Chen CR, Jin HL, Xu QJ, Yuan YL, Hu ZH, Liu Y, Lei HK. Development and validation of a nomogram for predicting postoperative venous thromboembolism risk in patients with hepatocellular carcinoma. World J Gastrointest Oncol 2025; 17(6): 105790 [DOI: 10.4251/wjgo.v17.i6.105790]
Corresponding Author of This Article
Hai-Ke Lei, PhD, Associate Professor, Chongqing Cancer Multi-omics Big Data Application Engineering Research Center, Chongqing University Cancer Hospital, No. 181 Hanyu Road, Shapingba District, Chongqing 400030, China. tohaike@163.com
Research Domain of This Article
Oncology
Article-Type of This Article
Retrospective Study
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/
Chun-Rong Chen, Department of Health Information Management, School of Public Health and Management, Chongqing Three Gorges Medical and Pharmaceutical College, Chongqing 404120, China
Hong-Liang Jin, Qian-Jie Xu, Yu-Liang Yuan, Ya Liu, Hai-Ke Lei, Chongqing Cancer Multi-omics Big Data Application Engineering Research Center, Chongqing University Cancer Hospital, Chongqing 400030, China
Zu-Hai Hu, Department of Health Statistics, School of Public Health, Chongqing Medical University, Chongqing 400016, China
Co-first authors: Chun-Rong Chen and Hong-Liang Jin.
Co-corresponding authors: Ya Liu and Hai-Ke Lei.
Author contributions: Chen CR and Jin HL contributed to formal analysis, methodology and writing-original draft, they contributed equally as co-first authors; Yuan YY contributed to data curation and cleaning; Xu QJ contributed to formal analysis and investigation; Hu ZH contributed to formal analysis; Liu Y contributed to resources, supervision and project administration; Lei HK conceived and designed the study, and contributed to conceptualization, review and editing; Liu Y and Lei HK contributed equally as co-corresponding authors; and all the authors contributed to the article and approved the final manuscript.
Institutional review board statement: The study was conducted according to the guidelines of the Declaration of Helsinki, and the Ethics Committee of Chongqing University Cancer Hospital reviewed and approved the research studies (Approval No. CZLS2023343-A).
Informed consent statement: Owing to the retrospective nature of the study, informed consent was waived.
Conflict-of-interest statement: All the authors report no relevant conflicts of interest for this article.
Data sharing statement: The datasets generated and/or analysed during the current study are not publicly available due to local legal requirements but are available from the corresponding author on reasonable request.
Open Access: This article is an open-access article that was selected by an in-house editor and fully peer-reviewed by external reviewers. It is distributed in accordance with the Creative Commons Attribution NonCommercial (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: https://creativecommons.org/Licenses/by-nc/4.0/
Corresponding author: Hai-Ke Lei, PhD, Associate Professor, Chongqing Cancer Multi-omics Big Data Application Engineering Research Center, Chongqing University Cancer Hospital, No. 181 Hanyu Road, Shapingba District, Chongqing 400030, China. tohaike@163.com
Received: February 7, 2025 Revised: April 14, 2025 Accepted: May 16, 2025 Published online: June 15, 2025 Processing time: 127 Days and 3.4 Hours
Abstract
BACKGROUND
Few studies have specifically modeled the risk of venous thromboembolism (VTE) for postoperative hepatocellular carcinoma (HCC) patients, although HCC is the third leading cause of cancer death worldwide. This study aimed to develop and validate a nomogram that accurately predicts the risk of VTE in patients after HCC surgery.
AIM
To develop and validate a nomogram to accurately predict the risk of VTE in postoperative HCC patients by integrating clinical and laboratory risk factors. The model seeks to provide a user-friendly tool for identifying high-risk individuals who may benefit from targeted anticoagulation therapy, thereby improving clinical decision-making and patient outcomes.
METHODS
Data from patients who underwent HCC surgery at Chongqing University Cancer Hospital in China were analyzed. Through univariate and multivariate logistic regression analyses, independent risk factors for VTE were identified and integrated into a nomogram. The predictive performance of the nomogram was assessed via receiver operating characteristic curves, calibration curves, decision curve analysis and other relevant metrics.
