Abstract
Ranasinghe KG, Kudo K, Syed F, Yballa C, Kramer JH, Miller BL, Rankin KP, Garcia PA, Kirsch HE, Vossel K, Jagust W, Rabinovici GD, Nagarajan SS. Nat Commun. 2025;16(1):7957. doi:10.1038/s41467-025-62798-4 A growing body of evidence shows that epileptic activity is frequently observed in patients with Alzheimer's disease (AD), implicating underlying excitatory–inhibitory imbalance. The distinction of whether the AD-epileptic phenotype represents a subset of patients or an underdiagnosed manifestation holds major therapeutic implications. Here, we quantified the excitatory–inhibitory imbalance in AD patients using magnetoencephalography and examined the relationships to AD pathophysiology—amyloid-beta and tau, and to epileptic activity. We used two metrics to quantify regional excitatory–inhibitory imbalance distinguishing between local hyperexcitability (Neural excitability, quantified by regional aperiodic spectral slope) and aberrant long-range synaptic input integration (Neural fragility, quantified by regional linear dynamic instability). We found that amyloid-beta correlated with higher neural fragility and higher neural excitability, while tau and hypometabolism uniquely correlated with higher neural excitability. Importantly, the AD-epileptic phenotype showed a distinctive increase in neural fragility. Our findings demonstrate that AD pathophysiology is associated with diverse mechanisms of excitatory–inhibitory imbalance and that AD-epileptic phenotype represents a distinct group of patients with greater impairments in long-range synaptic input integration.
Commentary
Epileptiform activity is increasingly recognized as a contributor to cognitive decline in Alzheimer's disease (AD) 1 and is associated with accumulation of AD-linked neuropathology.2,3 Amyloid-β (Aβ) and tau can promote hyperexcitability and hypersynchrony, 4 potentially setting the stage for epileptiform abnormalities and aberrant network activity to manifest. Furthermore, since Aβ production and tau release are neuronal activity-dependent,4,5 hyperactive networks could contribute to their pathological accumulation and spread. This vicious circle has important clinical implications: elucidating the underlying mechanisms could reveal ways to modulate or break it, improving symptoms and perhaps changing the course of the disease. However, even though subclinical epileptiform abnormalities are common in individuals with AD, 1 they can evade detection due to their preferential occurrence during sleep3,6 and because some of the epileptiform events that can be recorded through invasive means are undetectable in standard non-invasive recordings such as scalp electroencephalogram.3,7 In this sense, whether AD patients who have epileptiform activity and could benefit from antiseizure interventions are a vast majority or a select subset is unclear.
In this work, Ranasinghe et al 8 define the associations between Aβ and tau accumulation and excitation–inhibition imbalance in individuals with early-stage AD. Furthermore, the authors explore the important overarching question of whether epileptiform activity is a general hallmark of AD that often goes undetected or a distinctive feature of a particular subset of the AD population. Beyond shedding light on which patients are more likely to respond positively to antiseizure medications, the answer to this question could also lead to new therapeutic strategies. For example, in the same way that there exist “resilient” individuals whose cognition is preserved even in the presence of overt Aβ and tau pathology, 9 the absence of epileptiform activity in a fraction of individuals with AD might correspond to a form of resilience that could be promoted once better understood.
Ranasinghe et al 8 utilized data from resting state magnetoencephalography recordings to extract two parameters that gauge distinct aspects of the excitation–inhibition imbalance. These parameters were (1) neural excitability, which provides an estimate of hyperexcitability due to the intrinsic properties of local circuits, and (2) neural fragility, which estimates how vulnerable the excitation–inhibition balance in local circuits is to perturbations from long-range inputs. The authors assessed neural excitability and fragility in early-stage patients with AD and in control individuals, and investigated their relationship with Aβ and tau accumulation in positron emission tomography (PET) scans, cognitive performance using the Mini-Mental State Examination (MMSE), glucose hypometabolism as an indicator of neuronal loss and synaptic dysfunction, and the presence or absence of subclinical epileptiform discharges. Neural excitability and fragility, as well as Aβ and tau accumulation, were estimated for 210 cortical regions, and the spatial variation and regional association of these measures were assessed.
