Traditional radiomics models mainly rely on explicitly-designed handcrafted features from medical images. This paper aimed to investigate if deep features extracted via transfer learning can generate radiomics signatures for prediction of overall survival (OS) in patients with Glioblastoma Multiforme (GBM). This study comprised a discovery data set of 75 patients and an independent validation data set of 37 patients. A total of 1403 handcrafted features and 98304 deep features were extracted from preoperative multi-modality MR images. After feature selection, a six-deep-feature signature was constructed by using the least absolute shrinkage and selection operator (LASSO) Cox regression model. A radiomics nomogram was further presented by combining the signature and clinical risk factors such as age and Karnofsky Performance Score. Compared with traditional risk factors, the proposed signature achieved better performance for prediction of OS (C-index = 0.710, 95% CI: 0.588, 0.932) and significant stratification of patients into prognostically distinct groups (P < 0.001, HR = 5.128, 95% CI: 2.029, 12.960). The combined model achieved improved predictive performance (C-index = 0.739). Our study demonstrates that transfer learning-based deep features are able to generate prognostic imaging signature for OS prediction and patient stratification for GBM, indicating the potential of deep imaging feature-based biomarker in preoperative care of GBM patients.

In fully-automatic radiomics model for predicting overall survival (OS) of glioblastoma multiforme (GBM) patients, the effect of image standardization parameters such as voxel size, quantization method and gray level on model reproducibility and prognostic performance are still unclear. In this study, 45792 multiregional radiomics features were automatically extracted from multi-modality MR images with different voxel sizes, quantization methods, and gray levels. The feature reproducibility and prognostic performance were assessed. Multiparametric and fixed-parameter radiomics signatures were constructed based on a training cohort (60 patients). In an independent validation cohort (32 patients), the multiparametric signature achieved better performance for OS prediction (C-Index = 0.705, 95% CI: 0.672, 0.738) and significant stratification of patients into high- and low-risk groups (P = 0.0040, HR = 3.29, 95% CI: 1.40, 7.70), which outperformed the fixed-parameter signatures and conventional factors such as age, Karnofsky Performance Score and tumor volume. This study demonstrated that voxel size, quantization method and gray level had influence on reproducibility and prognosis of radiomics features for GBM OS prediction. An automatic method to determine the optimal parameter settings was provided. It indicated that multiparametric radiomics signature had the potential of offering better prognostic performance than fixed-parameter signatures.

• Multiregional and multiparametric MRI features reliably predicted MGMT methylation in multicentre cohorts. • All-relevant imaging features predicted MGMT methylation better than univariately-predictive and non-redundant features. • Combing clinical factors with radiomics features did not benefit the prediction performance.

PurposeIsocitrate dehydrogenase 1 (IDH1) has been proven as a prognostic and predictive marker in glioblastoma (GBM) patients. The purpose was to preoperatively predict IDH mutation status in GBM using multiregional radiomics features from multiparametric magnetic resonance imaging (MRI).MethodsIn this retrospective multicenter study, 225 patients were included. A total of 1614 multiregional features were extracted from enhancement area, non‐enhancement area, necrosis, edema, tumor core, and whole tumor in multiparametric MRI. Three multiregional radiomics models were built from tumor core, whole tumor, and all regions using an all‐relevant feature selection and a random forest classification for predicting IDH1. Four single‐region models and a model combining all‐region features with clinical factors (age, sex, and Karnofsky performance status) were also built. All models were built from a training cohort (118 patients) and tested on an independent validation cohort (107 patients).ResultsAmong the four single‐region radiomics models, the edema model achieved the best accuracy of 96% and the best F1‐score of 0.75 while the non‐enhancement model achieved the best area under the receiver operating characteristic curve (AUC) of 0.88 in the validation cohort. The overall performance of the tumor‐core model (accuracy 0.96, AUC 0.86 and F1‐score 0.75) and the whole‐tumor model (accuracy 0.96, AUC 0.88 and F1‐score 0.75) was slightly better than the single‐regional models. The 8‐feature all‐region radiomics model achieved an improved overall performance of an accuracy 96%, an AUC 0.90, and an F1‐score 0.78. Among all models, the model combining all‐region imaging features with age achieved the best performance of an accuracy 97%, an AUC 0.96, and an F1‐score 0.84.ConclusionsThe radiomics model built with multiregional features from multiparametric MRI has the potential to preoperatively detect the IDH1 mutation status in GBM patients. The multiregional model built with all‐region features performed better than the single‐region models, while combining age with all‐region features achieved the best performance.

