Search

OBJECTIVE The objective of this study was to assess and compare the potential of 5-aminolevulinic acid (5-ALA) and Raman spectroscopy (RS) in detecting tumor-infiltrated brain in patients with glioblastoma (GBM). METHODS Between July 2020 and October 2021, the authors conducted a prospective clinical trial with 15 patients who underwent neurosurgical treatment of newly diagnosed and histologically verified GBM. A solid contrast-enhancing tumor core and peritumoral tissue were investigated intraoperatively for cancer cells by using 5-ALA and RS to achieve pathology-tailored maximum resection. In each case, a minimum of 10 biopsies were sampled from navigation-guided areas. Two neuropathologists examined the biopsies for the presence of neoplastic cells. The detection performance of 5-ALA and RS alone and in combination was assessed. Pre- and postoperative MRI, Karnofsky Performance Status (KPS), and National Institutes of Health Stroke Scale (NIHSS) scores were compared, and median progression-free survival (PFS) was evaluated. RESULTS A total of 185 biopsy samples were harvested from the contrast-enhancing tumor core (n = 19) and peritumoral tissue (n = 166). In the tumor core, 5-ALA and RS each showed a sensitivity of 100%. In the peritumoral tissue, 5-ALA was less sensitive than RS in detecting cancer (46% vs 69%) but showed higher specificity (81% vs 57%). When the two methods were combined, the accuracy of tumor detection was increased by about 10%. Pathology-tailored resection led to a 52% increase in resection volume comparing the volume of preoperative contrast enhancement with the postoperative resection cavity on MRI (p = 0.0123). Eloquent brain involvement prevented gross-total resection in 4 patients. Four weeks after surgery, mean KPS (p = 0.7637) and NIHSS scores (p = 0.3146) were not significantly different from preoperative values. Of the 13 patients who had received postoperative chemoradiotherapy, 4 did not show any progression after a median follow-up of 14 months. The remaining 9 patients had a median PFS of 8 months. CONCLUSIONS According to the study data, RS is capable of detecting tumor-infiltrated brain with higher sensitivity but lower specificity than the current standard of 5-ALA. With further technological and workflow advancements, RS in combination with protoporphyrin IX fluorescence may contribute to pathology-tailored glioma resection in the future.

OBJECTIVE Intraoperative neuropathological assessment with conventional frozen sections supports the neurosurgeon in optimizing the surgical strategy. However, preparation and review of frozen sections can take as long as 45 minutes. Stimulated Raman histology (SRH) was introduced as a novel technique to provide rapid high-resolution digital images of unprocessed tissue samples directly in the operating room that are comparable to conventional histopathological images. Additionally, SRH images are simultaneously and easily accessible for neuropathological judgment. Recently, the first study showed promising results regarding the accuracy and feasibility of SRH compared with conventional histopathology. Thus, the aim of this study was to compare SRH with conventional H&E images and frozen sections in a large cohort of patients with different suspected central nervous system (CNS) tumors. METHODS The authors included patients who underwent resection or stereotactic biopsy of suspected CNS neoplasm, including brain and spinal tumors. Intraoperatively, tissue samples were safely collected and SRH analysis was performed directly in the operating room. To enable optimal comparison of SRH with H&E images and frozen sections, the authors created a digital databank that included images obtained with all 3 imaging modalities. Subsequently, 2 neuropathologists investigated the diagnostic accuracy, tumor cellularity, and presence of diagnostic histopathological characteristics (score 0 [not present] through 3 [excellent]) determined with SRH images and compared these data to those of H&E images and frozen sections, if available. RESULTS In total, 94 patients with various suspected CNS tumors were included, and the application of SRH directly in the operating room was feasible in all cases. The diagnostic accuracy based on SRH images was 99% when compared with the final histopathological diagnosis based on H&E images. Additionally, the same histopathological diagnosis was established in all SRH images (100%) when compared with that of the corresponding frozen sections. Moreover, the authors found a statistically significant correlation in tumor cellularity between SRH images and corresponding H&E images (p < 0.0005 and R = 0.867, Pearson correlation coefficient). Finally, excellent (score 3) or good (2) accordance between diagnostic histopathological characteristics and H&E images was present in 95% of cases. CONCLUSIONS The results of this retrospective analysis demonstrate the near-perfect diagnostic accuracy and capability of visualizing relevant histopathological characteristics with SRH compared with conventional H&E staining and frozen sections. Therefore, digital SRH histopathology seems especially useful for rapid intraoperative investigation to confirm the presence of diagnostic tumor tissue and the precise tumor entity, as well as to rapidly analyze multiple tissue biopsies from the suspected tumor margin. A real-time analysis comparing SRH images and conventional histological images at the time of surgery should be performed as the next step in future studies.

