Predicting disease-free survival in locally advanced rectal cancer using a prognostic model based on pretreatment
Fangying Chen1, Shaoting Zhang1,2, Caixia Fu3
1Department of Radiology, Changhai Hospital, Naval Medical University, NO.168 Changhai Road, Shanghai, 200433, China.
Purpose:
Disease-free survival (DFS) after neoadjuvant chemoradiotherapy (nCRT) is an important factor in affecting the quality of life and determining the subsequent treatment procedures for patients with locally advanced rectal cancer (LARC). This study aimed to develop a novel prognostic model for predicting the DFS in patients with LARC following nCRT and to verify its effectiveness.
Materials And Methods:
Patients with LARC who underwent magnetic resonance imaging (MRI) and nCRT at our institution between November 2017 and March 2022 were enrolled in this retrospective study. Clinicopathologic data and MRI indicators of all patients were collected and evaluated. All patients were divided into DFS and non-DFS groups according to the presence or absence of local recurrence or distant metastasis. The differences in the b-value threshold (bthreshold) and apparent diffusion coefficient (ADC) values between the DFS and non-DFS groups were compared. The Cox analyses were used to determine the risk factors in predicting the DFS. A merged model was established based on the risk factors, and a nomogram was constructed. The predictive performances of the merged model were validated using the receiver-operating characteristic (ROC) and decision curve analysis (DCA).
Results:
Of the 524 patients enrolled, 132 who underwent surgical resection post-nCRT were included in the final analysis. The post-neoadjuvant therapy pathological N stage, extramural venous invasion (EMVI), and bthreshold were independent factors in predicting the DFS. The C-index of the model was 0.688. The area under the curve (AUC) of the nomogram in predicting the 1-, 3-, and 5-year survival rates of patients was 0.731, 0.723, and 0.779, respectively. The DCA demonstrated that the merged model had a greater advantage than either the "all" or the "none" scheme when the threshold probability was between 0.1 and 0.65.
Conclusion:
A merged model based on the bthreshold value and clinicopathological features showed the potential to predict the prognosis of patients with LARC after nCRT and surgery.
