstrap resamples yielded an original C-index of 0.792, with an excellent optimism-corrected C-index of 0.779, demonstrating minimal overfitting. The calibration curves at 9-, 12-, and 18-month (Figure 7) exhibited high concordance between the nomogram-predicted probabilities and the actual Kaplan-Meier survival estimates.
Figure 6 Nomogram for individualized prediction of 9-, 12-, and 18-month overall survival.
Each variable is assigned a point score based on its Firth-penalized Cox regression coefficient; the sum of individual scores corresponds to the predicted survival probability at each time point. The nomogram incorporates interleukin (IL)-10 expression status, IL-17 expression status, programmed death-ligand 1 expression, Epstein-Barr virus status, and chemotherapy regimen. OS: Overall survival; EBV: Epstein-Barr virus; IL: Interleukin; PD-L1: Programmed death-ligand 1.
Figure 7 Calibration curves for the overall survival nomogram at 9-, 12-, and 18-month.
Patients were stratified into quartiles according to nomogram-predicted survival probability. Points represent the mean predicted probability vs Kaplan-Meier-observed survival for each quartile; error bars indicate 95% confidence intervals. The dashed diagonal represents perfect calibration. Original concordance index (C-index) = 0.792; 1000-bootstrap optimism-corrected C-index = 0.779 (optimism = 0.026). K-M: Kaplan-Meier; C-index: Concordance index.
Time-dependent ROC analysis of continuous cytokines
To evaluate the longitudinal dynamic predictive value of cytokines as continuous variables, time-dependent ROC analysis was performed for OS (Figure 8). The discriminatory power of continuous cytokine levels was generally modest across time points (e.g., the 9-month AUC for IL-10 was 0.534, 95%CI: 0.410-0.657). The apparent discrepancy between the modest AUCs of continuous values and the robust prognostic HRs of their dichotomized counterparts highlights a profound non-linear threshold effect. This finding underscores the clinical necessity of converting continuous cytokine levels into specific cut-off values for effective risk stratification.
Figure 8 Time-dependent receiver operating characteristic curves for continuous cytokine markers predicting overall survival at 9-, 12-, and 18-month.
A-C: Interleukin (IL)-10 (A), IL-17 (B), and IL-8 (C) are shown. Area under the curve (AUC) values with 95% confidence intervals were estimated using the inverse probability of censoring weighting method. The dashed diagonal indicates the random classifier reference line (AUC = 0.50). ROC: Receiver operating characteristic; AUC: Area under the curve; IL: Interleukin.
DISCUSSION
To our knowledge, this study represents the first extensive investigation into how cytokines influence the effectiveness of combination chemotherapy with sintilimab in treating AGC, and it aims to create a predictive model for assessing prognosis. The findings of our research indicated that patients with low levels of IL-6, IL-10, and IL-17 experienced better PFS than those with high levels of these cytokines. Conversely, IL-4 and IL-8 did not show a significant effect on PFS. Additionally, OS was markedly improved in the low-level groups of IL-6, IL-10, and IL-17; however, the OS for low-level groups of IL-4 and IL-8 did not show a significant difference when compared to their high-level counterparts. Furthermore, subgroup analysis concerning PD-L1 expression revealed that individuals with a CPS of 5 or higher had superior PFS and OS compared to those with a CPS lower than 5. Based on the Cox regression analysis, PD-L1 expression along with IL-10 and IL-17 Levels were recognized as independent prognostic factors affecting both PFS and OS, whereas IL-6 was identified solely as an independent prognostic factor for PFS in patients with AGC.
Gastric cancer is an inflammation-associated cancer, mainly caused by Helicobacter pylori infection, which can lead to chronic gastritis and develop into atrophy, metaplasia, dysplasia, and gastric cancer[26,27]. A growing number of studies have shown that various types of cancer are associated with chronic inflammation, and inflammation plays a dual role in tumor development. In some cases, they inhibit tumor growth by promoting the anti-tumor activity of cytotoxic T cells[28], which can limit the proliferation of transformed cells or tumor growth, and even achieve induced elimination. In contrast, in other cases, free radical-induced DNA damage in chronic inflammation helps create a favorable environment for tumor development, thereby promoting tumor occurrence and progression[28,29].
Cytokines can be roughly divided into two types: One type are cytokines that have antitumor effects, mainly secreted by type 1 helper T (Th1) cells, such as IL-2, IL-12, interferon-gamma, and tumor necrosis factor, which are responsible for cellular immunity; the other type are cytokines involved in the promotion of tumor progression, mainly secreted by type 2 helper T (Th2) cells, including IL-6, IL-8, IL-10, IL-17, which are responsible for inducing humoral immunity[30].
