arxiv: 2605.10278 · v1 · submitted 2026-05-11 · 💻 cs.LG

Recognition: 2 theorem links

· Lean Theorem

Predictive Radiomics for Evaluation of Cancer Immune SignaturE in Glioblastoma: the PRECISE-GBM study

Prajwal Ghimire , Junjie Li , Liu Yaou , Marc Modat , Thomas Booth

Authors on Pith no claims yet

Pith reviewed 2026-05-12 04:55 UTC · model grok-4.3

classification 💻 cs.LG

keywords radiogenomicsglioblastomaradiomicsimmune signaturemachine learningIDH-wildtypemacrophageMRI

0 comments

The pith

Radiogenomic models non-invasively predict the M0 macrophage immune signature in IDH-wildtype glioblastoma.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper develops radiogenomic models to identify radiological biomarkers for immune cell signatures in the glioblastoma tumor microenvironment. It extracts radiomic features from necrotic core, enhancing tumor, and edema regions of deep-learning auto-segmented MRI scans across multiple open datasets. These features train support vector machine and ensemble classifiers to predict seventeen immune scores derived from transcriptomic deconvolution using reference immune signature matrices. The models achieve stable performance specifically for the M0 macrophage subtype across held-out cohorts, with ensemble methods outperforming single classifiers. A sympathetic reader would care because the approach offers a potential non-invasive complement to tissue-based assessment for guiding immunotherapy in a disease where prognosis remains poor.

Core claim

Radiogenomic models non-invasively predicted the macrophage subtype M0 immune signature in IDH-wildtype glioblastoma. Radiomic features selected via nested cross-validated LASSO from shape, first-order, and higher-order statistics in auto-segmented MRI regions were used to train classifiers on transcriptomic-derived immune labels; these models maintained mean balanced accuracy of 0.67 and precision of 0.89 on three independent holdout datasets, with ensemble models outperforming support vector machines.

What carries the argument

LASSO-selected radiomic features from deep-learning auto-segmented necrotic, enhancing, and edema regions fed into support vector machine and ensemble classifiers trained against deconvoluted transcriptomic immune signatures.

Load-bearing premise

Radiomic features extracted from deep-learning auto-segmented MRI regions reliably capture the underlying immune cell infiltration as represented by transcriptomic deconvolution labels.

What would settle it

A new prospective cohort in which pre-operative MRI radiomic predictions are compared directly to matched post-operative biopsy transcriptomic immune signatures and show low correlation for the M0 macrophage label.

Figures

Figures reproduced from arXiv: 2605.10278 by Junjie Li, Liu Yaou, Marc Modat, Prajwal Ghimire, Thomas Booth.

read the original abstract

Background: Radiogenomics allows identification of radiological biomarkers for genomic phenotypes. In glioblastoma, these biomarkers could potentially complement patient stratification strategies. We aim to develop and analytically validate radiological biomarkers that capture immune cell signatures within IDH-wildtype glioblastoma microenvironment using radiogenomic analysis. Methods: This was a retrospective multicenter study using curated open-access anonymized imaging and genomic data from TCGA-GBM, CPTAC, IvyGAP, REMBRANDT and CGGA datasets. Imaging data consisted of MRI-based radiomic features extracted from necrotic core, enhancing and edema regions of deep learning-based auto-segmented tumors. Radiomic feature selections were performed using nested cross-validated LASSO. Support vector machine and ensemble models were trained using seventeen immune and cell-specific score labels extracted from deconvoluted transcriptomic data using pan-cancer and glioblastoma immune signature matrices as reference standards. Seventeen classifier models trained in three cross-cohort strategies were validated on three held-out datasets assessing stability and generalizability. Results: One-hundred-and-seventy-six patients were included in the study. The immune-related radiomic signatures obtained after feature selection were shape, first order and higher order radiomic features. Models predicting macrophage subtype immune signature showed stable mean performance on balanced accuracy (0.67) and precision (0.89) metrics for three independent holdout datasets with ensemble model outperforming support vector machine model. Conclusion: Radiogenomic models non-invasively predicted the macrophage subtype M0 immune signature in IDH-wildtype glioblastoma. These biomarkers have the potential to stratify patients for immunotherapy within prospective glioblastoma clinical trials.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper shows stable cross-cohort radiomics prediction of the M0 macrophage signature in GBM at 0.67 balanced accuracy using public data and nested CV, but auto-segmentation quality is unvalidated and performance remains modest.

read the letter

The main point for you is that this group trained SVM and ensemble models on radiomic features from deep-learning auto-segmented MRI subregions to predict seventeen immune signatures derived from transcriptomic deconvolution in IDH-wildtype glioblastoma. They report the strongest result for the M0 macrophage label, with mean balanced accuracy of 0.67 and precision of 0.89 holding up across three independent hold-out datasets after nested cross-validated LASSO feature selection and three different cross-cohort training schemes. The surviving features are mostly shape, first-order, and texture measures from necrotic core, enhancing tumor, and edema compartments. That is the concrete output. The work does a few things cleanly. It pulls together five public cohorts (TCGA-GBM, CPTAC, IvyGAP, REMBRANDT, CGGA), uses standard but careful nested CV to limit overfitting, and demonstrates that the ensemble beats the single SVM while performance stays consistent on held-out sets. For a retrospective radiogenomics study that is a reasonable level of validation and shows the authors took generalizability seriously. The soft spots are real but not fatal. The entire pipeline rests on auto-segmented tumor subregions with no reported Dice scores, manual comparison, or inter-observer metrics on the multi-center data. GBM boundaries are ambiguous enough that even small systematic shifts in segmentation can change the very features that survived LASSO; cross-validation and hold-out tests cannot detect that upstream bias. The 0.67 accuracy is also modest for the central claim, and the abstract gives no error bars, exact regularization values, or clear baseline comparisons. Those gaps make it harder to judge how much of the signal is genuine radiogenomic correlation versus segmentation artifact. This paper is for radiogenomics groups working on brain tumors who want to see a multi-cohort application to immune deconvolution labels. It is not a methodological advance but supplies targeted evidence that could be useful for planning follow-up studies with better segmentation or imaging modalities. I would send it to peer review. The public data, the hold-out strategy, and the cross-cohort testing are solid enough that referees can usefully ask for the missing segmentation checks and tighter performance reporting. It is not ready for acceptance as written, but it deserves that level of attention rather than a desk reject.

