Deep Learning Can Differentiate IDH-Mutant from IDH-Wild GBM.
Ontology highlight
ABSTRACT:
SUBMITTER: Pasquini L
PROVIDER: S-EPMC8069494 | biostudies-literature |
REPOSITORIES: biostudies-literature
Similar Datasets
Project description:This study aimed to investigate the potential of quantitative radiomic data extracted from conventional MR images in discriminating IDH-mutant grade 4 astrocytomas from IDH-wild-type glioblastomas (GBMs). A cohort of 57 treatment-naïve patients with IDH-mutant grade 4 astrocytomas (n = 23) and IDH-wild-type GBMs (n = 34) underwent anatomical imaging on a 3T MR system with standard parameters. Post-contrast T1-weighted and T2-FLAIR images were co-registered. A semi-automatic segmentation approach was used to generate regions of interest (ROIs) from different tissue components of neoplasms. A total of 1050 radiomic features were extracted from each image. The data were split randomly into training and testing sets. A deep learning-based data augmentation method (CTGAN) was implemented to synthesize 200 datasets from the training sets. A total of 18 classifiers were used to distinguish two genotypes of grade 4 astrocytomas. From generated data using 80% training set, the best discriminatory power was obtained from core tumor regions overlaid on post-contrast T1 using the K-best feature selection algorithm and a Gaussian naïve Bayes classifier (AUC = 0.93, accuracy = 0.92, sensitivity = 1, specificity = 0.86, PR_AUC = 0.92). Similarly, high diagnostic performances were obtained from original and generated data using 50% and 30% training sets. Our findings suggest that conventional MR imaging-based radiomic features combined with machine/deep learning methods may be valuable in discriminating IDH-mutant grade 4 astrocytomas from IDH-wild-type GBMs.
| S-EPMC9913426 | biostudies-literature
Project description:BackgroundNeoantigen based personalized immune therapies achieve promising results in melanoma and lung cancer, but few neoantigen based models perform well in IDH wild-type GBM, and the association between neoantigen intrinsic features and prognosis remain unclear in IDH wild-type GBM. We presented a novel neoantigen intrinsic feature-based deep learning model (neoDL) to stratify IDH wild-type GBMs into subgroups with different survivals.ResultsWe first derived intrinsic features for each neoantigen associated with survival, followed by applying neoDL in TCGA data cohort(AUC = 0.988, p value < 0.0001). Leave one out cross validation (LOOCV) in TCGA demonstrated that neoDL successfully classified IDH wild-type GBMs into different prognostic subgroups, which was further validated in an independent data cohort from Asian population. Long-term survival IDH wild-type GBMs identified by neoDL were found characterized by 12 protective neoantigen intrinsic features and enriched in development and cell cycle.ConclusionsThe model can be therapeutically exploited to identify IDH wild-type GBM with good prognosis who will most likely benefit from neoantigen based personalized immunetherapy. Furthermore, the prognostic intrinsic features of the neoantigens inferred from this study can be used for identifying neoantigens with high potentials of immunogenicity.
| S-EPMC8299600 | biostudies-literature
Project description:Glioblastoma (GBM) is the most common and aggressive primary brain malignancy. Adhesion G protein-coupled receptors (aGPCRs) have attracted interest for their functional role in gliomagenesis and their potential as treatment targets. To identify therapeutically targetable opportunities among aGPCR family members in an unbiased fashion, we analyzed expression levels of all aGPCRs in GBM and non-neoplastic brain tissue. Using bulk and single cell transcriptomic and proteomic data, we show that CD97 (ADGRE5), an aGPCR previously implicated in GBM pathogenesis, is the most promising aGPCR target in GBM, by virtue of its abundance in all GBM tumors and its de novo expression profile in GBM compared to normal brain tissue and neural progenitors. CD97 knockdown or knockout significantly reduces the tumor initiation capacity of patient-derived GBM cells (PDGC) in vitro and in vivo. Transcriptomic and metabolomic data from PDGCs suggest that CD97 promotes glycolytic metabolism. The oncogenic and metabolic effects of CD97 are mediated by the MAPK pathway. Activation of MAPK signaling depends on phosphorylation of the cytosolic C-terminus of CD97 and recruitment of -arrestin. Using single-cell RNA-sequencing and biochemical assays, we demonstrate that THY1/CD90 is the most likely CD97 ligand in GBM. Lastly, we show that targeting of GBM cells with an anti-CD97 antibody-drug conjugate in vitro selectively kills tumor cells but not human astrocytes or neural stem cells. Our studies identify CD97 as an important regulator of tumor metabolism in GBM, elucidate mechanisms of receptor activation and signaling, and provide strong scientific rationale for developing biologics to target CD97 in GBM.
