Abstract
PurposeTo evaluate whether multiparametric magnetic resonance imaging (MRI)-based logistic regression models can facilitate the early prediction of chemoradiotherapy response in patients with residual brain gliomas after surgery.Patients and MethodsA total of 84 patients with residual gliomas after surgery from January 2015 to September 2020 who were treated with chemoradiotherapy were retrospectively enrolled and classified as treatment-sensitive or treatment-insensitive. These patients were divided into a training group (from institution 1, 57 patients) and a validation group (from institutions 2 and 3, 27 patients). All preoperative and postoperative MR images were obtained, including T1-weighted (T1-w), T2-weighted (T2-w), and contrast-enhanced T1-weighted (CET1-w) images. A total of 851 radiomics features were extracted from every imaging series. Feature selection was performed with univariate analysis or in combination with multivariate analysis. Then, four multivariable logistic regression models derived from T1-w, T2-w, CET1-w and Joint series (T1+T2+CET1-w) were constructed to predict the response of postoperative residual gliomas to chemoradiotherapy (sensitive or insensitive). These models were validated in the validation group. Calibration curves, receiver operating characteristic (ROC) curves, and decision curve analysis (DCA) were applied to compare the predictive performances of these models.ResultsFour models were created and showed the following areas under the ROC curves (AUCs) in the training and validation groups: Model-Joint series (AUC, 0.923 and 0.852), Model-T1 (AUC, 0.835 and 0.809), Model-T2 (AUC, 0.784 and 0.605), and Model-CET1 (AUC, 0.805 and 0.537). These results indicated that the Model-Joint series had the best performance in the validation group, followed by Model-T1, Model-T2 and finally Model-CET1. The calibration curves indicated good agreement between the Model-Joint series predictions and actual probabilities. Additionally, the DCA curves demonstrated that the Model-Joint series was clinically useful.ConclusionMultiparametric MRI-based radiomics models can potentially predict tumor response after chemoradiotherapy in patients with postoperative residual gliomas, which may aid clinical decision making, especially to help patients initially predicted to be treatment-insensitive avoid the toxicity of chemoradiotherapy.
Highlights
Glioma is the most common tumor of the brain and is associated with high rates of disability and death, which is part because the tumor lesion is difficult to completely remove surgically due to its invasive growth characteristics
A total of 2553 (851×3) radiomics features were extracted from the Regions of interest (ROIs) of the T1-w, T2-w and contrast-enhanced T1-weighted (CET1-w) images, respectively, Radiomics Signature and Model Construction
There are few studies predicting the curative effect of chemoradiotherapy for residual gliomas before treatment, which is crucial for choosing the right therapeutic schedule
Summary
Glioma is the most common tumor of the brain and is associated with high rates of disability and death, which is part because the tumor lesion is difficult to completely remove surgically due to its invasive growth characteristics. According to the international guidelines for the treatment of neurological tumors, concurrent postoperative chemoradiotherapy is recommended for some grade 2, grade 3 and grade 4 glioma patients to treat residual tumor lesions. Radiotherapy and chemotherapy have a definite therapeutic effect on gliomas, but there are some negative effects, such as hair loss, vomiting, decreased immunity and high cost. Clinical practices have shown that not all gliomas are sensitive to chemoradiation due to the heterogeneity of tumor tissues. Some patients are unable to benefit from the treatment and unnecessarily suffer from the effects of chemoradiation. Identifying treatment-insensitive patients has become a critical area of research
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.