Applying machine learning to predict the risk of cancer cachexia in stage IV colorectal cancer patients

Applying machine learning to predict the risk of cancer cachexia in stage IV colorectal cancer patients | Research Square window.SnipcartSettings = { analytics: { enabled: false } }; (function() { var accessVector = localStorage.getItem('access_vector') || ''; window.dataLayer = window.dataLayer || []; if (accessVector) { window.dataLayer.push({ user: { profile: { profileInfo: { snid: accessVector } } } }); } })(); (function(w,d,s,l,i){w[l]=w[l]||[];w[l].push({'gtm.start':new Date().getTime(),event:'gtm.js'});var f=d.getElementsByTagName(s)[0],j=d.createElement(s),dl=l!='dataLayer'?'&l='+l:'';j.async=true;j.src='https://www.googletagmanager.com/gtm.js?id='+i+dl;f.parentNode.insertBefore(j,f);})(window,document,'script','dataLayer','GTM-K279D39R'); Browse Preprints In Review Journals COVID-19 Preprints AJE Video Bytes Research Tools Research Promotion AJE Professional Editing AJE Rubriq About Preprint Platform In Review Editorial Policies Our Team Advisory Board Help Center Sign In Submit a Preprint Cite Share Download PDF Research Article Applying machine learning to predict the risk of cancer cachexia in stage IV colorectal cancer patients Jing Wang, Baojuan Han, Kangjie Wang, Yaoxian Xiang, Lei Wu, Dong Yan, and 1 more This is a preprint; it has not been peer reviewed by a journal. https://doi.org/ 10.21203/rs.3.rs-4275850/v1 This work is licensed under a CC BY 4.0 License Status: Posted Version 1 posted You are reading this latest preprint version Abstract Background: Cancer cachexia (CC) is a multifactorial syndrome affectingadvanced cancer patients, significantly impacting their survival and quality of life. This study utilized artificial intelligence machine learning (ML) methods to evaluate the risk of cachexia in stage IV colorectal cancer (CRC) patients through clinical data, establishing a cachexia risk prediction model. Methods : We conducted a retrospective collection of data from stage IV CRC patients consecutively visiting Luhe Hospital from January 2015 to December 2022. LASSO regression was employed on this cohort to identify diverse risk variables for CC. Four MLalgorithms, namely, logistic regression model, random forest (RF), Extreme Gradient Boosting, and k-nearest neighbor algorithm, were employed to construct the predictive model. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity. Additionally, Shapley additive prediction (SHAP) values were utilized to elucidate the prediction model. Results : Out of 219 stage IV CRC patients, 119 cases (54.34%) developed malignant fluid. Among the four ML models, the RF model exhibited the most robust predictive performance, achieving the highest AUC (0.768, 95% confidence interval: 0.706–0.831). The RF model demonstrated accuracy, sensitivity, and specificity values of 0.744, 0.790, and 0.690, respectively. We employed SHAP correlation maps to explicate the influence of individual features on the output of RF prediction models. Conclusions: The RF model offers superior predictions, enhancing clinicians' ability to screen for CC, assess patient prognosis, and make informed decisions on targeted CC in stage IV cancer patients. Cancer cachexia Colorectal cancer Machine learning Prediction. Figures Figure 1 Figure 2 Figure 3 Figure 4 Figure 5 Figure 6 Introduction Cachexia is a complex systemic disease involving multiple metabolic pathways in various tissues and organs. It is characterized by systemic inflammation, progressive weight loss (WL), and depletion of adipose tissue and skeletal muscle. Traditional nutritional support has proven insufficient to fully reverse these effects[1].[1–4] Cancer cachexia (CC) is a multifactorial clinical complication associated with various malignant tumors in humans, leading to a decline in patients' quality of life, treatment response, and treatment tolerance[5, 6]. Moreover, up to 80% of cancer patients experience CC. Recent clinical and experimental evidence clearly indicates that CC is not inevitable[7, 8], and early screening is particularly important for preventing or delaying the occurrence of cachexia. However, significant obstacles exist in identifying patients at risk of CC[9]. Primarily, CC, being a form of malnutrition, lacks dedicated tools for risk assessment, unlike the tools available for identifying nutritional risks[10, 11]. Moreover, WL, a common symptom of cachexia, is the traditional and most widely used parameter caused by the consumption of skeletal muscle mass and loss of adipose tissue, but recent discussions highlight that complex syndromes associated with CC may not always result in significant total WL[12], a condition referred to as “hidden cachexia”[13]. Consequently, WL alone fails to classify all pathophysiological manifestations of cachexia, and its severity varies. Furthermore, many patients struggle to provide accurate information on the onset and extent of WL, limiting the efficacy of WL as an early identifier of CC risk. Recently, efforts have been made to develop clinical tools related to the occurrence of cachexia. Prokopchuk et al. introduced a tissue inhibitor of metalloproteinase-1/liver/cachexia score to predict the prognosis of gastrointestinal cancer patients[14], whereas Jafri et al. created a cachexia index based on clinical characteristics to assess the prognosis of non-small cell lung cancer patients[15]. Yang et al. conducted a serum and urine metabolomics study, unveiling a diagnostic model for CC[16]. Recently, machine learning (ML) has been applied in the medical field for tasks such as cancer classification, prediction, drug response, and treatment strategies. ML, by automatically learning from vast datasets, establishes implicit relationships between variables and outcomes and then builds an efficient model to predict the results of other data[17]. This study aims to determine whether cachexia serves as a prognostic factor for stage IV colorectal cancer (CRC), including cases of curable resection and metastatic CRC, influencing optimal treatment strategies from the perspective of host status. Four ML algorithms were employed to predict the risk of malignant tumors in stage IV CRC patients. Materials and methods Study population From January 1, 2015, to December 31, 2022, patients aged ≥ 18 years in our clinical database were retrospectively recruited. The inclusion criteria encompassed individuals diagnosed as consecutive stage IV CRC patients for the first time in our institution. Eligible patients were required to have stage IV CRC with histologically confirmed adenocarcinoma, excluding other histological types, anal cancer, or appendix cancer. Patients lacking complete clinical data for cachexia diagnosis or with a history of cancer were excluded. The study was conducted at the Cancer Center of Beijing Luhe Hospital affiliated with Capital Medical University of China, approved by our institutional review board, and adhered to the Declaration of Helsinki and its subsequent amendments to ethical standards(2022-LHKY-036-02). According to international consensus, individuals meeting the diagnostic criteria for CC were those with > 5% WL in the past 6 months, had BMI 2% [9](21296615). Data collection Within this cohort, data on candidate predictors, including demographics, clinical, nutritional, and oncological variables, were collected through medical records and patient interviews. The selection of predictive factors for CC was based on disciplinary knowledge and literature review. Demographic data comprised age, sex, and smoking and drinking status, while clinical data included disease complications, especially respiratory, cardiovascular, and diabetes-related complications; lymphocyte count; leukocyte count; Glucose (Glu); creatinine (Cr); and serum total cholesterol (TC) Nutritional data encompassed albumin, hemoglobin (Hb), and BMI calculated as weight (kg)/height (m 2 ). Oncology data included tumor location, histological type (low, medium, and high differentiation status), carcinoembryonic antigen (CEA), gene status, (primary and local) surgical treatment, and T and N stages based on the 8th edition of the American Joint Commission Cancer Staging System. The Modified Glasgow Prognostic Score (mGPS) was scored as follows: score 0 for CRP ≤ 1.0 mg/dL, score 1, CRP > 1.0 mg/dL and albumin ≥ 3.5 g/dL, and score 2, CRP > 1.0 mg/dL and albumin < 3.5 g/d. In addition to the collected data, long-term follow-up results, including overall and recurrence-free survival, were collected. The follow-up data for survival status were updated in June 2023. Statistical analysis Continuous variables are presented as median [interquartile range (IQR)], and categorical variables as frequency and proportion. The Mann–Whitney U test and χ2 test were employed. LASSO regression was conducted on all variables to identify CC risk factors. Shapley additive interpretation (SHAP) was used as an additive feature attribution method to visualize the association between features and the occurrence of cachexia. The horizontal axis represents the Shapley value, with values > 0 indicating a positive contribution to the occurrence of CC in CRC. The left vertical axis represents the opposite order of importance of features, whereas the right vertical axis represents the feature values from low to high. The performance of ML models was evaluated using ROC, model accuracy, sensitivity, and specificity. All statistical analyses were performed using R software (version 4.3.1) ( http://www.R-project.org ) and Python (version 3.11, Python Software Foundation). Results Demographics and clinical characteristics A total of 219 patients were included in this study (Figure 1). The median (IQR) age was 66 (12) years, with 140 patients (63.9%) being males. Among the 219 patients, 119 (54.3%) had CC. The rectum (43.8%) was the most common tumor site, followed by the left colon (36.1%) and right colon (20.1%). The majority of colon cancer patients had advanced clinical stages (T4: 43.8%; T3: 39.3%; T2: 9.13%; N1: 37.4%). Descriptive characteristics of the cohort study population are presented in Table 1. Fig1. Flow diagram illustrating recruitment of patients. Table1 Patient demographics and pathological characteristics. Variables No Cachexia Cachexia P-value (N=100) (N=119) Age,years 65(12) 67(11.5) 0.172 Gender Male 64 (64.0%) 76 (63.9%) 1.000 Female 36 (36.0%) 43 (36.1%) Co.morbidity No 31 (31.0%) 26 (21.8%) 0.167 Yes 69 (69.0%) 93 (78.2%) Respiratory.co.morbidity No 7 (7.0%) 6 (5.0%) 0.746 Yes 93 (93.0%) 113 (95.0%) Diabetes No 85 (85.0%) 83 (69.7%) 0.012 Yes 15 (15.0%) 36 (30.3%) Cardiovascular.co.morbidity No 89 (89.0%) 100 (84.0%) 0.386 Yes 11 (11.0%) 19 (16.0%) Smoking status Never 67 (67.0%) 73 (61.3%) 0.467 Former/current 33 (33.0%) 46 (38.7%) Drinking status Never 74 (74.0%) 91 (76.5%) 0.791 Former/current 26 (26.0%) 28 (23.5%) Primary. site left 40 (40.0%) 39 (32.8%) 0.540 rectal 41 (41.0%) 55 (46.2%) right 19 (19.0%) 25 (21.0%) T stage T1 9 (9.0%) 8 (6.7%) 0.493 T2 12 (12.0%) 8 (6.7%) T3 37 (37.0%) 49 (41.2%) T4 42 (42.0%) 54 (45.4%) N stage N0 41 (41.0%) 28 (23.5%) 0.014 N1 35 (35.0%) 47 (39.5%) N2 24 (24.0%) 44 (37.0%) Histologic type high 12 (12.0%) 7 (5.9%) 0.193 middle 81 (81.0%) 99 (83.2%) upper 7 (7.0%) 13 (10.9%) Number of metastases 1 57 (57.0%) 74 (62.2%) 0.521 >1 43 (43.0%) 45 (37.8%) Mutational status Wild type 58 (58.0%) 66 (55.5%) 0.892 RAS 39 (39.0%) 50 (42.0%) BRAF 3 (3.0%) 3 (2.5%) Primary lesion surgery No 25 (25.0%) 32 (26.9%) 0.870 Yes 75 (75.0%) 87 (73.1%) local treatment No 49 (49.0%) 68 (57.1%) 0.286 Yes 51 (51.0%) 51 (42.9%) CEA <5 35 (35.0%) 26 (21.8%) 0.044 ≥5 65 (65.0%) 93 (78.2%) Neutrophil‐lymphocyte ratio 2.865(2.69) 3.26( 2.845) 0.336 Platelet lymphocyte ratio 160.3(103.955) 180.63(162.06) 0.168 Hb, g/L 125(26.25) 119( 31.5) 0.018 Albumin, g/L 39.1(4.75) 39(5.3) 0.727 mGPS 0 65 (65.0%) 68 (57.1%) 0.200 1 22 (22.0%) 39 (32.8%) 2 13 (13.0%) 12 (10.1%) Glucose,mmol/L 5.235( 1.0125) 5.27(1.555) 0.250 Cr μmmol/L 67(21.5) 70( 22.5) 0.587 LDH,U/L 186.5(90.5) 183(68.5) 0.345 CRP,mg/L 3.45(14.7575) 6.85(30.43) 0.033 UA,Umol/L 266.5(92.5) 273( 123) 0.844 TG,mmol/L 1.055(0.69) 1.08(0.54) 0.849 CK ， U/L 74(52.5) 64(58) 0.428 Urea,mmol/L 4.615(2.0125) 4.83( 1.755) 0.202 TC,mmol/L 4.28( 1.095) 4.35(1.34) 0.312 HDL.C,mmol/L 1.06(0.425) 1.06(0.44) 0.846 Note : Data are median (IQR) or n (%). Abbreviations: Cr: creatinine; Glu：glucose; HDL.C: high density lipoprotein cholesterol; Hb: hemoglobin; mGPS: modified glasgow prognostic score; TG: triglycerides; CK: creatine kinase; TC: total cholesterol; UA: uric acid. ML model evaluation To effectively avoid redundancy or overfitting while selecting essential features, the LASSO regression model was employed. The 5-fold cross-validation method was applied to the iterative analysis, yielding a model with optimal performance and a minimal number of variables when λ was 0.047 (Log λ=-1.33). The LASSO logistic regression model identified 13 parameters with non-zero coefficients, including age, T stage, N stage, histologic type, metastases, primary site, local treatment, neutrophil-to-lymphocyte ratio (NLR), CEA, Hb, Glu, Cr, and TC (Figure 2). Subsequently, based on these 13 significant variables, four ML algorithms (LR, RF, XGBoost, and KNN) were employed to construct models, and their performance was assessed using ROC curves (Figure 2). The RF model demonstrated the highest accuracy in predicting the risk of malignancy in CRC, with an area under the receiver operating characteristic curve (AUC) of 0.768 (Figure 3). The accuracy, sensitivity, and specificity of the RF model were 0.744, 0.790, and 0.690, respectively (Table 2). A risk prediction model incorporating these variables was constructed through a random forest (RF) classifier. Following parameter debugging, a stabilized error rate was observed with ntree set to 100. The average decrease in Gini coefficient indicated that TC, followed by Hb, Glu, NLR, N stage, CEA, Cr, age, and metastases, were important features for CC risk assessment (Figure 4). Fig 2 Fig 2. Screening of variables based on Lasso regression. (A) The