2023-12-01 | GSE230389 | GEO
Project description:BackgroundConflicting data exist regarding the prognostic impact of the isocitrate dehydrogenase (IDH) mutation in intrahepatic cholangiocarcinoma (ICC), and limited data exist in patients with advanced-stage disease. Similarly, the clinical phenotype of patients with advanced IDH mutant (IDHm) ICC has not been characterized. In this study, we report the correlation of IDH mutation status with prognosis and clinicopathologic features in patients with advanced ICC.MethodsPatients with histologically confirmed advanced ICC who underwent tumor mutational profiling as a routine part of their care between 2009 and 2014 were evaluated. Clinical and pathological data were collected by retrospective chart review for patients with IDHm versus IDH wild-type (IDHwt) ICC. Pretreatment tumor volume was calculated on computed tomography or magnetic resonance imaging.ResultsOf the 104 patients with ICC who were evaluated, 30 (28.8%) had an IDH mutation (25.0% IDH1, 3.8% IDH2). The median overall survival did not differ significantly between IDHm and IDHwt patients (15.0 vs. 20.1 months, respectively; p = .17). The pretreatment serum carbohydrate antigen 19-9 (CA19-9) level in IDHm and IDHwt patients was 34.5 and 118.0 U/mL, respectively (p = .04). Age at diagnosis, sex, histologic grade, and pattern of metastasis did not differ significantly by IDH mutation status.ConclusionThe IDH mutation was not associated with prognosis in patients with advanced ICC. The clinical phenotypes of advanced IDHm and IDHwt ICC were similar, but patients with IDHm ICC had a lower median serum CA19-9 level at presentation.Implications for practicePrevious studies assessing the prognostic impact of the isocitrate dehydrogenase (IDH) gene mutation in intrahepatic cholangiocarcinoma (ICC) mainly focused on patients with early-stage disease who have undergone resection. These studies offer conflicting results. The target population for clinical trials of IDH inhibitors is patients with unresectable or metastatic disease, and the current study is the first to focus on the prognosis and clinical phenotype of this population and reports on the largest cohort of patients with advanced IDH mutant ICC to date. The finding that the IDH mutation lacks prognostic significance in advanced ICC is preliminary and needs to be confirmed prospectively in a larger study.
| S-EPMC4571807 | biostudies-literature
Project description:Microglia and macrophages are the largest component of the inflammatory infiltrate in glioblastoma (GBM). However, whether there are differences in their representation and activity in the prognostically-favorable isocitrate dehydrogenase (IDH)-mutated compared to -wild type GBMs is unknown. Studies on human specimens of untreated IDH-mutant GBMs are rare given they comprise 10% of all GBMs and often present at lower grades, receiving treatments prior to dedifferentiation that can drastically alter microglia and macrophage phenotypes. We were able to obtain large samples of four previously untreated IDH-mutant GBM. Using flow cytometry, immunofluorescence techniques with automated segmentation protocols that quantify at the individual-cell level, and comparison between single-cell RNA-sequencing (scRNA-seq) databases of human GBM, we discerned dissimilarities between GBM-associated microglia and macrophages (GAMMs) in IDH-mutant and -wild type GBMs. We found there are significantly fewer GAMM in IDH-mutant GBMs, but they are more pro-inflammatory, suggesting this contributes to the better prognosis of these tumors. Our pro-inflammatory score which combines the expression of inflammatory markers (CD68/HLA-A, -B, -C/TNF/CD163/IL10/TGFB2), Iba1 intensity, and GAMM surface area also indicates that more pro-inflammatory GAMMs are associated with longer overall survival independent of IDH status. Interrogation of scRNA-seq databases demonstrates microglia in IDH-mutants are mainly pro-inflammatory, while anti-inflammatory macrophages that upregulate genes such as FCER1G and TYROBP predominate in IDH-wild type GBM. Taken together, these observations are the first head-to-head comparison of GAMMs in treatment-naïve IDH-mutant versus -wild type GBMs. Our findings highlight biological disparities in the innate immune microenvironment related to IDH prognosis that can be exploited for therapeutic purposes.