Referee Report

2 major / 2 minor

Summary. The paper develops and validates radiogenomic models to predict immune cell signatures (particularly the M0 macrophage subtype) in IDH-wildtype glioblastoma using MRI radiomic features extracted from necrotic core, enhancing tumor, and edema regions defined by deep-learning auto-segmentation. It employs public multi-center datasets (TCGA-GBM, CPTAC, IvyGAP, REMBRANDT, CGGA), nested cross-validated LASSO for feature selection on 17 transcriptomic deconvolution-derived labels, and trains SVM plus ensemble classifiers, reporting stable hold-out performance (balanced accuracy 0.67, precision 0.89 for M0) across three independent validation sets.

Significance. If the central claim holds, the work offers a reproducible pipeline for non-invasive radiogenomic biomarkers of the GBM immune microenvironment that could aid stratification in immunotherapy trials. Strengths include use of public datasets, nested CV to mitigate overfitting, and explicit hold-out stability testing across cohorts. The modest performance and upstream methodological gaps limit immediate translational impact, but the approach is a useful addition to radiogenomics literature if segmentation reliability is demonstrated.

major comments (2)

[Methods] Methods section on image segmentation: The pipeline depends entirely on deep-learning auto-segmentation of necrotic core, enhancing tumor, and peritumoral edema without any reported quantitative validation (Dice/IoU scores, manual comparison, or inter-observer metrics) on the multi-center cohorts. GBM subregion boundaries are ambiguous on MRI; even small delineation errors propagate to shape, first-order, and texture features that survive LASSO selection and enter all downstream SVM/ensemble models. This is load-bearing for the claim that radiomic features capture immune infiltration, as cross-validation and hold-out stability cannot detect systematic segmentation bias.
[Results] Results and abstract: The reported mean balanced accuracy of 0.67 for M0 prediction is modest for a central claim, yet no error bars, exact LASSO regularization values, selected feature counts, or baseline comparisons (e.g., clinical variables or random classifiers) are provided. Without these, it is difficult to assess whether the radiogenomic signal exceeds what could arise from segmentation artifacts or class imbalance.

minor comments (2)

[Abstract] Abstract: The title and abstract use inconsistent capitalization ('SignaturE'); standardize to 'Signature'.
[Methods] Methods: Clarify the exact number of radiomic features extracted per compartment and the final count retained after nested LASSO across the three cross-cohort strategies.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive review. The comments identify important areas for clarification and strengthening of the manuscript. We address each major comment below and outline the corresponding revisions.

read point-by-point responses

Referee: [Methods] Methods section on image segmentation: The pipeline depends entirely on deep-learning auto-segmentation of necrotic core, enhancing tumor, and peritumoral edema without any reported quantitative validation (Dice/IoU scores, manual comparison, or inter-observer metrics) on the multi-center cohorts. GBM subregion boundaries are ambiguous on MRI; even small delineation errors propagate to shape, first-order, and texture features that survive LASSO selection and enter all downstream SVM/ensemble models. This is load-bearing for the claim that radiomic features capture immune infiltration, as cross-validation and hold-out stability cannot detect systematic segmentation bias.

Authors: We acknowledge that the manuscript does not report quantitative segmentation validation metrics (Dice/IoU or inter-observer) specifically on the TCGA-GBM, CPTAC, IvyGAP, REMBRANDT, and CGGA cohorts. The auto-segmentation relied on a published deep-learning model applied uniformly across all datasets. While identical processing reduces some sources of differential bias and the observed stability of model performance across independent cohorts provides indirect evidence of robustness, we agree this does not fully address potential systematic delineation errors. In revision we will (i) cite the original segmentation validation study, (ii) add an explicit limitations paragraph discussing segmentation variability in GBM, and (iii) include a sensitivity analysis on a randomly selected subset of cases where feasible. These changes will be reflected in the Methods and Discussion sections. revision: partial
Referee: [Results] Results and abstract: The reported mean balanced accuracy of 0.67 for M0 prediction is modest for a central claim, yet no error bars, exact LASSO regularization values, selected feature counts, or baseline comparisons (e.g., clinical variables or random classifiers) are provided. Without these, it is difficult to assess whether the radiogenomic signal exceeds what could arise from segmentation artifacts or class imbalance.

Authors: We agree that the current reporting is incomplete. The mean balanced accuracy of 0.67 (with precision 0.89) is indeed modest yet was stable across three fully independent hold-out cohorts; this exceeds the 0.5 expected from a random classifier and is accompanied by high precision, which is clinically relevant for identifying the M0 signature. In the revised manuscript we will: add standard-deviation error bars or cohort-specific ranges for all metrics; report the exact LASSO regularization parameters and the number of features retained after nested cross-validation for each model; and include baseline comparisons against (a) a random classifier and (b) models using only clinical variables (age, sex, KPS) where available. These additions will appear in the Results, supplementary tables, and abstract as appropriate. revision: yes

Circularity Check

0 steps flagged

No circularity: standard supervised radiogenomic mapping from independent transcriptomic labels

full rationale

The paper extracts radiomic features from DL-auto-segmented MRI subregions, applies nested LASSO feature selection, and trains SVM/ensemble classifiers to predict 17 immune scores obtained via separate transcriptomic deconvolution on held-out cohorts. This is a conventional supervised pipeline; the target labels are generated externally from RNA data and are not redefined or fitted from the radiomic inputs. No self-citations, uniqueness theorems, or ansatzes are used to justify the core mapping, and cross-cohort validation prevents reduction by construction. The derivation chain remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