variation characteristics of the coefficient of variables; (B) the selection process of the optimum value of the parameter λ in the Lasso regression model by cross-validation method. Each colored curve represents features relative to log（ λ) LASSO coefficient profile of the sequence. Fig 3 Fig 3.ROC curves of four machine learning algorithm models for predicting the risk of CC in patients with colorectal tumors. The RF (Random Forest) model showed the best performance with an AUC of 0.768. Fig 4 Fig 4. Variable Importance Based on Random Forest, ranking of out-of-bag variable importance. Table2.Performance Summary of Machine Learning Models. KNN, k-nearest neighbor algorithm; RF, random Forest; XGBoost, extreme gradient boosting. Model Accuracy Sensitivity Specificity Precision F1 score KNN 0.68 0.65 0.71 0.73 0.68 Logistic 0.69 0.71 0.67 0.72 0.71 RF 0.74 0.79 0.69 0.75 0.77 XGBoost 0.68 0.66 0.71 0.73 0.69 Abbreviations: KNN，k-nearest neighbor algorithm；RF,random forest; XGBoost, extreme gradient boosting. Model interpretation Figure 5 illustrates the importance of SHAP features based on the RF model. The average absolute SHAP values of the evaluated features, ranked from highest to lowest, depict the impact of features on prediction. The first five key features are TC, Hb, Glu, NLR, and N stage. The SHAP summary of the RF model visualizes the impact of features on the prediction model (Figure 5). Higher SHAP values of features correlate with a greater likelihood of CC. For instance, compared with patients with elevated Hb levels, those with decreased Hb levels during therapy more likely than not experience CC. Similarly, Patients with elevated CEA during treatment are far more likely experience CC compared to those with decreased CEA, in addition, the SHAP correlation map illustrates the impact of features on RF prediction (Figure 6). Patients with elevated TC levels (increased x-axis values) are associated with higher SHAP values, indicating a higher likelihood of CC (increased y-axis values). Additionally, a decrease in Hb (decrease in x-axis value) is associated with higher SHAP values. Fig5 Fig 5. SHAP summary chart based on 13 features of RF model. The higher the SHAP value of the feature, the higher the probability of malignant fluid quality occurring. Create a point for each feature attribute value of each patient's model, in order to assign a point for each patient on the line of each feature. The points are colored based on the corresponding patient's characteristic values and vertically accumulated to depict the density. Red represents higher feature values, while blue represents lower feature values. Fig 6 Fig 6. SHAP dependence plot of the RF model: (A) TC change, (B) Hb change, y-axis represents the SHAP values of features, and some feature values are displayed on the x-axis. Each point represents the SHAP value of each patient's characteristic, with red to blue indicating the value of the characteristic from high to low. A SHAP value exceeding zero for a specific feature indicates an increased risk of CC development. Discussion This study marks the first attempt to use ML models for predicting the risk of CC in stage IV CRC patients. We evaluated four ML prediction methods, utilizing patient clinical features for CC prediction. Among the models assessed, the RF model exhibited superior performance in single model prediction, achieving the highest AUC value. Additionally, we employed the SHAP method to elucidate the RF model, highlighting TC, Hb, Glu, NLR, and N stage as the most crucial features. Importantly, these variables are readily accessible from the medical records of stage IV cancer patients, providing clinicians with valuable routine data to assess these risk factors and identify patients at risk of CC. One notable strength of our study lies in the use of SHAP values to unveil the black box of ML. While previous risk scoring models identified several risk factors, including cancer site, cancer stage, time from symptom onset to hospitalization, appearance loss, BMI, skeletal muscle index, and NLR[18], our research highlights the significance of dynamic changes in metabolic indicators such as TC, Glu, and Cr, which traditional models often overlook. In this study, we employed metabolic indicators such as TC, Hb, Glu, and Cr to assess the risk of developing cachexia. Cancer cachexia is caused by the activation of energy rich compound mobilization processes [3, 19], such as increased liver gluconeogenesis [20]. Causing impaired glucose and lipid homeostasis[21], Cancer cachexia is also associated with insulin resistance, which enhances liver glycolysis [3, 22, 23] and lipid mobilization in white adipose tissue[24]. Insulin resistance in cancer patients is characterized by decreased insulin sensitivity or impaired glucose tolerance[25], and is suspected to increase during the progression of cachexia[26, 27]. In addition, elevated levels of glucagon[28] promote liver gluconeogenesis, thereby increasing blood sugar levels in cancer cachexia[28, 29]. However, others have also pointed out that there is no change in glucose levels in cancer cachexia[22, 30], which may reflect the complex metabolic dynamics of cancer cachexia[31]. Here, we investigated fasting blood glucose levels, whose damage is the second sign of insulin resistance (unrelated to impaired glucose tolerance)[32]. Our research findings support the view that insulin resistance increases in cancer cachexia. Patients with significantly elevated plasma fasting blood glucose levels also have an increased risk of developing cachexia. However, our research findings suggest that an increase in Glu is associated with the occurrence of cachexia, rather than with cancer staging. Previous studies have also shown that high fasting blood glucose levels during cancer diagnosis are associated with poor prognosis in non-small cell lung cancer (NSCLC) patients[33], supporting the view that elevated fasting blood glucose levels are not only an indicator of metabolic disorders or insulin resistance, but also a marker of poor prognosis. The increase in lipolysis in CC patients may be mediated by norepinephrine signals involved in the sympathetic nervous system[34, 35]. In addition, systemic inflammation associated with cachexia and/or activation of brown adipose tissue requires a high energy demand[34, 36], and brown adipose tissue can effectively remove energy rich compounds from normal physiological circulation[37]. Adipose tissue is an important regulatory factor for body composition and plays a crucial role in energy balance. White adipose tissue (WAT) and brown adipose tissue (BAT) regulate energy storage and body temperature, respectively. BAT maintains body temperature by generating heat, including increasing the expression of uncoupling protein 1 (UCP1) in adipose tissue and regulating glucose and lipid metabolism[38]. In addition, the thermogenic effect of BAT is believed to enhance resting energy expenditure and lipid mobilization[31]. The browning effect observed in metabolic disorders (such as obesity, diabetes and cancer) is proved to be harmful in CC[2]. Patients with cachexia typically exhibit elevated levels of circulating free fatty acids and glycerol, due to the extensive lipolysis of WAT by activating lipases such as hormone sensitive lipase and triglyceride lipase [39]. Therefore, the interactions between these metabolic processes may manifest as changes in plasma levels of energy rich compounds associated with cachexia[40]. The loss of skeletal muscle tissue is a key characteristic of cancer cachexia and the best studied aspect [1, 11, 41]. Muscle is the source of amino acids, which may be released for energy production during the process of catabolism[42]. Muscle homeostasis is maintained through a balance between the synthesis and degradation of muscle proteins[3]. However, when there is excessive degradation of protein and/or a decrease in protein synthesis in skeletal muscle, this imbalance can lead to muscle atrophy and the occurrence of cachexia [3, 41, 42].The SHAP summary chart indicates that changes in these metabolic characteristics are key predictive factors, indicating that as TC, Glu, and Cr increase, the risk of CC occurrence also increases. In addition, CC is associated with hyperlipidemia and decreased levels of branched chain amino acids[43]. It is currently unclear whether interventions based on these metabolic indicators can regulate CC correlation, and further research is needed. In our study, we used hemoglobin as a nutritional variable to assess the nutritional status and determine the risk of CC. While some reports suggest that low hemoglobin is a risk factor for CC and is associated with more complications and lower postoperative survival rates[44, 45], others indicate inconclusive results when using nutritional variables, including hemoglobin, to assess nutritional status, including CC[46]. Cancer patients with lower serum hemoglobin levels are more likely to die from this disease[47]. In CC patients, low hemoglobin levels are associated with mortality; therefore, hemoglobin can be used as a prognostic indicator. Our findings, however, indicate a significant association between serum hemoglobin levels and the risk of CC, with lower Hb levels indicating a higher risk of developing cachexia. Additionally, cancer-related indicators, such as cancer staging, were employed to assess the risk of CC in stage IV cancer patients. Given that cancer is the primary cause of cachexia, studying cancer-related factors for assessing cachexia risk in cancer patients holds significance. Our study identified a substantial correlation between cancer staging and CC, aligning with other studies[48]. Advanced-stage cancer patients often exhibit abnormal tumor biomarkers, further emphasizing the importance of detecting these biomarkers in routine cancer management. We questioned whether abnormal tumor biomarkers could serve as predictive indicators of cachexia, and our research indicated their potential relevance. Nevertheless, our study found that the histological type and differentiation of tumors had no significant impact on CC. This observation suggests that histological type may not be related to the patient's metabolic status and, consequently, may not significantly influence WL and cachexia. In conclusion, factors such as cancer stage and location, along with tumor-related indicators, offer valuable insights into assessing the risk of CC in stage IV CRC patients. One of the primary clinical features of CC is systemic inflammation, a key driver in its development and affects various tissues such as skeletal muscle, fat, brain, and liver[3]. Cancer-related pro-inflammatory cytokines, such as interleukin (IL)-6), IL-1, and tumor necrosis factor-α, contribute to inhibiting albumin production, a factor associated with the occurrence of CC[49]. Notably, NLR has recently been recognized as a new inflammatory evaluation indicator that can be easily obtained from blood routine. Moreover, NLR is a significant negative prognostic biomarker for cachexia patients[50]. In this study, NLR emerged as an independent risk factor for cancer cachexia, emphasizing the convenience of evaluating inflammation in clinical practice, especially in assessing the risk of cachexia. This suggests that anti-inflammatory interventions could potentially regulate systemic inflammation, aiding in muscle protection during treatment. Despite recommendations for nutrition and physical activity to maintain muscle mass, their efficacy might be limited in individuals with increased systemic inflammation[51, 52]. Our research has some limitations. First, due to the retrospective design and data collection from a single institution, there may be selection bias. Second, with evolving treatment strategies over the study period (2018–2022), our research may not entirely represent current medical practices. Third, the study's limited sample size from a single center in China necessitates additional validation in diverse centers to confirm the reliability of the ML model, and its applicability to patients outside China remains uncertain. To address these limitations, prospective trials on a larger and more diverse population with stricter inclusion criteria are essential. Finally, while nutritional management and the use of supplements or therapeutic diets are critical in evaluating the predictive utility of nutritional and inflammatory measures, this study lacks this information. The correlation between nutritional and inflammatory status and the prognosis of stage IV CRC warrants further investigation and validation in prospective cohort studies. Conclusion The ML model, utilizing easily accessible clinical data, effectively identifies stage IV CRC patients with CC. The prediction of ML model is explained through SHAP method, making them applicable in clinical settings. This aids clinicians in better screening for CC, adopting targeted interventions, and improving the understanding of patient prognosis. In daily clinical practice, clinical indicators should be considered to evaluate the host status of stage IV cancer patients and strengthen nutritional management. Declarations Acknowledgments Jing Wang: Conceptualization, Methodology, Software, Writing – original draft. Yaoxian Xiang: Data curation. Kangjie Wang: Software, Writing – original draft. Baojuan Han: Visualization. Lei Wu: Investigation. Dong Yan: Supervision. Jing wang and Baojuan Han: Validation, Writing – review & editing. Li Wang: Writing – review & editing. Funding: This work was financially supported by the capital health research and development of special (2022-2-7083); R&D Program of Beijing Municipal Education Commission (KM202010025005); Beijing Municipal Natural Science Foundation7222100 and 7232085; National Natural Science Foundation (No. 82203522). Competing interest The authors declare that they have no competing interests. Availability of data and materials All data generated during this study are included in this published article. Further inquiry data can be obtained by the corresponding author. Ethics Statement This