Psychological traits may uncover why Alzheimer’s biology differs between patients

By Vijay Kumar MalesuReviewed by Susha Cheriyedath, M.Sc.Jun 22 2026

A postmortem multi-omics study suggests that long-standing traits such as neuroticism, loneliness, and purpose in life may track distinct molecular routes through Alzheimer’s dementia, offering clues to disease heterogeneity that standard pathology alone does not explain.

Study: Associations of stable psychological traits with multi-omic subtypes of Alzheimer’s dementia. Image Credit: Inna Kot / Shutterstock

In a recent study published in the journal Translational Psychiatry, a group of researchers investigated whether long-standing psychological traits are associated with specific multi-omics molecular subtypes of Alzheimer’s dementia.

Background

Did you know that psychological factors such as neuroticism, loneliness, and a lack of purpose in life are strongly linked to a greater risk of cognitive decline and dementia? Previous research indicates that psychological variables may affect cognition and dementia risk, but biological mechanisms connecting these traits to Alzheimer’s disease (AD) remain poorly understood.

Traditional studies have focused on neuropathology, brain imaging, or single biological markers, producing mixed results. New multi-omics technologies enable researchers to examine multiple genetic, epigenetic, transcriptomic, proteomic, and metabolomic changes in an individual simultaneously at high resolution.

The present study examined whether psychological traits are associated with these complex molecular changes in Alzheimer’s dementia.

About the Study

The study included 822 deceased older adults from the Religious Orders Study (ROS) and the Rush Memory and Aging Project (MAP). Participants were free of known dementia at enrollment and underwent regular clinical evaluations throughout life.

After death, brain tissue from the dorsolateral prefrontal cortex was collected and analyzed using multi-omics platforms, including deoxyribonucleic acid (DNA) methylation, ribonucleic acid sequencing, protein expression profiling, and metabolomic assessments.

Researchers used a previously developed machine-learning framework, Multimodal Contrastive Trajectories Inference, to generate a molecular pseudotime representing progression from no cognitive impairment to AD dementia.

This framework had also identified three distinct molecular subtypes of Alzheimer’s dementia. The generalizability of the pseudotime model was examined in independent ROSMAP and Mount Sinai Brain Bank samples, whereas subtype generalizability was assessed only in an additional ROSMAP sample, as the Mount Sinai sample was too small for reliable subtype projection.

This research included neuropathologic assessment of AD pathology, cerebral amyloid angiopathy, Lewy bodies, transactive response DNA-binding protein (TDP-43) pathology, hippocampal sclerosis, infarcts, atherosclerosis, and arteriolosclerosis. In a subset of 278 participants, postmortem magnetic resonance imaging (MRI) was performed to obtain brain images. Furthermore, 802 polygenic risk scores covering 244 phenotypes and diseases were computed.

The psychological characteristics assessed at baseline or near baseline included neuroticism, depressive symptoms, loneliness, and purpose in life. Loneliness and purpose in life were available only in MAP participants, reducing the sample size for those analyses.

The statistical analysis examined associations among these psychological characteristics, molecular pseudotime, subtype assignment, neuropathology, brain morphology, and genetic risk, while controlling for age, sex, education level, time until death, and many other factors.

Study Results

The mean age at baseline across the full cohort (N = 822) was slightly over 80 years, while the mean age at death was nearly 90 years. Over 60% were female, the majority highly educated, and about a quarter carried the apolipoprotein E epsilon 4 allele, an established genetic risk factor for AD.

The molecular pseudotime successfully generalized to independent datasets. In both validation samples, higher pseudotime values were significantly associated with poorer cognitive status and more severe dementia, demonstrating that the molecular progression model remained robust across different populations. Subtype projection in the additional ROSMAP sample also showed significant concordance, supporting but not fully proving subtype portability.

The researchers then defined the biological characteristics of each subtype. Although DNA methylation and gene expression changes were common across all groups, individual subtypes exhibited highly distinct molecular signatures. AD1, AD2, and AD3 had 236, 186, and 147 distinct molecular features, respectively.

Metabolomic changes were most abundant in AD1, indicating a widespread disruption of cellular metabolism. 

Transcriptomic changes were enriched in AD2, suggesting significant differences in gene expression. AD3 showed predominant epigenomic changes, indicating greater engagement of gene-regulatory processes. The authors noted that these subtype labels are provisional and require further validation.

The strongest distinguishing molecular markers were metabolites rather than genes or proteins. Key features included phosphatidylcholine acyl-alkyl C38:4 in AD1, lysophosphatidylcholine acyl C20:3, and phosphatidylcholine acyl-acyl C36:6 in AD2, and glutamate in AD3.

There were similarities found across the three multi-omic molecular subtypes. Pathologic AD was present in more than 70% of participants across AD subtypes, and cerebral amyloid angiopathy was also common, with no significant subtype differences.

In addition, nearly 50% of the participants demonstrated a non-Alzheimer neurodegenerative pathologic condition. The frequency of infarcts, arteriolosclerosis, and atherosclerosis was not significantly different between subtypes.

Postmortem MRI showed greater brain atrophy in all AD subtypes compared with participants who had no cognitive impairment. AD1 displayed the most extensive cortical atrophy involving temporal, frontal, and parietal regions.

AD2 primarily showed temporal lobe involvement extending toward the temporal pole. 

AD3 exhibited the least severe atrophy, largely confined to the temporal lobe and insular cortex. Despite these patterns, differences between subtypes did not reach statistical significance.

According to genetic analyses, polygenic risk scores were significantly associated with molecular pseudotime, suggesting a link between genetic predisposition and disease progression. However, there was no significant difference in polygenic risk score between AD subtypes. Therefore, polygenic risk appeared more closely related to overall AD susceptibility than to subtype-specific molecular pathways.

After showing that traditional pathology, atrophy, and genetic risk did not clearly separate the molecular subtypes, the researchers examined whether psychological traits showed subtype-specific patterns.

Higher molecular pseudotime, indicating closer proximity to AD dementia, was associated with higher scores of neuroticism and loneliness, while depressive symptoms and purpose in life were not associated with pseudotime. Neuroticism was related to an increased likelihood of all AD types. 

The researchers found a lower likelihood of AD1 and AD2 among individuals with a greater sense of purpose. After accounting for neuropathologic changes, neuroticism remained associated with AD1 and AD3, loneliness remained associated with AD3, and purpose remained inversely associated with AD1 and AD2. However, after correction across the four psychological traits, only neuroticism with AD1 and purpose with AD1 and AD2 remained statistically significant.

These associations held in part even after accounting for neuropathologic changes, suggesting links beyond traditional pathological changes.

Conclusion

The findings suggest that stable psychological traits are associated with specific molecular subtypes of Alzheimer’s dementia. While genetic risk, neuropathology, and brain atrophy distinguished individuals with AD from cognitively healthy participants, they did not differentiate among molecular subtypes.

Psychological traits such as neuroticism, loneliness, and life purpose showed unique associations, although not all remained significant after multiple-comparison correction. Results show that polygenic risk is more closely related to overall AD susceptibility, while psychological traits may align with distinct biological pathways underlying the heterogeneity of the disease. However, the observational postmortem design means the study cannot prove that psychological traits caused these molecular changes. 

The authors also noted limitations, including reliance on self-reported psychological measures, smaller samples for loneliness and purpose, the lack of in vivo PET or plasma biomarkers, and omics data from a single brain region. These findings provide new insight into how long-term psychological characteristics may relate to the biological mechanisms underlying Alzheimer’s dementia.

Download your PDF copy by clicking here.