A Prediction Model for Colorectal Adenoma and Colorectal Cancer Based on Routine Test
10.3971/j.issn.1000-8578.2024.23.1169
- VernacularTitle:基于常规检验指标构建区分结直肠腺瘤与结直肠癌的预测模型
- Author:
Junsheng LIN
1
;
Ziling YING
;
Zhengyuan HUANG
;
Xianjin ZHU
;
Yingping CAO
;
Pingxia LU
Author Information
1. Department of Laboratory Medicine, Fujian Medical University Union Hospital, Fuzhou 350001, China
- Publication Type:Research Article
- Keywords:
Colorectal adenoma;
Colorectal cancer;
Prediction model;
Nomogram
- From:
Cancer Research on Prevention and Treatment
2024;51(5):353-360
- CountryChina
- Language:Chinese
-
Abstract:
Objective To analyze the routine test parameter levels of patients with colorectal adenoma and colorectal cancer, and develop a prediction model. Methods A total of 580 patients diagnosed with colorectal adenoma (117 patients) and colorectal cancer (463 patients) were included in the retrospective study. The patients were randomly divided into two groups according to a 7:3 ratio: a training set with 406 cases and a validation set with 174 cases. Logistic regression analysis was used to establish a prediction model, and a nomogram was drawn. The model′s discrimination, calibration, and clinical applicability were evaluated using receiver operating characteristic curve (ROC), calibration plot, and decision curve analysis (DCA). Results Univariate logistic regression analysis identified 13 potential predictors: age, fecal occult blood test (FOBT), fibrinogen (FIB), thrombin time (TT), albumin (ALB), white blood cell value (WBC), neutrophil count (NEUT#), hematocrit value (HCT), mean corpuscular hemoglobin (MCH), red cell distribution width (RDW), platelet count (PLT), mean platelet volume (MPV), and activated partial thromboplastin time (APTT). Multivariate logistic regression analysis showed MPV, FIB, ALB, FOBT, TT, and HCT were risk factors for colorectal cancer in patients with colorectal adenoma (P<0.05). A nomogram was constructed based on these predictors to build a prediction model. The AUC of the ROC curve was 0.915 for colorectal cancer in the training set and 0.836 in the validation set. Calibration plots demonstrated high prediction accuracy and good model calibration. DCA results indicated the prediction model provided greater net benefit compared with the extreme models at threshold probabilities of approximately 55%-95%. Conclusion The developed prediction model exhibits satisfactory discrimination, calibration, and clinical applicability. The model can serve as an auxiliary tool in distinguishing between colorectal adenoma and colorectal cancer in patients.
