Delirium is a common and serious condition in older adults, especially during hospitalization. It is an acute, fluctuating disturbance of attention and awareness, distinct from the gradual decline seen in dementia. Delirium increases the risk of long-term cognitive decline, institutionalization, and death. Despite its clinical impact, the biological factors that predispose individuals to delirium are unclear. Previous studies have been small, relied on heterogeneous definitions, and offered little insight into genetic or proteomic risk. 

A new study in Nature Aging analyzed data from more than one million participants across multiple ancestries from four major cohorts. The authors conducted the largest genetic investigation of delirium to date and combined it with proteomic analyses to identify circulating proteins associated with risk. They aimed to describe the genetic architecture of delirium, examine its overlap with related conditions, and highlight molecular pathways that may contribute to vulnerability in aging populations. 

Methods & Findings

Raptis et al. (2025) combined data from the UK Biobank, FinnGen, All of Us, and the Michigan Genomics Initiative. After standardizing phenotype definitions and applying quality controls, the final dataset included 1,059,130 participants and 11,931 delirium cases. The authors carried out a genome-wide association study (GWAS) in each cohort before performing a fixed-effects meta-analysis across all studies. 

The meta-analysis identified Apolipoprotein E (APOE) as the strongest genetic risk factor for delirium. The APOE ε4 allele showed a significant association with increased risk. Sensitivity analyses excluding individuals with Alzheimer’s disease (AD) or high AD polygenic risk showed that the association remained. 

The authors estimated genetic correlations between delirium and other traits, finding the strongest connection with AD. They also observed links with depression, neuroticism, and several sleep traits, but no significant association was found with intelligence or educational attainment. 

Because delirium and AD shared genetic architecture, the authors applied multi-trait analysis (MTAG) using summary statistics from both disorders. MTAG identified ten loci associated with delirium, five of which were consistent across individual cohorts. The nearest genes to these replicated loci were CR1, BIN1, CLU, MS4A4A, and TOMM40. Polygenic risk scores showed modest predictive performance in European groups and lower performance in other ancestries.

The study also analyzed plasma proteomics. The authors used Olink and SomaScan data from two cohorts to test associations between circulating proteins and delirium. They identified multiple proteins linked to risk, including those involved in inflammation, complement activation, and lipid metabolism. Several of these protein associations were consistent across Olink and SomaScan data, often with the same direction of effect. 

To test potential causal relationships, the authors used Mendelian randomization. They found evidence that higher levels of certain immune-related proteins may increase delirium risk. The number of proteins with suitable genetic instruments was limited, and results varied across methods. Finally, pathway enrichment integrating genetic and proteomic results highlighted immune processes, complement signaling, and lipid transport. 

What This Tells Us

The findings show that delirium has a measurable genetic component, with APOE emerging as the clearest signal. The persistence of the APOE association after excluding individuals with AD indicates that its effect on delirium is not explained only by underlying dementia. This suggests that some biological pathways linked to neurodegeneration may also influence acute cognitive vulnerability. 

However, the overlap between delirium and AD extends beyond APOE. The genetic correlation and MTAG loci associated with genes such as CR1, BIN1, CLU, MS4A4A, and TOMM40 show that both conditions share elements of immune regulation, complement activity, and lipid transport. These shared pathways are involved in mechanisms that affect brain resilience, inflammation, or microglial function in later life. 

The proteomic associations reinforce this. Proteins involved in inflammation and complement activation were linked to delirium risk, and Mendelian randomization suggested that some may play a causal role. Although only a small set of proteins had strong genetic instruments for Mendelian randomization, the agreement between genetic and proteomic signals supports the suggestion that immune activity contributes to susceptibility. 

Together, the results point to delirium as a condition shaped by both neurodegenerative and inflammatory processes. Genetic risk scores remain modest, indicating that genetics explains only part of individual vulnerability, but the combined analyses highlight pathways that may influence why some people develop delirium during illness or hospitalization while others do not. 

Outlook

Delirium remains understudied compared with other conditions common in older adults because it is transient, difficult to classify, and affects patients who are often underrepresented in research. Most work has focused on clinical risk factors, leaving limited insight into the biology that shapes vulnerability. This study provides a foundation for future research while highlighting remaining gaps. The modest performance of polygenic risk scores, the limited diversity across cohorts, and the small number of proteins suitable for causal testing show the limits of current data.

Greater representation of non-European populations will be important. Genetic architecture can differ across ancestries, and the lower predictive performance of risk scores outside European groups shows the need for broader sampling. More precise and consistent phenotyping is also needed, as current definitions vary across cohorts and health systems. 

The proteomic findings suggest that immune and complement pathways may contribute to delirium risk, but these signals require mechanistic work. Longitudinal data, experimental models, and integration with environmental exposures will help clarify how these pathways interact with age, illness, and hospital care. 

As populations age, delirium will place increasing pressure on health systems. A better understanding of its biological basis may support better prevention and treatment strategies, but this will require sustained investment. This work establishes a genomic and proteomic map that future research can use to explore mechanisms that influence cognitive vulnerability in older adults. 

Reference

Raptis V, Bhak Y, Cannings TI, MacLullich AMJ, Tenesa A. Dissecting the genetic and proteomic risk factors for delirium. Nat Aging. Published online November 24, 2025. doi:10.1038/s43587-025-01018-6