Abstract 204P
Background
Although neoadjuvant chemoradiotherapy followed by surgery is the standard treatment for esophageal cancer patients, most patients are unable to achieve pathological complete response with neoadjuvant therapy, resulting in poor outcomes. The aim of this study is to develop a method for selecting patients who can achieve pathological complete response through pre-neoadjuvant therapy chest-enhanced CT scans.
Methods
Two hundreds and one patients with esophageal cancer were enrolled and divided into a training set and a testing set in a 7:3 ratio. Radiomics features of intra-tumoral and peritumoral images were extracted from preoperative chest-enhanced CT scans of these patients. The features were dimensionally reduced in two steps. The selected intra-tumoral and peritumoral features, including marginal (with a distance of 0-3mm from the tumor) and adjacent (with a distance of 3-6mm from the tumor) ROI, were used to build models with four machine learning classifiers, including Support Vector Machine, XG-Boost, Random Forest and Naive Bayes. Models with satisfied accuracy and stability levels were considered to perform well. Finally, the performance of these well-performing models on the testing set was displayed using ROC curves.
Results
Among the 16 models, the best-performing models were the integrated (intra-tumoral and peritumoral features)-XGBoost and integrated-random forest models, which had average ROC AUCs of 0.906 and 0.918, respectively, with relative standard deviations (RSDs) of 6.26 and 6.89 in the training set. In the testing set, the AUCs were 0.845 and 0.871, respectively. There was no significant difference in the ROC curves between the two models. Table: 204P
The performance of the selected models on the testing set
Model | AUC (95% CI) | Specificity | Sensitivity |
Integrated-XGBoost | 0.845 (0.764, 0.928) | 0.864 | 0.777 |
Original-XGBoost | 0.759 (0.660, 0.857) | 0.900 | 0.592 |
Integrated-Random Forest | 0.871 (0.796, 0.946) | 0.682 | 0.933 |
Original-Random Forest | 0.795 (0.703, 0.887) | 0.825 | 0.673 |
Adjacent-Random Forest | 0.769 (0.671, 0.868) | 0.886 | 0.533 |
Integrated-Support Vector Machine | 0.719 (0.613, 0.825) | 0.795 | 0.622 |
Conclusions
The addition of peritumoral radiomics features to the radiomics analysis may improve the predictive performance of pathological response for esophageal cancer patients to neoadjuvant chemoradiotherapy.
Clinical trial identification
Editorial acknowledgement
Legal entity responsible for the study
The authors.
Funding
Has not received any funding.
Disclosure
All authors have declared no conflicts of interest.
Resources from the same session
16P - Patient and healthcare practitioner preferences in early-stage triple-negative breast cancer treatment: A discrete choice experiment
Presenter: Jiun-I Lai
Session: Poster Display
Resources:
Abstract
17P - Initial outcomes of the ACT Now PRIME CARE for breast cancer: Prevention of Breast canceR (screening/ stage shifting) utilizing Integrated MobilE Clinics and pAtient Reported online Evaluations and Education
Presenter: Herdee Gloriane Luna
Session: Poster Display
Resources:
Abstract
18P - Optimizing premenopausal HR+ HER2–ve eBC management in India: Insights from expert consensus
Presenter: Anitha Ramesh
Session: Poster Display
Resources:
Abstract
19P - Referral patterns among breast cancer patients in county-level hospitals in China
Presenter: Ping Lu
Session: Poster Display
Resources:
Abstract
20P - Real-world treatment of HER2+ and HR+/HER2- early breast cancer in county areas of China
Presenter: Ping Lu
Session: Poster Display
Resources:
Abstract
21P - Duration of breast cancer trials: Analysis of predicted versus actual completion date
Presenter: Daniëlle Verschoor
Session: Poster Display
Resources:
Abstract
22P - Impact of an online Asian genetic risk calculator on risk perception: Cancer-related distress and uptake of genetic counselling among Malaysian breast cancer patients (The ARiCa Study)
Presenter: HEAMANTHAA Padmanabhan
Session: Poster Display
Resources:
Abstract
23P - Consensus statements and expert recommendations for BRCAm breast cancer in the Asia-Pacific region (STREAM-AP)
Presenter: Soo Chin Lee
Session: Poster Display
Resources:
Abstract
24P - Germline genetic testing for hereditary cancer: A retrospective analysis in a single site referral centre in Malaysia
Presenter: Vivian Lee
Session: Poster Display
Resources:
Abstract
25P - Clinical presentations and prognostication of HER2-low breast cancer in Taiwan
Presenter: Bo-Fang Chen
Session: Poster Display
Resources:
Abstract