RESULTS
Of 905 postoperative HCC patients were included in the study. The nomogram incorporated eight independent risk factors for VTE: Karnofsky Performance Scale, base disease, cancer stage (tumor-node-metastasis), chemotherapy, D-dimer concentration, white blood cell count, hemoglobin, and fibrinogen. The C-index for the nomogram model was 0.825 in the training cohort and 0.820 in the validation cohort, indicating good discriminative ability. Calibration plots of the model revealed high concordance between the predicted probabilities and observed outcomes.
CONCLUSION
We developed and validated a novel nomogram that can accurately estimate the risk of VTE in individual postoperative HCC patients. This model can identify high-risk patients who may benefit from targeted anticoagulation therapy.
Core Tip: This study develops and validates a novel nomogram to predict postoperative venous thromboembolism (VTE) risk in hepatocellular carcinoma patients. Using data from 905 patients, we identified eight independent risk factors for VTE, including Karnofsky Performance Scale score, underlying disease, tumor-node-metastasis stage, chemotherapy, D-dimer, white blood cell count, hemoglobin, and fibrinogen. The nomogram demonstrated excellent discriminative ability (C-index: 0.825 in training, 0.820 in validation cohorts) and calibration. This model can accurately estimate individual VTE risk and aid in targeted anticoagulation strategies for high-risk patients.
Citation: Chen CR, Jin HL, Xu QJ, Yuan YL, Hu ZH, Liu Y, Lei HK. Development and validation of a nomogram for predicting postoperative venous thromboembolism risk in patients with hepatocellular carcinoma. World J Gastrointest Oncol 2025; 17(6): 105790
Hepatocellular carcinoma (HCC) is a cancer that often manifests subtly and typically progresses to a late stage by the time symptoms appear[1,2]. The global burden of HCC is substantial; in 2020 it was estimated to be the sixth most commonly diagnosed cancer and the second leading cause of cancer-related premature mortality. The incidence and mortality rates of HCC have increased in countries such as the United States, Australia, and some parts of Europe, where rates are relatively low, but have declined in China, Japan, and other East Asian nations[3]. Venous thromboembolism (VTE), including deep vein thrombosis and pulmonary embolism, is a recognized complication, especially in cancer patients undergoing abdominal surgery due to a high risk for VTE[4]. The annual incidence of VTE in the cancer population is estimated to be 1.3%, and the risk of VTE 30 days after liver resection is 2.2% to 4.6%, leading to increased mortality, severe complications and a lower quality of life[5-7]; this underscores the urgent need to explore factors associated with the high risk of VTE, especially in cancers such as HCC, which often lead to premature death.
Several established risk factors for VTE include trauma, fractures of the pelvis, hip, or long bones, pregnancy, immobility, and abdominal or pelvic surgeries[8]. Studies on VTE risk post-HCC surgery are scarce. A population-based retrospective cohort study by Ratib et al[9] estimated the relationships between patient characteristics postsurgery and VTE risk, associated risk factors, and their impact on survival within 30 and 90 days postliver resection, and the results suggested that VTE prophylaxis after liver surgery mirrors that of surgeries for other cancer types. There is a notable lack of understanding and data on VTE risk in patients with postoperative HCC in China, highlighting the need for predictive models to aid in clinical decision-making in this patient cohort.
Current models for identifying patients at high risk of VTE include the Caprini[10], Padua[11], Rogers[12], Ottawa[13] and Khorana[14] scores. Each scoring system is tailored to different patient populations. The Caprini score is suitable for surgical patients, and while the Ottawa score considers cancer type, disease stage, sex, and history of thrombosis, it does not adequately predict VTE recurrence in patients receiving anticoagulant treatment. The Khorana score, designed for medical and outpatient patients, has shown clinically significant discrepancies in discriminatory performance across different cancer types, necessitating the development of risk prediction tools specific to individual cancer types, including cancer-specific risk factors for VTE[15]. Given the unique aspects of postoperative HCC, existing models have limited efficacy in assessing VTE risk, making it crucial to improve risk assessment tools to identify postoperative HCC patients at high risk for VTE. The nomogram model is a visual tool that transforms traditional regression models into a visual risk assessment for each patient, making it user friendly[16]. There is substantial evidence of the nomogram’s effectiveness in accurately assessing VTE risk across various cancer types[17-20]. Based on our preliminary research, which indicates that surgery significantly affects the assessment of VTE[21], we identified a gap in predictive nomogram models for postoperative HCC patients. Consequently, we conducted this study specifically in patients who had undergone liver cancer surgery. The primary aim was to establish a nomogram model tailored for the postoperative HCC population in China, offering an accurate and reliable prediction of VTE risk probability.
