Connections across regional glymphatic clearance, neural activity and amyloid-β deposition in cortex

Article activity feed

1. 

   eLife

   Jun 30, 2026

   eLife Assessment

   The framework of the study - with the integration of multiple levels of analysis, glymphatic MRI, transcriptomics, functional MRI, and public amyloid maps, in one framework - is clever. The assertion that regional amyloid vulnerability may depend not just on neural activity alone, but on whether clearance is appropriately matched to activity is an interesting and novel concept. However, the chosen approach to imaging glymphatic clearance relies on indirect inferences from a small subgroup. In its current form, the main conclusions of this study are therefore incompletely supported.

   Read the original source
2. 

   eLife

   Jun 30, 2026

   Reviewer #1 (Public review):

   Summary:

   Regional differences in the brain's waste-clearance system may interact with neural activity to influence where amyloid-B accumulates. Using intrathecal GBCA administration to produce "Glymphatic MRI" in 96 subjects, the authors mapped cortical glymphatic influx and clearance and found distinct spatial patterns, with transcriptomic analyses linking better glymphatic function to neuronal cell types (through genes). In a subgroup with resting-state fMRI, regions with stronger resting-state activation generally showed higher contrast clearance, indicating a positive coupling between these processes. Notably, cortical regions where neural activity and glymphatic clearance were mismatched showed greater amyloid-β burden in a separate, publicly available PiB-PET dataset, suggesting that activity-clearance …

   Reviewer #1 (Public review):

   Summary:

   Regional differences in the brain's waste-clearance system may interact with neural activity to influence where amyloid-B accumulates. Using intrathecal GBCA administration to produce "Glymphatic MRI" in 96 subjects, the authors mapped cortical glymphatic influx and clearance and found distinct spatial patterns, with transcriptomic analyses linking better glymphatic function to neuronal cell types (through genes). In a subgroup with resting-state fMRI, regions with stronger resting-state activation generally showed higher contrast clearance, indicating a positive coupling between these processes. Notably, cortical regions where neural activity and glymphatic clearance were mismatched showed greater amyloid-β burden in a separate, publicly available PiB-PET dataset, suggesting that activity-clearance decoupling may contribute to regional vulnerability and neurodegeneration.

   Strengths:

   This is a rare and valuable dataset. Intrathecal contrast injection in ~100 subjects is quite a remarkable accomplishment alone, but the addition of resting-state fMRI, a correlative PiB cohort, and gene-expression pattern data is impressive.

   Weaknesses:

   This is a cross-sectional study, and we can't determine whether neural activity drives glymphatic clearance, whether glymphatic dysfunction alters neural activity, or whether both are shaped by a third factor. Language describing "flow", "influx", and "clearance" could be made more specific so the reader can more easily follow the methodological approach.

   Read the original source
3. 

   eLife

   Jun 30, 2026

   Reviewer #2 (Public review):

   Summary:

   In this study, Li et al. investigated the relationships among regional cortical tracer dynamics following intrathecal gadolinium administration, neural activity, and amyloid-β deposition in humans. Using serial MRI acquisitions after intrathecal gadodiamide administration in 96 participants, the authors characterized regional signal enhancement and clearance patterns across the human cortex. They integrated these imaging measures with transcriptomic data (Allen Human Brain Atlas), resting-state fMRI outcomes, and an external amyloid PET dataset. The authors report that regions with more efficient tracer clearance are enriched for genes related to synaptic organization and neuronal cell types, that tracer clearance patterns are in parts spatially coupled to spontaneous neural activity, and that …

   Reviewer #2 (Public review):

   Summary:

   In this study, Li et al. investigated the relationships among regional cortical tracer dynamics following intrathecal gadolinium administration, neural activity, and amyloid-β deposition in humans. Using serial MRI acquisitions after intrathecal gadodiamide administration in 96 participants, the authors characterized regional signal enhancement and clearance patterns across the human cortex. They integrated these imaging measures with transcriptomic data (Allen Human Brain Atlas), resting-state fMRI outcomes, and an external amyloid PET dataset. The authors report that regions with more efficient tracer clearance are enriched for genes related to synaptic organization and neuronal cell types, that tracer clearance patterns are in parts spatially coupled to spontaneous neural activity, and that regional mismatch between neural activity and tracer clearance is associated with increased amyloid burden according to the PET dataset.

