https://orcid.org/0000-0001-9479-0143
Figures
Abstract
Background
Postoperative delirium is the most common postoperative complication in older individuals. Genome-wide association studies (GWAS) can provide insights into how genetic factors influence postoperative risk. We examined the genetic architecture of postoperative delirium after major surgery and its relationship with related cognitive conditions (delirium of any type and Alzheimer’s disease, including the APOE ε4 allele).
Methods and findings
A case-control GWAS was performed in the UK Biobank to identify genetic variants associated with postoperative delirium, adjusted for age, sex, genetic chip, and the first 10 principal components. These results were then used in genetic correlation and polygenic risk score analyses to investigate shared genetic risk between postoperative delirium and a) delirium of all causes, and b) Alzheimer’s disease.
The GWAS (1,016 cases, 139,148 controls) identified seven Single Nucleotide Polymorphisms (SNPs) that mapped to four genes (APOE, TOMM40, APOC1, and PVRL2); p < 5 x 10−8. Five SNPs remained significant after excluding pre-existing dementia, and two after excluding subsequent dementia. The lead SNP was rs429358, a missense variant of APOE. Genetic correlation and polygenic risk score analyses revealed evidence of shared genetic architecture and risk between postoperative delirium and Alzheimer’s disease (rho 0.68, 95% CI [0.46, 0.81]; p < 0.001). After adjustment for age and sex, the APOE ε4 isoform had a dose-response effect on risk (odds ratios for one and two copies: 1.75, 95% CI [1.53, 2.0], and 4.19, 95% CI [3.25, 5.41], respectively; p < 0.001). The main limitations of the study include the reliance upon clinical coding for outcome definition and limited statistical power to detect small or modest genetic effects.
Conclusions
We identified genetic variants associated with increased risk of postoperative delirium. We also found evidence of shared genetic liability with Alzheimer’s disease via APOE, complementing recent large-scale studies in all-cause delirium. If validated, the findings have potential clinical applications, including preoperative risk stratification and early identification of pre-clinical Alzheimer’s disease risk.
Author summary
Why was this study done?
- Postoperative delirium is the most common postoperative complication in older individuals.
- Little is known about the genetic basis of postoperative delirium.
- Knowledge of the genetic risk factors could help with treatment or prevention in the future.
What did the researchers do and find?
- We performed a genome-wide association study of postoperative delirium including 1,016 cases and 139,148 controls.
- We found genetic variants associated with increased postoperative delirium risk.
- We also found genetic similarity between postoperative delirium, all-cause delirium, and Alzheimer’s disease.
What do these findings mean?
- Postoperative delirium has shared biology with Alzheimer’s disease and may represent unmasking of pre-clinical disease.
- Individuals with genetic risk for Alzheimer’s disease have an increased risk of postoperative delirium and vice versa.
- Due to the relatively small number of cases and the fact that we may not have detected all the true positive diagnoses in the dataset, we were unable to detect small or modest genetic effects in this study.
Citation: Armstrong RA, Yousefi P, Gibbison B, Khandaker GM, Gaunt TR (2026) The genetic architecture of postoperative delirium after major surgery and its relationship with nonpostoperative neurocognitive conditions: A genome-wide association study. PLoS Med 23(3): e1004963. https://doi.org/10.1371/journal.pmed.1004963
Academic Editor: Martin Schreiber, Uniformed Services University: Uniformed Services University of the Health Sciences, UNITED STATES OF AMERICA
Received: September 9, 2025; Accepted: February 12, 2026; Published: March 2, 2026
Copyright: © 2026 Armstrong et al. This is an open access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The individual-level participant data for the primary GWAS and PRS analyses are available from UK Biobank via their controlled access process (https://www.ukbiobank.ac.uk/use-our-data/). FinnGen GWAS summary statistics are available from the FinnGen website (https://www.finngen.fi/en). Author-generated code to support the analyses is available on Github (https://github.com/raarmstrong/gwas-postop-delirium/; DOI: 10.5281/zenodo.18196799). GWAS summary statistics for the primary analysis are available from the University of Bristol data repository, data.bris, at https://doi.org/10.5523/bris.1m83zai2e26yq2lro3tixz9kqq.