study was approved by the Beijing Luhe Hospital Ethic Committee for Clinical Investigation. References Argilés JM, Busquets S, Stemmler B, López-Soriano FJ: Cancer cachexia: understanding the molecular basis . Nature reviews Cancer 2014, 14 (11):754-762. Argilés JM, Stemmler B, López-Soriano FJ, Busquets S: Inter-tissue communication in cancer cachexia . Nature reviews Endocrinology 2018, 15 (1):9-20. Porporato PE: Understanding cachexia as a cancer metabolism syndrome . Oncogenesis 2016, 5 (2):e200. Tsoli M, Robertson G: Cancer cachexia: malignant inflammation, tumorkines, and metabolic mayhem . Trends Endocrinol Metab 2013, 24 (4):174-183. Fearon KC: Cancer cachexia and fat-muscle physiology . N Engl J Med 2011, 365 (6):565-567. Quan-Jun Y, Yan H, Yong-Long H, Li-Li W, Jie L, Jin-Lu H, Jin L, Peng-Guo C, Run G, Cheng G: Selumetinib Attenuates Skeletal Muscle Wasting in Murine Cachexia Model through ERK Inhibition and AKT Activation . Mol Cancer Ther 2017, 16 (2):334-343. Temel JS, Abernethy AP, Currow DC, Friend J, Duus EM, Yan Y, Fearon KC: Anamorelin in patients with non-small-cell lung cancer and cachexia (ROMANA 1 and ROMANA 2): results from two randomised, double-blind, phase 3 trials . Lancet Oncol 2016, 17 (4):519-531. Garcia JM, Boccia RV, Graham CD, Yan Y, Duus EM, Allen S, Friend J: Anamorelin for patients with cancer cachexia: an integrated analysis of two phase 2, randomised, placebo-controlled, double-blind trials . Lancet Oncol 2015, 16 (1):108-116. Fearon K, Strasser F, Anker SD, Bosaeus I, Bruera E, Fainsinger RL, Jatoi A, Loprinzi C, MacDonald N, Mantovani G et al : Definition and classification of cancer cachexia: an international consensus . The Lancet Oncology 2011, 12 (5):489-495. Fearon KC, Baracos VE: Cachexia in pancreatic cancer: new treatment options and measures of success . HPB : the official journal of the International Hepato Pancreato Biliary Association 2010, 12 (5):323-324. Fearon KC: Cancer cachexia: developing multimodal therapy for a multidimensional problem . European journal of cancer (Oxford, England : 1990) 2008, 44 (8):1124-1132. Prado CM, Lieffers JR, McCargar LJ, Reiman T, Sawyer MB, Martin L, Baracos VE: Prevalence and clinical implications of sarcopenic obesity in patients with solid tumours of the respiratory and gastrointestinal tracts: a population-based study . The Lancet Oncology 2008, 9 (7):629-635. Fearon K, Arends J, Baracos V: Understanding the mechanisms and treatment options in cancer cachexia . Nature reviews Clinical oncology 2013, 10 (2):90-99. Prokopchuk O, Hermann CD, Schoeps B, Nitsche U, Prokopchuk OL, Knolle P, Friess H, Martignoni ME, Krüger A: A novel tissue inhibitor of metalloproteinases-1/liver/cachexia score predicts prognosis of gastrointestinal cancer patients . Journal of cachexia, sarcopenia and muscle 2021, 12 (2):378-392. Jafri SH, Previgliano C, Khandelwal K, Shi R: Cachexia Index in Advanced Non-Small-Cell Lung Cancer Patients . Clinical Medicine Insights Oncology 2015, 9 :87-93. Yang QJ, Zhao JR, Hao J, Li B, Huo Y, Han YL, Wan LL, Li J, Huang J, Lu J et al : Serum and urine metabolomics study reveals a distinct diagnostic model for cancer cachexia . Journal of cachexia, sarcopenia and muscle 2018, 9 (1):71-85. Rajkomar A, Dean J, Kohane I: Machine Learning in Medicine . The New England journal of medicine 2019, 380 (14):1347-1358. Tan S, Xu J, Wang J, Zhang Z, Li S, Yan M, Tang M, Liu H, Zhuang Q, Xi Q et al : Development and validation of a cancer cachexia risk score for digestive tract cancer patients before abdominal surgery . Journal of cachexia, sarcopenia and muscle 2023, 14 (2):891-902. Petruzzelli M, Wagner EF: Mechanisms of metabolic dysfunction in cancer-associated cachexia . Genes & development 2016, 30 (5):489-501. Noguchi Y, Vydelingum NA, Brennan MF: The reversal of increased gluconeogenesis in the tumor-bearing rat by tumor removal and food intake . Surgery 1989, 106 (2):423-430; discussion 430-421. Petruzzelli M, Ferrer M, Schuijs MJ, Kleeman SO, Mourikis N, Hall Z, Perera D, Raghunathan S, Vacca M, Gaude E et al : Early Neutrophilia Marked by Aerobic Glycolysis Sustains Host Metabolism and Delays Cancer Cachexia . Cancers 2022, 14 (4). Dev R, Bruera E, Dalal S: Insulin resistance and body composition in cancer patients . Annals of oncology : official journal of the European Society for Medical Oncology 2018, 29 (suppl_2):ii18-ii26. Rui L: Energy metabolism in the liver . Comprehensive Physiology 2014, 4 (1):177-197. Jaworski K, Sarkadi-Nagy E, Duncan RE, Ahmadian M, Sul HS: Regulation of triglyceride metabolism. IV. Hormonal regulation of lipolysis in adipose tissue . American journal of physiology Gastrointestinal and liver physiology 2007, 293 (1):G1-4. Honors MA, Kinzig KP: The role of insulin resistance in the development of muscle wasting during cancer cachexia . Journal of cachexia, sarcopenia and muscle 2012, 3 (1):5-11. Tisdale MJ: Wasting in cancer . The Journal of nutrition 1999, 129 (1S Suppl):243s-246s. Jasani B, Donaldson LJ, Ratcliffe JG, Sokhi GS: Mechanism of impaired glucose tolerance in patients with neoplasia . British journal of cancer 1978, 38 (2):287-292. Bartlett DL, Charland SL, Torosian MH: Reversal of tumor-associated hyperglucagonemia as treatment for cancer cachexia . Surgery 1995, 118 (1):87-97. Tayek JA: A review of cancer cachexia and abnormal glucose metabolism in humans with cancer . Journal of the American College of Nutrition 1992, 11 (4):445-456. Schwarz S, Prokopchuk O, Esefeld K, Gröschel S, Bachmann J, Lorenzen S, Friess H, Halle M, Martignoni ME: The clinical picture of cachexia: a mosaic of different parameters (experience of 503 patients) . BMC cancer 2017, 17 (1):130. Fonseca G, Farkas J, Dora E, von Haehling S, Lainscak M: Cancer Cachexia and Related Metabolic Dysfunction . International journal of molecular sciences 2020, 21 (7). Petersen JL, McGuire DK: Impaired glucose tolerance and impaired fasting glucose--a review of diagnosis, clinical implications and management . Diabetes & vascular disease research 2005, 2 (1):9-15. Luo J, Chen YJ, Chang LJ: Fasting blood glucose level and prognosis in non-small cell lung cancer (NSCLC) patients . Lung cancer (Amsterdam, Netherlands) 2012, 76 (2):242-247. Tisdale MJ: Mechanisms of cancer cachexia . Physiological reviews 2009, 89 (2):381-410. Bartelt A, Heeren J: Adipose tissue browning and metabolic health . Nature reviews Endocrinology 2014, 10 (1):24-36. Beijer E, Schoenmakers J, Vijgen G, Kessels F, Dingemans AM, Schrauwen P, Wouters M, van Marken Lichtenbelt W, Teule J, Brans B: A role of active brown adipose tissue in cancer cachexia? Oncology reviews 2012, 6 (1):e11. Cohen P, Spiegelman BM: Brown and Beige Fat: Molecular Parts of a Thermogenic Machine . Diabetes 2015, 64 (7):2346-2351. Virtanen KA, Lidell ME, Orava J, Heglind M, Westergren R, Niemi T, Taittonen M, Laine J, Savisto NJ, Enerbäck S et al : Functional brown adipose tissue in healthy adults . The New England journal of medicine 2009, 360 (15):1518-1525. Han J, Meng Q, Shen L, Wu G: Interleukin-6 induces fat loss in cancer cachexia by promoting white adipose tissue lipolysis and browning . Lipids in health and disease 2018, 17 (1):14. Zwickl H, Zwickl-Traxler E, Pecherstorfer M: Is Neuronal Histamine Signaling Involved in Cancer Cachexia? Implications and Perspectives . Frontiers in oncology 2019, 9 :1409. Schmidt SF, Rohm M, Herzig S, Berriel Diaz M: Cancer Cachexia: More Than Skeletal Muscle Wasting . Trends in cancer 2018, 4 (12):849-860. Bonaldo P, Sandri M: Cellular and molecular mechanisms of muscle atrophy . Disease models & mechanisms 2013, 6 (1):25-39. Di Gangi IM, Mazza T, Fontana A, Copetti M, Fusilli C, Ippolito A, Mattivi F, Latiano A, Andriulli A, Vrhovsek U et al : Metabolomic profile in pancreatic cancer patients: a consensus-based approach to identify highly discriminating metabolites . Oncotarget 2016, 7 (5):5815-5829. Wallengren O, Lundholm K, Bosaeus I: Diagnostic criteria of cancer cachexia: relation to quality of life, exercise capacity and survival in unselected palliative care patients . Supportive care in cancer : official journal of the Multinational Association of Supportive Care in Cancer 2013, 21 (6):1569-1577. Arends J, Strasser F, Gonella S, Solheim TS, Madeddu C, Ravasco P, Buonaccorso L, de van der Schueren MAE, Baldwin C, Chasen M et al : Cancer cachexia in adult patients: ESMO Clinical Practice Guidelines(☆) . ESMO open 2021, 6 (3):100092. Martin L, Birdsell L, Macdonald N, Reiman T, Clandinin MT, McCargar LJ, Murphy R, Ghosh S, Sawyer MB, Baracos VE: Cancer cachexia in the age of obesity: skeletal muscle depletion is a powerful prognostic factor, independent of body mass index . Journal of clinical oncology : official journal of the American Society of Clinical Oncology 2013, 31 (12):1539-1547. Zhang XW, Zhang Q, Song MM, Zhang KP, Zhang X, Ruan GT, Yang M, Ge YZ, Tang M, Li XR et al : The prognostic effect of hemoglobin on patients with cancer cachexia: a multicenter retrospective cohort study . Supportive care in cancer : official journal of the Multinational Association of Supportive Care in Cancer 2022, 30 (1):875-885. Poisson J, Martinez-Tapia C, Heitz D, Geiss R, Albrand G, Falandry C, Gisselbrecht M, Couderc AL, Boulahssass R, Liuu E et al : Prevalence and prognostic impact of cachexia among older patients with cancer: a nationwide cross-sectional survey (NutriAgeCancer) . Journal of cachexia, sarcopenia and muscle 2021, 12 (6):1477-1488. Gupta D, Lis CG: Pretreatment serum albumin as a predictor of cancer survival: a systematic review of the epidemiological literature . Nutrition journal 2010, 9 :69. Zhang Q, Song MM, Zhang X, Ding JS, Ruan GT, Zhang XW, Liu T, Yang M, Ge YZ, Tang M et al : Association of systemic inflammation with survival in patients with cancer cachexia: results from a multicentre cohort study . Journal of cachexia, sarcopenia and muscle 2021, 12 (6):1466-1476. Golder AM, Sin LKE, Alani F, Alasadi A, Dolan R, Mansouri D, Horgan PG, McMillan DC, Roxburgh CS: The relationship between the mode of presentation, CT-derived body composition, systemic inflammatory grade and survival in colon cancer . Journal of cachexia, sarcopenia and muscle 2022, 13 (6):2863-2874. Collao N, Sanders O, Caminiti T, Messeiller L, De Lisio M: Resistance and endurance exercise training improves muscle mass and the inflammatory/fibrotic transcriptome in a rhabdomyosarcoma model . Journal of cachexia, sarcopenia and muscle 2023, 14 (2):781-793. Additional Declarations No competing interests reported. Cite Share Download PDF Status: Posted Version 1 posted You are reading this latest preprint version Research Square lets you share your work early, gain feedback from the community, and start making changes to your manuscript prior to peer review in a journal. As a division of Research Square Company, we’re committed to making research communication faster, fairer, and more useful. We do this by developing innovative software and high quality services for the global research community. Our growing team is made up of researchers and industry professionals working together to solve the most critical problems facing scientific publishing. Also discoverable on Platform About Our Team In Review Editorial Policies Advisory Board Help Center Resources Author Services Accessibility API Access RSS feed Manage Cookie Preferences © Research Square 2026 | ISSN 2693-5015 (online) Privacy Policy Terms of Service Do Not Sell My Personal Information {"props":{"pageProps":{"initialData":{"identity":"rs-4275850","acceptedTermsAndConditions":true,"allowDirectSubmit":true,"archivedVersions":[],"articleType":"Research Article","associatedPublications":[],"authors":[{"id":292591438,"identity":"fecc5bc1-544e-434f-be64-ef6a645adaaa","order_by":0,"name":"Jing Wang","email":"","orcid":"","institution":"Beijing Luhe hospital Affiliated to Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Jing","middleName":"","lastName":"Wang","suffix":""},{"id":292591439,"identity":"ca6aeb83-85f0-48ba-83e1-e1e337d33928","order_by":1,"name":"Baojuan Han","email":"","orcid":"","institution":"Beijing Luhe hospital Affiliated to Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Baojuan","middleName":"","lastName":"Han","suffix":""},{"id":292591440,"identity":"be899c9a-8e14-4390-89fb-cd8e13049ce5","order_by":2,"name":"Kangjie Wang","email":"","orcid":"","institution":"Beijing Luhe hospital Affiliated to Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Kangjie","middleName":"","lastName":"Wang","suffix":""},{"id":292591441,"identity":"af133dec-ea56-4123-a9ba-be091003a22b","order_by":3,"name":"Yaoxian Xiang","email":"","orcid":"","institution":"Beijing Luhe hospital Affiliated to Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Yaoxian","middleName":"","lastName":"Xiang","suffix":""},{"id":292591442,"identity":"544c9750-150e-487d-9165-4f8407b6bc17","order_by":4,"name":"Lei Wu","email":"","orcid":"","institution":"Beijing Luhe hospital Affiliated to Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Lei","middleName":"","lastName":"Wu","suffix":""},{"id":292591443,"identity":"9677e669-8f79-4e65-8649-359329ca777a","order_by":5,"name":"Dong Yan","email":"data:image/png;base64,iVBORw0KGgoAAAANSUhEUgAAAZAAAAAyAQMAAABI0h/eAAAABlBMVEX///8AAABVwtN+AAAACXBIWXMAAA7EAAAOxAGVKw4bAAAAyElEQVRIiWNgGAWjYFCCxPYPDAw2cvzMzIcfEKkluY2BISHNWLKdLc2ASC3pIC2HEzec51GQIEqDwfHEtse8P9KMjQ/zMBgw1NhEE9Zy5mG7MU+CjZzZYd4DDxiOpeU2ENRyI7FBmgfoF7PDfAkGjA2HidZyOHFzM4+BBLFa2sBaNjATq0XyzMNmwzlpacYSh4GBnECMX/iOpz988MYGGJX9hw8/+FBjQ1iLwgEGBiYeGC+BkHIQkAcayviDGJWjYBSMglEwcgEAdMxD8pStinMAAAAASUVORK5CYII=","orcid":"","institution":"Beijing Luhe hospital Affiliated to Capital Medical University","correspondingAuthor":true,"prefix":"","firstName":"Dong","middleName":"","lastName":"Yan","suffix":""},{"id":292591444,"identity":"086eeb6a-7056-479c-b40f-4a985cd4fc17","order_by":6,"name":"Li Wang","email":"","orcid":"","institution":"Beijing Luhe hospital Affiliated to Capital Medical University","correspondingAuthor":false,"prefix":"","firstName":"Li","middleName":"","lastName":"Wang","suffix":""}],"badges":[],"createdAt":"2024-04-16 11:57:15","currentVersionCode":1,"declarations":"","doi":"10.21203/rs.3.rs-4275850/v1","doiUrl":"https://doi.org/10.21203/rs.3.rs-4275850/v1","draftVersion":[],"editorialEvents":[],"editorialNote":"","failedWorkflow":false,"files":[{"id":55329381,"identity":"d985e286-afa4-430e-8367-c3c6af372364","added_by":"auto","created_at":"2024-04-25 18:58:59","extension":"png","order_by":1,"title":"Figure 1","display":"","copyAsset":false,"role":"figure","size":71759,"visible":true,"origin":"","legend":"\u003cp\u003eFlow diagram illustrating recruitment of patients.