   Strengths:

   The study addresses an important and very timely question about the interaction among neural activity, cerebrospinal fluid dynamics (waste clearance), and regional vulnerability to neurodegeneration. Integrating serial post-contrast MRI, transcriptomics, resting-state fMRI, and amyloid imaging is ambitious and conceptually very interesting. The spatial characterization of cortical tracer dynamics is potentially valuable for the field, particularly given the increasing interest in human glymphatic imaging approaches and intrathecal contrast MRI, which provides an opportunity to assess CSF tracer dynamics without confounding tracer signal from the blood. The imaging preprocessing pipeline includes normalization of regional cortical signal intensity to a reference region within each session before calculation of longitudinal percentage change, which helps reduce inter-session variability within individuals for conventional T1-weighted imaging. The transcriptomic analyses linking tracer dynamics to neuronal and synaptic gene expression patterns are also interesting. In addition, the manuscript addresses recent literature on neurovascular coupling, glymphatic function, and amyloid vulnerability.

   Weaknesses:

   Several issues limit the strength of the conclusions. One concern relates to the interpretation of repeated post-intrathecal contrast MRI measurements as direct indicators of glymphatic influx and clearance. The approach presented by the authors measures regional signal changes following intrathecal gadodiamide administration, but does not directly visualize paravascular flow or establish that the observed signal dynamics specifically reflect glymphatic transport mechanisms. Although it is widely accepted that CSF influx occurs primarily along periarterial spaces as part of the glymphatic system, and the terminology "glymphatic MRI" is increasingly used in the literature, the physiological processes contributing to delayed parenchymal enhancement, including CSF-interstitial exchange mediated by convective bulk flow and/or extracellular diffusion, as well as transient and, in the case of linear gadolinium agents, even long-term tracer retention remain incompletely resolved. Importantly, tracer kinetics may not directly reflect interstitial fluid kinetics, as solute transport may also be influenced by compartmental and extracellular barriers, diffusion constraints, and tissue retention effects. As currently written, several sections of the manuscript appear to overstate what can be directly inferred from the imaging data. This issue may be particularly relevant given the intrathecal use of gadodiamide (Omniscan), a linear gadolinium-based contrast agent with known long-lasting tissue retention due to lower kinetic stability compared to macrocyclic agents. Sustained signal at later imaging time points may therefore not only reflect impaired glymphatic clearance dynamics may also be influenced by tissue retention of contrast material, particularly in the context of neurological disease. In addition, the participant cohort is heterogeneous and includes individuals with neuroinflammatory and neurodegenerative diseases, peripheral neuropathy, and motor neuron disease. Although the authors argue that the spatial tracer patterns are relatively preserved across neurodegenerative groups, this heterogeneity complicates interpretation of imaging data and raises the possibility that disease-related factors and altered tracer-tissue interactions contribute to the observed effects. Thus, the rationale for interpreting a greater tracer signal at 39h as evidence of impaired glymphatic clearance should be explained more carefully, particularly given the highly heterogeneous patient population.

   In addition, the analyses linking spontaneous neural activity and tracer clearance are based on a very small rs-fMRI subgroup (n = 15), limiting the generalizability. The interpretation of the "mismatch" analysis also requires caution. The mismatch index was computed from z-scored fALFF and tracer clearance and is subsequently associated with amyloid burden derived from the external PET dataset rather than from the studied participants themselves. Therefore, the observed spatial associations should be interpreted with greater caution rather than as evidence for a direct mechanistic relationship. The cross-sectional nature of the analyses also limits conclusions regarding the directionality and temporal sequence of the relationships between neural activity, tracer dynamics, and amyloid burden. Several statements in the Discussion currently imply stronger causal or biological conclusions than are directly supported by the data.

