Re-convolving the compositional landscape of primary and recurrent glioblastoma reveals prognostic and targetable tissue states

Al‐Dalahmah, Osama; Argenziano, Michael; Kannan, Adithya; Mahajan, Aayushi; Furnari, Julia; Paryani, Fahad; Boyett, Deborah; Save, Akshay; Humala, Nelson; Khan, Fatima; Li, Juncheng; Lü, Hong; Sun, Yu; Tuddenham, John; Goldberg, Alexander; Dovas, Athanassios; Banu, Matei; Sudhakar, Tejaswi; Bush, Erin C.; Lassman, Andrew B.; McKhann, Guy M.; Gill, Brian J A; Youngerman, Brett E.; Sisti, Michael B.; Bruce, Jeffrey N.; Sims, Peter A.; Menon, Vilas; Canoll, Peter

doi:10.1038/s41467-023-38186-1

Cited by 21 publications

(30 citation statements)

References 68 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…This confirmed relationship (2). It is worth emphasizing that while these relationships were demonstrated using biopsy samples, they were presumed to exist in unlabeled samples too, because these relationships were supported by fundamental knowledge about the spatial landscape of neuropathological alterations in GBM (30).…”

Section: Resultsmentioning

confidence: 98%

“…Specifically, our domain knowledge, based on fundamental understanding of the neuropathological relationships between the three gene modules (30), reveals two key relationships between Pro and Inf conditional on the status of Neu ( Fig. 2a ): (1) We know that genes in the Neu module are enriched in normal brain tissue and depleted in lesional brain tissue.…”

Section: Resultsmentioning

confidence: 99%

“…This study involved a developmental cohort (A) and a test cohort (B). Cohort A was acquired as part of a retrospective, observational study designed to study patients who had undergone repeat surgical resection for recurrence of high-grade gliomas following chemotherapy and radiation therapy (30). Cohort B was acquired as part of a prospective, clinical trial for convection enhanced delivery of topotecan chemotherapy (46).…”

Section: Methodsmentioning

confidence: 99%

“…A variety of different clustering algorithms were used, which repeatedly found the optimal cluster number to be two or three. Considering a recent publication by our group that described a three-tissue-state model of glioma (30), we ultimately decided to proceed with the three-cluster result from hierarchical clustering with Ward's linkage. Three major clusters with mutually exclusive genes were identified, and these gene clusters were used as gene sets for downstream Gene Set Variation Analysis (GSVA) on a sample-by-sample basis, which produced three gene set enrichment scores for each MRI-localized biopsy.…”

Section: Patient Inclusion and Biopsy Acquisition This Study Involved...mentioning

confidence: 99%

“…The three gene modules identified through gene ontology analysis include: proliferative ( Pro ), associated with proliferation and cell cycle ontologies indicative of recurrent tumor; inflammatory ( Inf ), linked to cytokine production and immune response, representative of treatment-induced reactive cells; and neuronal ( Neu ), related to neuronal signaling, reflecting infiltrated brain tissue. Assessing the gene modules of GBM has significant clinical value and has drawn much attention recently (30). For recGBM, the ability to differentiate proliferative/recurrent tumor and treatment-induced reactive/inflammatory cells (two primary gene modules in our dataset) is crucial for evaluating treatment effectiveness.…”

Section: Introductionmentioning

confidence: 99%

See 4 more Smart Citations

Biologically-informed deep neural networks provide quantitative assessment of intratumoral heterogeneity in post-treatment glioblastoma

Wang

Argenziano

Yoon

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

Intratumoral heterogeneity presents a major challenge to diagnosis and treatment of glioblastoma (GBM). Such heterogeneity is further exacerbated upon the recurrence of GBM, where treatment-induced reactive changes produce additional intratumoral heterogeneity that is ambiguous to differentiate on clinical imaging. There is an urgent need to develop non-invasive approaches to map the heterogeneous landscape of histopathological alterations throughout the entire lesion for each patient. We propose to predictively fuse MRI with the underlying intratumoral heterogeneity in recurrent GBM using machine learning (ML) by leveraging unique image-localized biopsies with their associated locoregional MRI features. To this end, we develop BioNet, a biologically informed multi-task framework combining Bayesian neural networks and semi-supervised adversarial autoencoders, to predict regional distributions of three tissue-specific gene modules: proliferating tumor, reactive/inflammatory cells, and infiltrated brain tissue. BioNet provides insight into how to integrate implicit and hierarchical domain knowledge, which is difficult to incorporate into ML models through existing methods. The proposed architecture further addresses challenges in exploiting latent feature structures from limited labeled image-localized biopsy samples, which lead to improvements in prediction accuracy. BioNet performs signiﬁcantly better than existing methods on cross-validation and blind test datasets, shows generalizability that surpasses other models, and is adaptable to different types of data or tasks. Prediction maps of gene modules from BioNet provide accurate predictions of intratumoral heterogeneity, which can improve surgical planning and localization of diagnostic biopsies, as well as inform neuro-oncological treatment assessment for each patient. These results also highlight the emerging role of ML in precision medicine.