1 free parameters · 2 axioms · 0 invented entities

The central claim rests on the domain assumption that imaging features correlate with immune biology and on the validity of deconvolution as ground truth; no new entities are introduced.

free parameters (1)

LASSO regularization strength
Controls feature selection in nested cross-validation; specific value not reported in abstract.

axioms (2)

domain assumption Transcriptomic deconvolution using pan-cancer and GBM signature matrices accurately quantifies immune cell signatures
These scores serve as the training labels for all radiomic classifiers.
domain assumption Radiomic features from necrotic, enhancing, and edema regions reflect biological differences in immune infiltration
Core premise enabling the radiogenomic mapping.

pith-pipeline@v0.9.0 · 5610 in / 1327 out tokens · 58493 ms · 2026-05-12T04:55:34.427478+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Models predicting macrophage subtype immune signature showed stable mean performance on balanced accuracy (0.67) and precision (0.89) metrics for three independent holdout datasets with ensemble model outperforming support vector machine model.
IndisputableMonolith/Foundation/ArithmeticFromLogic.lean LogicNat recovery unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

Radiomic feature selections were performed using nested cross-validated LASSO... Support vector machine and ensemble models were trained using seventeen immune and cell-specific score labels

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

96 extracted references · 96 canonical work pages

[5]

Tumor microenvironment in glioblastoma: Current and emerging concepts

Sharma P, Aaroe A, Liang J, Puduvalli VK. Tumor microenvironment in glioblastoma: Current and emerging concepts. Neurooncol Adv. 2023;5(1):vdad009. Published 2023 Feb 23. doi:10.1093/noajnl/vdad009

work page doi:10.1093/noajnl/vdad009 2023
[6]

Challenges in glioblastoma immunotherapy: mechanisms of resistance and therapeutic approaches to overcome them

Habashy KJ, Mansour R, Moussalem C, Sawaya R, Massaad MJ. Challenges in glioblastoma immunotherapy: mechanisms of resistance and therapeutic approaches to overcome them. Br J Cancer. 2022;127(6):976-987. doi:10.1038/s41416-022-01864-w

work page doi:10.1038/s41416-022-01864-w 2022
[7]

Brain immunology and immunotherapy in brain tumours

Sampson JH, Gunn MD, Fecci PE, Ashley DM. Brain immunology and immunotherapy in brain tumours. Nat Rev Cancer. 2020;20(1):12-25. doi:10.1038/s41568-019-0224-7

work page doi:10.1038/s41568-019-0224-7 2020
[8]

Natural killer cell awakening: unleash cancer - immunity cycle against glioblastoma

Wang M, Zhou Z, Wang X, Zhang C, Jiang X. Natural killer cell awakening: unleash cancer - immunity cycle against glioblastoma. Cell Death Dis. 2022;13(7):588. Published 2022 Jul 8. doi:10.1038/s41419-022-05041-y

work page doi:10.1038/s41419-022-05041-y 2022
[9]

Immunotherapy for Glioblastoma: Current State, Challenges, and Future Perspectives

Yang M, Oh IY , Mahanty A, Jin WL, Yoo JS. Immunotherapy for Glioblastoma: Current State, Challenges, and Future Perspectives. Cancers (Basel). 2020;12(9):2334. Published 2020 Aug

work page 2020
[10]

doi:10.3390/cancers12092334 Downloaded from https://academic.oup.com/noa/advance-article/doi/10.1093/noajnl/vdag115/8666933 by guest on 06 May 2026 NOA-D-25-00604R2 29

work page doi:10.3390/cancers12092334 2026
[11]

Immunotherapy in glioblastoma treatment: Current state and future prospects

Rocha Pinheiro SL, Lemos FFB, Marques HS, Silva Luz M, de Oliveira Silva LG, Faria Souza Mendes Dos Santos C, da Costa Evangelista K, Calmon MS, Sande Loureiro M, Freire de Melo F. Immunotherapy in glioblastoma treatment: Current state and future prospects. World J Clin Oncol. 2023 Apr 24;14(4):138 -159. doi: 10.5306/wjco.v14.i4.138. PMID: 37124134; PMCID...

work page doi:10.5306/wjco.v14.i4.138 2023
[12]

Immunotherapy for glioblastoma: current state, challenges, and future perspectives

Liu Y , Zhou F, Ali H et al. Immunotherapy for glioblastoma: current state, challenges, and future perspectives. Cell Mol Immunol 2024;21:1354–1375. doi:10.1038/s41423-024-01226- x

work page doi:10.1038/s41423-024-01226- 2024
[13]

Neoadjuvant triplet immune checkpoint blockade in newly diagnosed glioblastoma

Long GV , Shklovskaya E, Satgunaseelan L, et al. Neoadjuvant triplet immune checkpoint blockade in newly diagnosed glioblastoma. Nat Med. Published online February 27, 2025. doi:10.1038/s41591-025-03512-1

work page doi:10.1038/s41591-025-03512-1 2025
[14]

A new paradigm for immunotherapy in glioblas toma

Ellenbogen Y , Zadeh G. A new paradigm for immunotherapy in glioblas toma. Nat Med. Published online March 31, 2025. doi:10.1038/s41591-025-03607-9

work page doi:10.1038/s41591-025-03607-9 2025
[15]

Louis, Arie Perry, Pieter Wesseling, Daniel J

Louis DN, Perry A, Wesseling P, et al. The 2021 WHO Classification of Tumors of the Central Nervous System: a summary. Neuro Oncol. 2021;23(8):1231 -1251. doi:10.1093/neuonc/noab106

work page doi:10.1093/neuonc/noab106 2021
[16]