\u003c/p\u003e","description":"","filename":"1.png","url":"https://assets-eu.researchsquare.com/files/rs-4275850/v1/7a4bc9260b5bfe0c4ae04507.png"},{"id":55329380,"identity":"37ab5ace-d0c9-47ba-bff3-682aff1aa872","added_by":"auto","created_at":"2024-04-25 18:58:59","extension":"png","order_by":2,"title":"Figure 2","display":"","copyAsset":false,"role":"figure","size":71651,"visible":true,"origin":"","legend":"\u003cp\u003eScreening of variables based on Lasso regression. (A) The variation characteristics of the coefficient of variables; (B) the selection process of the optimum value of the parameter λ in the Lasso regression model by cross-validation method. Each colored curve represents features relative to log（ λ) LASSO coefficient profile of the sequence.\u003c/p\u003e","description":"","filename":"2.png","url":"https://assets-eu.researchsquare.com/files/rs-4275850/v1/aead9d58971002778c97bf95.png"},{"id":55329383,"identity":"58dcf5c1-677e-4b56-a370-9452e8a13b46","added_by":"auto","created_at":"2024-04-25 18:58:59","extension":"png","order_by":3,"title":"Figure 3","display":"","copyAsset":false,"role":"figure","size":64824,"visible":true,"origin":"","legend":"\u003cp\u003eROC curves of four machine learning algorithm models for predicting the risk of CC in patients with colorectal tumors. The RF (Random Forest) model showed the best performance with an AUC of 0.768.\u003c/p\u003e","description":"","filename":"3.png","url":"https://assets-eu.researchsquare.com/files/rs-4275850/v1/26ed1d1465c685148da39dac.png"},{"id":55329382,"identity":"d5cdf2aa-f1f0-4c8a-bbdc-6fd9606c59c9","added_by":"auto","created_at":"2024-04-25 18:58:59","extension":"png","order_by":4,"title":"Figure 4","display":"","copyAsset":false,"role":"figure","size":40479,"visible":true,"origin":"","legend":"\u003cp\u003eVariable Importance Based on Random Forest, ranking of out-of-bag variable importance.\u003c/p\u003e","description":"","filename":"4.png","url":"https://assets-eu.researchsquare.com/files/rs-4275850/v1/6db4c33336870e194fd03ddd.png"},{"id":55329384,"identity":"77990348-c43a-48e6-b25f-89a3988d908a","added_by":"auto","created_at":"2024-04-25 18:59:00","extension":"png","order_by":5,"title":"Figure 5","display":"","copyAsset":false,"role":"figure","size":89532,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP summary chart based on 13 features of RF model. The higher the SHAP value of the feature, the higher the probability of malignant fluid quality occurring. Create a point for each feature attribute value of each patient's model, in order to assign a point for each patient on the line of each feature. The points are colored based on the corresponding patient's characteristic values and vertically accumulated to depict the density. Red represents higher feature values, while blue represents lower feature values.\u003c/p\u003e","description":"","filename":"5.png","url":"https://assets-eu.researchsquare.com/files/rs-4275850/v1/b4002e9b4035ad70d3b0f325.png"},{"id":55329385,"identity":"611863ac-73c2-486b-ac4a-32235587c9c2","added_by":"auto","created_at":"2024-04-25 18:59:00","extension":"png","order_by":6,"title":"Figure 6","display":"","copyAsset":false,"role":"figure","size":31331,"visible":true,"origin":"","legend":"\u003cp\u003eSHAP dependence plot of the RF model: (A) TC change, (B) Hb change, y-axis represents the SHAP values of features, and some feature values are displayed on the x-axis. Each point represents the SHAP value of each patient's characteristic, with red to blue indicating the value of the characteristic from high to low. A SHAP value exceeding zero for a specific feature indicates an increased risk of CC development.\u003c/p\u003e","description":"","filename":"6.png","url":"https://assets-eu.researchsquare.com/files/rs-4275850/v1/7e76d7af645303f587236f05.png"},{"id":58845218,"identity":"3b306863-1012-4e5f-8cdd-59ff7c382ecc","added_by":"auto","created_at":"2024-06-22 02:53:28","extension":"pdf","order_by":0,"title":"","display":"","copyAsset":false,"role":"manuscript-pdf","size":2146384,"visible":true,"origin":"","legend":"","description":"","filename":"manuscript.pdf","url":"https://assets-eu.researchsquare.com/files/rs-4275850/v1/155312c1-4c0a-48bd-a4be-ce40ab9b4871.pdf"}],"financialInterests":"No competing interests reported.","formattedTitle":"Applying machine learning to predict the risk of cancer cachexia in stage IV colorectal cancer patients","fulltext":[{"header":"Introduction","content":"\u003cp\u003eCachexia is a complex systemic disease involving multiple metabolic pathways in various tissues and organs. It is characterized by systemic inflammation, progressive weight loss (WL), and depletion of adipose tissue and skeletal muscle. Traditional nutritional support has proven insufficient to fully reverse these effects[1].[1\u0026ndash;4] Cancer cachexia (CC) is a multifactorial clinical complication associated with various malignant tumors in humans, leading to a decline in patients' quality of life, treatment response, and treatment tolerance[5, 6]. Moreover, up to 80% of cancer patients experience CC. Recent clinical and experimental evidence clearly indicates that CC is not inevitable[7, 8], and early screening is particularly important for preventing or delaying the occurrence of cachexia. However, significant obstacles exist in identifying patients at risk of CC[9].\u003c/p\u003e \u003cp\u003ePrimarily, CC, being a form of malnutrition, lacks dedicated tools for risk assessment, unlike the tools available for identifying nutritional risks[10, 11]. Moreover, WL, a common symptom of cachexia, is the traditional and most widely used parameter caused by the consumption of skeletal muscle mass and loss of adipose tissue, but recent discussions highlight that complex syndromes associated with CC may not always result in significant total WL[12], a condition referred to as \u0026ldquo;hidden cachexia\u0026rdquo;[13]. Consequently, WL alone fails to classify all pathophysiological manifestations of cachexia, and its severity varies. Furthermore, many patients struggle to provide accurate information on the onset and extent of WL, limiting the efficacy of WL as an early identifier of CC risk. Recently, efforts have been made to develop clinical tools related to the occurrence of cachexia. Prokopchuk et al. introduced a tissue inhibitor of metalloproteinase-1/liver/cachexia score to predict the prognosis of gastrointestinal cancer patients[14], whereas Jafri et al. created a cachexia index based on clinical characteristics to assess the prognosis of non-small cell lung cancer patients[15]. Yang et al. conducted a serum and urine metabolomics study, unveiling a diagnostic model for CC[16].\u003c/p\u003e \u003cp\u003eRecently, machine learning (ML) has been applied in the medical field for tasks such as cancer classification, prediction, drug response, and treatment strategies. ML, by automatically learning from vast datasets, establishes implicit relationships between variables and outcomes and then builds an efficient model to predict the results of other data[17]. This study aims to determine whether cachexia serves as a prognostic factor for stage IV colorectal cancer (CRC), including cases of curable resection and metastatic CRC, influencing optimal treatment strategies from the perspective of host status. Four ML algorithms were employed to predict the risk of malignant tumors in stage IV CRC patients.\u003c/p\u003e"},{"header":"Materials and methods","content":"\u003cdiv id=\"Sec3\" class=\"Section2\"\u003e \u003ch2\u003eStudy population\u003c/h2\u003e \u003cp\u003eFrom January 1, 2015, to December 31, 2022, patients aged\u0026thinsp;\u0026ge;\u0026thinsp;18 years in our clinical database were retrospectively recruited. The inclusion criteria encompassed individuals diagnosed as consecutive stage IV CRC patients for the first time in our institution. Eligible patients were required to have stage IV CRC with histologically confirmed adenocarcinoma, excluding other histological types, anal cancer, or appendix cancer. Patients lacking complete clinical data for cachexia diagnosis or with a history of cancer were excluded. The study was conducted at the Cancer Center of Beijing Luhe Hospital affiliated with Capital Medical University of China, approved by our institutional review board, and adhered to the Declaration of Helsinki and its subsequent amendments to ethical standards(2022-LHKY-036-02). According to international consensus, individuals meeting the diagnostic criteria for CC were those with \u0026gt;\u0026thinsp;5% WL in the past 6 months, had BMI\u0026thinsp;\u0026lt;\u0026thinsp;20 kg/m\u003csup\u003e2\u003c/sup\u003e, or exhibited skeletal muscle consumption consistent with muscle loss, with any degree of WL\u0026thinsp;\u0026gt;\u0026thinsp;2% [9](21296615).\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec4\" class=\"Section2\"\u003e \u003ch2\u003eData collection\u003c/h2\u003e \u003cp\u003eWithin this cohort, data on candidate predictors, including demographics, clinical, nutritional, and oncological variables, were collected through medical records and patient interviews. The selection of predictive factors for CC was based on disciplinary knowledge and literature review. Demographic data comprised age, sex, and smoking and drinking status, while clinical data included disease complications, especially respiratory, cardiovascular, and diabetes-related complications; lymphocyte count; leukocyte count; Glucose (Glu); creatinine (Cr); and serum total cholesterol (TC) Nutritional data encompassed albumin, hemoglobin (Hb), and BMI calculated as weight (kg)/height (m\u003csup\u003e2\u003c/sup\u003e). Oncology data included tumor location, histological type (low, medium, and high differentiation status), carcinoembryonic antigen (CEA), gene status, (primary and local) surgical treatment, and T and N stages based on the 8th edition of the American Joint Commission Cancer Staging System. The Modified Glasgow Prognostic Score (mGPS) was scored as follows: score 0 for CRP\u0026thinsp;\u0026le;\u0026thinsp;1.0 mg/dL, score 1, CRP\u0026thinsp;\u0026gt;\u0026thinsp;1.0 mg/dL and albumin\u0026thinsp;\u0026ge;\u0026thinsp;3.5 g/dL, and score 2, CRP\u0026thinsp;\u0026gt;\u0026thinsp;1.0 mg/dL and albumin\u0026thinsp;\u0026lt;\u0026thinsp;3.5 g/d.\u003c/p\u003e \u003cp\u003eIn addition to the collected data, long-term follow-up results, including overall and recurrence-free survival, were collected. The follow-up data for survival status were updated in June 2023.\u003c/p\u003e \u003c/div\u003e \u003cdiv id=\"Sec5\" class=\"Section2\"\u003e \u003ch2\u003eStatistical analysis\u003c/h2\u003e \u003cp\u003eContinuous variables are presented as median [interquartile range (IQR)], and categorical variables as frequency and proportion. The Mann\u0026ndash;Whitney U test and χ2 test were employed. LASSO regression was conducted on all variables to identify CC risk factors. Shapley additive interpretation (SHAP) was used as an additive feature attribution method to visualize the association between features and the occurrence of cachexia. The horizontal axis represents the Shapley value, with values\u0026thinsp;\u0026gt;\u0026thinsp;0 indicating a positive contribution to the occurrence of CC in CRC. The left vertical axis represents the opposite order of importance of features, whereas the right vertical axis represents the feature values from low to high. The performance of ML models was evaluated using ROC, model accuracy, sensitivity, and specificity. All statistical analyses were performed using R software (version 4.3.1) (\u003cspan class=\"ExternalRef\"\u003e\u003cspan class=\"RefSource\"\u003ehttp://www.R-project.org\u003c/span\u003e\u003cspan address=\"http://www.R-project.org\" targettype=\"URL\" class=\"RefTarget\"\u003e\u003c/span\u003e\u003c/span\u003e) and Python (version 3.11, Python Software Foundation).\u003c/p\u003e \u003c/div\u003e"},{"header":"Results","content":"\u003cp\u003e\u003cem\u003eDemographics and clinical characteristics\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eA total of 219 patients were included in this study (Figure 1). The median (IQR) age was 66 (12) years, with 140 patients (63.9%) being males. Among the 219 patients, 119 (54.3%) had CC. The rectum (43.8%) was the most common tumor site, followed by the left colon (36.1%) and right colon (20.1%). The majority of colon cancer patients had advanced clinical stages (T4: 43.8%; T3: 39.3%; T2: 9.13%; N1: 37.4%). Descriptive characteristics of the cohort study population are presented in Table 1.\u003c/p\u003e\n\u003cp\u003eFig1. Flow diagram illustrating recruitment of patients.\u003c/p\u003e\n\u003cp\u003eTable1 Patient demographics and pathological characteristics.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"645\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\" rowspan=\"2\"\u003e\n \u003cp\u003e\u003cstrong\u003eVariables\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e\u003cstrong\u003eNo Cachexia\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCachexia\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\" rowspan=\"2\"\u003e\n \u003cp\u003e\u003cstrong\u003eP-value\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"48.18181818181818%\"\u003e\n \u003cp\u003e\u003cstrong\u003e(N=100)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"51.81818181818182%\"\u003e\n \u003cp\u003e\u003cstrong\u003e(N=119)\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAge,years\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e65(12)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e67(11.