In this study, a correlation analysis was conducted to explore to the relationship between cytokines and clinical features. The results revealed a correlation between age and the levels of IL-6 and IL-10. The level of IL-6 increased with age, which may be related to the common chronic low-grade inflammatory state in the elderly population, known as “inflammatory aging”[31]. IL-10, as a key anti-inflammatory factor, is elevated in the elderly as a compensatory response to chronic inflammation, in order to prevent excessive inflammatory damage[32]. In addition, the expression level of PD-L1 was also correlated with IL-6 and IL-10. IL-6 is a pro-inflammatory cytokine associated with various cancer-related processes, including promotion of tumor growth and inhibition of antitumor immunity. IL-6 is known to activate the JAK/STAT signaling pathway, which may lead to increased expression of PD-L1 on tumor cells, thereby promoting immune escape by inhibiting T cell activity[33-35]. IL-10 is also an anti-inflammatory cytokine that can regulate immune responses by promoting the activity of regulatory T cells and inhibiting the production of pro-inflammatory cytokines. The presence of IL-10 in the TME can increase PD-L1 expression, which further helps to suppress effective immune responses against tumors[36,37]. This study found a positive correlation between CA199 and peripheral blood IL-10 levels. Although current research has examined CA199 and IL-10 expression in various situations, the direct correlation between the two has not been extensively documented in the literature. However, both of these biomarkers are involved in the processes of cancer and inflammation, indicating that potential interactions deserve further investigation.
IL-4 serves as a crucial mediator in maintaining the balance between Th1 and Th2 responses and possesses apoptotic properties. Despite its involvement in inflammation-related carcinogenesis within human organs, it exhibits a contradictory function in cancer[28,38]. In various models for prevention and treatment, IL-4 elicits the strongest immune response among a range of cytokines[39]. Furthermore, it can modulate the activity of tumor-associated fibroblasts, which play a significant role in promoting tumor growth and progression. This modulation subsequently hinders angiogenesis and obstructs the formation of blood vessels that sustain tumor growth. Collectively, these observations suggest that IL-4 functions as a potent antitumor cytokine and might serve as a promising therapeutic target for the clinical management of tumors[40,41]. Conversely, IL-6 is one of the most prevalent inflammatory cytokines. This glycopeptide triggers B lymphocytes to produce immunoglobulin[42]. Research indicates that IL-6 is pivotal in the inflammation connected with tumor development[43,44]. It primarily activates the STAT3 signaling pathways for transduction and transcription, thus encouraging tumor progression, invasion, and metastasis[45,46]. Consequently, it is linked to unfavorable prognoses and diminished survival rates among cancer patients. Another significant cytokine, IL-8, belongs to the family of neutrophil chemokines and is secreted by various cell types such as neutrophils, monocytes, fibroblasts, epithelial cells, endothelial cells, and mesothelial cells[47,48]. These cells react to inflammatory stimuli and contribute to the activation of neutrophils, which facilitates the recruitment of T lymphocytes and other non-specific inflammatory cells to the inflammation site. Thus, IL-8 is critical in processes related to inflammation and wound healing[47,49]. It is not surprising that malignant and tumor stromal cells across different tumor types are also capable of producing IL-8[50]. It enhances the activation of matrix metalloproteinase 2 and 9, promoting endothelial cell angiogenesis and bolstering the proliferation and survival of tumor cells[51,52].
Furthermore, IL-8 produced by tumor cells has the ability to activate immune-suppressive cells, including neutrophils and suppressor cells derived from bone marrow, and facilitates the conversion of epithelial cells into mesenchymal cells. This process not only diminishes the immune response of the body towards tumors but also boosts both the motility and invasive potential of tumor cells, thereby hastening tumor metastasis[53-55]. Nevertheless, the connection between IL-8 levels and patient prognosis in gastric cancer is still ambiguous, with some studies yielding conflicting outcomes[56,57]. Research by Wang et al[58] identified that elevated IL-8 expression serves as a risk factor for unfavorable prognosis in gastric cancer patients. Our findings observed a similar trend; however, the results lacked statistical significance, likely attributed to variations in our treatment protocols, patient profiles, and several other variables. IL-10 is a versatile immune regulatory cytokine that plays a vital role in preserving immune balance within mucosal tissues[59]. In gastric cancer, tumor-associated macrophages that express IL-10 facilitate immune evasion across several malignancies. Additionally, heightened levels of IL-10 can enhance activation and proliferation of CD8+ T cells in contexts of chronic inflammation and cancer. A study conducted by Wang et al[60] examining the interaction between five cytokines and immune efficacy as well as the prognosis of patients with AGC indicated that the expression levels of IL-6, IL-10, and PD-L1 were independent prognostic indicators for PFS, aligning with our observations. Regrettably, their findings did not discuss the predictive significance of cytokines concerning OS. Another investigation by Chen et al[61] regarding the levels and long-term outcomes associated with IL-17 in gastric adenocarcinoma demonstrated that patients with diminished IL-17 levels experienced poorer prognosis, contrasting with our results.
