Oligosarcomas, IDH-mutant are distinct and aggressive

Oligodendrogliomas are defined at the molecular level by the presence of an IDH mutation and codeletion of chromosomal arms 1p and 19q. In the past, case reports and small studies described gliomas with sarcomatous features arising from oligodendrogliomas, so called oligosarcomas. Here, we report a series of 24 IDH-mutant oligosarcomas from 23 patients forming a distinct methylation class. The tumors were recurrences from prior oligodendrogliomas or developed de novo. Precursor tumors of 12 oligosarcomas were histologically and molecularly indistinguishable from conventional oligodendrogliomas. Oligosarcoma tumor cells were embedded in a dense network of reticulin fibers, frequently showing p53 accumulation, positivity for SMA and CALD1, loss of OLIG2 and gain of H3K27 trimethylation (H3K27me3) as compared to primary lesions. In 5 oligosarcomas no 1p/19q codeletion was detectable, although it was present in the primary lesions. Copy number neutral LOH was determined as underlying mechanism. Oligosarcomas harbored an increased chromosomal copy number variation load with frequent CDKN2A/B deletions. Proteomic profiling demonstrated oligosarcomas to be highly distinct from conventional CNS WHO grade 3 oligodendrogliomas with consistent evidence for a smooth muscle differentiation. Expression of several tumor suppressors was reduced with NF1 being lost frequently. In contrast, oncogenic YAP1 was aberrantly overexpressed in oligosarcomas. Panel sequencing revealed mutations in NF1 and TP53 along with IDH1/2 and TERT promoter mutations. Survival of patients was significantly poorer for oligosarcomas as first recurrence than for grade 3 oligodendrogliomas as first recurrence. These results establish oligosarcomas as a distinct group of IDH-mutant gliomas differing from conventional oligodendrogliomas on the histologic, epigenetic, proteomic, molecular and clinical level. The diagnosis can be based on the combined presence of (a) sarcomatous histology, (b) IDH-mutation and (c) TERT promoter mutation and/or 1p/19q codeletion, or, in unresolved cases, on its characteristic DNA methylation profile.

MPC-8 Serum anti-zinc finger FYVE domain-containing protein 21 (ZFYVE21) autoantibody as a novel biomarker for oligodendroglioma IDH-mutant and 1p/19q co-deletion

Background: Glioma is one of the most challenging diseases to cure, and it would be beneficial to discover new serum biomarkers for early diagnosis. Moreover, zinc finger FYVE domain-containing protein 21 (ZFYVE21) was a regulator of tumor invasion and migration. In this study, we examined the levels of serum anti-ZFYVE21 antibodies in patients with glioma. Methods: This is a multicenter observational prospective study to discover a novel serum autologous antibody marker. We analyzed 286 pre-surgically collected sera of CNS tumors and compared them to healthy donors(HD). Bacterially expressed glutathione-S-transferase-fused ZFYVE21 protein was purified, and its antibody levels were measured by amplified luminescent proximity homogeneous assay-linked immunosorbent assay (AlphaLISA). Results: The anti-ZFYVE21 antibody levels were significantly elevated in patients with gliomas (P<0.001) than those in HD, instead of patients with other CNS tumors. Among gliomas, the highest sensitivity was observed for oligodendroglioma containing IDH mutation and 1p/19q co-deletion to HD (sensitivity: 72.00%, specificity: 67.71%, AUC: 0.7565, P<0.0001), while there is no significance in astrocytoma containing only IDH mutation. In comparing 1p/19q co-deleted oligodendroglioma with IDH-mutated astrocytoma, the sensitivity and specificity were 50% and 100%, respectively. Conclusion: Serum anti-ZFYVE21 antibodies might be a novel diagnostic marker distinguishing 1p/19q co-deleted oligodendroglioma from IDH-mutant astrocytoma.

NI-12 The ratio of T1-Weighted to T2-Weighted Signal Intensity and IDH mutation in glioma