Deep learning is increasingly utilized in medical image analysis and shows promise to substantially inform treatment decisions in clinical practice. While it finds broad applications in radiology or genomics, there are still relatively few applications of deep learning to the analysis of histological images. A disease in which a better understanding of underlying phenotype-genotype correlations seems crucial is glioblastoma, the most common malignant brain tumor in adults that features a dismal prognosis and limited therapeutic options. Using a large cohort of 276 glioblastoma patients with available histological, immunohistochemical, clinical and molecular data, we trained an Xception convolutional neural network to predict patient survival and transcriptional subtypes on histological images alone. We used 5-fold cross validation to evaluate the performance of the trained network. We achieve a high accuracy for predicting overall survival on univariable (p < 0.001) and multivariable (p = 0.01) analysis. Moreover, we observe a high accuracy for predicting the predominant transcriptional subtype with an AUC of 0.84, 0.81, and 0.71 for classical, mesenchymal and proneural subtypes, respectively (all p < 0.001). Finally, we plot heatmaps of the predicted transcriptional subtype and risk scores to directly link them to the histological phenotypes and the tumor microenvironment. Reassuringly, we find an association of the mesenchymal subtype with the perinecrotic niche and of the proneural subtype with the infiltrating edge. In conclusion, we introduce an easily adaptable convolutional neural network model for working with histological data, which can accurately predict outcomes and transcriptional subtypes in glioblastoma.

OBJECTIVE Early markers are urgently needed in low-grade glioma (LGG) evaluation to rapidly estimate the individual patient’s prognosis and to determine the optimal postoperative management. Generally, visible 5-aminolevulinic acid (5-ALA) fluorescence is present in only a few LGGs. Recently, the authors identified visible 5-ALA fluorescence as a powerful intraoperative marker for unfavorable outcome in LGG treatment. However, its precise histopathological correlate is unclear. Neoangiogenesis represents a crucial event in tumor evolution, and CD34 is an established marker for vascular endothelial progenitors potentially indicating tumor progression. The aim of this study was thus to correlate 5-ALA fluorescence and CD34 microvascularity as well as to investigate the prognostic value of CD34 in a large series of LGGs. METHODS In this retrospective study including 3 specialized centers, patients with histopathologically confirmed isocitrate dehydrogenase–mutated LGGs (WHO grade II) receiving 5-ALA prior to resection were included. During surgery, the presence of visible fluorescence was analyzed and one representative tumor sample from the area with the maximum fluorescence effect (tumor with focal fluorescence or nonfluorescing tumor) was selected for each LGG. All fluorescing or nonfluorescing tumor samples were stained for CD34 and semiquantitatively analyzed for microvascular proliferation patterns (physiological vessels, branching capillaries, or microvessel clusters) as well as automatically quantified for CD34 microvessel density (MVD) by standardized histomorphometry software. These semiquantitative/quantitative CD34 data were correlated to the fluorescence status and patient outcome including progression-free survival (PFS), malignant transformation–free survival (MTFS), and overall survival (OS). RESULTS In a total of 86 LGGs, visible fluorescence was found during surgery in 13 (15%) cases. First, the semiquantitative CD34 score significantly correlated with intraoperative fluorescence (p = 0.049). Accordingly, the quantitative CD34 MVD was significantly higher in tumors showing fluorescence (p = 0.03). Altogether, the semiquantitative CD34 score showed a strong correlation with quantitative CD34 MVD (p < 0.001). At a mean follow-up of 5.4 ± 2.6 years, microvessel clusters in semiquantitative analysis were a prognostic marker for poor PFS (p = 0.01) and MTFS (p = 0.006), but not OS (p = 0.28). Finally, quantitative CD34 MVD > 10 vessels/mm2 was a prognostic marker for poor PFS (p = 0.01), MTFS (p = 0.008), and OS (p = 0.049). CONCLUSIONS The data indicate that CD34 microvascularity is associated with intraoperative 5-ALA fluorescence and outcomes in patients with LGG. Thus, visible fluorescence in LGGs might indicate increased CD34 microvascularity, serving as an early prognostic marker for unfavorable patient outcome that is already available during surgery.

