Background
Surgery can lead to radical cure and long-term survival for individuals with esophageal squamous cell carcinoma (ESCC). Nevertheless, the survival rates markedly vary among patients. Accurately predicting surgical efficacy remains a pressing issue. This investigation sought to examine the predictive value of preoperative radiomics and the prognostic nutritional index for individuals with ESCC and to construct a comprehensive model for estimating the postoperative overall survival (OS) of individuals with ESCC.
Methods
This research conducted a retrospective examination of 466 individuals with ESCC from two medical centers. The data were arbitrarily categorized into a training cohort (TC, hospital 1, 246 cases), an internal validation cohort (IVC, hospital 1, 106 cases), and an external validation cohort (EVC, hospital 2, 114 cases). Upon demarcation of the area of interest, radiological features were extracted. The least absolute shrinkage and selection operator (LASSO) regression was utilized to identify the optimal radiomics features and calculate the radiomics score (RS). After the delineation of region of interest, radiological features were procured. Subsequently, the LASSO regression was employed to ascertain the optimal features and calculate the RS. The independent influencing factors acquired through Cox analyses were incorporated with the RS to establish a combined nomogram. The predictive capability of the model was examined utilizing the concordance index, time-dependent receiver operating characteristic (ROC) curves, calibration plots, and decision curve analysis.
Results
In the predictive model integrating preoperative radiomics with prognostic nutritional index and tumor–node–metastasis (TNM) stage for forecasting the 3-year OS rate, the area under the time-dependent ROC curve (AUC) was 0.812, 0.748, and 0.810 in the TC, IVC, and EVCs, respectively, thereby demonstrating outstanding prognostic significance. This was superior to the AUC values of the TNM stage prediction model in the TC, IVC, and EVCs, which were 0.717, 0.612, and 0.699, respectively. The concordance indexes of the combined model in the TC, IVC, and EVCs were 0.780, 0.760, and 0.764, respectively. The calibration and decision curves illustrated the nomogram’s remarkable calibration performance and clinical application value.
Conclusion
In this investigation, a predictive model was developed by integrating radiomics and the prognostic nutritional index. This model can predict the OS rate of postoperative patients with ESCC and could be employed as a tool for preoperative risk stratification.