Introduction: Prediction of IDH mutation status for Lower-grade glioma (LrGG) is clinically significant. The purpose of this study is to test the hypothesis that the T1-weighted image/T2-weighted image ratio (rT1/T2), an imaging surrogate developed for myelin integrity, is a useful MRI biomarker for predicting the IDH mutation status of LrGG. Methods: Twenty-five LrGG patients (IDHwt: 8, IDHmt: 17) at Asahikawa Medical University Hospital (AMUH) were used as an exploratory cohort. Twenty-nine LrGG patients (IDHwt: 13, IDHmt: 16) from Osaka International Cancer Institute (OICI) and 103 patients from the Cancer Imaging Archive (TCIA) / Cancer Genome Atlas (TCGA) dataset (IDHwt: 19, IDHmt: 84) were used as validation cohorts. rT1/T2 images were calculated from T1- and T2-weighted images using a recommended signal correction. The region-of-interest was defined on T2-weighted images, and the relationship between the mean rT1/T2 (mrT1/T2) and the IDH mutation status was investigated. Results: The mrT1/T2 was able to significantly predict the IDH mutation status for the AMUH exploratory cohort (AUC = 0.75, p = 0.048). The ideal cut-off for detecting mutant IDH was mrT1/T2 < 0.666 ~ 0.677, with a sensitivity of 58.8% and a specificity of 87.5%. This result was further validated by the OICI validation cohort (AUC = 0.75, p = 0.023) with a sensitivity of 56.3% and a specificity of 69.2%. On the other hand, the sensitivity was 42.9% and the specificity was 68.4 % for the TCIA validation cohort (AUC = 0.63, p = 0.068). Conclusion: Our results supported the hypothesis that mrT1/T2 could be a useful image surrogate to predict the IDH mutation status of LrGG using two domestic cohorts. The decline of the accuracy for the TCIA cohort should be further investigated.

MPC-13 The evaluation of the shift of trend in lower grade glioma diagnoses based on each era`s criteria

It is found that molecular characteristics in lower grade gliomas (LrGGs) such as codeletion of 1p/19q and IDH mutation was found to be more accurate to predict the patient`s clinical outcome compared to morphological diagnoses alone. Since the revision WHO2016 classification of LrGGs, molecular characteristics were implemented as diagnostic standard for LrGGs diagnoses. In the other hand, morphological diagnostic standard before WHO2016 classification era was determined by different considerations and therapeutic strategies. The malignancy grades were also majorly determined by morphological diagnoses only. This study re-evaluated 20 years of LrGG cases in single institution based on WHO2007 morphological criteria and compared them to the original institutional diagnoses from each era. The study samples were originally grade II-III diffuse glioma-diagnosed cases resected from 1990 to 2016. Biopsy cases were excluded. IDH mutation was analyzed by Sanger sequence and 1p/19 codeletion status was analyzed by Comparative Genome Hybridization (CGH). As the result 93 cases were collected and based on original diagnoses, more than 50% cases are astrocytomas. Compared to re-assessment by morphological diagnoses (WHO 2007), case numbers of astrocytoma diagnoses are decreased whereas oligodendroglioma and oligoastrocytoma case numbers are increased. But, based on WHO2016 criteria, the case number of astrocytomas is again found to be increased. From comparison between original institutional diagnoses and re-assessment results, it is found that there is a shift of trend from astrocytoma to oligodendroglioma and from grade II to grade III. Comparison between morphological diagnoses (WHO2007) and molecular (WHO2016) found that astrocytoma diagnoses remain unchanged meanwhile 45% of oligodendroglioma diagnoses were shifted into astrocytomas. There is a probability that there are high frequency of morphologically diagnosed oligodendroglioma tumors which are having molecular characteristics of astrocytoma. There is a trend that diagnosed grade II LrGGs are actually grade III based on re-assessment diagnosis.

AI and High-Grade Glioma for Diagnosis and Outcome Prediction: Do All Machine Learning Models Perform Equally Well?

Radiomic models outperform clinical data for outcome prediction in high-grade gliomas (HGG). However, lack of parameter standardization limits clinical applications. Many machine learning (ML) radiomic models employ single classifiers rather than ensemble learning, which is known to boost performance, and comparative analyses are lacking in the literature. We aimed to compare ML classifiers to predict clinically relevant tasks for HGG: overall survival (OS), isocitrate dehydrogenase (IDH) mutation, O-6-methylguanine-DNA-methyltransferase (MGMT) promoter methylation, epidermal growth factor receptor vIII (EGFR) amplification, and Ki-67 expression, based on radiomic features from conventional and advanced magnetic resonance imaging (MRI). Our objective was to identify the best algorithm for each task. One hundred fifty-six adult patients with pathologic diagnosis of HGG were included. Three tumoral regions were manually segmented: contrast-enhancing tumor, necrosis, and non-enhancing tumor. Radiomic features were extracted with a custom version of Pyradiomics and selected through Boruta algorithm. A Grid Search algorithm was applied when computing ten times K-fold cross-validation (K=10) to get the highest mean and lowest spread of accuracy. Model performance was assessed as AUC-ROC curve mean values with 95% confidence intervals (CI). Extreme Gradient Boosting (xGB) obtained highest accuracy for OS (74,5%), Adaboost (AB) for IDH mutation (87.5%), MGMT methylation (70,8%), Ki-67 expression (86%), and EGFR amplification (81%). Ensemble classifiers showed the best performance across tasks. High-scoring radiomic features shed light on possible correlations between MRI and tumor histology.

