Scholarlink

<p>Colorectal cancer and its surgical treatment often worsen patients’ nutritional status, increasing cachexia risk. Early postoperative cachexia prediction is crucial for timely intervention. This dual-center study included 1,125 colorectal cancer patients. Candidate predictors were screened by univariate logistic regression, with significant variables entered into multivariate analysis to develop the prediction model. Model performance was assessed using ROC curves, calibration curves and decision curve analysis in all three datasets. This study analyzed 863 colorectal cancer patients (Cohort 1), randomly divided 7:3 into training (<i>n</i> = 604) and validation sets (<i>n</i> = 259), with an additional 262 patients (Cohort 2) for external validation. Through rigorous statistical analysis, we identified seven key predictors for the final cachexia prediction model: preoperative BMI (OR = 1.16), chemotherapy (yes, OR = 3.59), anastomotic leakage (yes, OR = 10.00), hemoglobin (OR = 0.97), triglycerides (OR = 0.57), transferrin (OR = 0.97), and NLR (OR = 1.03). The model demonstrated outstanding predictive performance across all cohorts, with training set AUC = 0.939, validation set AUC = 0.917, and external validation set AUC = 0.910, along with excellent calibration and clinically meaningful decision curve analysis results. The developed model effectively predicts postoperative cachexia in colorectal cancer patients undergoing radical surgery, enabling early identification of high-risk individuals for timely nutritional intervention by surgical teams.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Dual-center development and validation of an immunoinflammatory-based preoperative model for predicting postoperative cachexia in colorectal cancer

  • 期刊论文

  • 发表日期：

    2026年3月6日

查看全文
  • Jiakui Liang,
  • Chenkai Zhang,
  • Fanyu Zhao,
  • Daorong Wang

摘要

Colorectal cancer and its surgical treatment often worsen patients’ nutritional status, increasing cachexia risk. Early postoperative cachexia prediction is crucial for timely intervention. This dual-center study included 1,125 colorectal cancer patients. Candidate predictors were screened by univariate logistic regression, with significant variables entered into multivariate analysis to develop the prediction model. Model performance was assessed using ROC curves, calibration curves and decision curve analysis in all three datasets. This study analyzed 863 colorectal cancer patients (Cohort 1), randomly divided 7:3 into training (n = 604) and validation sets (n = 259), with an additional 262 patients (Cohort 2) for external validation. Through rigorous statistical analysis, we identified seven key predictors for the final cachexia prediction model: preoperative BMI (OR = 1.16), chemotherapy (yes, OR = 3.59), anastomotic leakage (yes, OR = 10.00), hemoglobin (OR = 0.97), triglycerides (OR = 0.57), transferrin (OR = 0.97), and NLR (OR = 1.03). The model demonstrated outstanding predictive performance across all cohorts, with training set AUC = 0.939, validation set AUC = 0.917, and external validation set AUC = 0.910, along with excellent calibration and clinically meaningful decision curve analysis results. The developed model effectively predicts postoperative cachexia in colorectal cancer patients undergoing radical surgery, enabling early identification of high-risk individuals for timely nutritional intervention by surgical teams.