Immune Checkpoint Inhibition for Hypermutant Glioblastoma Multiforme Resulting From Germline Biallelic Mismatch Repair Deficiency

Bouffet E, Larouche V , Campbell BB, et al. Immune Checkpoint Inhibition for Hypermutant Glioblastoma Multiforme Resulting From Germline Biallelic Mismatch Repair Deficiency. J Clin Oncol. 2016;34(19):2206-2211. doi:10.1200/JCO.2016.66.6552

work page doi:10.1200/jco.2016.66.6552 2016
[17]

Immunogenomics of Hypermutated Glioblastoma: A Patient with Germline POLE Deficiency Treated with Checkpoint Blockade Immunotherapy

Johanns TM, Miller CA, Dorward IG, et al. Immunogenomics of Hypermutated Glioblastoma: A Patient with Germline POLE Deficiency Treated with Checkpoint Blockade Immunotherapy. Cancer Discov. 2016;6(11):1230-1236. doi:10.1158/2159-8290.CD-16-0575 Downloaded from https://academic.oup.com/noa/advance-article/doi/10.1093/noajnl/vdag115/8666933 by guest on 06 M...

work page doi:10.1158/2159-8290.cd-16-0575 2016
[18]

State of the neoadjuvant therapy for glioblastoma multiforme -Where do we stand? Neurooncol Adv

Nabian N, Ghalehtaki R , Zeinalizadeh M, Balaña C, Jablonska PA. State of the neoadjuvant therapy for glioblastoma multiforme -Where do we stand? Neurooncol Adv. 2024 Mar 5;6(1):vdae028. doi: 10.1093/noajnl/vdae028. PMID: 38560349; PMCID: PMC10981465

work page doi:10.1093/noajnl/vdae028 2024
[19]

Effect of Nivolumab vs Bevacizumab in Patients With Recurrent Glioblastoma: The CheckMate 143 Phase 3 Randomized Clinical Trial

Reardon DA, Brandes AA, Omuro A, et al. Effect of Nivolumab vs Bevacizumab in Patients With Recurrent Glioblastoma: The CheckMate 143 Phase 3 Randomized Clinical Trial. JAMA Oncol. 2020;6(7):1003-1010. doi:10.1001/jamaoncol.2020.1024

work page doi:10.1001/jamaoncol.2020.1024 2020
[20]

Tumor-associated macrophages: an effective player of the tumor microenvironment

Basak U, Sarkar T, Mukherjee S, Chakraborty S, Dutta A, Dutta S, Nayak D, Kaushik S, Das T, Sa G. Tumor-associated macrophages: an effective player of the tumor microenvironment. Front Immunol. 2023 Nov 16;14:1295257. doi: 10.3389/fimmu.2023.1295257. PMID: 38035101; PMCID: PMC10687432

work page doi:10.3389/fimmu.2023.1295257 2023
[21]

Glioblastoma Eradication Following Immune Checkpoint Blockade in an Orthotopic, Immunocompetent Model

Reardon DA, Gokhale PC, Klein SR, et al. Glioblastoma Eradication Following Immune Checkpoint Blockade in an Orthotopic, Immunocompetent Model. Cancer Immunol Res. 2016;4(2):124-135. doi:10.1158/2326-6066.CIR-15-0151

work page doi:10.1158/2326-6066.cir-15-0151 2016
[22]

Gliomas promote immunosuppression through induction of B7 -H1 expression in tumor -associated macrophages

Bloch O, Crane CA, Kaur R, Safaee M, Rutkowski MJ, Parsa AT. Gliomas promote immunosuppression through induction of B7 -H1 expression in tumor -associated macrophages. Clin Cancer Res. 2013;19(12):3165 -3175. doi:10.1158/1078 -0432.CCR-12- 3314

work page doi:10.1158/1078 2013
[23]

Durable therapeutic efficacy utilizing combinatorial blockade against IDO, CTLA -4, and PD -L1 in mice with brain tumors

Wainwright DA, Chang AL, Dey M, et al. Durable therapeutic efficacy utilizing combinatorial blockade against IDO, CTLA -4, and PD -L1 in mice with brain tumors. Clin Cancer Res. 2014;20(20):5290-5301. doi:10.1158/1078-0432.CCR-14-0514

work page doi:10.1158/1078-0432.ccr-14-0514 2014
[24]

Anti-PD-1 blockade and stereotactic radiation produce long - term survival in mice with intracranial gliomas

Zeng J, See AP, Phallen J, et al. Anti-PD-1 blockade and stereotactic radiation produce long - term survival in mice with intracranial gliomas. Int J Radiat Oncol Biol Phys. 2013;86(2):343-

work page 2013
[25]

doi:10.1016/j.ijrobp.2012.12.025 Downloaded from https://academic.oup.com/noa/advance-article/doi/10.1093/noajnl/vdag115/8666933 by guest on 06 May 2026 NOA-D-25-00604R2 31

work page doi:10.1016/j.ijrobp.2012.12.025 2012
[26]

Dual blockade of PD-1 and CTLA-4 generates long-lasting immunity against irradiated glioblastoma

De Martino M, Daviaud C, Lira MC, Hernandez-Zirofsky K, Vanpouille-Box C. Dual blockade of PD-1 and CTLA-4 generates long-lasting immunity against irradiated glioblastoma. Cancer Lett. 2025;628:217856. doi:10.1016/j.canlet.2025.217856

work page doi:10.1016/j.canlet.2025.217856 2025
[27]

Macrophage Polarization States in the Tumor Microenvironment

Boutilier AJ, Elsawa SF. Macrophage Polarization States in the Tumor Microenvironment. Int J Mol Sci. 2021 Jun 29;22(13):6995. doi: 10.3390/ijms22136995. PMID: 34209703; PMCID: PMC8268869

work page doi:10.3390/ijms22136995 2021
[28]