This could be attributed to several factors: (1) Their research focused exclusively on patients with adenocarcinoma, whereas our investigation encompassed all pathological types; (2) They utilized IHC for assessing IL-17 expression in cells, while we evaluated cytokine concentrations in peripheral blood; (3) Our patient cohort consisted solely of individuals with AGC; and (4) The participants in our study were postoperative cases who had not undergone antitumor therapies prior to surgery. Each of these factors may considerably influence prognosis. Furthermore, as cancer immunotherapy advances, the levels of PD-L1 expression have emerged as predictive biomarkers to facilitate anti-PD-1 immunotherapy[62]. Our findings also demonstrated that patients with a CPS of ≥ 5 exhibited a more favorable prognosis compared to those with a CPS of < 5, indicating that PD-L1 expression may enhance the effectiveness of combination chemotherapy using sintilimab. Nonetheless, these findings require validation through additional prospective clinical trials.
A notable finding in our study was the apparent discrepancy between the modest discriminatory power of continuous cytokine levels in time-dependent ROC analysis (AUCs ranging from approximately 0.25 to 0.58) and their profound prognostic significance when dichotomized (HRs ranging from 3 to 10). This non-linear dynamic suggests a strong threshold effect in the tumor immune microenvironment. Cytokine-mediated immunosuppression may not scale linearly with concentration; rather, once a specific critical concentration (cut-off value) is breached, the risk of disease progression or death escalates drastically. Consequently, converting these continuous variables into defined thresholds is not merely a statistical necessity, but a biologically and clinically relevant strategy, enabling physicians to rapidly stratify patients for real-world decision-making.
This study still has some limitations. First, as a single center, retrospective study, all samples were from a single hospital in a single region, which may lead to bias in the results. Second, we did not assess the source of cytokines and were unable to fully control the impact of inflammatory factors (such as concurrent infections or autoimmune adverse events) on the study. Inflammation can have a certain impact on cytokine levels and immune therapy prognosis. Additionally, we are concerned about the efficacy of combination chemotherapy with sintilimab, which means that further validation is necessary to validate conclusions regarding other PD-1/PD-L1 inhibitors and targeted drugs. In addition, excluding some patients from our study may have reduced its statistical power. Finally, the sample size we studied is relatively small, which may introduce bias into the results. Moreover, we acknowledge a potential real-world selection bias regarding PD-L1 expression, as a significantly high proportion of patients (79.2%) had a CPS ≥ 5, likely driven by initial reimbursement policies and physician preference for combination immunotherapy. Furthermore, the lack of an independent external validation cohort limits the immediate widespread generalizability of our nomogram, warranting future large-scale, multicenter prospective studies. Furthermore, we acknowledge that the specific cut-off values determined by ROC analysis are inherently data-dependent, and these optimal thresholds require further validation and calibration in independent external cohorts.
CONCLUSION
Initial cytokine levels may act as a biomarker to forecast the effectiveness of combination chemotherapy involving sintilimab for individuals with AGC. Notably, decreased levels of IL-6, IL-10, and IL-17 in peripheral blood prior to treatment show a significant correlation with extended PFS and OS (all P < 0.05). Additionally, the expressions of IL-10, IL-17, and PD-L1 serve as independent prognostic indicators for OS, while IL-6 is recognized as an independent prognostic factor for PFS in this patient population.