PATH-26. EVALUATING THE DISTRIBUTION AND PROGNOSIS OF IMMUNE CELL SIGNATURES OF THE TUMOR MICRO-ENVIRONMENT IN DIFFUSE GLIOMAS

The tumor micro-environment (TME) plays an important role in the biology of cancer, including gliomas. Single cell studies have highlighted the role of specific TME components in gliomas, and the methods to deconvolve bulk profiling data may serve to complement these studies on clinically annotated tumors. In this study, we estimated cell type proportions in 3 large glioma datasets (TCGA, CGGA-325, CGGA-693) using CIBERSORTx. Using a signature matrix comprising 22 immune cell types, we identified IDH mutation status-specific immune cell distributions and found that the proportions of 10 cell types were significantly different between IDHmut and IDHwt tumors across the 3 datasets. Looking further within IDHmut tumors, we found that monocytes were enriched in 1p/19q non-co-deleted tumors across the 3 glioma datasets, consistent with prior single cell studies. We then examined estimated gene expression among immune cell types relative to IDH mutation status and found clear separation of gene expression in 15 of 22 cell types in all 3 datasets. When we applied these 22 gene expression signatures in each tumor sample onto cluster-of-cluster analyses to identify tumor groups with distinct immune signature patterns, we found that samples were distributed largely according to the IDH status in all 3 datasets, confirming that immune cell expression is distinct based on IDH status. Among IDH-specific groups, cluster-of-cluster analyses showed that immune cell-based cluster groups had distinct survival outcomes, and that IDHwt samples were distributed significantly based on tumor grades as well as based on EGFR overexpression. Among IDHmut tumors, the distributions of tumor grade and 1p/19q co-deletion status were significantly different in the immune-based clusters in 2 of the 3 datasets examined. Overall, these results highlight the biological and clinical significance of the immune cell environment in gliomas, including distinctions based on IDH mutation status as well as prognosis within IDH-specific groups.

QOLP-08. THE LANDSCAPE OF EPILEPSY ASSOCIATED WITH DIFFUSELY INFILTRATIVE GLIOMAS

Seizures are among the most prevalent co-morbidities associated with glioma, and pose a serious threat to patients. Our prior work showed that IDH mutation (IDHmut) was associated with much greater seizure frequency at the time of initial glioma diagnosis. However, less is known about the variables that contribute to seizure risk throughout the course of disease. We therefore collected data from 247 patients with grade 2–4 glioma, and determined seizure risk using Kaplan-Meier survival probabilities and multivariable cox regression analyses. Median follow-up of IDH wildtype (IDHwt) and IDHmut glioma patients was 15 months and 36 months, respectively. Incidence of pre-operative seizures for IDHwt and IDHmut patients was 75/168 (45%) and 60/79 (76%), and incidence of post-operative seizures was 70/168 (42%) and 43/79 (54%), respectively. Patients who had a pre-operative seizure had a shorter time to their first post-operative seizure than patients who never had a pre-operative seizure in both IDHwt (P< 0.0001) and IDHmut (P= 0.039) cohorts. Among IDHmut glioma patients, those with subtotal resections developed post-operative seizures faster (median time to first seizure= 9.9 months) than those with gross-total resections (median not reached) (P= 0.0005), but a similar pattern was not observed in IDHwt glioma patients (P= 0.20). Those with IDHmut astrocytomas more quickly developed post-operative seizures (median= 11.1 months), compared to those with IDHwt astrocytomas (24.9 months) or IDHmut oligodendrogliomas (median not reached) (P= 0.033). Tumor progression closely followed post-operative seizures in patients with IDHwt gliomas when either their first post-operative seizure occurred longer than 6 months following resection, or when their post-operative seizures worsened in quality. These data suggest the best predictors of post-operative seizures are as follows: the presence of pre-operative seizures; extent of surgical resection; IDHmut status. These data will help clinicians better manage glioma patients by identifying those at greatest risk of seizures.

NIMG-36. VISUALIZATION OF TUMOR HETEROGENEITY AND PREDICTION OF ISOCITRATE DEHYDROGENASE MUTATION STATUS FOR HUMAN GLIOMAS BY USING MULTIPARAMETRIC PHYSIOLOGIC AND METABOLIC MRI