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.172\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eGender\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eMale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e64 (64.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e76 (63.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e1.000\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eFemale\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e36 (36.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e43 (36.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCo.morbidity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e31 (31.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e26 (21.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.167\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e69 (69.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e93 (78.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"55.26315789473684%\" colspan=\"2\"\u003e\n \u003cp\u003e\u003cstrong\u003eRespiratory.co.morbidity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e7 (7.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e6 (5.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.746\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e93 (93.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e113 (95.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eDiabetes\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e85 (85.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e83 (69.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.012\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e15 (15.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e36 (30.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"55.26315789473684%\" colspan=\"2\"\u003e\n \u003cp\u003e\u003cstrong\u003eCardiovascular.co.morbidity\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e89 (89.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e100 (84.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.386\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e11 (11.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e19 (16.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eSmoking status\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eNever\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e67 (67.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e73 (61.3%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.467\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eFormer/current\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e33 (33.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e46 (38.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eDrinking status\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eNever\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e74 (74.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e91 (76.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.791\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eFormer/current\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e26 (26.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e28 (23.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePrimary. site\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eleft\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e40 (40.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e39 (32.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.540\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003erectal\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e41 (41.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e55 (46.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eright\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e19 (19.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e25 (21.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eT stage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eT1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e9 (9.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e8 (6.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.493\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eT2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e12 (12.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e8 (6.7%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eT3\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e37 (37.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e49 (41.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eT4\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e42 (42.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e54 (45.4%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eN stage\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eN0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e41 (41.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e28 (23.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.014\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eN1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e35 (35.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e47 (39.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eN2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e24 (24.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e44 (37.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eHistologic type\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003ehigh\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e12 (12.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e7 (5.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.193\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003emiddle\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e81 (81.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e99 (83.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eupper\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e7 (7.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e13 (10.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eNumber of metastases\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e57 (57.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e74 (62.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.521\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u0026gt;1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e43 (43.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e45 (37.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eMutational status\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eWild type\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e58 (58.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e66 (55.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.892\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eRAS\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e39 (39.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e50 (42.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eBRAF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e3 (3.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e3 (2.5%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePrimary lesion surgery\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e25 (25.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e32 (26.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.870\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e75 (75.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e87 (73.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003elocal treatment\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eNo\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e49 (49.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e68 (57.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.286\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003eYes\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e51 (51.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e51 (42.9%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCEA\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u0026lt;5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e35 (35.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e26 (21.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.044\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u0026ge;5\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e65 (65.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e93 (78.2%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eNeutrophil‐lymphocyte ratio\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e2.865(2.69)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e3.26( 2.845)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.336\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003ePlatelet lymphocyte ratio\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e160.3(103.955)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e180.63(162.06)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.168\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eHb, g/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e125(26.25)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e119( 31.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.018\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eAlbumin, g/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e39.1(4.75)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e39(5.3)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.727\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003emGPS\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e0\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e65 (65.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e68 (57.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.200\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e1\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e22 (22.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e39 (32.8%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e2\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e13 (13.0%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e12 (10.1%)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e \u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eGlucose,mmol/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e5.235( 1.0125)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e5.27(1.555)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.250\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCr \u0026mu;mmol/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e67(21.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e70( 22.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.587\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eLDH,U/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e186.5(90.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e183(68.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.345\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCRP,mg/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e3.45(14.7575)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e6.85(30.43)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e\u003cstrong\u003e0.033\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eUA,Umol/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e266.5(92.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e273( 123)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.844\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eTG,mmol/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e1.055(0.69)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e1.08(0.54)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.849\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eCK\u003c/strong\u003e\u003cstrong\u003e，\u003c/strong\u003e\u003cstrong\u003eU/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e74(52.5)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e64(58)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.428\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eUrea,mmol/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e4.615(2.0125)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e4.83( 1.755)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.202\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eTC,mmol/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e4.28( 1.095)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e4.35(1.34)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.312\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"30.65015479876161%\"\u003e\n \u003cp\u003e\u003cstrong\u003eHDL.C,mmol/L\u003c/strong\u003e\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"24.613003095975234%\"\u003e\n \u003cp\u003e1.06(0.425)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"26.470588235294116%\"\u003e\n \u003cp\u003e1.06(0.44)\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.26625386996904%\"\u003e\n \u003cp\u003e0.846\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003e\u003cem\u003eNote\u003c/em\u003e: Data are median (IQR) or \u003cem\u003en\u003c/em\u003e (%).\u003c/p\u003e\n\u003cp\u003eAbbreviations: Cr: creatinine; Glu：glucose; HDL.C: high density lipoprotein\u0026nbsp;cholesterol; Hb: hemoglobin; mGPS: modified glasgow prognostic score; TG: triglycerides; CK: creatine kinase; TC: total cholesterol; UA: uric acid.