The complex role of tumor- infiltrating macrophages

Christofides A, Strauss L, Yeo A, Cao C, Charest A, Boussiotis V A. The complex role of tumor- infiltrating macrophages. Nat Immunol . 2022;23(8):1148 -1156. doi:10.1038/s41590-022- 01267-2

work page doi:10.1038/s41590-022- 2022
[29]

https://clinicaltrials.gov/study/NCT06816927

ClinicalTrials.gov. https://clinicaltrials.gov/study/NCT06816927

work page
[30]

Neoadjuvant anti -PD-1 immunotherapy promotes a survival benefit with intratumoral and systemic immune responses i n recurrent glioblastoma

Cloughesy TF, Mochizuki AY , Orpilla JR, et al. Neoadjuvant anti -PD-1 immunotherapy promotes a survival benefit with intratumoral and systemic immune responses i n recurrent glioblastoma. Nat Med. 2019;25(3):477-486. doi:10.1038/s41591-018-0337-7

work page doi:10.1038/s41591-018-0337-7 2019
[31]

Neoadjuvant anti-PD1 immunotherapy for surgically accessible recurrent glioblastoma: clinical and molecular outcomes of a s tage 2 single-arm expansion cohort

McFaline-Figueroa JR, Sun L, Youssef GC, et al. Neoadjuvant anti-PD1 immunotherapy for surgically accessible recurrent glioblastoma: clinical and molecular outcomes of a s tage 2 single-arm expansion cohort. Nat Commun. 2024;15(1):10757. Published 2024 Dec 30. doi:10.1038/s41467-024-54326-7

work page doi:10.1038/s41467-024-54326-7 2024
[32]

Glioblastoma-infiltrated innate immune cells resemble M0 macrophage phenotype

Gabrusiewicz K, Rodriguez B, Wei J, et al. Glioblastoma-infiltrated innate immune cells resemble M0 macrophage phenotype. JCI Insight. 2 016;1(2):e85841. doi:10.1172/jci.insight.85841

work page doi:10.1172/jci.insight.85841
[34]

Immunosuppressive mechanisms and therapeutic interventions shaping glioblastoma immunity

Moreno-Sanchez PM, Rezaeipour M, Inderberg EM, Platten M, Golebiewska A. Immunosuppressive mechanisms and therapeutic interventions shaping glioblastoma immunity. Nat Cancer. 2026;7(1):29-42. doi:10.1038/s43018-025-01097-9

work page doi:10.1038/s43018-025-01097-9 2026
[35]

-New frontiers in domain -inspired radiomics and radiogenomics: increasing role of molecular diagnostics in CNS tumor classification a nd grading following WHO CNS -5 updates

Singh G, Singh A, Bae J, et al. -New frontiers in domain -inspired radiomics and radiogenomics: increasing role of molecular diagnostics in CNS tumor classification a nd grading following WHO CNS -5 updates. Cancer Imaging. 2024;24(1):133. Published 2024 Oct 7. doi:10.1186/s40644-024-00769-6

work page doi:10.1186/s40644-024-00769-6 2024
[36]

MRI-derived radiomics assessing tumor -infiltrating macrophages enable prediction of immune-phenotype, immunotherapy response and survival in glioma

Chen D, Zhang R, Huang X, et al. MRI-derived radiomics assessing tumor -infiltrating macrophages enable prediction of immune-phenotype, immunotherapy response and survival in glioma. Biomark Res. 2024;12(1):14. Published 2024 Jan 31. doi:10.1186/s40364 -024- 00560-6

work page doi:10.1186/s40364 2024
[37]

Radiogenomics and imaging phenotypes in glioblastoma: novel observations and correlation with molecular chara cteristics

Ellingson BM. Radiogenomics and imaging phenotypes in glioblastoma: novel observations and correlation with molecular chara cteristics. Curr Neurol Neurosci Rep. 2015;15(1):506. doi:10.1007/s11910-014-0506-0

work page doi:10.1007/s11910-014-0506-0 2015
[38]

Radiogenomics to characterize the immune -related prognostic signature associated with biological functions in glioblastoma

Liu D, Chen J, Ge H, et al. Radiogenomics to characterize the immune -related prognostic signature associated with biological functions in glioblastoma. Eur Radiol. 2023;33( 1):209-

work page 2023
[39]

doi:10.1007/s00330-022-09012-x

work page doi:10.1007/s00330-022-09012-x
[40]

Radiomics for characterization of the glioma immune microenvironment

Khalili N, Kazerooni AF, Familiar A, et al. Radiomics for characterization of the glioma immune microenvironment. NPJ Precis Oncol. 2023;7(1):59. Published 2023 Jun 19. doi:10.1038/s41698-023-00413-9 Downloaded from https://academic.oup.com/noa/advance-article/doi/10.1093/noajnl/vdag115/8666933 by guest on 06 May 2026 NOA-D-25-00604R2 33

work page doi:10.1038/s41698-023-00413-9 2023
[41]

Applications of Radiomics and Radiogenomics in High -Grade Gliomas in the Era of Precision Medicine

Fathi Kazerooni A, Bagley SJ, Akbari H, Saxena S, Bagheri S, Guo J, Chawla S, Nabavizadeh A, Mohan S, Bakas S, Davatzikos C, Nasrallah MP. Applications of Radiomics and Radiogenomics in High -Grade Gliomas in the Era of Precision Medicine. Cancers (Basel). 2021 Nov 25;1 3(23):5921. doi: 10.3390/cancers13235921. PMID: 34885031; PMCID: PMC8656630

work page doi:10.3390/cancers13235921 2021
[42]

Radiomics in Glioblastoma: Current Status and Challenges Facing Clinical Implementation