MATERIALS AND METHODS
Patient population
From January 2022 to December 2024, 905 participants were recruited from the Chongqing University Cancer Hospital, China. Patients with at least one hospitalization were included and diagnosed with HCC. The inclusion criteria were as follows: (1) Postoperative pathology confirmed primary HCC; (2) No history of thrombotic disease; (3) No preoperative local treatments such as radiofrequency ablation or transarterial chemoembolization; and (4) A preoperative Eastern Cooperative Oncology Group performance status of 0-2. The exclusion criteria were as follows: (1) Early postoperative death due to severe complications; (2) Incomplete clinical pathological data; (3) Long-term preoperative use of antiplatelet or anticoagulant medications; and (4) Thrombus formation due to central venous catheter or peripherally inserted central catheters line placement (Figure 1).
Figure 1 Flow chart of the patients enrolled in the final study cohorts.
VTE: Venous thromboembolism; TNM: Tumor-node-metastasis; DCA: Decision curve analysis; ROC: Receiver operating characteristic.
Data collection and definitions
All the data were obtained from the hospital’s cancer big data research platform. The study included 21 research variables: Age, Karnofsky Performance Scale (KPS) score, sex, body mass index, type of HCC (pathological), presence of underlying disease (based disease), cancer stage (tumor-node-metastasis, TNM), white blood cell (WBC) count, hemoglobin (Hb) level, platelet count, D-dimer, activated partial thromboplastin time, prothrombin time, β2-microglobulin, lactate dehydrogenase, fibrinogen (FIB), absolute lymphocyte count, FIB degradation products, and receipt of radiotherapy, chemotherapy, immunotherapy, or targeted therapy. Underlying diseases included chronic conditions such as diabetes and hypertension. During data collection, relevant data for patients diagnosed with VTE were collected at the time of diagnosis. For patients who did not develop VTE, data were collected during the treatment period.
Diagnosis of VTE
Within one-month postsurgery for primary HCC, regardless of the presence of related VTE symptoms such as pit edema in the lower limbs, swelling, pain in the lower limbs, difficulty breathing, or chest pain, patients were diagnosed with VTE if unilateral or bilateral deep vein thrombosis was confirmed by lower limb vascular Doppler ultrasound or if pulmonary embolism was detected by pulmonary vascular imaging. The diagnoses were independently determined and reviewed by two experienced radiologists. In cases of disagreement, a third radiologist confirmed the diagnosis, which was considered final[22].
Statistical analysis
Continuous variables with a normal distribution are presented as mean ± SD. Differences between cohorts were assessed via t-tests for statistical significance. The median and interquartile range were reported for nonnormally distributed data, with nonparametric tests used to compare differences between cohorts. Categorical data are expressed as frequencies and percentages and were analyzed via the χ2-test. Logistic regression analysis was used to explore the relationships between clinical variables and VTE. Variables that achieved significance in the univariate logistic regression analysis (P < 0.05) were incorporated into the model. The discriminative ability of the nomogram model was evaluated via receiver operating characteristic curves and calibration plots. In addition, we used a generated calibration curve for decision curve analysis (DCA) to further evaluate the model. To validate the clinical applicability of the model, we compared the nomogram model with the existing VTE risk model Khorana. We randomly partitioned all patients into training and validation sets via R software’s “caret” package, with a fixed random seed ratio of 7:3. Missing values were judged and processed via the “mice” package. The scores of each variable were calculated via the “DynNom” package in R. The total score for each patient was calculated based on the score of each variable. The nomogram’s DCA was plotted via the “rmda” package. All the statistical analyses were conducted via R software version 4.1.2. The predetermined level of significance was set at 0.05.