Funding: This study was funded by a Wellcome Trust GW4-CAT PhD Programme for Health Professionals PhD Fellowship awarded to RAA [316275/Z/24/Z]. RAA, PY, GMK, and TRG are supported by the Medical Research Council Integrative Epidemiology Unit at the University of Bristol (RA, TRG: MC_UU_00032/3; PY: MC_UU_00032/4; GMK: MC_UU_0032/6). GMK acknowledges additional funding from the Wellcome Trust (grant numbers: 201486/Z/16/Z and 201486/B/16/Z), the Medical Research Council (grant numbers: MR/W014416/1; MR/S037675/1; MR/Z50354X/1; and MR/Z503745/1. PY, TRG, and GMK are also supported by the UK National Institute for Health and Care Research (NIHR) Bristol Biomedical Research Centre (grant number: NIHR 203315). The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript. https://wellcome.org/; https://www.ukri.org/councils/mrc/; https://www.bristolbrc.nihr.ac.uk/.
Competing interests: I have read the journal’s policy and the authors of this manuscript have the following competing interests: TRG receives funding from GlaxoSmithKline, Biogen and Roche for unrelated research. The other authors have declared that no competing interests exist.
Abbreviations: APOE, apolipoprotein E; ASA, American Society of Anesthesiologists; CCI, Charlson Comorbidity Index; FUMA GWAS, Functional Mapping and Annotation of Genome-Wide Association Studies; GWAS, genome-wide association studies; ICD-10, International Classification of Diseases, Tenth Revision; LAVA, Local Analysis of [co]Variant Association; MAGMA, Multi-marker Analysis of GenoMic Annotation; OPCS-4, Office of Population Censuses and Surveys Classification of Interventions and Procedures, Version 4; PRS, polygenic risk scores; SNPs, single nucleotide polymorphisms.
Introduction
Postoperative delirium is the most common postoperative complication in older individuals, with an incidence of up to 65% [1,2]. It has substantial health and resource implications including increased morbidity, mortality, and healthcare costs [1,3]. Hypotheses of the underlying mechanisms span a range of pathways including release of excitotoxic neurotransmitters, neuronal injury and inflammation, physiological stressors, metabolic derangements, electrolyte disorders, and genetic factors [1,2,4–8]. Postoperative delirium has been associated with an increased risk of cognitive decline and incident dementia [9], but there is conflicting evidence on how the acute occurrence of postoperative delirium interacts with longer-term cognitive decline and neurodegeneration [10].
Whilst the ageing global population and increasing numbers of older adults undergoing surgery have made perioperative brain health an increasing focus of research and clinical initiatives [11], the ability of clinical teams to produce personalised assessments of clinical risk for individuals remains limited. Improved detection of underlying biological risk in conjunction with established clinical risk factors could aid in preoperative risk prediction, facilitate better-informed shared-decision-making and allow targeting of multicomponent interventions to prevent delirium [11] with potential reductions in morbidity, mortality, and resource utilisation.
Population-based genome-wide association studies (GWAS) are a powerful approach for identifying genetic risk factors associated with complex conditions of multifactorial aetiology. Their results can aid prediction but also identify underlying biological mechanisms and potential targets for intervention. They have been extensively applied to a wide range of clinical phenotypes [12], but rarely to postoperative delirium specifically. Previous GWAS that have included postoperative neurocognitive complications have failed to identify any genome-wide significant associations [13–15]—most likely because they suffered from limited sample sizes and heterogeneity of outcomes.
This study aimed to evaluate the association between genetic variants genome-wide and postoperative delirium following major surgery, and to explore whether postoperative delirium shares genetic risk factors with related conditions, specifically delirium of all causes and Alzheimer’s disease. The limitations of previous studies are addressed by defining a major surgery cohort in a large-scale national biobank and focussing on a single outcome phenotype.