show abstract

Section: Resultsmentioning

confidence: 98%

Section: Resultsmentioning

confidence: 99%

Section: Methodsmentioning

confidence: 99%

Section: Patient Inclusion and Biopsy Acquisition This Study Involved...mentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

See 3 more Smart Citations

Biologically-informed deep neural networks provide quantitative assessment of intratumoral heterogeneity in post-treatment glioblastoma

Wang

Argenziano

Yoon

et al. 2022

Preprint

Self Cite

View full text Add to dashboard Cite

show abstract

Genetic and epigenetic instability as an underlying driver of progression and aggressive behavior in IDH-mutant astrocytoma

Richardson,

Walker,

Hambardzumyan

et al. 2024

Acta Neuropathol

View full text Add to dashboard Cite

In recent years, the classification of adult-type diffuse gliomas has undergone a revolution, wherein specific molecular features now represent defining diagnostic criteria of IDH-wild-type glioblastomas, IDH-mutant astrocytomas, and IDH-mutant 1p/19q-codeleted oligodendrogliomas. With the introduction of the 2021 WHO CNS classification, additional molecular alterations are now integrated into the grading of these tumors, given equal weight to traditional histologic features. However, there remains a great deal of heterogeneity in patient outcome even within these established tumor subclassifications that is unexplained by currently codified molecular alterations, particularly in the IDH-mutant astrocytoma category. There is also significant intercellular genetic and epigenetic heterogeneity and plasticity with resulting phenotypic heterogeneity, making these tumors remarkably adaptable and robust, and presenting a significant barrier to the design of effective therapeutics. Herein, we review the mechanisms and consequences of genetic and epigenetic instability, including chromosomal instability (CIN), microsatellite instability (MSI)/mismatch repair (MMR) deficits, and epigenetic instability, in the underlying biology, tumorigenesis, and progression of IDH-mutant astrocytomas. We also discuss the contribution of recent high-resolution transcriptomics studies toward defining tumor heterogeneity with single-cell resolution. While intratumoral heterogeneity is a well-known feature of diffuse gliomas, the contribution of these various processes has only recently been considered as a potential driver of tumor aggressiveness. CIN has an independent, adverse effect on patient survival, similar to the effect of histologic grade and homozygous CDKN2A deletion, while MMR mutation is only associated with poor overall survival in univariate analysis but is highly correlated with higher histologic/molecular grade and other aggressive features. These forms of genomic instability, which may significantly affect the natural progression of these tumors, response to therapy, and ultimately clinical outcome for patients, are potentially measurable features which could aid in diagnosis, grading, prognosis, and development of personalized therapeutics.

show abstract

Spatial transcriptomics in cancer research and potential clinical impact: a narrative review

Cilento,

Sweeney,

Butler

2024

J Cancer Res Clin Oncol

View full text Add to dashboard Cite

Spatial transcriptomics (ST) provides novel insights into the tumor microenvironment (TME). ST allows the quantification and illustration of gene expression profiles in the spatial context of tissues, including both the cancer cells and the microenvironment in which they are found. In cancer research, ST has already provided novel insights into cancer metastasis, prognosis, and immunotherapy responsiveness. The clinical precision oncology application of next-generation sequencing (NGS) and RNA profiling of tumors relies on bulk methods that lack spatial context. The ability to preserve spatial information is now possible, as it allows us to capture tumor heterogeneity and multifocality. In this narrative review, we summarize precision oncology, discuss tumor sequencing in the clinic, and review the available ST research methods, including seqFISH, MERFISH (Vizgen), CosMx SMI (NanoString), Xenium (10x), Visium (10x), Stereo-seq (STOmics), and GeoMx DSP (NanoString). We then review the current ST literature with a focus on solid tumors organized by tumor type. Finally, we conclude by addressing an important question: how will spatial transcriptomics ultimately help patients with cancer?

show abstract

Re-convolving the compositional landscape of primary and recurrent glioblastoma reveals prognostic and targetable tissue states

Cited by 21 publications

References 68 publications

Biologically-informed deep neural networks provide quantitative assessment of intratumoral heterogeneity in post-treatment glioblastoma

Biologically-informed deep neural networks provide quantitative assessment of intratumoral heterogeneity in post-treatment glioblastoma

Genetic and epigenetic instability as an underlying driver of progression and aggressive behavior in IDH-mutant astrocytoma

Spatial transcriptomics in cancer research and potential clinical impact: a narrative review

Contact Info

Product

Resources

About