A model based on endorectal ultrasonography predicts lateral lymph node metastasis in low and middle rectal cancer

To investigate the risk factors for lymph node (LN) metastasis in low and middle rectal tumors, construct a predictive model and test the model's diagnostic efficacy.

The clinical and pathological data of 172 patients with rectal cancer confirmed by surgery were retrospectively evaluated, among whom 61 patients were finally included in this study. Patients were divided into positive groups and negative groups based on LN metastasis, and risk factors that might affect LN metastasis were analyzed. Finally, a risk predictive model was constructed based on the weights of each risk factor.

Compared with pathology, the efficacy of diagnosing LN metastasis only according to conventional endorectal ultrasonography (ERUS) features of LN was not high, with sensitivity 67%, specificity 86%, positive predictive value 76%, negative predictive value 80%, and accuracy 79%. Univariate analysis showed that circumferential angle of the tumor, ultrasonic T- stage (UT stage), conventional ultrasound features diagnosis of LN metastasis, strain ratio (SR) of tumor were risk factors for LN metastasis, while vascular resistance index of rectal tumor was protective factor. Multivariate analysis showed that UT stage (OR = 7.188, p = 0.049), conventional ultrasound features diagnosis of LN metastasis (OR = 8.010, p = 0.025) and SR (OR = 5.022, p = 0.031) were independent risk factors for LN metastasis. These risk factors were included in logistic regression analysis and the model was established, Y = -7.3 + 1.9 X10 + 2.1 X11 + 1.6 X13 (Y = Logit[P], P: LN metastasis rate, X10: UT stage, X11: conventional ultrasound features diagnosis of LN metastasis, X13: SR). The receiver operating characteristic (ROC) curve was used to test the model's predictive efficacy, the area under the curve was 0.95, sensitivity: 95%, specificity: 87%. Hosmer-Lemeshow goodness of fit test showed X2 = 6.015, p = 0.65 (p > 0.05), indicating that the model had a high predictive value.

Evaluation of perirectal LN metastasis only based on conventional ERUS features of LN was not effective enough. UT stage of tumor, conventional ultrasound features diagnosis of LN metastasis and SR were independent risk factors for LN metastasis. The predictive model had good assessment efficacy and had certain clinical application value.