Preoperative prediction of isocitrate dehydrogenase mutation status is clinically meaningful, but remains challenging. This study aimed to predict the isocitrate dehydrogenase (IDH) status of gliomas by using the machine learning voxel-wise clustering method of multiparametric physiologic and metabolic magnetic resonance imaging (MRI) and to show the association of the created cluster labels with the glucose metabolism status of the tumors. Sixty-nine patients with diffuse glioma were scanned by pH-sensitive MRI, diffusion-weighted imaging, fluid-attenuated inversion recovery, and contrast-enhanced T1-weighted imaging at 3 T. An unsupervised two-level clustering approach, including the generation of a self-organizing map followed by the K-means clustering, was used for voxel-wise feature extraction from the acquired images. The logarithmic ratio of the labels in each class within tumor regions was applied to a support vector machine to differentiate IDH mutation status. Bootstrapping and leave-one-out cross-validation were used to calculate the area under the curve (AUC) of receiver operating characteristic curves, accuracy, sensitivity, and specificity for evaluating performance. Targeted biopsies were performed for 14 patients to explore the relationship between clustered labels and the expression of key glycolytic proteins determined using immunohistochemistry. The highest prediction performance to differentiate IDH status was found for 10-class clustering, with a mean AUC, accuracy, sensitivity, and specificity of 0.94, 0.91, 0.90, and 0.91, respectively. The tissues with labels 7 + 8 + 9 + 10 showed high expression levels of hypoxia-inducible factor 1-alpha, glucose transporter 3, and hexokinase 2, which are typical of IDH wild-type glioma, whereas those with labels 1 showed low expression of these proteins. Our machine learning model successfully predicted the IDH mutation status of gliomas, and the resulting clusters properly reflected the metabolic status of the tumors.

PATH-19. TERT PROMOTER MUTATION, NOT H3K27M MUTATION IS A PROGNOSTIC FACTOR FOR ADULT THALAMIC GLIOMAS

Purpose Thalamic diffuse glioma is classified as WHO grade 4 as Diffuse midline glioma, H3K27M mutation if H3K27M mutation was found regardless of its histological findings, but the significance of H3K27M mutation is not clear compared with pediatric cases. We aimed to find genetic prognostic factors in adult thalamic diffuse gliomas. METHODS Pathological diagnosis, genetic abnormalities, and clinical course of adult newly diagnosed thalamic gliomas diagnosed and treated at our institution from July 2007 to March 2020 were retrospectively analyzed. RESULTS The number of cases was 41 (24 males, 17 females), median age was 47 years (20-75 years). Tumor localization was 20 cases on the left, 14 cases on the right, and 7 cases on both sides. The pathological diagnosis was GBM 15 cases, DMG 15 cases, AA-IDH WT 7 cases, DA-IDH WT 4 cases, all of which were IDH wild type, and none of them had IDH mutation and 1p/19q co-deletion. H3K27M mutations were found in 15 cases and TERT promoter mutations were found in 12 cases, both of which were completely mutually exclusive. Tumor resection and biopsy was performed in 33 and 8 cases, respectively, and the median removal rate was 95% for those who underwent tumor resection. The median PFS and OS of all cases were 14.3 months and 38 months, respectively, and the median OS by pathological diagnosis was GBM 12.4 months, DMG 47.4 months, AA-IDH WT 37.3 months, DA-IDH WT not reached. The median OS in the H3K27M mutant group (47.4 months) was significantly better (p=0.02) than that in the TERT promoter mutation group (13.5 months). CONCLUSION There was no IDH mutation in adult thalamic gliomas, the H3K27M mutation and the TERT promoter mutation were mutually exclusive. The H3K27M mutation was not a prognostic factor, but the TERT promoter mutation was the strongest prognostic factor.

PATH-22. GENETIC VARIATION BETWEEN IDH MUTANT AND IDH WILD-TYPE GLIOMA

OBJECTIVE The prognosis of IDH mutant glioma was significantly different from that of IDH wild-type glioma. In order to explore the differences between them at the genetic level, we analyzed the genetic results of IDH mutant and IDH wild-type glioma. METHODS This study analyzed the clinical data and genetic results of 45 glioma patients from Jan. 2017 to Dec. 2019, exploring relevant prognostic indicators and the difference in genetic profile between IDH mutant glioma and IDH wild-type glioma. RESULTS 45 patients were included in this study, including 15 IDH mutant glioma patients and 30 IDH wild-type glioma patients. Genetic analysis showed that there was no difference in tumor mutation burden (TMB) and microsatellite instability (MSI) between IDH mutant glioma and IDH wild-type glioma. But somatic mutation between IDH mutant and IDH wild-type glioma was different. The expressions of IDH1, CIC, SYNE1 and TP53 were different in the two groups, among which IDH1 and CIC were more significantly different. Copy number variations (CNV) was also different between IDH mutant glioma and IDH wild-type glioma. STIL occured more frequently in IDH wild-type gliomas. Genetic analysis also showed the difference in variant allel frequence (VAF). IDH mutant gliomas were more likely to be combined with ATRX and TP53 mutations, while IDH wild-type gliomas, in addition to the combination of TP53 mutations, often also combined with the mutations of NF1, BRAF and PTEN. In survival analysis, glioma with IDH mutation has a good prognosis, and IDH wild-type patients have a poor prognosis. In IDH wild-type patients, patients with PTEN mutation have a worse prognosis. CONCLUSION There is an obvious genetic difference between IDH mutation and IDH wild-type glioma, and PTEN mutation is a poor prognostic factor for IDH wild-type patients.