Diagnostic Efficiency of Diffusion Sequences and a Clinical Nomogram for Detecting Lymph Node Metastases from Rectal Cancer

Abstract

First, to evaluate and compare three different diffusion sequences (i.e., standard DWI, IVIM, and DKI) for nodal staging. Second, to combine the DWI, and anatomic information to assess metastatic lymph node (LN). We retrospectively identified 136 patients of rectal adenocarcinoma who met the inclusion criteria. Three diffusion sequences (i.e., standard DWI, IVIM, and DKI) were performed, and quantitative parameters were evaluated. Univariate and multivariate analyses were used to assess the associations between the anatomic and DWI information and LN pathology. Multivariate logistic regression was used to identify independent risk factors. A nomogram model was established, and the model performance was evaluated by the concordance index (c-index) and calibration curve. There was a statistical difference in variables (LN long diameter, LN short diameter, LN boundary, LN signal, peri-LN signal intensity, ADC-1000, ADC-1400, ADC-2000, Kapp and D) between metastatic and non-metastatic LN for training and validation cohorts (p < 0.05). The ADC value derived from b=1000 mm/s (ADC-1000) showed the relative higher AUC (AUC=0.780) than the ADC value derived from b=1400 mm/s (ADC-1400) (AUC=0.703). The predictive accuracy of the nomogram measured by the c-index was 0.854 and 0.812 in the training and validation cohort, respectively. The IVIM and DKI model's diagnostic efficiency was not significantly improved compared to conventional DWI. The diagnostic accuracy of metastatic LN can be enhanced using the nomogram model, leading to a rational therapeutic choice.