Methods
Study populations
The data source for the primary analysis was UK Biobank, a long-term prospective cohort study that provides a large-scale biomedical database and research resource containing genetic, lifestyle and health information from half a million UK participants recruited between 2006 and 2010 (http://www.ukbiobank.ac.uk). The UK Biobank study was approved by the North-West Multi-centre Research Ethics Committee and all participants provided written informed consent [16]. Additionally, publicly available GWAS summary statistics from the FinnGen study were used as an external cohort for genetic correlation analyses. FinnGen is a large-scale genomics initiative that has analysed over 500,000 Finnish biobank samples and correlated genetic variation with health data to understand disease mechanisms and predispositions. The project is a collaboration between research organisations and biobanks within Finland and international industry partners [17].
Genotyping
UK Biobank genotyping was performed by Affymetrix using two purpose-designed arrays. Genotype data were quality controlled, phased and ~96 million genotypes were imputed using the Haplotype Reference Consortium and UK10K haplotype resources. FinnGen genotyping used a ThermoFisher Axiom custom chip array with imputation to a reference dataset derived from Finnish whole-genome sequences. Further information on the methods and QC pipelines is available [16,18].
Inclusion and exclusion criteria
Eligible participants were those undergoing major, inpatient surgery after their date of enrolment in the UK Biobank. Linked hospital inpatient data is available for all UK Biobank participants with follow-up to 31 May 2022 for Wales, 31 August 2022 for Scotland and 31 October 2022 for England [19]. Major surgery was defined by Office of Population Censuses and Surveys Classification of Interventions and Procedures, Version 4 (OPCS-4) code using the Bupa Schedule of Procedures and Abbott classification [20,21], and diagnoses of delirium were identified using International Classification of Diseases, Tenth Revision (ICD-10) code F05. Exclusion criteria included planned day case surgery and a previous diagnosis of delirium (see S1 Appendix for more detail on phenotype definition and exclusions).
Outcome definition
The primary outcome was a first diagnosis of delirium within 30 days of surgery, including diagnoses made during readmission. Controls were defined as those undergoing major surgery with no delirium diagnosis within 30 days. The 30-day window was used to align with recommendations on the nomenclature of postoperative delirium [22] and other widely-used sources [23].
Genome-wide Association Study (GWAS)
Case-control GWAS adjusted for age, sex, genetic chip, and the first 10 principal components was performed using regenie (v3.2.2) [24]. Participants with missing data for these variables were excluded. Quality control and ancestry definition were performed by previously established workflows and methods [25]. Post-hoc analyses were conducted using Functional Mapping and Annotation of Genome-Wide Association Studies (FUMA GWAS [26]). Independent significant Single Nucleotide Polymorphisms (SNPs) were defined as those meeting a genome-wide significance threshold of p < 5 x 10−8 and independent of each other at r2 < 0.6. Lead SNPs were identified from independent significant SNPs at r2 < 0.1. Gene-based analysis was performed using Multi-marker Analysis of GenoMic Annotation (MAGMA [27]).
Sensitivity analyses
The primary GWAS analysis did not include additional clinical covariates as they may be mediators of genetic risk or colliders given the selection of a surgical patient cohort. To explore the robustness of the primary analysis the following sensitivity analyses were performed: a) after excluding participants who had a previous dementia diagnosis; b) excluding participants who had a previous dementia diagnosis or were diagnosed with dementia after the index procedure; c) stratified by cardiothoracic/noncardiothoracic surgery; d) adjusted for Charlson Comorbidity Index (0–1 versus ≥ 2, [28]); and e) conditional on Apolipoprotein E (APOE) ε4 haplotype count (0, 1 or 2) to identify associated variants independent of APOE.
Post-GWAS follow-up analyses
Data-driven exploratory follow-up analyses were informed by the results of the primary GWAS analysis. Two approaches were taken to formally assess shared genetic architecture between postoperative delirium and a) delirium of any type; b) Alzheimer’s Disease: genetic correlation analysis and polygenic risk scoring.