Chaddad A, Kucharczyk MJ, Daniel P, Sabri S, Jean -Claude BJ, Niazi T, Abdulkarim B. Radiomics in Glioblastoma: Current Status and Challenges Facing Clinical Implementation. Front Oncol. 2019 May 21;9:374. doi: 10.3389/fonc.2019.00374. PMID: 31165039; PMCID: PMC6536622

work page doi:10.3389/fonc.2019.00374 2019
[43]

Radiomics for precision medicine in glioblastoma

Aftab K, Aamir FB, Mallick S, et al. Radiomics for precision medicine in glioblastoma. J Neurooncol. 2022;156(2):217-231. doi:10.1007/s11060-021-03933-1

work page doi:10.1007/s11060-021-03933-1 2022
[44]

Ghimire P, Kinnersley B, Golestan K, Arumugam P, Houlston R, Ashkan K, Modat M, Booth CT. Radiogenomic biomarkers for immunotherapy in glioblastoma: a systematic review of magnetic resonance imaging studies, Neuro-Oncology Advances, 2024;, vdae055, https://doi.org/10.1093/noajnl/vdae055

work page doi:10.1093/noajnl/vdae055 2024
[45]

The FDA NIH Biomarkers, EndpointS, and other Tools (BEST) resource in neuro -oncology

Cagney DN, Sul J, Huang RY , Ligon KL, Wen PY , Alexander BM. The FDA NIH Biomarkers, EndpointS, and other Tools (BEST) resource in neuro -oncology. Neuro Oncol . 2018;20(9):1162-1172. doi:10.1093/neuonc/nox242

work page doi:10.1093/neuonc/nox242 2018
[46]

The Cancer Genome Atlas Glioblastoma Multiforme Collection (TCGA-GBM) (Version 5) [Data set]

Scarpace L, Mikkelsen, T, Cha, S, et al. The Cancer Genome Atlas Glioblastoma Multiforme Collection (TCGA-GBM) (Version 5) [Data set]. The Cancer Imaging Archive.2016

work page 2016
[48]

Data from Ivy Glioblastoma Atlas Project (IvyGAP) [Data set]

Shah N, Feng X, Lankerovich M, Puchalski RB, Keogh B. Data from Ivy Glioblastoma Atlas Project (IvyGAP) [Data set]. The Cancer Imaging Archive. 2016. https://doi.org/10.7937/K9/TCIA.2016.XLwaN6nL

work page doi:10.7937/k9/tcia.2016.xlwan6nl 2016
[49]

Data From REMBRANDT [Data set]

Scarpace L, Flanders AE, Jain R, Mikkelsen T, Andrews, DW. Data From REMBRANDT [Data set]. The Cancer Imaging Archive. 2019. https://doi.org/10.7937/K9/TCIA.2015.588OZUZB

work page doi:10.7937/k9/tcia.2015.588ozuzb 2019
[50]

Chinese Glioma Genome Atlas (CGGA): A Comprehensive Resource with Functional Genomic Data from Chinese Glioma Patients

Zhao Z, Zhang KN, Wang Q, et al. Chinese Glioma Genome Atlas (CGGA): A Comprehensive Resource with Functional Genomic Data from Chinese Glioma Patients. Genomics, Proteomics & Bioinformatics. 2021 Feb;19(1):1-12

work page 2021
[51]

Generative Adversarial Networks to Synthesize Missing T1 and FLAIR MRI Sequences for Use in a Multisequence Brain Tumor Segmentation Model

Conte GM, Weston AD, V ogelsang DC, et al. Generative Adversarial Networks to Synthesize Missing T1 and FLAIR MRI Sequences for Use in a Multisequence Brain Tumor Segmentation Model. Radiology.2021;299(2):313-323. doi:10.1148/radiol.2021203786

work page doi:10.1148/radiol.2021203786 2021
[52]

Predicting transcriptional outcomes of novel multigene perturba- tions with GEARS

Newman AM, Steen CB, Liu CL, et al. Determining cell type abundance and expression from bulk tissues with digital cytometry. Nat Biotechnol. 2019;37(7):773-782. doi:10.1038/s41587- 019-0114-2

work page doi:10.1038/s41587- 2019
[53]

GBMdeconvoluteR accurately infers proportions of neoplastic and immune cell populations from bulk glioblastoma transcriptomics data

Ajaib S, Lodha D, Pollock S, et al. GBMdeconvoluteR accurately infers proportions of neoplastic and immune cell populations from bulk glioblastoma transcriptomics data. Neuro Oncol. 2023;25(7):1236-1248. doi:10.1093/neuonc/noad021

work page doi:10.1093/neuonc/noad021 2023
[55]

GMMSampling: a new model -based, data difficulty-driven resampling method for multi -class imbalanced data

Naglik, I., Lango, M. GMMSampling: a new model -based, data difficulty-driven resampling method for multi -class imbalanced data. Mach Learn 2023;113:5183 -5202. doi:10.1007/s10994-023-06416-8

work page doi:10.1007/s10994-023-06416-8 2023
[56]

Retrieved April 4, 2024, from https://www.nitrc.org/projects/deepbratumia/

NITRC: DeepBraTumIA: Tool/Resource Info. Retrieved April 4, 2024, from https://www.nitrc.org/projects/deepbratumia/

work page 2024
[57]

The LUMIERE dataset: Longitudinal Glioblastoma MRI with expert RANO evaluation

Suter Y , Knecht U, Valenzuela W, Notter M, He wer E, Schucht P, Wiest R, Reyes M. The LUMIERE dataset: Longitudinal Glioblastoma MRI with expert RANO evaluation. Sci Data. 2022 Dec 15;9(1):768. doi: 10.1038/s41597 -022-01881-7. PMID: 36522344; PMCID: PMC9755255

work page doi:10.1038/s41597 2022
[58]