   Despite these limitations, the study presents an interesting dataset and proposes a framework for understanding regional vulnerability to protein accumulation in neurodegeneration. This work hopefully motivates further investigation into the important relationships among neural activity, CSF dynamics, and neurodegeneration in humans.

   Read the original source
4. 

   eLife

   Jun 30, 2026

   Reviewer #3 (Public review):

   This manuscript addresses an interesting and timely question: whether regional glymphatic clearance in the human cortex is spatially coupled to neural activity and whether a mismatch between activity and clearance may help explain regional vulnerability to amyloid-β deposition. The authors use intrathecal gadolinium-based glymphatic MRI in 96 participants, derive cortical influx and clearance maps, integrate these with Allen Human Brain Atlas transcriptomic data, and then relate regional clearance to resting-state fMRI measures in a smaller subgroup. They further compare the resulting activity-clearance mismatch map with an open-source ¹¹C-PiB amyloid PET dataset. The overall concept is attractive because it attempts to connect glymphatic physiology, neuronal activity, and proteopathy at the regional level …

   Reviewer #3 (Public review):

   This manuscript addresses an interesting and timely question: whether regional glymphatic clearance in the human cortex is spatially coupled to neural activity and whether a mismatch between activity and clearance may help explain regional vulnerability to amyloid-β deposition. The authors use intrathecal gadolinium-based glymphatic MRI in 96 participants, derive cortical influx and clearance maps, integrate these with Allen Human Brain Atlas transcriptomic data, and then relate regional clearance to resting-state fMRI measures in a smaller subgroup. They further compare the resulting activity-clearance mismatch map with an open-source ¹¹C-PiB amyloid PET dataset. The overall concept is attractive because it attempts to connect glymphatic physiology, neuronal activity, and proteopathy at the regional level of the human brain, an important and understudied area.

   The main strength of the study is the use of direct intrathecal contrast-enhanced MRI to generate cortical maps of glymphatic tracer dynamics. This is a technically demanding approach and provides a richer spatial readout than indirect MRI proxies of glymphatic function. The authors show that the cortical tracer signal increases from 4.5 h to 15 h and then decreases by 39 h, allowing them to interpret the early signal as reflecting influx and the persistent signal at 39 h as impaired clearance. They further identify regional patterns, with faster influx in medial prefrontal/insular areas and slower clearance in dorsal prefrontal and parietal surface regions. The analysis is visually clear, and the use of cortical gradients is a useful way to reduce complex regional data into interpretable spatial axes.

   The multimodal integration is also interesting. The transcriptomic analysis suggests that regions with faster glymphatic clearance are enriched for synaptic organisation and neuronal activity-related pathways, while regions with slower clearance show enrichment for metabolic and mitochondrial pathways. The cell-type enrichment analysis further implicates excitatory and inhibitory neurons, oligodendrocyte lineage cells, microglia and, to a lesser extent, astrocytes. This provides a plausible biological bridge between regional neural activity and clearance function, and the sensitivity analysis using ReHo in addition to fALFF is a useful robustness check.

   However, the manuscript should be more careful in its causal interpretation. The study is cross-sectional and largely correlative in space. The finding that regions with higher spontaneous neural activity tend to show better glymphatic clearance is intriguing, but it does not establish that neural activity drives clearance in these participants. Conversely, it remains possible that better tissue integrity, vascular function, CSF access, cortical geometry, vascular density, or disease composition jointly influence both fMRI measures and tracer clearance. The authors do acknowledge some of these limitations, but the abstract and discussion should more consistently frame the findings as associations rather than evidence of an activity-clearance mechanism in humans.