Genetic correlation analysis
Genetic correlation analyses were performed between the UKBiobank GWAS results presented here and the external FinnGen study cohort. GWAS summary statistics (Freeze 9, https://r9.finngen.fi/) were obtained for a) delirium (any cause); and b) Alzheimer’s disease. Local Analysis of [co]Variant Association (LAVA [29]) was used to analyse bivariate local genetic correlation between these phenotypes and postoperative delirium (using the summary statistics from the primary GWAS analysis in UK Biobank). The genome is partitioned into 2,495 blocks for LAVA analysis, so a Bonferroni-adjusted p-value of 0.05/2,495 (2 x 10−5) was used to define statistical significance.
Polygenic risk score analysis
Polygenic risk scores (PRS) calculate a weighted sum of risk based on alleles associated with a disease or outcome phenotype. The Polygenic Score Catalog (https://www.pgscatalog.org/) was used to identify polygenic risk scores for Alzheimer’s disease. No polygenic risk scores were identified for delirium. The pgsc_calc workflow [30] was used to download, variant match and calculate individual polygenic scores for the primary UK Biobank cohort. Individual scores were then centred and scaled before being summed and averaged. Odds ratios for the association between increasing quintile of polygenic risk score for Alzheimer’s disease and postoperative delirium were estimated by logistic regression with adjustment for age and sex using the glm function in R (version 4.4.0). Sensitivity analyses were performed with additional adjustment for Charlson Comorbidity Score (0–1 versus ≥ 2, [28]). A Bonferroni correction for the number of comparisons against the reference quintile was applied, resulting in a significance threshold of p < 0.0125 (0.05/4).
APOE genotype analysis
It has been recommended that the APOE region of chromosome 19 should be excluded from Alzheimer’s disease polygenic risk scores, with the APOE ε4 allele included as an additional covariate in modelling, due to its outsize effect relative to other regions across the genome [31]. The subset of polygenic scores identified above which did not include the APOE region (defined as a 10 kilobase window either side of region from the start position of TOMM40, chr19:45,394,477, to the stop position of APOC1, chr19:45,422,606, build 37/hg19 [31]) were therefore considered in an additional analysis, adjusted for age, sex and APOE ε4 allele count (0, 1, 2).
This study is reported as per the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guideline (S1 STROBE Checklist).
Results
Participant characteristics
A total of 1,016 cases of postoperative delirium and 139,148 controls were identified (Fig 1). Individuals who developed postoperative delirium tended to be older, male, and had a higher prevalence of comorbidities (Table 1). Compared to controls, a higher proportion of patients in the postoperative delirium group underwent emergency (532 (52%) versus 29,530 patients (21%)) and cardiothoracic surgery (172 (17%) versus 5,041 patients (3.6%)).
Genome-wide significant associations for postoperative delirium
Seven independent, statistically significant SNPs were identified within a single genetic risk region on chromosome 19 (Fig 2 and Table 2). The Q-Q plot showed no evidence of systematic inflation (genomic inflation factor (λ) 1.01, Fig 3). The lead SNP was rs429358, a missense variant in the APOE gene that, together with rs7412, defines the ε2/ε3/ε4 isoforms linked to Alzheimer’s disease risk. A total of 21 genes were mapped to this region, including four (APOE, APOC1, TOMM40, and PVRL2) that were significantly associated in MAGMA gene-level analysis (Fig 4 and Table B in S2 Appendix). The Q-Q plot showed no evidence of systematic inflation (genomic inflation factor (λ) 1.0, Fig 5).
The red line represents a genome-wide significant p value of 5 x 10−8.
Genomic inflation factor (λ) 1.01.
The red line represents a genome-wide significant p value of 2.5 x 10−6.
Genomic inflation factor (λ) 1.0.
GWAS results after excluding existing or incident dementia
After excluding individuals with a pre-existing diagnosis of dementia (56 cases (5.7%) and 375 controls (0.3%)), 5 of the 6 independent significant SNPs remained significant. However, rs75627662 (a downstream variant of APOE) no longer reached genome-wide significance. Three genes, APOE, APOC1, and TOMM40, remained significant on MAGMA testing (Figs A and B in S2 Appendix). When patients who went on to be diagnosed with dementia during UK Biobank follow-up were also excluded (141 cases, 13.9%, and 2,358 controls, 1.7%), two SNPs remained genome-wide significant, rs429358 and rs157592. The MAGMA results were unchanged (Figs C and D in S2 Appendix).
