Integrating morphologic and molecular histopathological features through whole slide image registration and deep learning

Background Modern molecular pathology workflows in neuro-oncology heavily rely on the integration of morphologic and immunohistochemical patterns for analysis, classification, and prognostication. However, despite the recent emergence of digital pathology platforms and artificial intelligence-driven computational image analysis tools, automating the integration of histomorphologic information found across these multiple studies is challenged by large files sizes of whole slide images (WSIs) and shifts/rotations in tissue sections introduced during slide preparation. Methods To address this, we develop a workflow that couples different computer vision tools including scale-invariant feature transform (SIFT) and deep learning to efficiently align and integrate histopathological information found across multiple independent studies. We highlight the utility and automation potential of this workflow in the molecular subclassification and discovery of previously unappreciated spatial patterns in diffuse gliomas. Results First, we show how a SIFT-driven computer vision workflow was effective at automated WSI alignment in a cohort of 107 randomly selected surgical neuropathology cases (97/107 (91%) showing appropriate matches, AUC = 0.96). This alignment allows our AI-driven diagnostic workflow to not only differentiate different brain tumor types, but also integrate and carry out molecular subclassification of diffuse gliomas using relevant immunohistochemical biomarkers (IDH1-R132H, ATRX). To highlight the discovery potential of this workflow, we also examined spatial distributions of tumors showing heterogenous expression of the proliferation marker MIB1 and Olig2. This analysis helped uncovered an interesting and unappreciated association of Olig2 positive and proliferative areas in some gliomas (r = 0.62). Conclusion This efficient neuropathologist-inspired workflow provides a generalizable approach to help automate a variety of advanced immunohistochemically compatible diagnostic and discovery exercises in surgical neuropathology and neuro-oncology.

Tumor Mutation Burden, Expressed Neoantigens and the Immune Microenvironment in Diffuse Gliomas

Background: A consistent correlation between tumor mutation burden (TMB) and tumor immune microenvironment has not been observed in gliomas as in other cancers. Methods: Driver germline and somatic mutations, TMB, neoantigen, and immune cell signatures were analyzed using whole exome sequencing (WES) and transcriptome sequencing of tumor and WES of matched germline DNA in a cohort of 66 glioma samples (44 IDH-mutant and 22 IDH-wildtype). Results: Fourteen samples revealed a hypermutator phenotype (HMP). Eight pathogenic (P) or likely pathogenic (LP) germline variants were detected in 9 (19%) patients. Six of these 8 genes were DNA damage repair genes. P/LP germline variants were found in 22% of IDH-mutant gliomas and 12.5% of IDH-wildtype gliomas (p = 0.7). TMB was correlated with expressed neoantigen but showed an inverse correlation with immune score (R = −0.46, p = 0.03) in IDH-wildtype tumors and no correlation in IDH-mutant tumors. The Antigen Processing and Presentation (APP) score correlated with immune score and was surprisingly higher in NHMP versus HMP samples in IDH-wildtype gliomas, but higher in HMP versus NHMP in IDH-mutant gliomas. Conclusion: TMB was inversely correlated with immune score in IDH-wildtype gliomas and showed no correlation in IDH-mutant tumors. APP was correlated with immune score and may be further investigated as a biomarker for response to immunotherapy in gliomas. Studies of germline variants in a larger glioma cohort are warranted.

NI-10 Reclassification of diffuse gliomas based on molecular diagnosis -evaluation of methionine uptake and treatment outcome-

Object: The revised 2016 WHO Classification of Tumours of the Central Nervous System incorporates genetic alterations into the classification system, with the goal of creating more homogenous disease categories with greater prognostic value. In this study, we reclassified diffuse gliomas with molecular diagnosis and examined for 11C-methionine uptake and prognosis. Methods. 182 diffuse glioma patients (Grade II in 42 patients, Grade III in 61 patients, Grade IV in 77 patients) treated at Tokyo Medical and Dental University Hospital from 2000 to 2018 were included in this study. The IDH1/2, ATRX and 1p19q status were analyzed using tumor samples. The tumor-to-normal ratio (T/N) of 11 C-methionine uptake was calculated by dividing the mean standardized uptake value (SUV) for the tumor by the mean SUV of the normal brain. Result. By molecular diagnosis, 11 diffuse astrocytomas and 17 anaplastic astrocytomas were diagnosed as “IDH-mutant”, while 14 diffuse astrocytomas and 29 anaplastic astrocytomas were diagnosed as “IDH-wild”. 5 out of 77 grade IV tumors had IDH mutation. 4 tumors were diagnosed as “Diffuse midline glioma, H3 K27M-mutant”. In the 32 oligodendroglial tumors, 12 oligodendrogliomas and 9 anaplastic oligodendrogliomas were diagnosed as “IDH-mutant and 1p/19q-codeleted”. The median T/N ratios in oligodendroglial tumors with “IDH-mutant and 1p/19q-codeleted” were significantly higher than those in astrocytic tumors with “IDH-mutant”. On the other hand, in tumors with the same genetic background, higher grade tumor has significant higher T/N ratio. Kaplan-Meier survival analysis revealed that oligodendroglial tumors with “IDH-mutant and 1p/19q-codeleted” had significantly better outcomes regardless of WHO grade. Overall survival was 90.9% at 5 years and 77.9% at 10 years in oligodendroglial tumors with “IDH-mutant and 1p/19q-codeleted”. IDH wild tumors had significantly worse outcomes.Conclusions. The results indicated that diffuse glioma categories reclassified with molecular classification correlate with the T/N ratio of methionine and the prognosis.

Machine learning modeling of genome-wide copy number alteration signatures reliably predicts IDH mutational status in adult diffuse glioma

Knowledge of 1p/19q-codeletion and IDH1/2 mutational status is necessary to interpret any investigational study of diffuse gliomas in the modern era. While DNA sequencing is the gold standard for determining IDH mutational status, genome-wide methylation arrays and gene expression profiling have been used for surrogate mutational determination. Previous studies by our group suggest that 1p/19q-codeletion and IDH mutational status can be predicted by genome-wide somatic copy number alteration (SCNA) data alone, however a rigorous model to accomplish this task has yet to be established. In this study, we used SCNA data from 786 adult diffuse gliomas in The Cancer Genome Atlas (TCGA) to develop a two-stage classification system that identifies 1p/19q-codeleted oligodendrogliomas and predicts the IDH mutational status of astrocytic tumors using a machine-learning model. Cross-validated results on TCGA SCNA data showed near perfect classification results. Furthermore, our astrocytic IDH mutation model validated well on four additional datasets (AUC = 0.97, AUC = 0.99, AUC = 0.95, AUC = 0.96) as did our 1p/19q-codeleted oligodendroglioma screen on the two datasets that contained oligodendrogliomas (MCC = 0.97, MCC = 0.97). We then retrained our system using data from these validation sets and applied our system to a cohort of REMBRANDT study subjects for whom SCNA data, but not IDH mutational status, is available. Overall, using genome-wide SCNAs, we successfully developed a system to robustly predict 1p/19q-codeletion and IDH mutational status in diffuse gliomas. This system can assign molecular subtype labels to tumor samples of retrospective diffuse glioma cohorts that lack 1p/19q-codeletion and IDH mutational status, such as the REMBRANDT study, recasting these datasets as validation cohorts for diffuse glioma research.