Cerebral arteriovenous malformations (AVMs) are characterized by a network of abnormal blood vessels where the feeding arteries are directly connected to the draining veins, exhibiting high-flow, low-resistance hemodynamic features (1, 2). AVMs may rupture and cause bleeding, leading to symptoms such as seizures and loss of consciousness (3, 4). However, for unruptured AVMs, due to the rarity of symptoms related to focal neurological deficits (4, 5), there has been limited attention on whether AVMs cause neurological functional changes and fiber alterations.

Unruptured AVMs usually do not exhibit these symptoms, even when the lesion is located in eloquent areas (6, 7), which is different from brain tumors, ischemic stroke, and intracranial hemorrhage (8). One possible explanation is that AVM lesions are congenital, and potential functional reorganization in these regions (9, 10) distinguishes them from acquired conditions like brain tumors. There is ongoing discussion about whether the reorganization associated with AVMs occurs at a functional level or is based on structural changes. Some reports suggest that preoperative fiber changes in AVM patients may be related to their functional prognosis (11), indicating unique alterations in fiber function and supporting the hypothesis of brain reorganization in AVM patients. Given the findings from unimodal studies of structural and functional imaging may involve structure–function relationships, and magnetic resonance diffusion tensor imaging (DTI) is currently the only method capable of tracing white matter fiber tracts in vivo (12). Researchers have utilized DTI to observe the impact of AVM on fiber tracts. However, previous studies have primarily focused on the relationship between AVM and the surrounding fibers (13, 14), and there is insufficient reporting on the concurrent changes in function within the fiber connectivity regions.

Resting-state functional MRI (rs-fMRI) is a commonly used non-invasive imaging technique that has been widely applied in the study of various diseases, such as Alzheimer’s disease, stroke, mild cognitive impairment, and gliomas (15–18). Amplitude of low-frequency fluctuation (ALFF) analysis is a data-driven method in rs-fMRI that investigates regional spontaneous activity and measures the intensity of blood oxygen-level dependent (BOLD) signals in low-frequency oscillations within local brain regions (19). Changes in ALFF may represent underlying abnormalities in brain activity (20, 21). Previous studies have used rs-fMRI to explore the relationship between AVM and changes/reorganization in language function (5, 22). However, these studies have only observed the functional changes without delving into the causes and processes underlying these changes.

To delve deeper into the intricacies of brain reorganization in relation to AVMs, it is imperative to examine the potential associations between AVM and alterations in both neural function and fiber connectivity. We conducted a retrospective study of AVM patients who visited our hospital between 2020 and 2023, all of whom underwent both fMRI and DTI examinations. This study also included demographic and imaging data from healthy controls available in open databases for comparison. To our knowledge, this study is the first to simultaneously demonstrate differences in brain function and fiber tract alternations between AVM patients and normal brains.

Our findings reveal concurrent alterations in white matter integrity, characterized by functional network reorganization. Previous research has established that arteriovenous malformations (AVMs) can affect deep white matter fiber tracts within eloquent areas (9, 10), even if the lesions are not located in these areas (41). Our study extends this understanding by demonstrating concurrent functional and structural alterations associated with AVMs—a dimension often overlooked in prior investigations. The simultaneous changes we observed in cortical function and fibers, along with alterations in TW-FC (time-varying functional connectivity), suggest that AVMs modify key structural nodes within functional connectivity pathways. These findings align with reports of functional reorganization following AVM resection, where the recruitment of homologous regions and changes in white matter tracts were pivotal for post-surgical language recovery (42). Several studies have reported altered functional and structural connectivity in AVMs. Using fMRI and DTI, Liu et al. observed significant functional reorganization following AVM resection, with recruitment of the right hemispheric homologous regions of Broca’s and Wernicke’s areas and changes in related white matter tracts. This reorganization was crucial for language recovery post-surgery (42). A study on AVMs located in Broca’s area utilized diffusion imaging and graph theory to analyze structural brain networks (43). The findings indicated altered topological properties, such as connectivity strength and efficiency, in patients compared to controls. These changes highlight how AVMs can disrupt the structural connectivity of critical language regions, leading to potential functional impairments. But above studies have not integrated these modalities to examine how structural changes contribute to functional reshaping. Our work represents one of the first comprehensive analyses of both functional and fiber reconstruction in AVMs. Using TW-FC analysis, we provided novel evidence that AVM-related changes in the visual network (VN) and sensorimotor network (SMN) may reflect the brain’s attempt to reorganize function through fiber tract modification. Preoperative TW-FC mapping revealing networks exhibiting significant hyperconnectivity or altered TW-FC may assist clinically identifying “at-risk” regions potentially heavily reliant on compensatory mechanisms. Surgical or radiosurgical interventions targeting AVMs embedded within or adjacent to these networks may necessitate more aggressive functional mapping and preservation procedures (e.g., intraoperative electrophysiological monitoring) to mitigate the risk of postoperative complications. Future studies must correlate TW-FC metrics with standardized neuropsychological testing and long-term clinical outcomes to establish their predictive value and specificity. Therefore, TW-FC holds potential as a more sensitive biomarker for identifying AVM-induced functional or structural alterations; however, robust validation in larger, prospectively designed cohorts remains essential prior to clinical implementation.