GWAS results in cardiothoracic and noncardiothoracic surgery groups
The results for the noncardiothoracic surgery subgroup (798 cases, 126,973 controls) were similar to those in the overall cohort with five independent significant SNPs identified in the same risk locus on chromosome 19 and three of the four genes remained significant on MAGMA testing (APOE, APOC1, and TOMM40). The lead SNP was the same as in the primary GWAS and the other variants either overlapped with the primary analysis or were in strong linkage disequilibrium (r2 > 0.97) with them (Table C in S2 Appendix). The cardiothoracic analysis was likely underpowered and returned no genome-wide significant results at p < 5 x 10−8. Multiple SNPs reached a lower indicative threshold of significance including the lead SNP from the primary analysis, rs429358 (p = 3.8 x 10−7; Table D in S2 Appendix).
GWAS results adjusted for CCI
A total of 495 (48.7%) cases and 26,659 (19.2%) controls had a Charlson Comorbidity Index (CCI) ≥2 (χ2 = 562.42, p < 0.001). GWAS results after adjustment for CCI were consistent with the primary analysis, returning the same seven independent significant SNPs on chromosome 19 (Table E in S2 Appendix).
GWAS results conditional on APOE ε4 allele count
APOEε4 allele counts differed between cases and controls (0, 1 and 2 copies in 566 (57.4%), 348 (35.3%), and 71 (7.2%) cases versus 95,489 (71.6%), 34,754 (26.1%), and 3,082 (2.3%) controls; χ2 = 159.37, p < 0.001). After adjustment for APOE ε4 allele count, all the independent significant SNPs from the primary GWAS analysis lost statistical significance (Fig E in S2 Appendix).
Genetic correlation with Alzheimer’s disease
Significant local genetic correlation was identified on chromosome 19 between postoperative delirium and both delirium of all causes and Alzheimer’s disease (Table 3). The correlated region (chr19:45040933-45893307) encompassed the independent significant SNPs from the primary GWAS.
Association of genetic predisposition for Alzheimer’s disease with postoperative delirium
Prior to averaging, 42 PRS for Alzheimer’s disease were identified and passed variant matching (Table F in S2 Appendix). Increased polygenic risk for Alzheimer’s disease was associated with increased risk of postoperative delirium in the UK Biobank cohort after adjusting for age and sex (adjusted OR PRS quintile 4: 1.49 (95% CI [1.2, 1.85]), PRS quintile 5: 2.29 (95% CI [1.87, 2.8]); p < 0.001; Table 4). These results were largely unchanged in a sensitivity analysis adjusting for CCI (Table G in S2 Appendix).
Effect of APOE genotype on the association of Alzheimer’s disease risk with postoperative delirium
Fifteen of the polygenic risk scores identified for Alzheimer’s disease contained no variants in the APOE region (APOE-independent scores, Table F in S2 Appendix). The effect size of increasing PRS quintile was reduced compared to APOE-inclusive scores and the presence of one or more APOE ε4 isoforms was associated with increased risk of postoperative delirium (Table 4). Similar results were seen with adjustment for CCI (Table G in S2 Appendix).
Discussion
In this study, using a large national biobank with linked healthcare data, we identified a genetic risk locus for postoperative delirium on chromosome 19 which included seven independent significant SNPs and four mapped genes meeting genome-wide significance. Follow-up analyses in an independent cohort confirmed the presence of shared genetic risk between postoperative delirium and Alzheimer’s disease.
