Loneliness linked to smaller brain memory regions when Alzheimer’s biomarker is elevated

Loneliness has long been treated as a matter of mental health and social well-being, but a new study suggests its imprint on the aging brain may depend on something far more biological: the presence of preclinical Alzheimer’s disease pathology. In one of the largest neuroimaging investigations of its kind, researchers analyzing baseline data from the IGNITE study found that greater loneliness was associated with smaller volumes in two critically important medial temporal lobe regions, the hippocampal subfield CA3 and the entorhinal cortex, but only among older adults whose blood showed elevated levels of plasma phosphorylated tau 217, a sensitive biomarker of amyloid-related Alzheimer’s pathology. Among participants below the biomarker threshold, loneliness showed no such regional associations at all. The findings, published in GeroScience, offer a biomarker-informed framework that may finally explain why previous loneliness-brain studies have produced such conflicting results.

The research team, led by Isabel Martín-Fuentes and Kirk Erickson, examined 601 cognitively unimpaired older adults aged 65 to 80 years enrolled in IGNITE, a multi-site exercise intervention trial conducted at the University of Pittsburgh, Northeastern University, and the University of Kansas Medical Center. All participants passed comprehensive neuropsychological testing confirming intact cognition, and none had major depression, a crucial design feature given the known overlap between loneliness, depressive symptoms, and hippocampal structure. Loneliness was measured with the revised UCLA Loneliness Scale, a 20-item questionnaire rated on a four-point scale from never to often, yielding scores from 20 to 80. The sample averaged 36.86 points, indicating moderate perceived loneliness, and scores did not differ by pTau217 status or sex. Participants were predominantly female, 72.2 percent, with an average age of 69.8 years and 16.3 years of education.

Technically, the study went beyond the total hippocampal volume measures that dominate prior literature. The hippocampus is not a unitary structure but a mosaic of anatomically and functionally distinct subfields, including cornu ammonis regions CA1 through CA3, the dentate gyrus, and the subiculum, each differing in cytoarchitecture and functional roles. The entorhinal cortex, though not part of the hippocampus proper, serves as the critical input-output gateway between neocortex and hippocampus and is among the earliest sites of Alzheimer’s-related neurodegeneration. To capture these regions, the researchers acquired harmonized 3 Tesla MRI across sites, including a high-resolution T2-weighted turbo spin echo sequence with 0.4 by 0.4 by 2.0 millimeter voxels optimized for subfield segmentation. The Automated Segmentation of Hippocampal Subfields pipeline, which applies a multi-atlas machine learning approach, extracted bilateral volumes for each subfield, while FreeSurfer estimated total hippocampal volume and intracranial volume. Monthly phantom scans and weekly quality-control checks monitored scanner consistency throughout enrollment.

The biological innovation came from plasma pTau217, measured with the ALZpath assay on a Quanterix HD-X platform. This blood-based biomarker shows strong correspondence with amyloid-beta PET positivity, including in preclinical stages of disease. Using an empirically derived, Aβ-PET-calibrated threshold of greater than 0.46 pg/mL previously established in this cohort, the team classified 582 participants with usable biomarker data as pTau217-positive or negative. Notably, 28.9 percent, 168 individuals, were pTau217-positive, meaning they carried blood-based evidence consistent with incipient amyloid pathology despite being cognitively intact. Prevalence was higher in males than females, 36.2 versus 26.0 percent. This stratification allowed the researchers to test whether loneliness-related structural differences emerge only under specific biological conditions rather than uniformly across all older adults.

The results revealed a striking biomarker-dependent pattern. Greater loneliness was associated with smaller total hippocampal volume after adjustment for age, sex, education, body mass index, study site, intracranial volume, and social network size, with a standardized coefficient of −0.077. In unstandardized terms, each point on the UCLA scale corresponded to roughly 6.5 cubic millimeters less hippocampal tissue, and a one-standard-deviation increase in loneliness corresponded to approximately 61 cubic millimeters smaller total hippocampal volume. However, no individual subfield showed a significant direct association with loneliness, with coefficients ranging from −0.068 to 0.015 across CA1, CA2, CA3, the dentate gyrus, the subiculum, and the entorhinal cortex. The regional story only became visible when pTau217 status entered the models as a moderator.