Computational Radiomics System to Decode the Radiographic Phenotype

van Griethuysen JJM, Fedorov A, Parmar C, et al. Computational Radiomics System to Decode the Radiographic Phenotype. Cancer Res. 2017;77(21):e104 -e107. doi:10.1158/0008 - 5472.CAN-17-0339

work page doi:10.1158/0008 2017
[59]

Harmonization of cortical thickness measurements across scanners and sites

Fortin JP, Cullen N, Sheline YI, et al. Harmonization of cortical thickness measurements across scanners and sites. Neuroimage. 2018;167:104-120. doi:10.1016/j.neuroimage.2017.11.024

work page doi:10.1016/j.neuroimage.2017.11.024 2018
[60]

A comprehensive survey on support vector machine classification: Applications, challenges and trends

Cervantes J, García-Lamont F, Rodríguez-Mazahua L, López A. A comprehensive survey on support vector machine classification: Applications, challenges and trends. Neurocomputing. 2020;408:189-215. doi:10.1016/j.neucom.2019.10.118

work page doi:10.1016/j.neucom.2019.10.118 2020
[61]

Radiomics signature: A potential biomarker for the prediction of MGMT promoter methylation in glioblastoma

Xi YB, Guo F, Xu ZL, et al. Radiomics signature: A potential biomarker for the prediction of MGMT promoter methylation in glioblastoma. J Magn Reson Imaging. 2018;47(5):1380 -

work page 2018
[62]

doi:10.1002/jmri.25860

work page doi:10.1002/jmri.25860
[63]

Optimizing a machine learning based glioma grading system using multi-parametric MRI histogram and texture features

Zhang X, Yan LF, Hu YC, et al. Optimizing a machine learning based glioma grading system using multi-parametric MRI histogram and texture features. Oncotarget. 2017;8(29):47816 - 47830. doi:10.18632/oncotarget.18001 Downloaded from https://academic.oup.com/noa/advance-article/doi/10.1093/noajnl/vdag115/8666933 by guest on 06 May 2026 NOA-D-25-00604R2 36

work page doi:10.18632/oncotarget.18001 2017
[64]

Radiomics strategy for glioma grading using texture features from multiparametric MRI

Tian Q, Yan LF, Zhang X, et al. Radiomics strategy for glioma grading using texture features from multiparametric MRI. J Magn Reson Imaging. 2018;48(6):1518 -1528. doi:10.1002/jmri.26010

work page doi:10.1002/jmri.26010 2018
[65]

Differentiation between glioblastoma, brain metastasis and subtypes using radiomic s analysis

Artzi M, Bressler I, Ben Bashat D. Differentiation between glioblastoma, brain metastasis and subtypes using radiomic s analysis. J Magn Reson Imaging. 2019;50(2):519 -528. doi:10.1002/jmri.26643

work page doi:10.1002/jmri.26643 2019
[66]

The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation

Chicco D, Jurman G. The advantages of the Matthews correlation coefficient (MCC) over F1 score and accuracy in binary classification evaluation. BMC Genomics. 2020 Jan 2;21(1):6. doi: 10.1186/s12864-019-6413-7

work page doi:10.1186/s12864-019-6413-7 2020
[67]

Bootstrapping the out-of-sample predictions for efficient and accurate cross -validation

Tsamardinos I, Greasidou E, Borboudakis G. Bootstrapping the out-of-sample predictions for efficient and accurate cross -validation. Mach Learn. 2018;107(12):1895 -1922. doi: 10.1007/s10994-018-5714-4. Epub 2018 May 9

work page doi:10.1007/s10994-018-5714-4 2018
[68]

Assessing Model Selection Uncertainty Using a Bootstrap Approach: An update

Lubke GH, Campbell I, McArtor D, Miller P, Luningham J, van den Berg SM. Assessing Model Selection Uncertainty Using a Bootstrap Approach: An update. Struct Equ Modeling. 2017;24(2):230-245. doi: 10.1080/10705511.2016.1252265. Epub 2016 Dec 5. PMID: 28652682; PMCID: PMC5482523

work page doi:10.1080/10705511.2016.1252265 2017
[69]

Necroptosis-based glioblastoma prognostic subtypes: implications for TME remodeling and therapy response

Khan M, Huang X, Ye X, et al. Necroptosis-based glioblastoma prognostic subtypes: implications for TME remodeling and therapy response. Ann Med. 2024;56(1):2405079. doi:10.1080/07853890.2024.2405079

work page doi:10.1080/07853890.2024.2405079 2024
[70]

A Systematic Pan-Cancer Analysis of MEIS1 in Human Tumors as Prognostic Biomarker and Immunotherapy Target

Li H, Tang Y , Hua L, et al. A Systematic Pan-Cancer Analysis of MEIS1 in Human Tumors as Prognostic Biomarker and Immunotherapy Target. J Clin Med. 2023;12(4):1646. Published 2023 Feb 18. doi:10.3390/jcm12041646 Downloaded from https://academic.oup.com/noa/advance-article/doi/10.1093/noajnl/vdag115/8666933 by guest on 06 May 2026 NOA-D-25-00604R2 37

work page doi:10.3390/jcm12041646 2023
[71]

BACH1 as a potential target for immunotherapy in glioblastomas

Yuan F, Cong Z, Cai X, et al. BACH1 as a potential target for immunotherapy in glioblastomas. Int Immunopharmacol. 2022;103:108451. doi:10.1016/j.intimp.2021.108451

work page doi:10.1016/j.intimp.2021.108451 2022
[72]

Redox Regulator GLRX Is Associated With Tumor Immunity in Glioma

Chang Y , Li G, Zhai Y , et al. Redox Regulator GLRX Is Associated With Tumor Immunity in Glioma. Front Immunol. 2020;11:58 0934. Published 2020 Nov 30. doi:10.3389/fimmu.2020.580934

work page doi:10.3389/fimmu.2020.580934 2020
[73]