Previous GWAS which have included postoperative delirium as an outcome have failed to find any genome-wide significant results, but were likely underpowered [13–15]. Our study, to the best of our knowledge the largest GWAS of postoperative delirium to date, is substantially better powered than previous analyses and implicates APOE—an established genetic determinant of Alzheimer’s disease—in postoperative delirium risk. The lead SNP identified (rs429358) remained significant when patients with a pre-existing or subsequent diagnosis of dementia were excluded, as did the significance of the APOE gene on MAGMA testing. We have identified additional genes associated with postoperative delirium (APOC1, TOMM40, and PVRL2), which were previously reported to be associated with a range of neurocognitive phenotypes including dementia, cognitive function and decline, brain biomarkers, and cerebral imaging metrics (Table B and Fig F in S2 Appendix). However, a sensitivity GWAS conditional on APOE ε4 allele count resulted in the loss of significance of all SNPs in the APOE region suggesting APOE is the sole contributing factor at this genetic risk locus. We did not replicate the results of an earlier candidate gene study that identified an association between the cholinergic genes CHRM2 and CHRM4 and postoperative delirium [14] (all three identified variants showed null effects, Table H in S2 Appendix). However, we used a genome-wide approach which has methodological advantages including increased genetic coverage, reduced bias and lower false positive rates and therefore represents a more robust analysis [32].
Post-GWAS follow-up analyses showed that genetic predisposition for Alzheimer’s disease is associated with risk of perioperative delirium, including a dose-response effect of APOE ε4 allele count. Previous studies investigating the association between APOE ε4 and postoperative delirium have shown mixed results. A recent study of 19,331 patients (2086 of whom had postoperative delirium) found APOE ε4 carriers to be at increased risk of neurocognitive disorders compared with noncarriers [33]. However, the neurocognitive traits in this study were defined using Phecodes—which include a wider range of medical diagnoses than the more specific ones used in our study [34]—and the authors did not report the time point at which outcomes were assessed. Several other studies have found no association [35–37] whilst one small study (29 cases of postoperative delirium) did report increased risk with the ε4 allele [38].
Taken together, these data suggest shared genetic architecture between postoperative delirium, delirium in general, and Alzheimer’s disease via APOE. This is supported by a large-scale genetic meta-analysis of all-cause delirium which identified APOE as a strong delirium risk factor and showed that incorporating ε4 status into prediction models improved incident delirium prediction [39]. The study identified a further four risk loci for delirium which we did not replicate in our postoperative delirium cohort, likely due to differences in statistical power and analytical approach. We did, however, find that APOE-independent PRS for Alzheimer’s disease were associated with postoperative delirium, mirroring the smaller contribution made by other genomic regions. A role for APOE-related pathways in postoperative neurocognitive outcomes has biological plausibility through effects on neuroinflammation and blood-brain barrier integrity [40]. In the perioperative setting, APOE ε4 carriers have shown greater postoperative decreases in functional connectivity in key Alzheimer’s disease risk regions [41]. Further validation of these findings may lead to potential clinical applications including preoperative risk stratification based on associated variants or existing polygenic risk scores (e.g., for Alzheimer’s disease) and the identification of potential therapeutic targets for the reduction of postoperative delirium risk. Alternatively, the acute occurrence of postoperative delirium might represent an unmasking of underlying neurocognitive vulnerability or an undiagnosed chronic condition, e.g., pre-clinical Alzheimer’s disease. If genetic liability for Alzheimer’s disease extends to postoperative delirium, patients found to be at increased risk of delirium may benefit from additional monitoring long-term.