RESULTS
Demographic baseline characteristics
A total of 905 postoperative HCC patients were included in this study. The average age was 56.04 ± 11.40 years, with 716 male patients (79.12%). There were 797 cases (88.07%) of non-VTE and 108 (11.93%) of VTE. Basic disease was present in 252 patients (27.85%), TNM stage III-IV tumors were present in 580 patients (64.09%), and chemotherapy was used in 86 patients (9.50%). The average KPS was 84.67 ± 9.48, the WBC count was 5.85 ± 2.38 on average, the average Hb count was 133.23 ± 22.36, and the D-dimer median and quarterback spacing were 0.52 (0.24, 1.46). Among all the VTE patients, 84 were male. The average age was 56.86 ± 10.49 years. Baseline comparisons among the queue, KPS, disease-based, TNM, targeted therapy, chemotherapy, immunotherapy, D-dimer, β2-microglobulin, Hb, and FIB degradation products variables revealed statistically significant differences (Table 1). Using random split sample methodology, 634 patients were assigned to the training cohort, and the remaining 271 were assigned to the validation cohort, maintaining a 7:3 split ratio. As shown in Table 1, there were no significant differences between the two cohorts.
Table 1 Patient demographic and clinicopathological characteristics, n (%).
Significant predictive factors in the training cohort
In the training cohort, VTE occurred in 77 patients (12.15%), whereas 557 patients (87.85%) did not develop VTE. Figure 2 displays the results of the logistic regression analysis after variable selection. The significant predictive factors for VTE determined through univariate analysis included the KPS, disease, TNM, and history of chemotherapy, among nine others (P < 0.2). The results of the multivariate stepwise regression revealed that KPS, disease status, TNM stage, history of chemotherapy, D-dimer level, WBC count, Hb level, and FIB level were risk factors for VTE. Specifically, disease [OR = 2.34 (1.26-4.34), P = 0.007], TNM stage [OR = 2.15 (1.07-4.55), P = 0.036], and history of chemotherapy [OR = 3.27 (1.81-6.10), P < 0.001] significantly increased the risk of VTE.
Figure 2 Multivariate logistic regression analysis of venous thromboembolism risk factors in the training cohort.
KPS: Karnofsky Performance Scale; TNM: Tumor-node-metastasis; WBC: White blood cell count; Hb: Hemoglobin; FIB: Fibrinogen; OR: Odds ratio; CI: Confidence interval.
Construction and calibration of the nomogram model
A nomogram was developed based on significant predictive factors found in the multivariate logistic regression analysis and factors with clinical relevance (Figure 3A). Each variable was scored, and the total score was calculated by summing the scores of each factor, with the cumulative score on the scale corresponding to the estimated probability of VTE. KPS had the most significant impact on predicting VTE, followed by FIB, D-dimer, WBC, Hb, chemotherapy, disease-based, and TNM, which moderately influenced the prediction of postoperative VTE in HCC patients. We also developed a user-friendly web-based calculator that can be accessed through the following link: https://cuch.shinyapps.io/HCC_VTE/, which allows for a rapid calculation of the VTE risk for patients. For example, for a patient with a KPS score of 80, FIB level of 2 g/L, D-dimer concentration of 2 mg/L, WBC count of 4 × 109/L, and Hb level of 166 g/L, who did not receive chemotherapy, was diagnosed with disease, and was at TNM stage III, the estimated risk of VTE is 0.130 according to our model. The calibration curves indicated excellent concordance between the expected and actual rates of VTE in the training cohort and reasonable concordance in the validation cohort (Figure 3B).
Figure 3 Construction and validation of the model.
A: Nomogram for predicting the risk of venous thromboembolism in patients; B: Receiver operating characteristic curve of the nomogram for predicting venous thromboembolism risk in the training cohort and validation cohort; C: Training cohorts’ risk of venous thromboembolism nomogram correction chart; D: Validation cohorts’ risk.
Validation and calibration of the graphs
In the training cohort, the constructed nomogram for predicting VTE had a C-index of 0.825 (95% confidence interval: 0.765-0.885), whereas in the validation cohort, the C-index was 0.820 (95% confidence interval: 0.737-0.904) (Figure 3C and D). Calibration plots demonstrated optimal concordance between the predicted and observed VTE rates in the training (Figure 3C) and validation cohorts (Figure 3D).