Shi Y, Su H, Song Y, Jiang W, Sun X, Qian W, Zhang W, Gao Y, Jin Z, Zhou J, Jin C, Zou L, Qiu L, Li W, Yang J, Hou M, Zeng S, Zhang Q, Hu J, Zhou H, Xiong Y, Liu P. Safety and activity of sintilimab in patients with relapsed or refractory classical Hodgkin lymphoma (ORIENT-1): a multicentre, single-arm, phase 2 trial.Lancet Haematol. 2019;6:e12-e19.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 206][Cited by in RCA: 201][Article Influence: 28.7][Reference Citation Analysis (0)]
Zhou C, Wu L, Fan Y, Wang Z, Liu L, Chen G, Zhang L, Huang D, Cang S, Yang Z, Zhou J, Zhou C, Li B, Li J, Fan M, Cui J, Li Y, Zhao H, Fang J, Xue J, Hu C, Sun P, Du Y, Zhou H, Wang S, Zhang W. Sintilimab Plus Platinum and Gemcitabine as First-Line Treatment for Advanced or Metastatic Squamous NSCLC: Results From a Randomized, Double-Blind, Phase 3 Trial (ORIENT-12).J Thorac Oncol. 2021;16:1501-1511.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 94][Cited by in RCA: 255][Article Influence: 51.0][Reference Citation Analysis (0)]
Ren Z, Xu J, Bai Y, Xu A, Cang S, Du C, Li Q, Lu Y, Chen Y, Guo Y, Chen Z, Liu B, Jia W, Wu J, Wang J, Shao G, Zhang B, Shan Y, Meng Z, Wu J, Gu S, Yang W, Liu C, Shi X, Gao Z, Yin T, Cui J, Huang M, Xing B, Mao Y, Teng G, Qin Y, Wang J, Xia F, Yin G, Yang Y, Chen M, Wang Y, Zhou H, Fan J; ORIENT-32 study group. Sintilimab plus a bevacizumab biosimilar (IBI305) versus sorafenib in unresectable hepatocellular carcinoma (ORIENT-32): a randomised, open-label, phase 2-3 study.Lancet Oncol. 2021;22:977-990.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 918][Cited by in RCA: 887][Article Influence: 177.4][Reference Citation Analysis (4)]
Lu Z, Wang J, Shu Y, Liu L, Kong L, Yang L, Wang B, Sun G, Ji Y, Cao G, Liu H, Cui T, Li N, Qiu W, Li G, Hou X, Luo H, Xue L, Zhang Y, Yue W, Liu Z, Wang X, Gao S, Pan Y, Galais MP, Zaanan A, Ma Z, Li H, Wang Y, Shen L; ORIENT-15 study group. Sintilimab versus placebo in combination with chemotherapy as first line treatment for locally advanced or metastatic oesophageal squamous cell carcinoma (ORIENT-15): multicentre, randomised, double blind, phase 3 trial.BMJ. 2022;377:e068714.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 345][Cited by in RCA: 332][Article Influence: 83.0][Reference Citation Analysis (5)]
Xu J, Jiang H, Pan Y, Gu K, Cang S, Han L, Shu Y, Li J, Zhao J, Pan H, Luo S, Qin Y, Guo Q, Bai Y, Ling Y, Yang J, Yan Z, Yang L, Tang Y, He Y, Zhang L, Liang X, Niu Z, Zhang J, Mao Y, Guo Y, Peng B, Li Z, Liu Y, Wang Y, Zhou H; ORIENT-16 Investigators. Sintilimab Plus Chemotherapy for Unresectable Gastric or Gastroesophageal Junction Cancer: The ORIENT-16 Randomized Clinical Trial.JAMA. 2023;330:2064-2074.
[RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)][Cited by in Crossref: 316][Cited by in RCA: 353][Article Influence: 117.7][Reference Citation Analysis (1)]
Yin WB, Wang ZA. [The value of level detection of carcinoembryonic antigen combined with fasting plasma glucose for evaluating the prognosis of advanced gastric cancer].Xiandai Zhongliu Yixue. 2023;31:282-286.
[PubMed] [DOI] [Full Text]
Zhou P, Qu H, Shi HW, Sun GR, He QS. [Predictive value of CA125 in peritoneal metastasis and prognosis of patients with gastric carcinoma].Zhonghua Weichang Waike Zazhi. 2014;17:1027-1030.
[PubMed] [DOI]
Zhu YB, Ge SH, Zhang LH, Wang XH, Xing XF, Du H, Hu Y, Li YA, Jia YN, Lin Y, Fan B, Ji JF. [Clinical value of serum CEA, CA19-9, CA72-4 and CA242 in the diagnosis and prognosis of gastric cancer].Zhonghua Weichang Waike Zazhi. 2012;161-164.
[PubMed] [DOI]
Unver N, Delgado O, Zeleke K, Cumpian A, Tang X, Caetano MS, Wang H, Katayama H, Yu H, Szabo E, Wistuba II, Moghaddam SJ, Hanash SM, Ostrin EJ. Reduced IL-6 levels and tumor-associated phospho-STAT3 are associated with reduced tumor development in a mouse model of lung cancer chemoprevention with myo-inositol.Int J Cancer. 2018;142:1405-1417.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 26][Cited by in RCA: 35][Article Influence: 3.9][Reference Citation Analysis (0)]
Wang XY, Dai HW, Liu XL. [Peripheral blood cytokines predict the efficacy and prognosis of anti-PD-1 therapy for advanced gastric cancer].Xiandai Zhongliu Yixue. 2023;31:2872-2878.
[PubMed] [DOI] [Full Text]
