Machine learning identifies experimental brain metastasis subtypes based on their influence on neural circuits

Aug 2, 2026

Patients with brain metastases often develop neurocognitive impairment, yet tumor mass effect alone cannot explain differences in how brain circuits are disrupted. A multidimensional analysis tested multiple preclinical brain metastasis models from distinct primary sources and oncogenic profiles to separate heterogeneous effects on local-field potential oscillations from features such as tumor size or glial response. Transcriptomic and mutational scoring revealed model-specific molecular programs linked to impaired neuronal crosstalk, while machine-learning analysis of brain activity readouts predicted metastasis presence and subtype.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages25

Reading Time1 h 3 min

AccessFree to download

Summary

Heterogeneity beyond tumor mass effect

Molecular programs from transcriptomic and mutational profiles

Predicting presence and subtype with machine learning

Introduction

Clinical impact and limits of current explanations

Rationale for coupling electrophysiology and calcium imaging

Inter-model heterogeneity and computational diagnosis potential

Article Machine learning identiﬁes experimental brain metastasis subtypes based on their inﬂuence on neural circuits Authors Alberto Sanchez-Aguilera, Mariam Masmudi-Martn, Andrea Navas-Olive, ..., F´atima Al-Shahrour, Liset Menendez de la Prida, Manuel Valiente Correspondence [lmprida@cajal.csic.es](lmprida@cajal.csic.es) (L.M.d.l.P.), [mvaliente@cnio.es](mvaliente@cnio.es) (M.V.) In brief Patients with brain metastasis experience neurocognitive impairment. Until now, the mass effect of the tumor was the only underlying cause. Sanchez-Aguilera et al. demonstrate that, independently on the size, number, and location, a machine learning approach correctly classiﬁes different models of brain metastasis based on their impact on brain activity. d Brain metastasis experimental models recapitulate neuronal impact heterogeneity d The underlying mechanism cannot be explained by the tumor mass effect d A molecular signature is enriched in models imposing high neural impact d Altered brain activity patterns predict the presence and subtype of metastasis Sanchez-Aguilera et al., 2023, Cancer Cell 41, 1–13 September 11, 2023 ª 2023 The Author(s) . Published by Elsevier Inc. [https://doi.org/10.1016/j.ccell.2023.07.010](https://doi.org/10.1016/j.ccell.2023.07.010) ll Please cite this article in press as: Sanchez-Aguilera et al., Machine learning identiﬁes experimental brain metastasis subtypes based on their inﬂuence on neural circuits, Cancer Cell (2023), [https://doi.org/10.1016/j.ccell.2023.07.010](https://doi.org/10.1016/j.ccell.2023.07.010) ll OPEN ACCESS Article Machine learning identiﬁes experimental brain metastasis subtypes based on their inﬂuence on neural circuits Alberto Sanchez-Aguilera,1,7,8 Mariam Masmudi-Mart´ın,2,8 Andrea Navas-Olive,1 Patricia Baena,2 Carolina Hern´andez-Oliver,2 Neibla Priego,2 Llus Cord´on-Barris,2 Laura Alvaro-Espinosa,2 Santiago Garca,3 Sonia Martnez,4 Miguel Lafarga,5 RENACER, Michael Z Lin,6 F´atima Al-Shahrour,3 Liset Menendez de la Prida,1,* and Manuel Valiente2,9,* 1Instituto Cajal, CSIC, 28002 Madrid, Spain 2Brain Metastasis Group, CNIO, 28029 Madrid, Spain 3Bioinformatics Unit, CNIO, Madrid, Spain 4Experimental Therapeutics Programme, CNIO, 28029 Madrid, Spain 5Department of Anatomy and Cell Biology and CIBERNED, University of Cantabria-IDIVAL, 39011 Santander, Spain 6Departments of Neurobiology and Bioengineering, Stanford University, Stanford, CA 94305-5090, USA 7Present address: Department of Physiology, Faculty of Medicine, Universidad Complutense de Madrid, Madrid 28040, Spain 8These authors contributed equally 9Lead contact *Correspondence: [lmprida@cajal.csic.es](lmprida@cajal.csic.es) (L. M.d. l. P.), [mvaliente@cnio.es](mvaliente@cnio.es) (M.V.) [https://doi.org/10.1016/j.ccell.2023.07.010](https://doi.org/10.1016/j.ccell.2023.07.010) SUMMARY A high percentage of patients with brain metastases frequently develop neurocognitive symptoms; however, understanding how brain metastasis co-opts the function of neuronal circuits beyond a tumor mass effect remains unknown. We report a comprehensive multidimensional modeling of brain functional analyses in the context of brain metastasis. By testing different preclinical models of brain metastasis from various primary sources and oncogenic proﬁles, we dissociated the heterogeneous impact on local ﬁeld potential oscillatory activity from cortical and hippocampal areas that we detected from the homogeneous inter-model tumor size or glial response. In contrast, we report a potential underlying molecular program responsible for impairing neuronal crosstalk by scoring the transcriptomic and mutational proﬁles in a model-speciﬁc manner. Additionally, measurement of various brain activity readouts matched with machine learning strategies conﬁrmed model-speciﬁc alterations that could help predict the presence and subtype of metastasis. INTRODUCTION Brain metastases have a dram

FAQs

What new explanation does the study propose for neurocognitive impairment in brain metastases?

It shows that effects on neural circuits vary across experimental models independently of tumor mass effect, indicating that heterogeneity is driven by factors beyond tumor size.

How did the researchers characterize differences between brain metastasis models?

They used multidimensional functional analyses, including in vivo electrophysiology and ex vivo calcium imaging, across models from different primary sources and oncogenic profiles.

What role does machine learning play in the study?

Machine-learning strategies applied to brain activity readouts confirmed model-specific alterations and enabled prediction of metastasis presence and subtype.

Aug 2, 2026

Upload by Emma Wilson