We found simultaneous alterations in both function and fibers within the VN and SMN in AVMs. However, only a minority of patients exhibited clinical presentations of neurological functional deficit. Reports indicate that less than a quarter of AVMs experience progressive neurological deficits (44, 45), which aligns with our findings. This also aligns with animal models showing lesion-induced myelin remodeling facilitates functional rewiring (46), and in humans, such gradual plasticity may enable functional preservation during slowly evolving lesions. When comparing patients with functional impairments to other patients, we found only one additional brain region with decreased function that was not among the previously identified 12 ROIs (Supplementary Table S2; Supplementary Figure S1). This suggests that changes in these two brain networks are not directly responsible for functional impairments. One explanation is that damage to connected cortical areas and along fiber tracts leads to hyperconnectivity and reorganization (47). Further, enhanced functional connectivity was also observed in the SMN in our study. Considering the relatively few local neurological symptoms in AVMs, this functional hyperconnectivity may be a compensatory brain strategy to offset functional deficits (48, 49). During recovery, disinhibition phenomena may occur (50), and the enhanced cortical signals could reflect cortical disinhibition and neuroplasticity. Finally, the bidirectional TW-FC shifts may indicate pathological network destabilization rather than beneficial reorganization, as seen in other neurological disorders (51). These alternative interpretations highlight the multifactorial complexity of AVM-related brain changes and emphasize the need for longitudinal multimodal studies to disentangle these possibilities. Without neuropsychological testing, we cannot confirm whether these connectivity changes genuinely improve functional output. Importantly, increased synchrony may represent maladaptive reorganization rather than beneficial compensation, as similar hyperconnectivity patterns precede pathological states like epilepsy (52). Therefore, we propose these alterations represent potential adaptive responses, acknowledging that the clinical significance requires validation through behavioral correlation. Additionally, the dominant hemisphere was previously found receives relatively less interhemispheric inhibition from the non-dominant hemisphere (53). Our study also found that cortical ALFF changes were greater in the dominant hemisphere.

Cortical functional changes may play a crucial role in AVM-related epilepsy (54–56). Epilepsy is a common symptom among AVMs (22.7–60.9%) (57–59). In this study, we observed a significant decrease in cortical ALFF values in AVMs presented with seizures (p < 0.05) (Supplementary Figure S2). Currently, there are conflicting findings regarding whether ALFF value changes in epileptic brains are increased or decreased (54, 56, 60). Some studies posit that hyperactivity could be a compensatory mechanism or a result of abnormal synchronization within the brain networks. Conversely, other research indicates decreased ALFF values, suggesting a possible reduction in neural activity due to impaired functional connectivity or neuronal loss in regions affected by seizures. Based on the results of this study, we hypothesize that epilepsy in AVM patients may be caused by local functional connectivity disruptions that arise during the process of functional reorganization. Further research is essential to understand how these networks’ functional and structural integrity is compromised in AVM patients with epilepsy, which may reveal new insights into the pathophysiology of seizures in AVM patients and help identify potential therapeutic targets.

In this study, we visualized the modifications of fiber tracts using DTI while previous studies have discussed the relationship between fiber tracts affected by AVMs and clinical symptoms (13, 14), our study is the first to combine Resting-state fMRI and DTI to investigate AVM-induced changes. AVMs can alter surrounding fiber tracts, which is related to their pathophysiological behavior, including bleeding and ongoing dynamic evolution (61). Not only functional reshaping but also reconstruction of fiber tracts near AVMs has been observed (22), consistent with the results of our study. Additionally, as congenital lesions, AVMs may alter neurodevelopmental trajectories, making it difficult to distinguish between true reorganization and innate neuroanatomical differences (58). However, due to the limited number of AVMs, we were unable to further discuss the relationship between AVMs and nearby fiber reconstruction and functional changes based on the location of the AVMs. Larger-scale, multicenter studies may be needed in the future to discuss the impact of AVMs on fiber tracts more comprehensively.

However, our study also has limitations. Firstly, the healthy controls in our study were obtained from a public database, which may introduce biases due to factors such as equipment, center, ethnicity, etc., reducing the robustness of our conclusions. Different scanning devices in different centers may lead to the bias of ALFF and FA. Educational attainment (unavailable in public datasets) correlates with functional connectivity strength, potentially confounding network-level comparisons. These limitations need cautious interpretation of absolute effect sizes. Future multicenter studies with standardized protocols are essential to confirm these results. Secondly, the relatively small patient cohort hindered classification based on clinical data such as lesion location, lesion size and patient symptoms, impeding further exploration of AVM-related changes in function and structure. Future studies are essential to discuss these characters in a larger cohort. Lastly, we did not use scales to assess the cognitive abilities of patients, which may lead to an underestimation of clinical symptoms related to brain function in patients.

In summary, our study revealed a profound association between AVMs and a spectrum of alterations in both brain function and fiber reorganization. These changes extend beyond focal regions, reverberating across neural networks to significantly reshape functional connectivity in key hubs. We observed concurrent functional and fiber connectivity changes between the VN and SMN in AVMs, suggesting the brain’s innate capacity to marshal compensatory mechanisms. Furthermore, concurrent functional and structural connectivity changes between the visual network VN and SMN suggest the innate capacity to deploy compensatory mechanisms of brain. Our works paved the way for future research dissecting the interplay between AVMs and brain plasticity and emphasize the need for nuanced approaches. Future studies should correlate AVM neurofunctional and structural features with clinical symptomatology to broaden the utility of neurofunctional and structural biomarkers in AVM management.