SUMMARY Glioblastoma multiforme (GBM) is among the most aggressive of human cancers. Although differentiation therapy has been proposed as a potential approach to treat GBM, the mechanisms of induced differentiation remain poorly defined. Here, we established an induced differentiation model of GBM using cAMP activators that specifically directed GBM differentiation into astroglia. Transcriptomic and proteomic analyses revealed that oxidative phosphorylation and mitochondrial biogenesis are involved in induced differentiation of GBM. Dibutyryl cyclic AMP (dbcAMP) reverses the Warburg effect, as evidenced by increased oxygen consumption and reduced lactate production. Mitochondrial biogenesis induced by activation of the CREB-PGC1α pathway triggers metabolic shift and differentiation. Blocking mitochondrial biogenesis using mdivi1 or by silencing PGC1α abrogates differentiation; conversely, overexpression of PGC1α elicits differentiation. In GBM xenograft models and patient-derived GBM samples, cAMP activators also induce tumor growth inhibition and differentiation. Our data show that mitochondrial biogenesis and metabolic switch to oxidative phosphorylation drive the differentiation of tumor cells.

BackgroundAs a neurotrophic factor, prosaposin (PSAP) can exert neuroprotective and neurotrophic effects. It is involved in the occurrence and development of prostate and breast cancer. However, there is no research about the role of PSAP in glioma.MethodsThe PSAP overexpressed or silenced glioma cells or glioma stem cells were established based on Lentiviral vector transfection. Cell viability assay, Edu assay, neurosphere formation assay and xenograft experiments were used to detect the proliferative ability. Western blot, Elisa and luciferase reporter assays were used to detect the possible mechanism.FindingsOur study firstly found that PSAP was highly expressed and secreted in clinical glioma specimens, glioma stem cells, and glioma cell lines. It was associated with poor prognosis. We found that PSAP significantly promoted the proliferation of glioma stem cells and cell lines. Moreover, PSAP promoted tumorigenesis in subcutaneous and orthotopic models of this disease. Furthermore, GSEA and KEGG analysis predicted that PSAP acts through the TLR4 and NF-κB signaling pathways, which was confirmed by western blot, immunoprecipitation, immunofluorescence, and use of the TLR4-specific inhibitor TAK-242.InterpretationThe findings of this study suggest that PSAP can promote glioma cell proliferation via the TLR4/NF-κB signaling pathway and may be an important target for glioma treatment.FundThis work was funded by (Nos. 81101917, 81270036, 81201802, 81673025), (No. LR2014023), and (Nos. 20170541022, 20172250290). The funders did not play a role in manuscript design, data collection, data analysis, interpretation nor writing of the manuscript.

No abstract

Background MRI is one of the most important techniques to assess the treatment response of gliomas. However, differentiating tumor recurrence (TuR) from treatment effects (TrE) remains challenging. Purpose To compare the diagnostic performance of MR diffusion‐weighted imaging (DWI), arterial spin labeling (ASL), proton MR spectroscopy (MRS), and amide proton transfer (APT) imaging in differentiating between TuR and TrE in posttreatment glioma patients. Study Type Prospective. Population Thirty patients with suspected tumor progression. Field Strength/Sequence DWI, ASL, proton MRS, and APT imaging were performed at 3T MR. Assessment MR indices, including ADC, relative cerebral blood flow (rCBF), ratios of Cho/Cr, Cho/NAA, and NAA/Cr and APT‐weighted (APTw) effect were obtained from DWI, ASL, proton MRS, and APT imaging, respectively. Indices were measured in the contralateral normal‐appearing white matter and lesions defined on the Gd‐enhanced T1w image. TuR or TrE was either determined histologically or clinically from longitudinal MRI follow‐up for at least 6 months. Statistical Tests The diagnostic performance of the indices was evaluated using Student's t‐test, receiver operating characteristic (ROC) curve, and multivariate logistic regression analyses. Results Among the 30 patients, 16 were diagnosed as having TuR and the rest having TrE. The recurrent tumors showed a significantly higher APTw effect (1.56 ± 1.14%) and rCBF (1.44 ± 0.61) compared with lesions representing treatment effects (–0.44 ± 1.34% and 0.72 ± 0.25, respectively, with P < 0.001). The areas under the curve (AUCs) were 0.87 and 0.90 for APTw and rCBF, respectively, in differentiating between TuR and TrE. Combining APTw and rCBF achieved a higher AUC of 0.93. MRS index ratios of Cho/Cr (P = 0.25), Cho/NAA (P = 0.16), and NAA/Cr (P = 0.86) and ADC (P = 0.37) showed no significant differences between TuR and TrE lesions, with AUCs lower than 0.70. Data Conclusion Compared with DWI and MRS, ASL and APT imaging techniques showed better diagnostic capability in distinguishing TuR from TrE. Level of Evidence: 1 Technical Efficacy: Stage 4 J. Magn. Reson. Imaging 2020;51:1154–1161.