The strengths of this study include the use of a large prospective cohort with linked healthcare data that is subject to rigorous quality control and validation procedures. We defined our exposure using published criteria and employed robust statistical methods to deal with the large case-control imbalance. We were also able to extend our findings through a range of post-GWAS analyses using an independent validation cohort. The study does have some limitations. The diagnosis of delirium relied upon ICD-10 clinical coding and was not prospectively screened for. Validation studies have shown that ICD-10 coding in administrative databases has low sensitivity but excellent specificity for postoperative delirium compared to prospective clinical screening [42,43]. As such, there will be some misclassification bias, particularly true cases which are included in the control group. This will likely have attenuated the observed genetic effects and reduced our statistical power to detect small or modest effects. Whilst it is also possible that some cases of delirium preceded surgery, the risk of this was minimised as far as possible. It is also possible that some cases in the 30-day follow-up period were related to other factors, for example, infection or hospitalisation, but these occurred in the context of recovery from major surgery so were included in line with recommendations on the nomenclature of postoperative cognitive changes. Whilst UK Biobank represents a large cohort of patients there is a recognised “healthy volunteer” selection bias [44] and so the results may not be entirely generalisable to the wider UK population. Any secondary data analysis relying on routinely-collected administrative data is limited by the fields available, with some potentially relevant factors such as the American Society of Anesthesiologists (ASA) physical status grade, surgical urgency and type of anaesthesia not recorded. Finally, whilst rates of some comorbidities and surgical characteristics differed between cases and controls, these variables may lie on the causal pathway or act as selection-related colliders in a major surgical cohort. We therefore did not adjust for them in the primary GWAS but examined their influence through a range of sensitivity analyses which support the robustness of our findings.
In conclusion, we have used a large national biobank with linked healthcare data to identify seven SNPs and four genes associated with increased risk of postoperative delirium. Data from an independent cohort suggests there is shared genetic architecture and risk between postoperative delirium, delirium more broadly, and Alzheimer’s disease. If validated, these findings have potential clinical benefit through risk stratification based on genotype or polygenic risk score data and early identification of future Alzheimer’s disease risk in the perioperative setting.
Supporting information
S1 STROBE Checklist. Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) checklist.
Checklist of items that should be included in reports of observational studies; licensed under CC BY 4.0. Checklist available from https://www.strobe-statement.org/checklists/.
https://doi.org/10.1371/journal.pmed.1004963.s001
(DOCX)
S2 Appendix. Tables and figures.
Table A. Additional participant characteristics for case-control GWAS of postoperative delirium. Comorbidities assessed at time of surgery. Operative categories are listed as Abbott/Bupa classification. Table B. Mapped genes from primary case-control GWAS of postoperative delirium. minGwasP represents the minimum p-value associated with the gene; IndSigSNPs lists the independent significant SNPs tagged to that gene. Table C. Independent significant SNPs in noncardiothoracic surgery case-control GWAS of postoperative delirium (sensitivity analysis). rsID: Reference SNP cluster ID. Table D. Top 10 SNPs from case-control GWAS of postoperative delirium in cardiothoracic cohort (sensitivity analysis). rsID: Reference SNP cluster ID. Table E. Independent significant single-nucleotide polymorphisms passing genome-wide significance in case-control GWAS of postoperative delirium after adjustment for age, sex, chip, first 10 PCs and Charlson Comorbidity Index (sensitivity analysis). rsID: Reference SNP cluster ID. Table F. Polygenic risk scores from PGS Catalog passing matching and included in polygenic risk score analyses. *APOE-independent scores. Table G. Odds ratio for postoperative delirium by Alzheimer’s disease polygenic risk score (PRS) quintile. Results are shown for all 42 polygenic risk scores, adjusted for age and sex, and the subset which did not include the APOE region (15 APOE-independent scores), adjusted for age, sex, APOE ε4 genotype, and Charlson Comorbidity Index (0−1 versus ≥2). Table H. Results from primary analysis (case-control GWAS of postoperative delirium) for variants identified in candidate gene study by Heinrich and colleagues rsID: Reference SNP cluster ID. Fig A. Manhattan plot for case-control GWAS of postoperative delirium after previous dementia diagnoses excluded. The red line represents a genome-wide significant p value of 5 x 10−8. Fig B. Manhattan plot for MAGMA testing for case-control GWAS of postoperative delirium after previous dementia diagnoses excluded. Fig C. Manhattan plot for case-control GWAS of postoperative delirium after previous or subsequent dementia diagnoses excluded. The red line represents a genome-wide significant p value of 5 x 10−8. Fig D. Manhattan plot for MAGMA testing for case-control GWAS of postoperative delirium after previous or subsequent dementia diagnoses excluded. Fig E. Manhattan plot for case-control GWAS of postoperative delirium conditional on APOE ε4 allele count (adjusted for age, sex, chip, and first 10 PCs). Fig F. GWAS catalog reported genes for case-control GWAS of postoperative delirium (primary analysis).
