[Prediction of 1p/19q codeletion status in diffuse lower-grade glioma using multimodal MRI radiomics]

OBJECTIVE

To develop a noninvasive method for prediction of 1p/19q codeletion in diffuse lower-grade glioma (DLGG) based on multimodal magnetic resonance imaging (MRI) radiomics.

METHODS

We collected MRI data from 104 patients with pathologically confirmed DLGG between October, 2015 and September, 2022. A total of 535 radiomics features were extracted from T₂WI, T₁WI, FLAIR, CE-T₁WI and DWI, including 70 morphological features, 90 first order features, and 375 texture features. We constructed logistic regression (LR), logistic regression least absolute shrinkage and selection operator (LRlasso), support vector machine (SVM) and Linear Discriminant Analysis (LDA) radiomics models and compared their predictive performance after 10-fold cross validation. The MRI images were reviewed by two radiologists independently for predicting the 1p/19q status. Receiver operating characteristic curves were used to evaluate classification performance of the radiomics models and the radiologists.

RESULTS

The 4 radiomics models (LR, LRlasso, SVM and LDA) achieved similar area under the curve (AUC) in the validation dataset (0.833, 0.819, 0.824 and 0.819, respectively; P>0.1), and their predictive performance was all superior to that of resident physicians of radiology (AUC=0.645, P=0.011, 0.022, 0.016, 0.030, respectively) and similar to that of attending physicians of radiology (AUC=0.838, P>0.05).

CONCLUSION

Multiparametric MRI radiomics models show good performance for noninvasive prediction of 1p/19q codeletion status in patients with in diffuse lower-grade glioma.

Classification of diffuse lower-grade glioma based on immunological profiling

Transcriptomic data derived from bulk sequencing have been applied to delineate the tumor microenvironment (TME) and define immune subtypes in various cancers, which may facilitate the design of immunotherapy treatment strategies. We herein gathered published gene expression data from diffuse lower-grade glioma (LGG) patients to identify immune subtypes. Based on the immune gene profiles of 402 LGG patients from The Cancer Genome Atlas, we performed consensus clustering to determine robust clusters of patients, and evaluated their reproducibility in three Chinese Glioma Genome Atlas cohorts. We further integrated immunogenomics methods to characterize the immune environment of each subtype. Our analysis identified and validated three immune subtypes-Im1, Im2, and Im3-characterized by differences in lymphocyte signatures, somatic DNA alterations, and clinical outcomes. Im1 had a higher infiltration of CD8+ T cells, Th17, and mast cells. Im2 was defined by elevated cytolytic activity, exhausted CD8+ T cells, macrophages, higher levels of aneuploidy, and tumor mutation burden, and these patients had worst outcome. Im3 displayed more prominent T helper cell and APC coinhibition signatures, with elevated pDCs and macrophages. Each subtype was associated with distinct somatic alterations. Moreover, we applied graph structure learning-based dimensionality reduction to the immune landscape and revealed significant intracluster heterogeneity with Im2 subtype. Finally, we developed and validated an immune signature with better performance of prognosis prediction. Our results demonstrated the immunological heterogeneity within diffuse LGG and provided valuable stratification for the design of future immunotherapy.

A five-microRNA signature for individualized prognosis evaluation and radiotherapy guidance in patients with diffuse lower-grade glioma

