CT-based deep learning radiomics and hematological biomarkers in the assessment of pathological complete response to neoadjuvant chemoradiotherapy in patients with esophageal squamous cell carcinoma: A two-center study

• Published in: Translational oncology Vol. 39; p. 101804
• Main Authors: Zhang, Meng, Lu, Yukun, Sun, Hongfu, Hou, Chuanke, Zhou, Zichun, Liu, Xiao, Zhou, Qichao, Li, Zhenjiang, Yin, Yong
• Format: Journal Article
• Language: English
• Published: Elsevier Inc 01-01-2024; Neoplasia Press; Elsevier
• Subjects: Computed tomography; Esophageal squamous cell carcinoma; Machine learning; Neoadjuvant chemoradiotherapy; Original Research; Pathological complete response; Radiomics; NRI; LoG; LR; PDGFs; FGF2; AUC; CNNs; GLDM; IDI; RS2; RS1; cCR; Computed tomography; LASSO; ML; HGB; MO; DCA; nCRT; ROC; ESCC; GLCM; GLRLM; IMRT; Pathological complete response; SVM; GLSZM; Neoadjuvant chemoradiotherapy; Radiomics; NLR; NPV; XGBoost; PLR; PPV; mRMR; PLT; WBC; pCR; ROIs; NGTDM; VEGFA; LMR; PF; Machine learning; ALB; ALC; Esophageal squamous cell carcinoma; TP; NEU
• ISSN: 1936-5233; 1936-5233
• DOI: 10.1016/j.tranon.2023.101804

•A two-center study showed that deep learning radiomics analysis of pre- and post-nCRT CT images could improve the pCR prediction of patients with ESCC.•The combined model was superior to the clinical and radiomics models in predicting pCR in locally advanced ESCC and the LR classifier performed best in the current study.•Decision curves demonstrated that the novel predictive model based on deep learning and handcrafted radiomics features combined with hematological parameters has great clinical utility.
To evaluate and validate CT-based models using pre- and posttreatment deep learning radiomics features and hematological biomarkers for assessing esophageal squamous cell carcinoma (ESCC) pathological complete response (pCR) after neoadjuvant chemoradiotherapy (nCRT).
This retrospective study recruited patients with biopsy-proven ESCC who underwent nCRT from two Chinese hospitals between May 2017 and May 2022, divided into a training set (hospital I, 111 cases), an internal validation set (hospital I, 47 cases), and an external validation set (hospital II, 33 cases). We used minimum redundancy maximum relevance (mRMR) and least absolute shrinkage and selection operator (LASSO) as feature selection methods and three classifiers as model construction methods. The assessment of models was performed using area under the receiver operating characteristic (ROC) curve (AUC) and decision curve analysis (DCA).
A total 190 patients were included in our study (60.8 ± 7.08 years, 133 men), and seventy-seven of them (40.5 %) achieved pCR. The logistic regression (LR)-based combined model incorporating neutrophil to lymphocyte ratio, lymphocyte to monocyte ratio, albumin, and radscores performed well both in the internal and external validation sets with AUCs of 0.875 and 0.857 (95 % CI, 0.776–0.964; 0.731–0.984, P <0.05), respectively. DCA demonstrated that nomogram was useful for pCR prediction and produced clinical net benefits.
The incorporation of radscores and hematological biomarkers into LR-based model improved pCR prediction after nCRT in ESCC. Enhanced pCR predictability may improve patients selection before surgery, providing clinical application value for the use of active surveillance.