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eML model evaluation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eTo effectively avoid redundancy or overfitting while selecting essential features, the LASSO regression model was employed. The 5-fold cross-validation method was applied to the iterative analysis, yielding a model with optimal performance and a minimal number of variables when \u0026lambda; was 0.047 (Log \u0026lambda;=-1.33). The LASSO logistic regression model identified 13 parameters with non-zero coefficients, including age, T stage, N stage, histologic type, metastases, primary site, local treatment, neutrophil-to-lymphocyte ratio (NLR), CEA, Hb, Glu, Cr, and TC (Figure 2). Subsequently, based on these 13 significant variables, four\u0026nbsp;ML\u0026nbsp;algorithms (LR, RF, XGBoost, and KNN)\u0026nbsp;were employed to construct models, and their performance was assessed using ROC curves (Figure 2). The RF model demonstrated the highest accuracy in predicting the risk of malignancy in CRC, with an\u0026nbsp;area under the receiver operating characteristic curve (AUC)\u0026nbsp;of 0.768 (Figure 3). The accuracy, sensitivity, and specificity of the RF model were 0.744, 0.790, and 0.690, respectively (Table 2). A risk prediction model incorporating these variables was constructed through a random forest (RF) classifier. Following parameter debugging, a stabilized error rate was observed with ntree set to 100. The average decrease in Gini coefficient indicated that TC, followed by Hb, Glu, NLR, N stage, CEA, Cr, age, and metastases, were important features for CC risk assessment (Figure 4).\u003c/p\u003e\n\u003cp\u003eFig 2\u003c/p\u003e\n\u003cp\u003eFig 2. Screening of variables based on Lasso regression. (A) The variation characteristics of the coefficient of variables; (B) the selection process of the optimum value of the parameter \u0026lambda; in the Lasso regression model by cross-validation method. Each colored curve represents features relative to log（\u0026nbsp;\u0026lambda;) LASSO coefficient profile of the sequence.\u003c/p\u003e\n\u003cp\u003eFig 3\u003c/p\u003e\n\u003cp\u003eFig 3.ROC curves of four machine learning algorithm models for predicting the risk of CC in patients with colorectal tumors. The RF (Random Forest) model showed the best performance with an AUC of 0.768.\u003cstrong\u003e\u0026nbsp;\u003c/strong\u003e\u003c/p\u003e\n\u003cp\u003eFig 4\u003c/p\u003e\n\u003cp\u003eFig 4. Variable Importance Based on Random Forest, ranking of out-of-bag variable importance.\u003c/p\u003e\n\u003cp\u003eTable2.Performance Summary of Machine Learning Models. KNN, k-nearest neighbor algorithm; RF, random Forest; XGBoost,\u0026nbsp;extreme gradient boosting.\u003c/p\u003e\n\u003ctable border=\"0\" cellspacing=\"0\" cellpadding=\"0\" width=\"463\"\u003e\n \u003ctbody\u003e\n \u003ctr\u003e\n \u003ctd width=\"16.163793103448278%\"\u003e\n \u003cp\u003eModel\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.379310344827587%\"\u003e\n \u003cp\u003eAccuracy\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.887931034482758%\"\u003e\n \u003cp\u003eSensitivity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.103448275862068%\"\u003e\n \u003cp\u003eSpecificity\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.948275862068966%\"\u003e\n \u003cp\u003ePrecision\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.517241379310345%\"\u003e\n \u003cp\u003eF1 score\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"16.163793103448278%\"\u003e\n \u003cp\u003eKNN\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.379310344827587%\"\u003e\n \u003cp\u003e0.68\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.887931034482758%\"\u003e\n \u003cp\u003e0.65\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.103448275862068%\"\u003e\n \u003cp\u003e0.71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.948275862068966%\"\u003e\n \u003cp\u003e0.73\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.517241379310345%\"\u003e\n \u003cp\u003e0.68\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"16.163793103448278%\"\u003e\n \u003cp\u003eLogistic\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.379310344827587%\"\u003e\n \u003cp\u003e0.69\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.887931034482758%\"\u003e\n \u003cp\u003e0.71\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.103448275862068%\"\u003e\n \u003cp\u003e0.67\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.948275862068966%\"\u003e\n \u003cp\u003e0.72\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.517241379310345%\"\u003e\n \u003cp\u003e0.71\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"16.163793103448278%\"\u003e\n \u003cp\u003eRF\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.379310344827587%\"\u003e\n \u003cp\u003e0.74\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.887931034482758%\"\u003e\n \u003cp\u003e0.79\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.103448275862068%\"\u003e\n \u003cp\u003e0.69\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.948275862068966%\"\u003e\n \u003cp\u003e0.75\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.517241379310345%\"\u003e\n \u003cp\u003e0.77\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003ctr\u003e\n \u003ctd width=\"16.163793103448278%\"\u003e\n \u003cp\u003eXGBoost\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"16.379310344827587%\"\u003e\n \u003cp\u003e0.68\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"17.887931034482758%\"\u003e\n \u003cp\u003e0.66\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"18.103448275862068%\"\u003e\n \u003cp\u003e0.71\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.948275862068966%\"\u003e\n \u003cp\u003e0.73\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003ctd width=\"15.517241379310345%\"\u003e\n \u003cp\u003e0.69\u0026nbsp;\u003c/p\u003e\n \u003c/td\u003e\n \u003c/tr\u003e\n \u003c/tbody\u003e\n\u003c/table\u003e\n\u003cp\u003eAbbreviations: KNN，k-nearest neighbor algorithm；RF,random forest; XGBoost, extreme gradient boosting.\u003c/p\u003e\n\u003cp\u003e\u003cem\u003eModel interpretation\u003c/em\u003e\u003c/p\u003e\n\u003cp\u003eFigure 5 illustrates the importance of SHAP features based on the RF model. The average absolute SHAP values of the evaluated features, ranked from highest to lowest, depict the impact of features on prediction. The first five key features are TC, Hb, Glu, NLR, and N stage. The SHAP summary of the RF model visualizes the impact of features on the prediction model (Figure 5). Higher SHAP values of features correlate with a greater likelihood of CC. For instance, compared with patients with elevated Hb levels, those with decreased Hb levels during therapy more\u0026nbsp;likely\u0026nbsp;than\u0026nbsp;not experience CC. Similarly, Patients with elevated CEA during treatment are far more likely experience CC compared to those with decreased CEA, in addition, the SHAP correlation map illustrates the impact of features on RF prediction (Figure 6). Patients with elevated TC levels (increased x-axis values) are associated with higher SHAP values, indicating a higher likelihood of CC (increased y-axis values). Additionally, a decrease in Hb (decrease in x-axis value) is associated with higher SHAP values.\u003c/p\u003e\n\u003cp\u003eFig5\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eFig 5. SHAP summary chart based on 13 features of RF model. The higher the SHAP value of the feature, the higher the probability of malignant fluid quality occurring. Create a point for each feature attribute value of each patient\u0026apos;s model, in order to assign a point for each patient on the line of each feature. The points are colored based on the corresponding patient\u0026apos;s characteristic values and vertically accumulated to depict the density. Red represents higher feature values, while blue represents lower feature values.\u003c/p\u003e\n\u003cp\u003eFig 6\u003c/p\u003e\n\u003cp\u003eFig 6. SHAP dependence plot of the RF model: (A) TC change, (B) Hb change, y-axis represents the SHAP values of features, and some feature values are displayed on the x-axis. Each point represents the SHAP value of each patient\u0026apos;s characteristic, with red to blue indicating the value of the characteristic from high to low. A SHAP value exceeding zero for a specific feature indicates an increased risk of CC development.\u003c/p\u003e"},{"header":"Discussion","content":"\u003cp\u003eThis study marks the first attempt to use ML models for predicting the risk of CC in stage IV CRC\u0026nbsp;patients. We evaluated four ML prediction methods, utilizing patient clinical features for CC prediction. Among the models assessed, the RF model exhibited superior performance in single model prediction, achieving the highest AUC value. Additionally, we employed the SHAP method to elucidate the RF model, highlighting TC, Hb, Glu, NLR, and N stage as the most crucial features. Importantly, these variables are readily accessible from the medical records of stage IV cancer patients, providing clinicians with valuable routine data to assess these risk factors and identify patients at risk of CC.\u003c/p\u003e\n\u003cp\u003eOne notable strength of our study lies in the use of SHAP values to unveil the black box of ML. While previous risk scoring models identified several risk factors, including cancer site, cancer stage, time from symptom onset to hospitalization, appearance loss, BMI, skeletal muscle index, and NLR[18], our research highlights the significance of dynamic changes in metabolic indicators such as TC, Glu, and Cr, which traditional models often overlook. In this study, we employed metabolic indicators such as TC, Hb, Glu, and Cr to assess the risk of developing cachexia.\u0026nbsp;Cancer cachexia is caused by the activation of energy rich compound mobilization processes\u0026nbsp;[3, 19], such as increased liver gluconeogenesis\u0026nbsp;[20]. Causing impaired glucose and lipid homeostasis[21], Cancer cachexia is also associated with insulin resistance, which enhances liver glycolysis\u0026nbsp;[3, 22, 23]\u0026nbsp;and lipid mobilization in white adipose tissue[24]. Insulin resistance in cancer patients is characterized by decreased insulin sensitivity or impaired glucose tolerance[25], and is suspected to increase during the progression of cachexia[26, 27]. In addition, elevated levels of glucagon[28]\u0026nbsp;promote liver gluconeogenesis, thereby increasing blood sugar levels in cancer cachexia[28, 29]. However, others have also pointed out that there is no change in glucose levels in cancer cachexia[22, 30], which may reflect the complex metabolic dynamics of cancer cachexia[31]. Here, we investigated fasting blood glucose levels, whose damage is the second sign of insulin resistance (unrelated to impaired glucose tolerance)[32]. Our research findings support the view that insulin resistance increases in cancer cachexia. Patients with significantly elevated plasma fasting blood glucose levels also have an increased risk of developing cachexia. However, our research findings suggest that an increase in Glu is associated with the occurrence of cachexia, rather than with cancer staging. Previous studies have also shown that high fasting blood glucose levels during cancer diagnosis are associated with poor prognosis in non-small cell lung cancer (NSCLC) patients[33], supporting the view that elevated fasting blood glucose levels are not only an indicator of metabolic disorders or insulin resistance, but also a marker of poor prognosis.\u0026nbsp;\u003c/p\u003e\n\u003cp\u003eThe increase in lipolysis in CC patients may be mediated by norepinephrine signals involved in the sympathetic nervous system[34, 35]. In addition, systemic inflammation associated with cachexia and/or activation of brown adipose tissue requires a high energy demand[34, 36], and brown adipose tissue can effectively remove energy rich compounds from normal physiological circulation[37]. Adipose tissue is an important regulatory factor for body composition and plays a crucial role in energy balance. White adipose tissue (WAT) and brown adipose tissue (BAT) regulate energy storage and body temperature, respectively. BAT maintains body temperature by generating heat, including increasing the expression of uncoupling protein 1 (UCP1) in adipose tissue and regulating glucose and lipid metabolism[38]. In addition, the thermogenic effect of BAT is believed to enhance resting energy expenditure and lipid mobilization[31]. The browning effect observed in metabolic disorders (such as obesity, diabetes and cancer) is proved to be harmful in CC[2]. Patients with cachexia typically exhibit elevated levels of circulating free fatty acids and glycerol, due to the extensive lipolysis of WAT by activating lipases such as hormone sensitive lipase and triglyceride lipase\u0026nbsp;[39]. Therefore, the interactions between these metabolic processes may manifest as changes in plasma levels of energy rich compounds associated with cachexia[40]. The loss of skeletal muscle tissue is a key characteristic of cancer cachexia and the best studied aspect\u0026nbsp;[1, 11, 41]. Muscle is the source of amino acids, which may be released for energy production during the process of catabolism[42]. Muscle homeostasis is maintained through a balance between the synthesis and degradation of muscle proteins[3]. However, when there is excessive degradation of protein and/or a decrease in protein synthesis in skeletal muscle, this imbalance can lead to muscle atrophy and the occurrence of cachexia\u0026nbsp;[3, 41, 42].The SHAP summary chart indicates that changes in these metabolic characteristics are key predictive factors, indicating that as TC, Glu, and Cr increase, the risk of CC occurrence also increases. In addition, CC is associated with hyperlipidemia and decreased levels of branched chain amino acids[43]. It is currently unclear whether interventions based on these metabolic indicators can regulate CC correlation, and further research is needed.