   The most important limitation is the small size of the fMRI subgroup. Although the whole glymphatic MRI cohort includes 96 participants, the key activity-clearance analysis is based on only 15 individuals, including 11 with peripheral neuropathy and 4 with motor neuron disease. This is a very small and clinically heterogeneous sample on which to build a central conclusion about regional neural activity and glymphatic clearance. The authors show that the 39 h PC map in the fMRI subgroup resembles the whole-cohort map, which is helpful, but this does not address whether the fALFF-clearance relationship is robust at the individual level. The paper would be strengthened by reporting subject-level stability, leave-one-out analyses, and whether the association persists after excluding the four motor neuron disease cases.

   A second major concern is the interpretation of the amyloid analysis. The ¹¹C-PiB map is derived from an external open-source Alzheimer's disease dataset, not from the same participants who underwent glymphatic MRI and fMRI. Therefore, the association between activity-clearance mismatch and amyloid burden is a spatial correspondence across group-average maps, not an individual-level relationship. This is valuable for hypothesis generation, but should not be presented as evidence that a mismatch in the present cohort predicts amyloid deposition. The authors should clearly state that this analysis tests whether mismatch regions overlap with known amyloid-prone cortical regions, rather than directly linking mismatch to amyloidosis in individual participants.

   The definition of "mismatch" also needs clarification. The text defines the mismatch index as the negative absolute difference between z-fALFF and z-39h PC, and states that higher scores indicate greater mismatch. Because the index is negative, values closer to zero would normally indicate a smaller absolute difference rather than a greater mismatch. This should be checked carefully and corrected if necessary. More broadly, because a higher 39 h PC indicates worse clearance, the interpretation of match and mismatch categories is not intuitive. The authors should provide a clearer schematic and ensure that the mathematical definition, biological interpretation and figure labelling are fully aligned.

   Several technical confounds require more attention. Intrathecal gadolinium MRI is influenced by CSF dynamics, posture, sleep, circadian timing, renal clearance, age, intracranial pathology, and potentially diagnosis-specific differences. The authors acquired scans at fixed time points and noted that patients slept as usual, but individual sleep duration, sleep quality, posture, and daytime activity were not objectively measured. Given that the central claim concerns glymphatic clearance, these are not minor confounders. The authors should consider adjusting for age, sex, diagnosis, vascular risk factors, and relevant clinical variables where possible, and be more explicit about how heterogeneous disease indications may influence cortical tracer kinetics.

   The statistics are generally good. However, many correlations are performed across 400 cortical parcels, which are not independent biological samples. The paper would benefit from clearer separation between participant-level inference and region-level spatial inference. For example, the fALFF-clearance and mismatch-amyloid analyses are regional map correlations, not correlations across individuals. This should be clearly stated throughout. The authors should also report effect sizes and confidence intervals more consistently, and explain how multiple comparisons were controlled across transcriptomic, cell-type, fMRI, ReHo and amyloid analyses.

   The transcriptomic analysis is useful but should be presented as indirect. AHBA data come from six post-mortem brains; only the left hemisphere was used, and the donors were healthy and younger than the clinical cohort. Therefore, these data capture intrinsic regional gene-expression patterns rather than disease-state expression in the same individuals. The authors should avoid implying that the transcriptomic findings directly explain glymphatic function in their participants. The current discussion partly acknowledges this, but the framing in the abstract and results could be more cautious.

   There are also several points of presentation that should be improved. The manuscript should consistently distinguish glymphatic influx, glymphatic clearance, CSF tracer retention, and waste clearance. A 39 h residual gadolinium signal is a useful proxy for delayed clearance, but it is not the same as direct measurement of amyloid or tau clearance. The language around "waste clearance" and "amyloidosis" should therefore be precise. The authors should also clarity whether "higher clearance" corresponds to lower 39 h PC across all analyses, as this inversion is easy for readers to misinterpret.

   Read the original source
5. 

   Version published to 10.64898/2026.04.23.720377 on bioRxiv

   Apr 25, 2026