Despite the prognostic value of IDH and other gene mutations found in diffuse glioma, markers that judge individual prognosis of patients with diffuse lower‐grade glioma (LGG) are still lacking. This study aims to develop an expression‐based microRNA signature to provide survival and radiotherapeutic response prediction for LGG patients. MicroRNA expression profiles and relevant clinical information of LGG patients were downloaded from The Cancer Genome Atlas (TCGA; the training group) and the Chinese Glioma Genome Atlas (CGGA; the test group). Cox regression analysis, random survival forests‐variable hunting (RSFVH) screening and receiver operating characteristic (ROC) were used to identify the prognostic microRNA signature. ROC and TimeROC curves were plotted to compare the predictive ability of IDH mutation and the signature. Stratification analysis was conducted in patients with radiotherapy information. Gene ontology (GO) analysis and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway analysis were performed to explore the biological function of the signature. We identified a five‐microRNA signature that can classify patients into low‐risk or high‐risk group with significantly different survival in the training and test datasets (P < 0.001). The five‐microRNA signature was proved to be superior to IDH mutation in survival prediction (AUCtraining = 0.688 vs 0.607). Stratification analysis found the signature could further divide patients after radiotherapy into two risk groups. GO and KEGG analyses revealed that microRNAs from the prognostic signature were mainly enriched in cancer‐associated pathways. The newly discovered five‐microRNA signature could predict survival and radiotherapeutic response of LGG patients based on individual microRNA expression.

Factor analysis for survival time prediction with informative censoring and diverse covariates

Fulfilling the promise of precision medicine requires accurately and precisely classifying disease states. For cancer, this includes prediction of survival time from a surfeit of covariates. Such data presents an opportunity for improved prediction, but also a challenge due to high dimensionality. Furthermore, disease populations can be heterogeneous. Integrative modeling is sensible, as the underlying hypothesis is that joint analysis of multiple covariates provides greater explanatory power than separate analyses. We propose an integrative latent variable model that combines factor analysis for various data types and an exponential proportional hazards (EPH) model for continuous survival time with informative censoring. The factor and EPH models are connected through low-dimensional latent variables that can be interpreted and visualized to identify subpopulations. We use this model to predict survival time. We demonstrate this model's utility in simulation and on four Cancer Genome Atlas datasets: diffuse lower-grade glioma, glioblastoma multiforme, lung adenocarcinoma, and lung squamous cell carcinoma. These datasets have small sample sizes, high-dimensional diverse covariates, and high censorship rates. We compare the predictions from our model to three alternative models. Our model outperforms in simulation and is competitive on real datasets. Furthermore, the low-dimensional visualization for diffuse lower-grade glioma displays known subpopulations.

Prognostic relevance of genetic alterations in diffuse lower-grade gliomas

Background

Diffuse lower-grade gliomas (LGGs) are genetically classified into 3 distinct subtypes based on isocitrate dehydrogenase (IDH) mutation status and codeletion of chromosome 1p and 19q (1p/19q). However, the subtype-specific effects of additional genetic lesions on survival are largely unknown.

Methods

Using Cox proportional hazards regression modeling, we investigated the subtype-specific effects of genetic alterations and clinicopathological factors on survival in each LGG subtype, in a Japanese cohort of LGG cases fully genotyped for driver mutations and copy number variations associated with LGGs (n = 308). The results were validated using a dataset from 414 LGG cases available from The Cancer Genome Atlas (TCGA).

Results

In Oligodendroglioma, IDH-mutant and 1p/19q codeleted, NOTCH1 mutations (P = 0.0041) and incomplete resection (P = 0.0019) were significantly associated with shorter survival. In Astrocytoma, IDH-mutant, PIK3R1 mutations (P = 0.0014) and altered retinoblastoma pathway genes (RB1, CDKN2A, and CDK4) (P = 0.013) were independent predictors of poor survival. In IDH-wildtype LGGs, co-occurrence of 7p gain, 10q loss, mutation in the TERT promoter (P = 0.024), and grade III histology (P < 0.0001) independently predicted poor survival. IDH-wildtype LGGs without any of these factors were diagnosed at a younger age (P = 0.042), and were less likely to have genetic lesions characteristic of glioblastoma, in comparison with other IDH-wildtype LGGs, suggesting that they likely represented biologically different subtypes. These results were largely confirmed in the cohort of TCGA.

Conclusions

Subtype-specific genetic lesions can be used to stratify patients within each LGG subtype. enabling better prognostication and management.