| S-EPMC6517100 | biostudies-literature
Project description:ObjectiveThe purpose of this study was to investigate the feasibility of applying handcrafted radiomics (HCR) and deep learning-based radiomics (DLR) for the accurate preoperative classification of glioblastoma (GBM) and solitary brain metastasis (BM).MethodsA retrospective analysis of the magnetic resonance imaging (MRI) data of 140 patients (110 in the training dataset and 30 in the test dataset) with GBM and 128 patients (98 in the training dataset and 30 in the test dataset) with BM confirmed by surgical pathology was performed. The regions of interest (ROIs) on T1-weighted imaging (T1WI), T2-weighted imaging (T2WI), and contrast-enhanced T1WI (T1CE) were drawn manually, and then, HCR and DLR analyses were performed. On this basis, different machine learning algorithms were implemented and compared to find the optimal modeling method. The final classifiers were identified and validated for different MRI modalities using HCR features and HCR + DLR features. By analyzing the receiver operating characteristic (ROC) curve, the area under the curve (AUC), accuracy, sensitivity, and specificity were calculated to evaluate the predictive efficacy of different methods.ResultsIn multiclassifier modeling, random forest modeling showed the best distinguishing performance among all MRI modalities. HCR models already showed good results for distinguishing between the two types of brain tumors in the test dataset (T1WI, AUC = 0.86; T2WI, AUC = 0.76; T1CE, AUC = 0.93). By adding DLR features, all AUCs showed significant improvement (T1WI, AUC = 0.87; T2WI, AUC = 0.80; T1CE, AUC = 0.97; p < 0.05). The T1CE-based radiomic model showed the best classification performance (AUC = 0.99 in the training dataset and AUC = 0.97 in the test dataset), surpassing the other MRI modalities (p < 0.05). The multimodality radiomic model also showed robust performance (AUC = 1 in the training dataset and AUC = 0.84 in the test dataset).ConclusionMachine learning models using MRI radiomic features can help distinguish GBM from BM effectively, especially the combination of HCR and DLR features.
| S-EPMC8195660 | biostudies-literature
Project description:This study aimed to find any ambiguous genetic outlier for "oligodendroglioma, IDH-mutant and 1p/19q-codeleted (O_IDH_mut)" and "astrocytoma, IDH-mutant (A_IDH_mut)" and to redefine the genetic landscape and prognostic factors of IDH-mutant gliomas. Next-generation sequencing (NGS) using a brain tumor-targeted gene panel, methylation profiles, and clinicopathological features were analyzed for O_IDH_mut (n = 74) in 70 patients and for A_IDH_mut (n = 95) in 90 patients. 97.3% of O_IDH_mut and 98.9% of A_IDH_mut displayed a classic genomic landscape. Combined CIC (75.7%) and/or FUBP1 (45.9%) mutations were detected in 93.2% and MGMTp methylation in 95.9% of O_IDH_mut patients. In A_IDH_mut, TP53 mutations were found in 86.3% and combined ATRX (82.1%) and TERTp (6.3%) mutations in 88.4%. Although there were 3 confusing cases, NOS (not otherwise specified) category, based on genetic profiles, but they were clearly classified by combining histopathology and DKFZ methylation classifier algorithms. The patients with MYCN amplification and/or CDKN2A/2B homozygous deletion in the A_IDH_mut category had a worse prognosis than those without these gene alterations and MYCN-amplified A_IDH_mut showed the worst prognosis. However, there was no prognostic genetic marker in O_IDH_mut. In histopathologically or genetically ambiguous cases, methylation profiles can be used as an objective tool to avoid a diagnosis of NOS or NEC (not elsewhere classified), as well as for tumor classification. The authors have not encountered a case of true mixed oligoastrocytoma using an integrated diagnosis of histopathological, genetic and methylation profiles. MYCN amplification, in addition to CDKN2A/2B homozygous deletion, should be included in the genetic criteria for CNS WHO grade 4 A_IDH_mut.