\u003c/p\u003e\n\u003cp\u003eIn our study, we used hemoglobin as a nutritional variable to assess the nutritional status and determine the risk of CC. While some reports suggest that low hemoglobin is a risk factor for CC and is associated with more complications and lower postoperative survival rates[44, 45], others indicate inconclusive results when using nutritional variables, including hemoglobin, to assess nutritional status, including CC[46]. Cancer patients with lower serum hemoglobin levels are more likely to die from this disease[47]. In CC patients, low hemoglobin levels are associated with mortality; therefore, hemoglobin can be used as a prognostic indicator. Our findings, however, indicate a significant association between serum hemoglobin levels and the risk of CC, with lower Hb levels indicating a higher risk of developing cachexia. Additionally, cancer-related indicators, such as cancer staging, were employed to assess the risk of CC in stage IV cancer patients. Given that cancer is the primary cause of cachexia, studying cancer-related factors for assessing cachexia risk in cancer patients holds significance. Our study identified a substantial correlation between cancer staging and CC, aligning with other studies[48]. Advanced-stage cancer patients often exhibit abnormal tumor biomarkers, further emphasizing the importance of detecting these biomarkers in routine cancer management. We questioned whether abnormal tumor biomarkers could serve as predictive indicators of cachexia, and our research indicated their potential relevance. Nevertheless, our study found that the histological type and differentiation of tumors had no significant impact on CC. This observation suggests that histological type may not be related to the patient's metabolic status and, consequently, may not significantly influence WL and cachexia. In conclusion, factors such as cancer stage and location, along with tumor-related indicators, offer valuable insights into assessing the risk of CC in stage IV CRC\u0026nbsp;patients.\u003c/p\u003e\n\u003cp\u003eOne of the primary clinical features of CC is systemic inflammation, a key driver in its development and affects various tissues such as skeletal muscle, fat, brain, and liver[3]. Cancer-related pro-inflammatory cytokines, such as interleukin (IL)-6), IL-1, and tumor necrosis factor-α, contribute to inhibiting albumin production, a factor associated with the occurrence of CC[49]. Notably, NLR has recently been recognized as a new inflammatory evaluation indicator that can be easily obtained from blood routine. Moreover, NLR is a significant negative prognostic biomarker for cachexia\u0026nbsp;patients[50]. In this study, NLR emerged as an independent risk factor for cancer cachexia, emphasizing the convenience of evaluating inflammation in clinical practice, especially in assessing the risk of cachexia. This suggests that anti-inflammatory interventions could potentially regulate systemic inflammation, aiding in muscle protection during treatment. Despite recommendations for nutrition and physical activity to maintain muscle mass, their efficacy might be limited in individuals with increased systemic inflammation[51, 52].\u003c/p\u003e\n\u003cp\u003eOur research has some limitations. First, due to the retrospective design and data collection from a single institution, there may be selection bias. Second, with evolving treatment strategies over the study period (2018–2022), our research may not entirely represent current medical practices. Third, the study's limited sample size from a single center in China necessitates additional validation in diverse centers to confirm the reliability of the ML model, and its applicability to patients outside China remains uncertain. To address these limitations, prospective trials on a larger and more diverse population with stricter inclusion criteria are essential. Finally, while nutritional management and the use of supplements or therapeutic diets are critical in evaluating the predictive utility of nutritional and inflammatory measures, this study lacks this information. The correlation between nutritional and inflammatory status and the prognosis of stage IV CRC warrants further investigation and validation in prospective cohort studies.\u003c/p\u003e"},{"header":"Conclusion","content":"\u003cp\u003eThe ML model, utilizing easily accessible clinical data, effectively identifies stage IV CRC patients with CC. The prediction of ML model is explained through SHAP method, making them applicable in clinical settings. This aids clinicians in better screening for CC, adopting targeted interventions, and improving the understanding of patient prognosis. In daily clinical practice, clinical indicators should be considered to evaluate the host status of stage IV cancer patients and strengthen nutritional management.\u003c/p\u003e"},{"header":"Declarations","content":"\u003cp\u003eAcknowledgments\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eJing Wang:\u003c/strong\u003e Conceptualization, Methodology, Software, Writing \u0026ndash; original draft. \u003cstrong\u003eYaoxian Xiang:\u003c/strong\u003e Data curation. \u0026nbsp;\u003cstrong\u003eKangjie Wang:\u003c/strong\u003eSoftware, Writing \u0026ndash; original draft. \u003cstrong\u003eBaojuan Han:\u003c/strong\u003e Visualization. \u003cstrong\u003eLei Wu:\u003c/strong\u003e Investigation. \u003cstrong\u003eDong Yan:\u003c/strong\u003e Supervision. \u003cstrong\u003eJing wang and Baojuan Han:\u003c/strong\u003e Validation, Writing \u0026ndash; review \u0026amp; editing. \u003cstrong\u003eLi Wang:\u003c/strong\u003e Writing \u0026ndash; review \u0026amp; editing.\u003c/p\u003e\n\u003cp\u003eFunding:\u0026nbsp;This work was financially supported by the capital health research\u0026nbsp;and development of special (2022-2-7083); R\u0026amp;D Program of Beijing Municipal Education Commission (KM202010025005); Beijing Municipal Natural Science Foundation7222100 and 7232085;\u0026nbsp;National Natural Science Foundation (No. 82203522).\u003c/p\u003e\n\u003cp\u003eCompeting interest\u003c/p\u003e\n\u003cp\u003eThe authors declare that they have no competing interests.\u003c/p\u003e\n\u003cp\u003eAvailability of data and materials\u003c/p\u003e\n\u003cp\u003eAll data generated during this study are included in this published article. Further inquiry data can be obtained by the corresponding author.\u003c/p\u003e\n\u003cp\u003eEthics Statement\u003c/p\u003e\n\u003cp\u003eThis study was approved by the Beijing Luhe Hospital Ethic Committee for Clinical Investigation.\u003c/p\u003e"},{"header":"References","content":"\u003col\u003e\n \u003cli\u003eArgil\u0026eacute;s JM, Busquets S, Stemmler B, L\u0026oacute;pez-Soriano FJ: \u003cstrong\u003eCancer cachexia: understanding the molecular basis\u003c/strong\u003e. \u003cem\u003eNature reviews Cancer\u0026nbsp;\u003c/em\u003e2014, \u003cstrong\u003e14\u003c/strong\u003e(11):754-762.\u003c/li\u003e\n \u003cli\u003eArgil\u0026eacute;s JM, Stemmler B, L\u0026oacute;pez-Soriano FJ, Busquets S: \u003cstrong\u003eInter-tissue communication in cancer cachexia\u003c/strong\u003e. \u003cem\u003eNature reviews Endocrinology\u0026nbsp;\u003c/em\u003e2018, \u003cstrong\u003e15\u003c/strong\u003e(1):9-20.\u003c/li\u003e\n \u003cli\u003ePorporato PE: \u003cstrong\u003eUnderstanding cachexia as a cancer metabolism syndrome\u003c/strong\u003e. \u003cem\u003eOncogenesis\u0026nbsp;\u003c/em\u003e2016, \u003cstrong\u003e5\u003c/strong\u003e(2):e200.\u003c/li\u003e\n \u003cli\u003eTsoli M, Robertson G: \u003cstrong\u003eCancer cachexia: malignant inflammation, tumorkines, and metabolic mayhem\u003c/strong\u003e. \u003cem\u003eTrends Endocrinol Metab\u0026nbsp;\u003c/em\u003e2013, \u003cstrong\u003e24\u003c/strong\u003e(4):174-183.\u003c/li\u003e\n \u003cli\u003eFearon KC: \u003cstrong\u003eCancer cachexia and fat-muscle physiology\u003c/strong\u003e. \u003cem\u003eN Engl J Med\u0026nbsp;\u003c/em\u003e2011, \u003cstrong\u003e365\u003c/strong\u003e(6):565-567.\u003c/li\u003e\n \u003cli\u003eQuan-Jun Y, Yan H, Yong-Long H, Li-Li W, Jie L, Jin-Lu H, Jin L, Peng-Guo C, Run G, Cheng G: \u003cstrong\u003eSelumetinib Attenuates Skeletal Muscle Wasting in Murine Cachexia Model through ERK Inhibition and AKT Activation\u003c/strong\u003e. \u003cem\u003eMol Cancer Ther\u0026nbsp;\u003c/em\u003e2017, \u003cstrong\u003e16\u003c/strong\u003e(2):334-343.\u003c/li\u003e\n \u003cli\u003eTemel JS, Abernethy AP, Currow DC, Friend J, Duus EM, Yan Y, Fearon KC: \u003cstrong\u003eAnamorelin in patients with non-small-cell lung cancer and cachexia (ROMANA 1 and ROMANA 2): results from two randomised, double-blind, phase 3 trials\u003c/strong\u003e. \u003cem\u003eLancet Oncol\u0026nbsp;\u003c/em\u003e2016, \u003cstrong\u003e17\u003c/strong\u003e(4):519-531.\u003c/li\u003e\n \u003cli\u003eGarcia JM, Boccia RV, Graham CD, Yan Y, Duus EM, Allen S, Friend J: \u003cstrong\u003eAnamorelin for patients with cancer cachexia: an integrated analysis of two phase 2, randomised, placebo-controlled, double-blind trials\u003c/strong\u003e. \u003cem\u003eLancet Oncol\u0026nbsp;\u003c/em\u003e2015, \u003cstrong\u003e16\u003c/strong\u003e(1):108-116.\u003c/li\u003e\n \u003cli\u003eFearon K, Strasser F, Anker SD, Bosaeus I, Bruera E, Fainsinger RL, Jatoi A, Loprinzi C, MacDonald N, Mantovani G\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003eDefinition and classification of cancer cachexia: an international consensus\u003c/strong\u003e. \u003cem\u003eThe Lancet Oncology\u0026nbsp;\u003c/em\u003e2011, \u003cstrong\u003e12\u003c/strong\u003e(5):489-495.\u003c/li\u003e\n \u003cli\u003eFearon KC, Baracos VE: \u003cstrong\u003eCachexia in pancreatic cancer: new treatment options and measures of success\u003c/strong\u003e. \u003cem\u003eHPB : the official journal of the International Hepato Pancreato Biliary Association\u0026nbsp;\u003c/em\u003e2010, \u003cstrong\u003e12\u003c/strong\u003e(5):323-324.\u003c/li\u003e\n \u003cli\u003eFearon KC: \u003cstrong\u003eCancer cachexia: developing multimodal therapy for a multidimensional problem\u003c/strong\u003e. \u003cem\u003eEuropean journal of cancer (Oxford, England : 1990)\u0026nbsp;\u003c/em\u003e2008, \u003cstrong\u003e44\u003c/strong\u003e(8):1124-1132.\u003c/li\u003e\n \u003cli\u003ePrado CM, Lieffers JR, McCargar LJ, Reiman T, Sawyer MB, Martin L, Baracos VE: \u003cstrong\u003ePrevalence and clinical implications of sarcopenic obesity in patients with solid tumours of the respiratory and gastrointestinal tracts: a population-based study\u003c/strong\u003e. \u003cem\u003eThe Lancet Oncology\u0026nbsp;\u003c/em\u003e2008, \u003cstrong\u003e9\u003c/strong\u003e(7):629-635.\u003c/li\u003e\n \u003cli\u003eFearon K, Arends J, Baracos V: \u003cstrong\u003eUnderstanding the mechanisms and treatment options in cancer cachexia\u003c/strong\u003e. \u003cem\u003eNature reviews Clinical oncology\u0026nbsp;\u003c/em\u003e2013, \u003cstrong\u003e10\u003c/strong\u003e(2):90-99.\u003c/li\u003e\n \u003cli\u003eProkopchuk O, Hermann CD, Schoeps B, Nitsche U, Prokopchuk OL, Knolle P, Friess H, Martignoni ME, Kr\u0026uuml;ger A: \u003cstrong\u003eA novel tissue inhibitor of metalloproteinases-1/liver/cachexia score predicts prognosis of gastrointestinal cancer patients\u003c/strong\u003e. \u003cem\u003eJournal of cachexia, sarcopenia and muscle\u0026nbsp;\u003c/em\u003e2021, \u003cstrong\u003e12\u003c/strong\u003e(2):378-392.\u003c/li\u003e\n \u003cli\u003eJafri SH, Previgliano C, Khandelwal K, Shi R: \u003cstrong\u003eCachexia Index in Advanced Non-Small-Cell Lung Cancer Patients\u003c/strong\u003e. \u003cem\u003eClinical Medicine Insights Oncology\u0026nbsp;\u003c/em\u003e2015, \u003cstrong\u003e9\u003c/strong\u003e:87-93.\u003c/li\u003e\n \u003cli\u003eYang QJ, Zhao JR, Hao J, Li B, Huo Y, Han YL, Wan LL, Li J, Huang J, Lu J\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003eSerum and urine metabolomics study reveals a distinct diagnostic model for cancer cachexia\u003c/strong\u003e. \u003cem\u003eJournal of cachexia, sarcopenia and muscle\u0026nbsp;\u003c/em\u003e2018, \u003cstrong\u003e9\u003c/strong\u003e(1):71-85.\u003c/li\u003e\n \u003cli\u003eRajkomar A, Dean J, Kohane I: \u003cstrong\u003eMachine Learning in Medicine\u003c/strong\u003e. \u003cem\u003eThe New England journal of medicine\u0026nbsp;\u003c/em\u003e2019, \u003cstrong\u003e380\u003c/strong\u003e(14):1347-1358.\u003c/li\u003e\n \u003cli\u003eTan S, Xu J, Wang J, Zhang Z, Li S, Yan M, Tang M, Liu H, Zhuang Q, Xi Q\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003eDevelopment and validation of a cancer cachexia risk score for digestive tract cancer patients before abdominal surgery\u003c/strong\u003e. \u003cem\u003eJournal of cachexia, sarcopenia and muscle\u0026nbsp;\u003c/em\u003e2023, \u003cstrong\u003e14\u003c/strong\u003e(2):891-902.\u003c/li\u003e\n \u003cli\u003ePetruzzelli M, Wagner EF: \u003cstrong\u003eMechanisms of metabolic dysfunction in cancer-associated cachexia\u003c/strong\u003e. \u003cem\u003eGenes \u0026amp; development\u0026nbsp;\u003c/em\u003e2016, \u003cstrong\u003e30\u003c/strong\u003e(5):489-501.\u003c/li\u003e\n \u003cli\u003eNoguchi Y, Vydelingum NA, Brennan MF: \u003cstrong\u003eThe reversal of increased gluconeogenesis in the tumor-bearing rat by tumor removal and food intake\u003c/strong\u003e. \u003cem\u003eSurgery\u0026nbsp;\u003c/em\u003e1989, \u003cstrong\u003e106\u003c/strong\u003e(2):423-430; discussion 430-421.\u003c/li\u003e\n \u003cli\u003ePetruzzelli M, Ferrer M, Schuijs MJ, Kleeman SO, Mourikis N, Hall Z, Perera D, Raghunathan S, Vacca M, Gaude E\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003eEarly Neutrophilia Marked by Aerobic Glycolysis Sustains Host Metabolism and Delays Cancer Cachexia\u003c/strong\u003e. \u003cem\u003eCancers\u0026nbsp;\u003c/em\u003e2022, \u003cstrong\u003e14\u003c/strong\u003e(4).\u003c/li\u003e\n \u003cli\u003eDev R, Bruera E, Dalal S: \u003cstrong\u003eInsulin resistance and body composition in cancer patients\u003c/strong\u003e. \u003cem\u003eAnnals of oncology : official journal of the European Society for Medical Oncology\u0026nbsp;\u003c/em\u003e2018, \u003cstrong\u003e29\u003c/strong\u003e(suppl_2):ii18-ii26.\u003c/li\u003e\n \u003cli\u003eRui L: \u003cstrong\u003eEnergy metabolism in the liver\u003c/strong\u003e. \u003cem\u003eComprehensive Physiology\u0026nbsp;\u003c/em\u003e2014, \u003cstrong\u003e4\u003c/strong\u003e(1):177-197.\u003c/li\u003e\n \u003cli\u003eJaworski K, Sarkadi-Nagy E, Duncan RE, Ahmadian M, Sul HS: \u003cstrong\u003eRegulation of triglyceride metabolism. IV. Hormonal regulation of lipolysis in adipose tissue\u003c/strong\u003e. \u003cem\u003eAmerican journal of physiology Gastrointestinal and liver physiology\u0026nbsp;\u003c/em\u003e2007, \u003cstrong\u003e293\u003c/strong\u003e(1):G1-4.