DCA and validation
DCA was used to evaluate the clinical utility of the nomogram model by calculating the net benefit across various risk threshold levels. The decision curves indicated that when the threshold probabilities ranged from 4% to 96%, the use of the nomogram to decide on treatment provided a more significant net benefit than the use of all or none, with similar results verified in the validation cohort (Figure 4A and B). The decision curves demonstrated that the nomogram performed well and could feasibly make beneficial clinical decisions. We compared the nomogram and Khorana in terms of accuracy, sensitivity, specificity, and area under the curve (Figure 4C and D). In both the training and validation cohorts, the nomogram demonstrated better sensitivity and area under the curve than did the Khorana nomogram. Although Khorana had higher predictive accuracy, this came at the expense of sensitivity. In real clinical practice, Khorana is not suitable for assessing VTE risk, it is only suitable for ruling it out. For evaluating the magnitude of VTE risk, the nomogram is more suitable for use in clinical practice.
Figure 4 Decision curve analysis curve of the graph model.
A: Training cohort; B: Validation cohort; C: Training cohort. Comparison of the nomogram model with Khorana; D: Validation cohort. VTE: Venous thromboembolism; AUC: Area under the curve.
DISCUSSION
VTE in postoperative HCC patients is a severe complication that can lead to persistent ascites and liver failure[23,24]. Identifying and understanding the risk factors associated with VTE is crucial. However, few studies have specifically modeled the risk of VTE for postoperative HCC patients, despite HCC being the third leading cause of cancer death worldwide. The aim of this study was to establish a comprehensive, multifactorial predictive nomogram for VTE risk in HCC patients, identify critical risk factors and provide valuable insights for preventing and treating thrombosis in these patients.
The nomogram model transforms complex regression equations into a simple, intuitive graphical format for straightforward interpretation, and it accurately predicts the probability of various outcomes. In this study, we developed a nomogram for predicting VTE risk, including factors such as the KPS, base disease status, TNM stage, chemotherapy, D-dimer concentration, WBC count, Hb level, and FIB. The model demonstrated excellent identification, calibration, and predictive value; it was internally validated, showing significant utility in clinical identification and decision-making. Additionally, our study incorporated variables related to the hemostatic system and underlying diseases, which are not typically included in similar studies.
Our findings indicate that a lower KPS is a significant risk factor for VTE[25]. Reduced physical activity is correlated with a higher risk of VTE; the KPS, a simple yet valuable method for evaluating daily performance, is widely used in medical oncology. Studies have shown that bed rest for more than three days is significantly associated with increased VTE incidence[26,27]. Reduced mobility means increased bed rest, and immobility by increasing venous stasis increases the likelihood of VTE. Consistent with results from previous studies[28-30], we observed that lower FIB levels and higher D-dimer concentrations were associated with an increased likelihood of VTE in individuals diagnosed with HCC. FIB, a significant protein in the coagulation process, and D-dimer, a degradation product of cross-linked fibrin, rapidly increase in acute thrombosis and are elevated in conditions such as cancer and infection. Therefore, coagulation levels reflect the degree of thrombus formation and fibrinolysis processes[31], making them essential indicators for predicting VTE during treatment. Patients who have undergone chemotherapy are at increased risk of developing VTE[32]. A study by Khorana et al[14] indicated that cancer patients have a significantly higher rate of VTE within the first 12 months after starting chemotherapy than noncancer patients do. This effect can be attributed to the cumulative impact of chemotherapeutic drugs, which cause sustained damage to vascular endothelial cells, increase the levels of procoagulant proteins, decrease the levels of endogenous anticoagulants, and activate platelets.
Our study revealed that both Hb levels and WBCs are risk factors for VTE in postoperative HCC patients. Studies have shown a significant reduction in Hb after liver surgery[31], which is part of the hemostasis control mechanism; thus, disturbances in Hb levels can lead to thrombotic complications. Elevated WBC levels indicate changes in inflammatory markers within the body, and cohort studies suggest that white cells may play a causative role in cancer-associated VTE[33]. Patients with underlying disease or in advanced stages of cancer are at greater risk for VTE[34]; our study revealed that patients with underlying disease and those with stage III or IV cancer were more than twice as likely to develop VTE. Previous cohort studies have shown that the risk of VTE increases with cancer stage, with adjusted relative risks calculated for stages I, II, III, and IV at 2.9, 2.9, 7.5, and 17.1, respectively[35].