The authors found that the distribution of neural excitability and fragility measures followed distinct spatial gradients across the cortex. Although the spatial distributions were similar in individuals with or without AD, individuals with AD had higher excitability and fragility values in a number of regions. Interestingly, the areas in which excitability and fragility were increased with respect to controls were largely nonoverlapping, indicating that the cortical regions in which the excitation–inhibition balance favors hyperexcitability, and the regions that are more susceptible to perturbations by inputs from distant areas are different. Also, whereas measures of neural excitability and fragility were not related to each other in healthy elderly controls, they were positively correlated in individuals with AD, suggesting that these measures might be similarly vulnerable to AD pathophysiology. Increased excitability and fragility predicted worse MMSE performance in individuals with AD, demonstrating that these measures can provide direct readouts of the severity of the disease.
Regional comparison of neural excitability and fragility with Aβ and tau accumulation revealed that Aβ is associated with higher neural excitability and fragility, whereas tau is associated with higher neural excitability, but not fragility. These findings suggest that Aβ and tau have unique and dissociable pathological signatures in the context of hyperexcitability and epilepsy that can be distinguished. In part, arriving at these conclusions was possible due to the careful consideration of controls and covariates, which allowed the authors to rule out interactions and possible confounders such as Aβ burden driving the observed associations between tau and the excitation–inhibition balance. This approach was instrumental in teasing apart the correlates of discrete but closely intertwined phenomena, and it enabled the finding that neural excitability, but not fragility, was associated with brain glucose hypometabolism, which is considered an indicator of AD-related atrophy and neurodegeneration. 10 Glucose hypometabolism has known associations with tau pathology, 10 which in turn was found by Ranasinghe et al 8 to predict increased neural excitability. Notably, the relationship between hypometabolism and neural excitability persisted even after the authors controlled for Aβ and tau burdens, indicating that factors other than neuropathology itself contribute to the association between hypometabolism and neural excitability.
Subclinical epileptiform discharges were detected in ∼40% of AD patients in which the presence of epileptiform activity was assessed, similar to the proportion found in a previous study. 1 When individuals were grouped according to “epileptic” and “non-epileptic” phenotypes, those with epileptiform abnormalities had similar neural excitability profiles but increased neural fragility compared to their non-epileptic counterparts. Considering that both groups had increased excitability and fragility with respect to controls, this result reveals that hyperexcitability might be a generalized hallmark of AD that only gives rise to overt epileptiform activity in a subset of patients with circuits that are more locally excitable and particularly susceptible to perturbations from external inputs.
Ranasinghe et al 8 skillfully dissect and expose the complexity of the excitation–inhibition imbalance in AD, which extends well beyond the mere presence of epileptiform activity, and the way it interacts with different tangible aspects of AD pathophysiology. In light of the prevalence of epileptiform abnormalities in the context of AD1,8 and the increasing body of evidence that demonstrates their association with worse cognitive outcomes and pathology,1–3 this is a timely work that furthers our understanding of the mechanisms underlying the heterogeneity of AD. Future research in this area could explore the temporal progression and trajectories of the excitation–inhibition imbalance along the AD continuum, from the early-stage cases reported in the present work to the later stages of disease, as hyperexcitability transitions into hypoexcitability. Moreover, since epileptiform activity is most frequent during nonrapid eye movement (NREM) sleep in individuals with AD, 6 understanding how measures of neural excitability and fragility might differ during sleep could add a new dimension to this research. Excitability and fragility were measured in awake individuals in this work; further investigating whether these parameters exhibit different characteristics, regional distribution, and associations with other variables during sleep may reveal whether the spatial complexity of the excitation–inhibition imbalance in AD is subject to temporal dynamics associated with the sleep–wake states. This approach could provide insight into the question of why epileptiform activity is much more prevalent in NREM than in rapid eye movement (REM) sleep in AD, 6 and perhaps inspire novel diagnostic or therapeutic avenues.
