MRI-based bone marrow radiomics for predicting cytogenetic abnormalities in multiple myeloma
1Department of Radiology, The First Affiliated Hospital of Soochow University, Suzhou, 215006, Jiangsu, China.
Clinical Radiology
|January 18, 2024
Summary
This study developed a magnetic resonance imaging (MRI) radiomics signature to non-invasively predict high-risk cytogenetic abnormalities in multiple myeloma (MM) patients. The FS-T2WI sequence-based signature showed promising predictive performance in both training and validation cohorts.
Area of Science:
- Radiology
- Oncology
- Medical Imaging Analysis
Background:
- Multiple Myeloma (MM) is a hematologic malignancy characterized by diverse cytogenetic abnormalities.
- Accurate prediction of cytogenetic risk stratification is crucial for guiding treatment decisions in MM.
- Current methods for assessing cytogenetics can be invasive and may not capture the full spectrum of disease heterogeneity.
Purpose of the Study:
- To develop and validate a radiomics signature using MRI images for predicting cytogenetic abnormalities in MM.
- To assess the performance of radiomics signatures derived from different MRI sequences (T1WI and FS-T2WI) in distinguishing high-risk (HRC) from standard-risk (SRC) cytogenetics.
- To establish a non-invasive tool for prognostication in newly diagnosed MM patients.
Main Methods:
- Retrospective analysis of 105 newly diagnosed MM patients with MRI data (T1WI and FS-T2WI sequences).
- Manual segmentation of Volumes of Interest (VOI) on FS-T2WI and T1WI sequences.
- Extraction and selection of radiomics features using reproducibility, redundancy analysis, and the LASSO algorithm to build predictive signatures.
- Validation of signature performance using ROC curves, AUC, sensitivity, and specificity.
Main Results:
- A radiomics signature was developed using 11 significant features from the FS-T2WI sequence and 4 from the T1WI sequence.
- The FS-T2WI-based radiomics signature demonstrated superior performance, achieving an AUC of 0.896 in the training cohort and 0.729 in the validation cohort.
- In the validation cohort, the FS-T2WI signature achieved a sensitivity of 0.833 and specificity of 0.667 for predicting HRC.
Conclusions:
- The developed MRI-based radiomics signature serves as a non-invasive and convenient tool for predicting cytogenetic abnormalities in MM.
- This approach can aid in risk stratification and potentially personalize treatment strategies for MM patients.
- Further prospective studies are warranted to confirm these findings and integrate radiomics into clinical practice.