This study had several limitations. First, all the patients were recruited from a single center, inevitably introducing bias and weakening the statistical power. Second, there are inherent limitations associated with a retrospective study design. Third, our study did not discuss genetic information, leaving some gaps in the research content. Therefore, a well-designed prospective, multicenter, large-sample study with detailed genetic information is needed to address these critical issues.
CONCLUSION
We established and validated a new nomogram for predicting the risk of VTE in postoperative HCC patients. This model can accurately estimate the risk of VTE in individual postoperative HCC patients and identify those who may benefit from targeted anticoagulation strategies.
Footnotes
Provenance and peer review: Unsolicited article; Externally peer reviewed.
Peer-review model: Single blind
Specialty type: Oncology
Country of origin: China
Peer-review report’s classification
Scientific Quality: Grade B, Grade B, Grade B
Novelty: Grade B, Grade B, Grade B
Creativity or Innovation: Grade B, Grade B, Grade B
Scientific Significance: Grade B, Grade B, Grade B
P-Reviewer: Wang N; Yang F S-Editor: Wei YF L-Editor: A P-Editor: Guo X
Giustozzi M, Valerio L, Agnelli G, Becattini C, Fronk EM, Klok FA, Konstantinides SV, Vedovati MC, Cohen AT, Barco S. Sex-specific differences in the presentation, clinical course, and quality of life of patients with acute venous thromboembolism according to baseline risk factors. Insights from the PREFER in VTE.Eur J Intern Med. 2021;88:43-51.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 7][Cited by in RCA: 14][Article Influence: 3.5][Reference Citation Analysis (0)]
van Es N, Ventresca M, Di Nisio M, Zhou Q, Noble S, Crowther M, Briel M, Garcia D, Lyman GH, Macbeth F, Griffiths G, Iorio A, Mbuagbaw L, Neumann I, Brozek J, Guyatt G, Streiff MB, Baldeh T, Florez ID, Gurunlu Alma O, Agnelli G, Ageno W, Marcucci M, Bozas G, Zulian G, Maraveyas A, Lebeau B, Lecumberri R, Sideras K, Loprinzi C, McBane R, Pelzer U, Riess H, Solh Z, Perry J, Kahale LA, Bossuyt PM, Klerk C, Büller HR, Akl EA, Schünemann HJ; IPDMA Heparin Use in Cancer Patients Research Group. The Khorana score for prediction of venous thromboembolism in cancer patients: An individual patient data meta-analysis.J Thromb Haemost. 2020;18:1940-1951.
[RCA] [PubMed] [DOI] [Full Text][Cited by in Crossref: 54][Cited by in RCA: 77][Article Influence: 15.4][Reference Citation Analysis (0)]
Huang X, Luo Z, Liang W, Xie G, Lang X, Gou J, Liu C, Xu X, Fu D. ASO Visual Abstract: Survival Nomogram for Young Breast Cancer Patients Based on the SEER Database and an External Validation Cohort.Ann Surg Oncol. 2022;29:5784-5785.
[RCA] [PubMed] [DOI] [Full Text][Reference Citation Analysis (0)]
Yoshikawa T, Sano T, Terashima M, Yamaguchi K, Bando E, Kawabata R, Yabusaki H, Shinohara H, Oba MS, Kimura T, Takita A, Sasako M. Incidence and risk factors for venous thromboembolism in the Cancer-VTE Registry stomach cancer subcohort.Gastric Cancer. 2023;26:493-503.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 9][Reference Citation Analysis (0)]
Li XP, Wang YY, Sun YS, Zhang LJ, Zhao XY, Liu ZQ, Jiang HM, Zha JD, Zhang XJ, Yan JN, Pan HY. Preoperative and postoperative clinical signatures of postgastrectomy venous thromboembolism in patients with gastric cancer: A retrospective cohort study.Asian J Surg. 2023;46:1556-1563.
[RCA] [PubMed] [DOI] [Full Text][Cited by in RCA: 2][Reference Citation Analysis (0)]
