Development of an MRI-Based Radiomics System for Personalized Radiotherapy Recommendation in LGG
Poster No:
1088
Submission Type:
Abstract Submission
Authors:
Zihan Wang1, Hongbo Bao1, Jie Hu1, Zeya Yan1, Renwu Zhang1, Ruiyang Wang1, Tingting Pu1, Yang Wang1, Lei Wang2, Yinyan Wang1
Institutions:
1Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University, Beijing, Beijing, 2Department of Neurosurgery, Guiqian International General Hospital, Guiyang, Guizhou
First Author:
Zihan Wang
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Co-Author(s):
Hongbo Bao
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Jie Hu
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Zeya Yan
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Renwu Zhang
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Ruiyang Wang
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Tingting Pu
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Yang Wang
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Yinyan Wang
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Department of Neurosurgery, Beijing Tiantan Hospital, Capital Medical University
Beijing, Beijing
Introduction:
The standard treatment protocol for lower-grade gliomas (LGG) includes surgery and postoperative radiotherapy(Stupp et al., 2005). However, the inability to identify patients sensitive to radiotherapy renders this universal approach ineffective for non-responders, causing unnecessary physical and psychological burdens.
In clinical practice, radiosensitivity is assessed using the RANO criteria, which tracks tumor volume changes through MRI scans(van den Bent et al., 2011). Radiomics, a rapidly evolving technique that extracts quantitative features from medical images, has facilitated the development of predictive models for personalized therapy(Gillies, Kinahan, & Hricak, 2016). While AI models have been developed for predicting treatment response of temozolomide for high-risk LGG (Wu et al., 2024) , no models exist specifically for predicting LGG radiosensitivity.
This study aims to develop a radiomics-based personalized radiotherapy recommendation system for LGG patients, delivering tailored treatment based on radiotherapy sensitivity while minimizing treatment-related toxicity.
In clinical practice, radiosensitivity is assessed using the RANO criteria, which tracks tumor volume changes through MRI scans(van den Bent et al., 2011). Radiomics, a rapidly evolving technique that extracts quantitative features from medical images, has facilitated the development of predictive models for personalized therapy(Gillies, Kinahan, & Hricak, 2016). While AI models have been developed for predicting treatment response of temozolomide for high-risk LGG (Wu et al., 2024) , no models exist specifically for predicting LGG radiosensitivity.
This study aims to develop a radiomics-based personalized radiotherapy recommendation system for LGG patients, delivering tailored treatment based on radiotherapy sensitivity while minimizing treatment-related toxicity.
Methods:
We retrospectively included LGG patients who underwent craniotomy and postoperative radiotherapy between 2005 and 2023 from four medical centers for model development and validation. Additionally, a clinical trial was launched to recruit patients from Tiantan Hospital between January and October 2024 for prospective validation. Longitudinal MRI data were collected at four time points: the last scan before surgery, the first scan after surgery, the last scan before radiotherapy, and one month after radiotherapy. MRI and RNA sequencing data from LGG patients in the TCGA and CGGA databases were also analyzed to explore the model's biological basis. Radiosensitivity was defined based on tumor volume changes and clinical characteristics before and after radiotherapy, following the RANO criteria. The radiomics pipeline involved image preprocessing, tumor delineation, and feature extraction and filtration. Six radiomic models were constructed using the XGBoost algorithm based on features from individual MRI sequences and their combinations. The optimal model was selected using DeLong tests and the IDI index, and a hybrid model incorporating independent clinical predictors of radiosensitivity was developed. The decision-making process of the model was visualized using the SHAP algorithm. Radiomics scores (RadScore) were calculated for each patient in the exploration set, grouped by median value, and analyzed for prognostic differences. Sequencing data were used to quantify immune microenvironment components and identify biological pathways associated with radiosensitivity.
·Figure 1: Flowchart illustrating the recruitment of eligible patients from four medical centers in China and two international databases.
Results:
Among 21,520 glioma patients from 4 medical centers, 1,653 met the inclusion criteria. After excluding cases with incomplete or poor-quality images, 1,303 patients were included. Of these, 963 from the Tiantan center were randomized into a training set (n = 770) and an internal validation set (n = 193) in an 8:2 ratio, while 99 patients from 3 other centers formed the external validation set. The prospective validation and exploration sets included 65 and 379 patients, respectively. The optimal radiomic and hybrid models achieved AUCs of 0.85 and 0.86 (internal), 0.85 and 0.88 (external), and 0.86 and 0.88 (prospective validation). Patients in the low-RadScore group had significantly worse overall survival, with tumors showing higher infiltration of immunosuppressive cells, such as M2 macrophages and T-reg cells, and upregulated radioresistance-related processes, including hypoxia and the p53 pathway.
Conclusions:
This study developed and validated a robust LGG personalized radiotherapy recommendation system using multi-center datasets to improve treatment outcomes and reduce toxicity. Immune microenvironment remodeling and metabolic reprogramming form the biological foundation of the system. It holds promise for redefining current radiotherapy strategies and promoting individualized treatment for glioma.
·Figure2: Development of a robust LGG personalized radiotherapy recommendation system, underpinned by metabolic reprogramming and immune microenvironment remodeling
Modeling and Analysis Methods:
Classification and Predictive Modeling 1
Novel Imaging Acquisition Methods:
Anatomical MRI 2
Keywords:
Computing
Data analysis
Machine Learning
Modeling
MRI
1|2Indicates the priority used for review
Abstract Information
By submitting your proposal, you grant permission for the Organization for Human Brain Mapping (OHBM) to distribute your work in any format, including video, audio print and electronic text through OHBM OnDemand, social media channels, the OHBM website, or other electronic publications and media.
The Open Science Special Interest Group (OSSIG) is introducing a reproducibility challenge for OHBM 2025. This new initiative aims to enhance the reproducibility of scientific results and foster collaborations between labs. Teams will consist of a “source” party and a “reproducing” party, and will be evaluated on the success of their replication, the openness of the source work, and additional deliverables. Click here for more information. Propose your OHBM abstract(s) as source work for future OHBM meetings by selecting one of the following options:
Please indicate below if your study was a "resting state" or "task-activation” study.
Healthy subjects only or patients (note that patient studies may also involve healthy subjects):
Was this research conducted in the United States?
Were any human subjects research approved by the relevant Institutional Review Board or ethics panel? NOTE: Any human subjects studies without IRB approval will be automatically rejected.
Were any animal research approved by the relevant IACUC or other animal research panel? NOTE: Any animal studies without IACUC approval will be automatically rejected.
Please indicate which methods were used in your research:
For human MRI, what field strength scanner do you use?
Which processing packages did you use for your study?
Provide references using APA citation style.
Stupp, R. (2005). Radiotherapy plus concomitant and adjuvant temozolomide for glioblastoma. N Engl J Med, 352(10), 987-996.
van den Bent, M. J.(2011). Response assessment in neuro-oncology (a report of the RANO group): assessment of outcome in trials of diffuse low-grade gliomas. Lancet Oncol, 12(6), 583-593.
Wu, G. (2024). Study of radiochemotherapy decision-making for young high-risk low-grade glioma patients using a macroscopic and microscopic combined radiomics model. Eur Radiol, 34(5), 2861-2872.