(1) Background: Recent developments in 7T magnetic resonance spectroscopic imaging (MRSI) made the acquisition of high-resolution metabolic images in clinically feasible measurement times possible. The amino acids glutamine (Gln) and glycine (Gly) were identified as potential neuro-oncological markers of importance. For the first time, we compared 7T MRSI to amino acid PET in a cohort of glioma patients. (2) Methods: In 24 patients, we co-registered 7T MRSI and routine PET and compared hotspot volumes of interest (VOI). We evaluated dice similarity coefficients (DSC), volume, center of intensity distance (CoI), median and threshold values for VOIs of PET and ratios of total choline (tCho), Gln, Gly, myo-inositol (Ins) to total N-acetylaspartate (tNAA) or total creatine (tCr). (3) Results: We found that Gln and Gly ratios generally resulted in a higher correspondence to PET than tCho. Using cutoffs of 1.6-times median values of a control region, DSCs to PET were 0.53 ± 0.36 for tCho/tNAA, 0.66 ± 0.40 for Gln/tNAA, 0.57 ± 0.36 for Gly/tNAA, and 0.38 ± 0.31 for Ins/tNAA. (4) Conclusions: Our 7T MRSI data corresponded better to PET than previous studies at lower fields. Our results for Gln and Gly highlight the importance of future research (e.g., using Gln PET tracers) into the role of both amino acids.

The clinical impact of 5-aminolevulinic acid (5-ALA) fluorescence during resection of brain metastases is not yet clear.. Recent data demonstrated significantly lower incidence of visible fluorescence in cerebral melanoma metastases (CMM) compared to other brain metastases (BM). The aim of this study was to investigate if characteristic melanoma features such as pigmentation, intratumoural hemosiderin and bleeding have an influence on visible fluorescence in CMM.A retrospective study of two neurosurgical centers was performed including adult patients with resection of CMM after preoperative administration of 5-ALA. Data on the fluorescence status (visible or no fluorescence), the fluorescence quality (strong, vague, none) and fluorescence homogeneity (homogeneous or heterogeneous) of each CMM were collected. The amount of melanin, hemosiderin and intratumoural bleeding was semi-quantitatively determined and automated computer-based calculation of the relative pigmented area was performed in fluorescing and non-fluorescing CMM samples.Altogether, 29 CMM were surgically removed after 5-ALA administration. Visible fluorescence was detected in 8 CMM (28%), whereas no fluorescence was detected in 21 CMM (72%). In detail, 3 tumors (10%) showed strong fluorescence, 5 tumors (17%) revealed vague fluorescence and in 21 tumors (72%) no fluorescence was found. In total, 8 fluorescing and 25 non-fluorescing CMM samples were investigated. According to the semi-quantitatively calculated fluorescence status, no statistically significant difference in the median amount of melanin (p = 0.242), hemosiderin (p = 0.603) and bleeding (p = 0.762) between CMM samples with and without visible fluorescence was found. Moreover, the automatically assessed relative pigmented area did not show a statistically significant difference between samples with visible and no fluorescence (p = 0.966).Our data indicate that 5-ALA fluorescence is not dependent on the amount of pigmentation, intratumoural hemosiderin and bleeding in CMM. We thus assume that other factors are responsible for the low rate of visible fluorescence in CMM.

Currently, approximately 150 different brain tumour types are defined by the WHO. Recent endeavours to exploit machine learning and deep learning methods for supporting more precise diagnostics based on the histological tumour appearance have been hampered by the relative paucity of accessible digital histopathological datasets. While freely available datasets are relatively common in many medical specialties such as radiology and genomic medicine, there is still an unmet need regarding histopathological data. Thus, we digitized a significant portion of a large dedicated brain tumour bank based at the Division of Neuropathology and Neurochemistry of the Medical University of Vienna, covering brain tumour cases from 1995–2019. A total of 3,115 slides of 126 brain tumour types (including 47 control tissue slides) have been scanned. Additionally, complementary clinical annotations have been collected for each case. In the present manuscript, we thoroughly discuss this unique dataset and make it publicly available for potential use cases in machine learning and digital image analysis, teaching and as a reference for external validation.