Influencing factors and prediction model construction of perineural or lymphovascular invasion in patients with color-ectal cancer

Li Manman; Yang Tao; Chen Jiang; Zeng Lian; Luo Zicheng; Xiao Tianbao

doi:10.3969/j.issn.1671-6450.2025.04.018

Abstract：Objective To investigate the influencing factors of peripheral perineural or lymphovascular invasion(PNI/LVI)in patients with colorectal cancer(CRC)and construct a corresponding prediction model.Methods Three hun-dred patients admitted to the Department of Anus and Intestines of the First Affiliated Hospital of Guizhou University of Traditional Chinese Medicine with confirmed diagnosis of CRC and undergoing radical surgery from January 2020 to January 2024 were selected for the study,and were randomly split into the training set(n=210)and the validation set(n=90)ac-cording to the ratio of 7∶3.Patients in the training set were included in the positive(n=80)and negative(n=130)groups according to the presence of PNI/LVI.The clinical data of the training set,validation set,and positive and negative groups were compared.Multifactorial Logistic regression was used to analyze the influencing factors of PNI/LVI in patients with CRC and to construct a column-line diagram prediction model.The validation set data were used to perform an internal test of the column-line diagram model,and the efficacy of the model was assessed using the subject's work characteristics(ROC)curves,calibration curves,and decision curves.Results There was no statistically significant difference in the comparison of clinical data between the training set and validation set(P＞0.05).The proportions of T stage T3-T4,N stage N1-N2,ulcerative type,low differentiation degree,lymph node metastasis,and cancer nodules in the positive group were higher than those in the negative group(x2/P=17.037/＜0.001,33.094/＜0.001,-/0.006,18.674/＜0.001,51.822/＜0.001,47.223/＜0.001).Multifactorial Logistic regression analysis showed that T stage T3 to T4,lymph node metastasis,and cancerous nod-ules were independent risk factors for PNI/LVI in CRC patients[OR(95％CI)=7.158(1.474-34.756),8.464(2.657-26.963),4.617(2.042-10.440)].A column-line graph model was constructed,and the area under the ROC curve(AUC)of the model was 0.832(95％CI=0.800-0.885)in the training set and 0.653(95％CI=0.535-0.771)in the validation set.The calibration curve of the model suggests that the model has good agreement between the ideal and actual values in the training set(P=0.748)and validation set(P=0.132).The decision curve suggests that the model leads to greater clinical benefits for patients in both the training and validation sets.Conclusion Staging T3-T4,lymph node metastasis,and cancer nodules are inde-pendent risk factors for PNI/LVI infiltration in CRC patients,and the column-line diagram model developed based on the above risk factors has good predictive value and practicality.