Radiomics and Qualitative Features From Multiparametric MRI Predict Molecular Subtypes in Patients With Lower-Grade Glioma

Abstract

BackgroundIsocitrate dehydrogenase (IDH) mutation and 1p19q codeletion status have been identified as significant markers for therapy and prognosis in lower-grade glioma (LGG). The current study aimed to construct a combined machine learning-based model for predicting the molecular subtypes of LGG, including (1) IDH wild-type astrocytoma (IDHwt), (2) IDH mutant and 1p19q non-codeleted astrocytoma (IDHmut-noncodel), and (3) IDH-mutant and 1p19q codeleted oligodendroglioma (IDHmut-codel), based on multiparametric magnetic resonance imaging (MRI) radiomics, qualitative features, and clinical factors.MethodsA total of 335 patients with LGG (WHO grade II/III) were retrospectively enrolled. The sum of 5,929 radiomics features were extracted from multiparametric MRI. Selected robust, non-redundant, and relevant features were used to construct a random forest model based on a training cohort (n = 269) and evaluated on a testing cohort (n = 66). Meanwhile, preoperative MRIs of all patients were scored in accordance with Visually Accessible Rembrandt Images (VASARI) annotations and T2-fluid attenuated inversion recovery (T2-FLAIR) mismatch sign. By combining radiomics features, qualitative features (VASARI annotations and T2-FLAIR mismatch signs), and clinical factors, a combined prediction model for the molecular subtypes of LGG was built.ResultsThe 17-feature radiomics model achieved area under the curve (AUC) values of 0.6557, 0.6830, and 0.7579 for IDHwt, IDHmut-noncodel, and IDHmut-codel, respectively, in the testing cohort. Incorporating qualitative features and clinical factors into the radiomics model resulted in improved AUCs of 0.8623, 0.8056, and 0.8036 for IDHwt, IDHmut-noncodel, and IDHmut-codel, with balanced accuracies of 0.8924, 0.8066, and 0.8095, respectively.ConclusionThe combined machine learning algorithm can provide a method to non-invasively predict the molecular subtypes of LGG preoperatively with excellent predictive performance.

Improving Noninvasive Classification of Molecular Subtypes of Adult Gliomas With Diffusion‐Weighted MR Imaging: An Externally Validated Machine Learning Algorithm

“…In this study, we constructed a single three-class radiomic model to classify molecular subtypes of adult diffuse gliomas based on a large cohort of patients with both conventional There is a growing number of studies leveraging machine learning radiomics to predict molecular subtypes defined by IDH mutations and 1p/19q codeletions markers or a combination of these since they were first introduced in the 2016 WHO classifications of gliomas. 1,8,13,17,[22][23][24] However, due to limited sample size, most of these studies only focused on classifying WHO grade II-III lower-grade gliomas or adopted binary classifiers to predict molecular subtypes in tiered steps. The relative large sample size of the current datasets enabled us to train a three-class radiomic model able to directly predict molecular subtypes in adult infiltrative gliomas irrespective of WHO grades.…”

“…Three-directional DWI is a widely used sequence in clinical MR protocols on gliomas, and it works on the assumption that the motion of tissue free water decreases with increasing tumor cellularity. [15][16][17]25 Previous studies have provided evidence that quantitative metrics of ADC derived from DWI are predictive of molecular subtypes of lower-grade gliomas. 15,16 A recent study also included conventional sequences combined with DWI to construct radiomic models for predicting molecular subtypes in lower-grade gliomas.…”

“…14 In addition, a growing number of studies have demonstrated that features derived from DWI have predictive values of IDH mutations and 1p/19q codeletion, especially in lower-grade gliomas. [15][16][17] Nonetheless, there is a lack of studies focusing on adult diffuse gliomas (WHO grade II, III, IV) with a large sample size to investigate the added values and generalizability of DWI-based radiomics in predicting molecular subtypes of gliomas.…”

The combination of radiomics features and VASARI standard to predict glioma grade

“…Chen et al. ( 19 ) combined radiomics with qualitative features (VASARI annotations and T2-FLAIR mismatch signs) to predict molecular subtypes in patients with lower-grade glioma. The AUC of the model containing radiomics and qualitative features was higher than the AUC of the model containing radiomics alone, with 0.8623 versus 0.6557.…”

Editorial for “Improving Noninvasive Classification of Molecular Subtypes of Adult Gliomas With Diffusion‐Weighted MR Imaging: An Externally Validated Machine Learning Algorithm”

“…In particular, diffusion weighted MRI (DWI) reflects the Brownian motion of water molecules represented, at a voxel level, by the computed Apparent Diffusion Coefficient (ADC) map. This technique was successfully investigated by previous studies, 6,7 where the most promising results in predicting glioma molecular subtypes were reported for radiomic-based models relying on combined clinical, anatomical MRI, and ADC features.…”