Remote Neuroinflammation in Newly Diagnosed Glioblastoma Correlates with Unfavorable Clinical Outcome.

PURPOSE: Current therapy strategies still provide only limited success in the treatment of glioblastoma, the most frequent primary brain tumor in adults. In addition to the characterization of the tumor microenvironment, global changes in the brain of patients with glioblastoma have been described. However, the impact and molecular signature of neuroinflammation distant of the primary tumor site have not yet been thoroughly elucidated. EXPERIMENTAL DESIGN: We performed translocator protein (TSPO)-PET in patients with newly diagnosed glioblastoma (n = 41), astrocytoma WHO grade 2 (n = 7), and healthy controls (n = 20) and compared TSPO-PET signals of the non-lesion (i.e., contralateral) hemisphere. Back-translation into syngeneic SB28 glioblastoma mice was used to characterize Pet alterations on a cellular level. Ultimately, multiplex gene expression analyses served to profile immune cells in remote brain. RESULTS: Our study revealed elevated TSPO-PET signals in contralateral hemispheres of patients with newly diagnosed glioblastoma compared to healthy controls. Contralateral TSPO was associated with persisting epileptic seizures and shorter overall survival independent of the tumor phenotype. Back-translation into syngeneic glioblastoma mice pinpointed myeloid cells as the predominant source of contralateral TSPO-PET signal increases and identified a complex immune signature characterized by myeloid cell activation and immunosuppression in distant brain regions. CONCLUSIONS: Neuroinflammation within the contralateral hemisphere can be detected with TSPO-PET imaging and associates with poor outcome in patients with newly diagnosed glioblastoma. The molecular signature of remote neuroinflammation promotes the evaluation of immunomodulatory strategies in patients with detrimental whole brain inflammation as reflected by high TSPO expression.

Feasibility of radiomic feature harmonization for pooling of [18F]FET or [18F]GE-180 PET images of gliomas.

INTRODUCTION: Large datasets are required to ensure reliable non-invasive glioma assessment with radiomics-based machine learning methods. This can often only be achieved by pooling images from different centers. Moreover, trained models should perform with high accuracy when applied to data from different centers. In this study, the impact of reconstruction settings and segmentation methods on radiomic features derived from amino acid and TSPO PET images of glioma patients was examined. Additionally, the ability to model and thus reduce feature differences was investigated. METHODS: [18F]FET and [18F]GE-180 PET data were acquired from 19 glioma patients. For each acquisition, 10 reconstruction settings and 9 segmentation methods were included to emulate multicentric data. Statistical robustness measures were calculated before and after ComBat harmonization. Differences between features due to setting variations were assessed using Friedman test, coefficient of variation (CV) and inter-rater reliability measures, including intraclass and Spearman's rank correlation coefficients and Fleiss' Kappa. RESULTS: According to Friedman analyses, most features (>60%) showed significant differences. Yet, CV and inter-rater reliability measures indicated higher robustness. ComBat resulted in almost complete harmonization (>87%) according to Friedman test and little to no improvement according to CV and inter-rater reliability measures. [18F]GE-180 features were more sensitive to reconstruction settings than [18F]FET features. CONCLUSIONS: According to Friedman test, feature distributions could be successfully aligned using ComBat. However, depending on settings, changes in patient ranks were observed for some features and could not be eliminated by harmonization. Thus, for clinical utilization it is recommended to exclude affected features.