That moderation analysis produced the study’s primary novel finding. Significant loneliness-by-pTau217 interactions emerged for CA3, with a coefficient of −0.215, and for the entorhinal cortex, with a coefficient of −0.167. Simple slopes clarified the pattern: among pTau217-positive participants, greater loneliness was inversely associated with CA3 volume, β = −0.230, p = 0.001, whereas among pTau217-negative participants there was no association, β = −0.016. For the entorhinal cortex, the estimated association was negative among pTau217-positive participants, β = −0.122, though it did not reach statistical significance, while the estimated association trended positive among pTau217-negative participants. The interactions corresponded to between-group slope differences of 6.65 and 20.32 cubic millimeters per standard deviation of loneliness for CA3 and ERC, respectively. No moderation appeared for CA1, CA2, the dentate gyrus, or the subiculum. Both regional interactions survived the prespecified false discovery rate threshold of 0.20 and the more stringent 0.10 threshold, though neither met 0.05, and sensitivity analyses adjusting for APOE ε4 carriership left the findings essentially unchanged.

Why CA3 and the entorhinal cortex specifically? The authors point to converging evidence implicating both regions in Alzheimer’s-related and stress-related processes. CA3 supports pattern completion and contextual memory retrieval, functions that shape how prior experiences are recalled and interpreted, including socially relevant situations. The entorhinal cortex shows early vulnerability to amyloid-related pathology and occupies a strategic position within hippocampal networks. The researchers propose that pTau217 positivity may mark a state of reduced structural resilience, in which modest psychosocial effects become detectable at the subfield level. Models of the lonely brain, which describe hypervigilance to social threat, negative affective bias toward social cues, and altered self-processing, offer one framework: these neurocognitive features could plausibly influence how individuals appraise social information and encode salient experiences, and in biologically vulnerable individuals, loneliness-related processes such as inflammatory signaling and stress may interact with existing synaptic and metabolic vulnerability to amplify subfield-specific volume differences.

An intriguing complementary interpretation suggests loneliness may partly reflect an early, internally driven awareness of change rather than solely an external social condition. Individuals with biomarker evidence of incipient amyloid pathology may notice subtle internal signals, such as increased cognitive effort or reduced efficiency, that do not meet clinical thresholds but still alter subjective self-perception. These cues could heighten sensitivity to social experiences and increase perceived disconnection even before clinicians detect impairment. Consistent with this account, the associations persisted after adjusting for social network size, derived from Cohen’s Social Network Index, indicating they are unlikely to be explained by objective social isolation. Loneliness, the distressing subjective experience arising when perceived social needs remain unmet, can be high despite frequent contact, and the findings suggest how social experience is internally appraised may matter more than network composition when biomarker-defined vulnerability is present.

The authors are careful about interpretation. Effect sizes were modest, and statistical significance does not imply established clinical meaningfulness at the individual level, since no thresholds exist for volumetric differences of these magnitudes. The cross-sectional design precludes causal inference: loneliness may contribute to medial temporal vulnerability, biomarker-related brain changes may predispose individuals to feel lonelier, or both may operate simultaneously. Loneliness did not differ by pTau217 status, making it less likely that loneliness increases susceptibility to amyloid pathology. The sample was also predominantly female, highly educated, and largely white, and automated segmentation carries measurement variability, particularly for smaller structures. Still, the study’s strengths are substantial: a large well-characterized sample, harmonized high-resolution imaging, validated segmentation with quality control, and the integration of a sensitive blood-based biomarker that earlier loneliness-brain studies lacked. Because loneliness is potentially modifiable, the findings raise the question of whether reducing loneliness during preclinical stages could influence brain structure or cognitive trajectories, potentially alongside physical activity, which has been linked to reductions in loneliness. Longitudinal and interventional studies will be needed to determine whether these modest structural associations translate into clinically relevant outcomes, but the work establishes a clear rationale for treating loneliness as a psychosocial factor whose relationship with brain health may vary according to Alzheimer’s-related biological context.

Subject of Research: Biomarker-dependent associations between loneliness and medial temporal lobe volumes in cognitively unimpaired older adults from the IGNITE cohort

Article Title: Loneliness relates to smaller hippocampal subfield volumes in individuals with elevated plasma pTau217: results from IGNITE

Article References: Martín-Fuentes, I., Inagaki, T. K., Solis-Urra, P., Oberlin, L. E., Sewell, K. R., Karikari, T. K., Zeng, X., Kang, C., Morris, J. K., Ripperger, H. S., Grove, G. A., Huang, H., Sutton, B. P., Wan, L., Vidoni, E. D., Marsland, A. L., Hillman, C. H., McAuley, E., Kramer, A. F., … Erickson, K. I. (2026). Loneliness relates to smaller hippocampal subfield volumes in individuals with elevated plasma pTau217: results from IGNITE. GeroScience. https://doi.org/10.1007/s11357-026-02614-z

Image Credits: AI Generated

DOI: 10.1007/s11357-026-02614-z

Keywords: loneliness, hippocampus, pTau217, Alzheimer's disease, entorhinal cortex, CA3, brain volume, MRI, blood biomarker, aging, IGNITE, GeroScience