Microglial Activity Patterns Track Tau Buildup in Early Alzheimer’s Disease

Researchers in Munich have found that the spatial geography of the brain’s inflammatory activity carries detailed information about where tau, the misfolded protein that spreads through the Alzheimer’s disease brain, is likely to accumulate. In a prospective multimodal neuroimaging study of 56 participants, a team at Ludwig Maximilian University Hospital Munich showed that two distinct signatures of microglial activation, measured with TSPO positron emission tomography, relate to tau deposition in fundamentally different ways, and that this relationship is organized by the brain’s own functional network architecture.

The study, published in Alzheimer’s Research & Therapy, enrolled 35 amyloid-positive individuals across the Alzheimer’s disease spectrum and 21 amyloid-negative healthy controls. Amyloid status was classified primarily using the cerebrospinal fluid Aβ42/Aβ40 ratio in 45 participants, with expert visual assessment of baseline amyloid PET serving as a backup in 11 participants for whom eligible CSF was unavailable. At baseline, all participants underwent TSPO PET using the tracer [18F]GE-180, resting-state functional MRI, structural MRI, and clinical assessment. A longitudinal subgroup returned approximately 18 months later: 24 participants underwent follow-up tau PET with [18F]PI-2620, and 18 underwent repeat TSPO PET, yielding 16 participants with complete longitudinal TSPO and follow-up tau data.

Microglia, the resident immune cells of the brain, express the 18 kDa translocator protein, or TSPO, a mitochondrial membrane protein whose density rises with glial activation. TSPO PET therefore provides an in vivo marker of one component of neuroinflammation. The relationship between microglial activation and Alzheimer’s pathology has long been ambiguous: experimental work shows microglia can phagocytose aggregated amyloid-beta and tau, and early inflammatory responses are thought to be protective, yet chronic activation may eventually facilitate tau spread and plaque formation. Whether activation is a secondary response to pathology or an active driver remains contested, and the new study was designed to test whether functional connectivity structures the link between longitudinal microglial dynamics and tau.

The first substantive finding was, in a sense, a null result. At baseline, mean TSPO binding did not differ significantly between the Alzheimer’s disease spectrum group and controls: mean SUVR was 0.96 in patients versus 0.98 in controls, a difference that was not statistically significant. Over the roughly 18-month follow-up, TSPO change increased modestly in patients, with a mean change of 0.041, and decreased in controls, with a mean of −0.06, but again without a statistically significant group difference. Higher baseline TSPO was associated with less subsequent TSPO increase in both groups, a plateau-like pattern compatible with regression to the mean, which the authors acknowledge as a possible partial explanation.

The spatial picture was more revealing. Regions with higher baseline TSPO signal showed higher tau at follow-up in both patients and controls. At the subject level in the patient group, both higher baseline TSPO and higher longitudinal TSPO change were associated with higher tau, with exponentiated coefficients of 1.91 and 1.82 respectively, both statistically significant. In the reciprocal direction, higher regional tau was associated with greater TSPO change in both groups, though the association was numerically steeper in controls; this group difference was not statistically significant at the participant level, and the authors caution that no causal directionality can be inferred. They also note that in amyloid-negative older adults, the tau pattern observed is compatible with primary age-related tauopathy rather than an Alzheimer’s-specific cascade.

The study’s central contribution came from combining the PET data with resting-state functional connectivity, computed from correlations between regional time series across 200 cortical parcels of the Schaefer atlas. The team defined epicenters as the top 20 percent of regions by a given tracer signal, then measured each region’s connectivity to that epicenter set. Connectivity to baseline-TSPO epicenters was associated with higher tau in both groups. Connectivity to TSPO-change epicenters, however, was associated with lower tau in patients but not in controls, a group-by-connectivity interaction with a p value below 0.001, corroborated by a participant-level comparison of individual slopes with a p value of 0.012. When both connectivity predictors were entered together, the change-epicenter association in patients was attenuated but remained significant, and it proved robust across epicenter thresholds from 5 to 20 percent.