\u003c/li\u003e\n \u003cli\u003eHonors MA, Kinzig KP: \u003cstrong\u003eThe role of insulin resistance in the development of muscle wasting during cancer cachexia\u003c/strong\u003e. \u003cem\u003eJournal of cachexia, sarcopenia and muscle\u0026nbsp;\u003c/em\u003e2012, \u003cstrong\u003e3\u003c/strong\u003e(1):5-11.\u003c/li\u003e\n \u003cli\u003eTisdale MJ: \u003cstrong\u003eWasting in cancer\u003c/strong\u003e. \u003cem\u003eThe Journal of nutrition\u0026nbsp;\u003c/em\u003e1999, \u003cstrong\u003e129\u003c/strong\u003e(1S Suppl):243s-246s.\u003c/li\u003e\n \u003cli\u003eJasani B, Donaldson LJ, Ratcliffe JG, Sokhi GS: \u003cstrong\u003eMechanism of impaired glucose tolerance in patients with neoplasia\u003c/strong\u003e. \u003cem\u003eBritish journal of cancer\u0026nbsp;\u003c/em\u003e1978, \u003cstrong\u003e38\u003c/strong\u003e(2):287-292.\u003c/li\u003e\n \u003cli\u003eBartlett DL, Charland SL, Torosian MH: \u003cstrong\u003eReversal of tumor-associated hyperglucagonemia as treatment for cancer cachexia\u003c/strong\u003e. \u003cem\u003eSurgery\u0026nbsp;\u003c/em\u003e1995, \u003cstrong\u003e118\u003c/strong\u003e(1):87-97.\u003c/li\u003e\n \u003cli\u003eTayek JA: \u003cstrong\u003eA review of cancer cachexia and abnormal glucose metabolism in humans with cancer\u003c/strong\u003e. \u003cem\u003eJournal of the American College of Nutrition\u0026nbsp;\u003c/em\u003e1992, \u003cstrong\u003e11\u003c/strong\u003e(4):445-456.\u003c/li\u003e\n \u003cli\u003eSchwarz S, Prokopchuk O, Esefeld K, Gr\u0026ouml;schel S, Bachmann J, Lorenzen S, Friess H, Halle M, Martignoni ME: \u003cstrong\u003eThe clinical picture of cachexia: a mosaic of different parameters (experience of 503 patients)\u003c/strong\u003e. \u003cem\u003eBMC cancer\u0026nbsp;\u003c/em\u003e2017, \u003cstrong\u003e17\u003c/strong\u003e(1):130.\u003c/li\u003e\n \u003cli\u003eFonseca G, Farkas J, Dora E, von Haehling S, Lainscak M: \u003cstrong\u003eCancer Cachexia and Related Metabolic Dysfunction\u003c/strong\u003e. \u003cem\u003eInternational journal of molecular sciences\u0026nbsp;\u003c/em\u003e2020, \u003cstrong\u003e21\u003c/strong\u003e(7).\u003c/li\u003e\n \u003cli\u003ePetersen JL, McGuire DK: \u003cstrong\u003eImpaired glucose tolerance and impaired fasting glucose--a review of diagnosis, clinical implications and management\u003c/strong\u003e. \u003cem\u003eDiabetes \u0026amp; vascular disease research\u0026nbsp;\u003c/em\u003e2005, \u003cstrong\u003e2\u003c/strong\u003e(1):9-15.\u003c/li\u003e\n \u003cli\u003eLuo J, Chen YJ, Chang LJ: \u003cstrong\u003eFasting blood glucose level and prognosis in non-small cell lung cancer (NSCLC) patients\u003c/strong\u003e. \u003cem\u003eLung cancer (Amsterdam, Netherlands)\u0026nbsp;\u003c/em\u003e2012, \u003cstrong\u003e76\u003c/strong\u003e(2):242-247.\u003c/li\u003e\n \u003cli\u003eTisdale MJ: \u003cstrong\u003eMechanisms of cancer cachexia\u003c/strong\u003e. \u003cem\u003ePhysiological reviews\u0026nbsp;\u003c/em\u003e2009, \u003cstrong\u003e89\u003c/strong\u003e(2):381-410.\u003c/li\u003e\n \u003cli\u003eBartelt A, Heeren J: \u003cstrong\u003eAdipose tissue browning and metabolic health\u003c/strong\u003e. \u003cem\u003eNature reviews Endocrinology\u0026nbsp;\u003c/em\u003e2014, \u003cstrong\u003e10\u003c/strong\u003e(1):24-36.\u003c/li\u003e\n \u003cli\u003eBeijer E, Schoenmakers J, Vijgen G, Kessels F, Dingemans AM, Schrauwen P, Wouters M, van Marken Lichtenbelt W, Teule J, Brans B: \u003cstrong\u003eA role of active brown adipose tissue in cancer cachexia?\u003c/strong\u003e \u003cem\u003eOncology reviews\u0026nbsp;\u003c/em\u003e2012, \u003cstrong\u003e6\u003c/strong\u003e(1):e11.\u003c/li\u003e\n \u003cli\u003eCohen P, Spiegelman BM: \u003cstrong\u003eBrown and Beige Fat: Molecular Parts of a Thermogenic Machine\u003c/strong\u003e. \u003cem\u003eDiabetes\u0026nbsp;\u003c/em\u003e2015, \u003cstrong\u003e64\u003c/strong\u003e(7):2346-2351.\u003c/li\u003e\n \u003cli\u003eVirtanen KA, Lidell ME, Orava J, Heglind M, Westergren R, Niemi T, Taittonen M, Laine J, Savisto NJ, Enerb\u0026auml;ck S\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003eFunctional brown adipose tissue in healthy adults\u003c/strong\u003e. \u003cem\u003eThe New England journal of medicine\u0026nbsp;\u003c/em\u003e2009, \u003cstrong\u003e360\u003c/strong\u003e(15):1518-1525.\u003c/li\u003e\n \u003cli\u003eHan J, Meng Q, Shen L, Wu G: \u003cstrong\u003eInterleukin-6 induces fat loss in cancer cachexia by promoting white adipose tissue lipolysis and browning\u003c/strong\u003e. \u003cem\u003eLipids in health and disease\u0026nbsp;\u003c/em\u003e2018, \u003cstrong\u003e17\u003c/strong\u003e(1):14.\u003c/li\u003e\n \u003cli\u003eZwickl H, Zwickl-Traxler E, Pecherstorfer M: \u003cstrong\u003eIs Neuronal Histamine Signaling Involved in Cancer Cachexia? Implications and Perspectives\u003c/strong\u003e. \u003cem\u003eFrontiers in oncology\u0026nbsp;\u003c/em\u003e2019, \u003cstrong\u003e9\u003c/strong\u003e:1409.\u003c/li\u003e\n \u003cli\u003eSchmidt SF, Rohm M, Herzig S, Berriel Diaz M: \u003cstrong\u003eCancer Cachexia: More Than Skeletal Muscle Wasting\u003c/strong\u003e. \u003cem\u003eTrends in cancer\u0026nbsp;\u003c/em\u003e2018, \u003cstrong\u003e4\u003c/strong\u003e(12):849-860.\u003c/li\u003e\n \u003cli\u003eBonaldo P, Sandri M: \u003cstrong\u003eCellular and molecular mechanisms of muscle atrophy\u003c/strong\u003e. \u003cem\u003eDisease models \u0026amp; mechanisms\u0026nbsp;\u003c/em\u003e2013, \u003cstrong\u003e6\u003c/strong\u003e(1):25-39.\u003c/li\u003e\n \u003cli\u003eDi Gangi IM, Mazza T, Fontana A, Copetti M, Fusilli C, Ippolito A, Mattivi F, Latiano A, Andriulli A, Vrhovsek U\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003eMetabolomic profile in pancreatic cancer patients: a consensus-based approach to identify highly discriminating metabolites\u003c/strong\u003e. \u003cem\u003eOncotarget\u0026nbsp;\u003c/em\u003e2016, \u003cstrong\u003e7\u003c/strong\u003e(5):5815-5829.\u003c/li\u003e\n \u003cli\u003eWallengren O, Lundholm K, Bosaeus I: \u003cstrong\u003eDiagnostic criteria of cancer cachexia: relation to quality of life, exercise capacity and survival in unselected palliative care patients\u003c/strong\u003e. \u003cem\u003eSupportive care in cancer : official journal of the Multinational Association of Supportive Care in Cancer\u0026nbsp;\u003c/em\u003e2013, \u003cstrong\u003e21\u003c/strong\u003e(6):1569-1577.\u003c/li\u003e\n \u003cli\u003eArends J, Strasser F, Gonella S, Solheim TS, Madeddu C, Ravasco P, Buonaccorso L, de van der Schueren MAE, Baldwin C, Chasen M\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003eCancer cachexia in adult patients: ESMO Clinical Practice Guidelines(☆)\u003c/strong\u003e. \u003cem\u003eESMO open\u0026nbsp;\u003c/em\u003e2021, \u003cstrong\u003e6\u003c/strong\u003e(3):100092.\u003c/li\u003e\n \u003cli\u003eMartin L, Birdsell L, Macdonald N, Reiman T, Clandinin MT, McCargar LJ, Murphy R, Ghosh S, Sawyer MB, Baracos VE: \u003cstrong\u003eCancer cachexia in the age of obesity: skeletal muscle depletion is a powerful prognostic factor, independent of body mass index\u003c/strong\u003e. \u003cem\u003eJournal of clinical oncology : official journal of the American Society of Clinical Oncology\u0026nbsp;\u003c/em\u003e2013, \u003cstrong\u003e31\u003c/strong\u003e(12):1539-1547.\u003c/li\u003e\n \u003cli\u003eZhang XW, Zhang Q, Song MM, Zhang KP, Zhang X, Ruan GT, Yang M, Ge YZ, Tang M, Li XR\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003eThe prognostic effect of hemoglobin on patients with cancer cachexia: a multicenter retrospective cohort study\u003c/strong\u003e. \u003cem\u003eSupportive care in cancer : official journal of the Multinational Association of Supportive Care in Cancer\u0026nbsp;\u003c/em\u003e2022, \u003cstrong\u003e30\u003c/strong\u003e(1):875-885.\u003c/li\u003e\n \u003cli\u003ePoisson J, Martinez-Tapia C, Heitz D, Geiss R, Albrand G, Falandry C, Gisselbrecht M, Couderc AL, Boulahssass R, Liuu E\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003ePrevalence and prognostic impact of cachexia among older patients with cancer: a nationwide cross-sectional survey (NutriAgeCancer)\u003c/strong\u003e. \u003cem\u003eJournal of cachexia, sarcopenia and muscle\u0026nbsp;\u003c/em\u003e2021, \u003cstrong\u003e12\u003c/strong\u003e(6):1477-1488.\u003c/li\u003e\n \u003cli\u003eGupta D, Lis CG: \u003cstrong\u003ePretreatment serum albumin as a predictor of cancer survival: a systematic review of the epidemiological literature\u003c/strong\u003e. \u003cem\u003eNutrition journal\u0026nbsp;\u003c/em\u003e2010, \u003cstrong\u003e9\u003c/strong\u003e:69.\u003c/li\u003e\n \u003cli\u003eZhang Q, Song MM, Zhang X, Ding JS, Ruan GT, Zhang XW, Liu T, Yang M, Ge YZ, Tang M\u003cem\u003e\u0026nbsp;et al\u003c/em\u003e: \u003cstrong\u003eAssociation of systemic inflammation with survival in patients with cancer cachexia: results from a multicentre cohort study\u003c/strong\u003e. \u003cem\u003eJournal of cachexia, sarcopenia and muscle\u0026nbsp;\u003c/em\u003e2021, \u003cstrong\u003e12\u003c/strong\u003e(6):1466-1476.\u003c/li\u003e\n \u003cli\u003eGolder AM, Sin LKE, Alani F, Alasadi A, Dolan R, Mansouri D, Horgan PG, McMillan DC, Roxburgh CS: \u003cstrong\u003eThe relationship between the mode of presentation, CT-derived body composition, systemic inflammatory grade and survival in colon cancer\u003c/strong\u003e. \u003cem\u003eJournal of cachexia, sarcopenia and muscle\u0026nbsp;\u003c/em\u003e2022, \u003cstrong\u003e13\u003c/strong\u003e(6):2863-2874.\u003c/li\u003e\n \u003cli\u003eCollao N, Sanders O, Caminiti T, Messeiller L, De Lisio M: \u003cstrong\u003eResistance and endurance exercise training improves muscle mass and the inflammatory/fibrotic transcriptome in a rhabdomyosarcoma model\u003c/strong\u003e. \u003cem\u003eJournal of cachexia, sarcopenia and muscle\u0026nbsp;\u003c/em\u003e2023, \u003cstrong\u003e14\u003c/strong\u003e(2):781-793.\u003c/li\u003e\n\u003c/ol\u003e"}],"fulltextSource":"","fullText":"","funders":[],"hasAdminPriorityOnWorkflow":false,"hasManuscriptDocX":true,"hasOptedInToPreprint":true,"hasPassedJournalQc":"","hasAnyPriority":false,"hideJournal":true,"highlight":"","institution":"","isAcceptedByJournal":false,"isAuthorSuppliedPdf":false,"isDeskRejected":"","isHiddenFromSearch":false,"isInQc":false,"isInWorkflow":false,"isPdf":false,"isPdfUpToDate":true,"isWithdrawnOrRetracted":false,"journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true},"keywords":"Cancer cachexia, Colorectal cancer, Machine learning, Prediction. ","lastPublishedDoi":"10.21203/rs.3.rs-4275850/v1","lastPublishedDoiUrl":"https://doi.org/10.21203/rs.3.rs-4275850/v1","license":{"name":"CC BY 4.0","url":"https://creativecommons.org/licenses/by/4.0/"},"manuscriptAbstract":"\u003cp\u003e\u003cstrong\u003eBackground:\u003c/strong\u003e Cancer cachexia (CC) is a multifactorial syndrome affectingadvanced cancer patients, significantly impacting their survival and quality of life. This study utilized artificial intelligence machine learning (ML) methods to evaluate the risk of cachexia in stage IV colorectal cancer (CRC) patients through clinical data, establishing a cachexia risk prediction model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eMethods\u003c/strong\u003e: We conducted a retrospective collection of data from stage IV CRC patients consecutively visiting Luhe Hospital from January 2015 to December 2022. LASSO regression was employed on this cohort to identify diverse risk variables for CC. Four MLalgorithms, namely, logistic regression model, random forest (RF), Extreme Gradient Boosting, and k-nearest neighbor algorithm, were employed to construct the predictive model. Model performance was assessed using the area under the receiver operating characteristic curve (AUC), accuracy, sensitivity, and specificity. Additionally, Shapley additive prediction (SHAP) values were utilized to elucidate the prediction model.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eResults\u003c/strong\u003e: Out of 219 stage IV CRC patients, 119 cases (54.34%) developed malignant fluid. Among the four ML models, the RF model exhibited the most robust predictive performance, achieving the highest AUC (0.768, 95% confidence interval: 0.706–0.831). The RF model demonstrated accuracy, sensitivity, and specificity values of 0.744, 0.790, and 0.690, respectively. We employed SHAP correlation maps to explicate the influence of individual features on the output of RF prediction models.\u003c/p\u003e\n\u003cp\u003e\u003cstrong\u003eConclusions:\u003c/strong\u003e The RF model offers superior predictions, enhancing clinicians' ability to screen for CC, assess patient prognosis, and make informed decisions on targeted CC in stage IV cancer patients.\u003c/p\u003e","manuscriptTitle":"Applying machine learning to predict the risk of cancer cachexia in stage IV colorectal cancer patients","msid":"","msnumber":"","nonDraftVersions":[{"code":1,"date":"2024-04-25 18:58:54","doi":"10.21203/rs.3.rs-4275850/v1","editorialEvents":[{"type":"communityComments","content":0}],"status":"published","journal":{"display":true,"email":"[email protected]","identity":"researchsquare","isNatureJournal":false,"hasQc":true,"allowDirectSubmit":true,"externalIdentity":"","sideBox":"","snPcode":"","submissionUrl":"/submission","title":"Research Square","twitterHandle":"researchsquare","acdcEnabled":true,"dfaEnabled":false,"editorialSystem":"","reportingPortfolio":"","inReviewEnabled":false,"inReviewRevisionsEnabled":true}}],"origin":"","ownerIdentity":"989772e8-cb78-4c2d-8437-75a681aa6add","owner":[],"postedDate":"April 25th, 2024","published":true,"recentEditorialEvents":[],"rejectedJournal":[],"revision":"","amendment":"","status":"posted","subjectAreas":[],"tags":[],"updatedAt":"2024-06-22T02:53:16+00:00","versionOfRecord":[],"versionCreatedAt":"2024-04-25 18:58:54","video":"","vorDoi":"","vorDoiUrl":"","workflowStages":[]},"version":"v1","identity":"rs-4275850","journalConfig":"researchsquare"},"__N_SSP":true},"page":"/article/[identity]/[[...version]]","query":{"redirect":"/article/rs-4275850","identity":"rs-4275850","version":["v1"]},"buildId":"8U1c8b4HqxoKbykW_rLl7","isFallback":false,"isExperimentalCompile":false,"dynamicIds":[84888],"gssp":true,"scriptLoader":[]}