Drug-free mannosylated liposomes inhibit tumor growth by promoting the polarization of tumor -associated macrophages

Ye J, Yang Y , Dong W, et al. Drug-free mannosylated liposomes inhibit tumor growth by promoting the polarization of tumor -associated macrophages. Int J Nanomedicine. 2019;14:3203-3220. Published 2019 May 2. doi:10.2147/IJN.S207589

work page doi:10.2147/ijn.s207589 2019
[74]

Glioblastoma genetic drivers dictate the function of tumor - associated macrophages/microglia and responses to CSF1R inhibition

Rao R, Han R, Ogurek S, et al. Glioblastoma genetic drivers dictate the function of tumor - associated macrophages/microglia and responses to CSF1R inhibition. Neuro Oncol. 2022;24(4):584-597. doi:10.1093/neuonc/noab228

work page doi:10.1093/neuonc/noab228 2022
[75]

The tumor microenvironment underlies acquired resistance to CSF -1R inhibition in gliomas

Quail DF, Bowman RL, Akkari L, et al. The tumor microenvironment underlies acquired resistance to CSF -1R inhibition in gliomas. Science. 2016;352(6288):aad3018. doi:10.1126/science.aad3018

work page doi:10.1126/science.aad3018 2016
[76]

Blockade of CD73 delays gliob lastoma growth by modulating the immune environment

Azambuja JH, Schuh RS, Michels LR, et al. Blockade of CD73 delays gliob lastoma growth by modulating the immune environment. Cancer Immunol Immunother. 2020;69(9):1801 -

work page 2020
[77]

doi:10.1007/s00262-020-02569-w

work page doi:10.1007/s00262-020-02569-w
[78]

CCR2 of Tumor Microenvironmental Cells Is a Relevant Modulator of Glioma Biology

Felsenstein M, Blank A, Bungert AD, et al. CCR2 of Tumor Microenvironmental Cells Is a Relevant Modulator of Glioma Biology. Cancers (Basel). 2020;12(7):1882. Published 2020 Jul

work page 2020
[79]

doi:10.3390/cancers12071882

work page doi:10.3390/cancers12071882
[80]

Remodeling tumor immune microenvironment (TIME) for glioma therapy using multi -targeting liposomal codelivery

Zheng Z, Zhang J, Jiang J, et al. Remodeling tumor immune microenvironment (TIME) for glioma therapy using multi -targeting liposomal codelivery. J Immunother Cancer. 2020;8(2):e000207. doi:10.1136/jitc-2019-000207 Downloaded from https://academic.oup.com/noa/advance-article/doi/10.1093/noajnl/vdag115/8666933 by guest on 06 May 2026 NOA-D-25-00604R2 38

work page doi:10.1136/jitc-2019-000207 2020
[81]

Synergistic immunotherapy of glioblastoma by dual targeting of IL-6 and CD40

Yang F, He Z, Duan H, et al. Synergistic immunotherapy of glioblastoma by dual targeting of IL-6 and CD40. Nat Commun. 2021;12(1):3424. Published 2021 Jun 8. doi:10.1038/s41467 - 021-23832-3

work page doi:10.1038/s41467 2021
[82]

CCL8 secreted by tumor-associated macrophages promotes invasion and stemness of glioblastoma cells via ERK1/2 signaling

Zhang X, Chen L, Dang WQ, et al. CCL8 secreted by tumor-associated macrophages promotes invasion and stemness of glioblastoma cells via ERK1/2 signaling. Lab Invest. 2020;100(4):619-629. doi:10.1038/s41374-019-0345-3

work page doi:10.1038/s41374-019-0345-3 2020
[83]

Tumor -associated macrophage -related strategies for glioma immunotherapy

Tang F, Wang Y , Zeng Y , Xiao A, Tong A, Xu J. Tumor -associated macrophage -related strategies for glioma immunotherapy. NPJ Precis Oncol . 2023;7(1):78. Published 2023 Aug

work page 2023
[84]

doi:10.1038/s41698-023-00431-7

work page doi:10.1038/s41698-023-00431-7
[85]

Phase 1b/2 study of orally administered pexidartinib in combination wi th radiation therapy and temozolomide in patients with newly diagnosed glioblastoma

Mendez JS, Cohen AL, Eckenstein M, et al. Phase 1b/2 study of orally administered pexidartinib in combination wi th radiation therapy and temozolomide in patients with newly diagnosed glioblastoma. Neurooncol Adv . 2024;6(1):vdae202. Published 2024 Nov 22. doi:10.1093/noajnl/vdae202

work page doi:10.1093/noajnl/vdae202 2024
[86]

Repolarization of glioblastoma macrophage cells u sing non-agonistic Dectin -1 ligand encapsulating TLR -9 agonist: plausible role in regenerative medicine against brain tumor

Tiwari RK, Singh S, Gupta CL, et al. Repolarization of glioblastoma macrophage cells u sing non-agonistic Dectin -1 ligand encapsulating TLR -9 agonist: plausible role in regenerative medicine against brain tumor. Int J Neurosci . 2021;131(6):591 -598. doi:10.1080/00207454.2020.1750393

work page doi:10.1080/00207454.2020.1750393 2021
[87]

Oncolytic DNX-2401 virotherapy plus pembrolizumab in recurrent glioblastoma: a phase 1/2 trial

Nassiri F, Patil V , Yefet LS, et al. Oncolytic DNX-2401 virotherapy plus pembrolizumab in recurrent glioblastoma: a phase 1/2 trial. Nat Med . 2023;29(6):1370 -1378. doi:10.1038/s41591-023-02347-y

work page doi:10.1038/s41591-023-02347-y 2023

Showing first 80 references.