| S-EPMC10130138 | biostudies-literature
Project description:Mutations in isocitrate dehydrogenase genes IDH1 and IDH2 are frequently found in diffuse and anaplastic astrocytic and oligodendroglial tumours as well as in secondary glioblastomas. As IDH is a very important prognostic, diagnostic and therapeutic biomarker for glioma, it is of paramount importance to determine its mutational status. The haematoxylin and eosin (H&E) staining is a valuable tool in precision oncology as it guides histopathology-based diagnosis and proceeding patient's treatment. However, H&E staining alone does not determine the IDH mutational status of a tumour. Deep learning methods applied to MRI data have been demonstrated to be a useful tool in IDH status prediction, however the effectiveness of deep learning on H&E slides in the clinical setting has not been investigated so far. Furthermore, the performance of deep learning methods in medical imaging has been practically limited by small sample sizes currently available. Here we propose a data augmentation method based on the Generative Adversarial Networks (GAN) deep learning methodology, to improve the prediction performance of IDH mutational status using H&E slides. The H&E slides were acquired from 266 grade II-IV glioma patients from a mixture of public and private databases, including 130 IDH-wildtype and 136 IDH-mutant patients. A baseline deep learning model without data augmentation achieved an accuracy of 0.794 (AUC = 0.920). With GAN-based data augmentation, the accuracy of the IDH mutational status prediction was improved to 0.853 (AUC = 0.927) when the 3,000 GAN generated training samples were added to the original training set (24,000 samples). By integrating also patients' age into the model, the accuracy improved further to 0.882 (AUC = 0.931). Our findings show that deep learning methodology, enhanced by GAN data augmentation, can support physicians in gliomas' IDH status prediction.
| S-EPMC7206037 | biostudies-literature
Project description:Mutations in isocitrate dehydrogenases 1 and 2 (IDHmut) are present in a variety of cancers, including glioma, acute myeloid leukemia (AML), melanoma, and cholangiocarcinoma. These mutations promote hypermethylation, yet it is only a favorable prognostic marker in glioma, for reasons that are unclear. We hypothesized that the patterns of DNA methylation, and transcriptome profiles, would vary among IDHmut cancers, especially gliomas. Using Illumina 450K and RNA-Seq data from The Cancer Genome Atlas, we show that of 365,092 analyzed CpG sites, 70,591 (19%) were hypermethylated in IDHmut gliomas compared to wild-type (IDHwt) gliomas, and only 3%, 2%, and 4% of CpG sites were hypermethylated in IDHmut AML, melanoma, and cholangiocarcinoma, relative to each of their IDHwt counterparts. Transcriptome differences showed pro-malignant genes that appear to be unique to IDHmut gliomas. However, genes involved in differentiation and immune response were suppressed in all IDHmut cancers. Additionally, IDHmut caused a greater degree of hypermethylation in undifferentiated neural progenitor cells than in mature astrocytes. These data suggest that the extent and targets of IDHmut-induced genomic hypermethylation vary greatly according to the cellular context and may help explain why IDHmut is only a favorable prognostic marker in gliomas.
| S-EPMC6586617 | biostudies-literature
Project description:BackgroundIsocitrate dehydrogenase (IDH)-mutant tumors exhibit an altered metabolic state and are critically dependent upon nicotinamide adenine dinucleotide (NAD+) for cellular survival. NAD+ steady-state levels can be influenced by both biosynthetic and consumptive processes. Here, we investigated activation of sirtuin (SIRT) enzymes, which consume NAD+ as a coenzyme, as a potential mechanism to reduce cellular NAD+ levels in these tumors.MethodsThe effect of inhibition or activation of sirtuin activity, using (i) small molecules, (ii) clustered regularly interspaced short palindromic repeat/CRISPR associated protein 9 gene editing, and (iii) inducible overexpression, was investigated in IDH-mutant tumor lines, including patient-derived IDH-mutant glioma lines.ResultsWe found that Sirt1 activation led to marked augmentation of NAD+ depletion and accentuation of cytotoxicity when combined with inhibition of nicotinamide phosphoribosyltransferase (NAMPT), consistent with the enzymatic activity of SIRT1 as a primary cellular NAD+ consumer in IDH-mutant cells. Activation of Sirt1 through either genetic overexpression or pharmacologic Sirt1-activating compounds (STACs), an existing class of well-tolerated drugs, led to inhibition of IDH1-mutant tumor cell growth.ConclusionsActivation of Sirt1 can selectively target IDH-mutant tumors. These findings indicate that relatively nontoxic STACs, administered either alone or in combination with NAMPT inhibition, could alter the growth trajectory of IDH-mutant gliomas while minimizing toxicity associated with cytotoxic chemotherapeutic regimens.
| S-EPMC7850026 | biostudies-literature