One interpretation offered by the authors is that baseline TSPO epicenters mark regions embedded in tau-vulnerable networks, whereas TSPO-change epicenters identify regions where new inflammatory activity sits outside the dominant tau-associated network architecture. In healthy aging, tau-rich regions show stronger TSPO increases, a predominantly local coupling; in the Alzheimer’s spectrum, longitudinal TSPO increases appear spatially distinct from connectivity-associated regional tau. Experimental studies lend this framework plausibility: activated microglia can internalize and degrade tau and amyloid-beta, but dysfunctional microglia can facilitate tau spread, and tau itself induces microglial senescence, potentially producing the spatial decoupling observed in patients.

Crucially, the key association replicated out of sample. Baseline-TSPO epicenters were defined from 15 patients who contributed no tau PET data, and TSPO-change epicenters from 6 patients with repeat TSPO PET. In the primary validation using baseline fMRI, connectivity to baseline-TSPO epicenters was associated with higher regional tau in both patients and controls. Among five patients independent of both epicenter definitions, greater connectivity to baseline-TSPO epicenters was associated with higher tau, and greater connectivity to TSPO-change epicenters with lower tau, both highly significant. A sensitivity analysis using follow-up fMRI in all 11 patients with tau PET showed that the change-epicenter term was attenuated and no longer significant in the combined models, underscoring the fragility of this result in such small subgroups. The team also showed that connectivity to pseudo-change epicenters, derived purely from the cross-sectional baseline difference between patients and controls, mirrored the tau-predictive pattern of the actual longitudinal epicenters.

The authors are explicit about the study’s limits. The central longitudinal analyses rested on only five patients and 11 controls with complete repeat TSPO and follow-up tau data, and region-level mixed-effects models, however sophisticated, do not increase the number of independent participants; the connectivity findings should be considered exploratory and hypothesis-generating. The complete-data patient subgroup comprised four individuals with CDR 0.5 and one with CDR 1, with no preclinical participant, limiting inference across the disease continuum. The [18F]GE-180 tracer has relatively limited blood-brain barrier penetration compared with other second-generation TSPO tracers, warranting caution about absolute signal levels, though the primary findings rest on within-group spatial associations. No partial-volume correction was applied, the cerebellar reference region cannot be assumed entirely unaffected, low-affinity TSPO binders were excluded, and residual confounding by vascular risk, systemic inflammation, and medications cannot be excluded.

Even with those caveats, the study offers a new lens on neuroinflammation in Alzheimer’s disease: not a uniform rise or fall in immune activity, but a spatially reorganized inflammatory landscape whose relationship to tau depends on which microglial signature, and which network neighborhood, is examined. The findings suggest that connectivity-informed neuroinflammatory measures could eventually help characterize the spatial distribution of tau, and the authors propose that future work test whether this dissociation intensifies in later disease stages, replicates with more specific microglial tracers targeting molecules such as TREM2 or P2RY12, and predicts responses to immunomodulatory interventions. For now, the message is one of pattern rather than mechanism: in early Alzheimer’s disease, where the brain’s immune cells are active, and how those regions are wired, says something about where tau is heading.

Subject of Research: Longitudinal TSPO PET and functional connectivity imaging of microglial activation in relation to tau deposition in early Alzheimer's disease

Article Title: Microglial dynamics associate with the spatial pattern of tau deposition in early Alzheimer’s disease

Article References: Öchsner, M., Brendel, M., Franzmeier, N., Trappmann, L., Zaganjori, M., Ersoezlue, E., Kurz, C., Haeckert, J., Utecht, J., Schirge, P. M., Papazov, B., Pogarell, O., Janowitz, D., Buerger, K., Ewers, M., Palleis, C., Weidinger, E., Biechele, G., Schuster, S., … Rauchmann, B.-S. (2026). Microglial dynamics associate with the spatial pattern of tau deposition in early Alzheimer’s disease. Alzheimer's Research & Therapy, 18(1), Article 232. https://doi.org/10.1186/s13195-026-02230-1

Image Credits: AI Generated

DOI: 10.1186/s13195-026-02230-1

Keywords: Alzheimer's disease, microglia, TSPO PET, tau, neuroinflammation, functional connectivity, [18F]GE-180, [18F]PI-2620, resting-state fMRI, amyloid, biomarkers, longitudinal imaging