https://doi.org/10.1371/journal.pmed.1004963.s003
(DOCX)
Acknowledgments
This research has been conducted using the UK Biobank Resource under Application Number 128,619. This includes linked data provided by patients and collected by the NHS as part of their care and support and data assets made available by National Safe Haven as part of the Data and Connectivity National Core Study, led by Health Data Research UK in partnership with the Office for National Statistics and funded by UK Research and Innovation (grant ref MC_PC_20058).
Quality Control filtering of the UK Biobank data used the process described by R. Mitchell, G. Hemani, T. Dudding, L. Corbin, S. Harrison, and L. Paternoster [https://doi.org/10.5523/bris.1ovaau5sxunp2cv8rcy88688v].
This work was carried out using the computational and data storage facilities of the Advanced Computing Research Centre, University of Bristol (http://www.bristol.ac.uk/acrc/).
The views expressed are those of the authors and not necessarily those of the UK NIHR or the Department of Health and Social Care.
References
- 1. Evered L, Atkins K, Silbert B, Scott DA. Acute peri-operative neurocognitive disorders: a narrative review. Anaesthesia. 2022;77 Suppl 1:34–42. pmid:35001385
- 2. Inouye SK, Westendorp RGJ, Saczynski JS. Delirium in elderly people. Lancet. 2014;383(9920):911–22. pmid:23992774
- 3. Gou RY, Hshieh TT, Marcantonio ER, Cooper Z, Jones RN, Travison TG, et al. One-year medicare costs associated with delirium in older patients undergoing major elective surgery. JAMA Surg. 2021;156(5):430–42. pmid:33625501
- 4. Maclullich AMJ, Ferguson KJ, Miller T, de Rooij SEJA, Cunningham C. Unravelling the pathophysiology of delirium: a focus on the role of aberrant stress responses. J Psychosom Res. 2008;65(3):229–38. pmid:18707945
- 5. Cerejeira J, Firmino H, Vaz-Serra A, Mukaetova-Ladinska EB. The neuroinflammatory hypothesis of delirium. Acta Neuropathol. 2010;119(6):737–54. pmid:20309566
- 6. Rump K, Adamzik M. Epigenetic mechanisms of postoperative cognitive impairment induced by anesthesia and neuroinflammation. Cells. 2022;11(19):2954. pmid:36230916
- 7. Swarbrick CJ, Partridge JSL. Evidence-based strategies to reduce the incidence of postoperative delirium: a narrative review. Anaesthesia. 2022;77 Suppl 1:92–101. pmid:35001376
- 8. Schenning KJ, Deiner SG. Postoperative delirium in the geriatric patient. Anesthesiol Clin. 2015;33(3):505–16. pmid:26315635
- 9. Nadelson MR, Sanders RD, Avidan MS. Perioperative cognitive trajectory in adults. Br J Anaesth. 2014;112(3):440–51. pmid:24384981
- 10. Goldberg TE, Chen C, Wang Y, Jung E, Swanson A, Ing C, et al. Association of delirium with long-term cognitive decline: a meta-analysis. JAMA Neurol. 2020;77(11):1373–81. pmid:32658246
- 11. Peden CJ, Miller TR, Deiner SG, Eckenhoff RG, Fleisher LA, Members of the Perioperative Brain Health Expert Panel. Improving perioperative brain health: an expert consensus review of key actions for the perioperative care team. Br J Anaesth. 2021;126(2):423–32. pmid:33413977
- 12. Abdellaoui A, Yengo L, Verweij KJH, Visscher PM. 15 years of GWAS discovery: realizing the promise. Am J Hum Genet. 2023;110(2):179–94. pmid:36634672
- 13. Rickenbacher M, Reinbold CS, Herms S, Hoffmann P, Cichon S, Wueest AS, et al. Genome-wide association study of postoperative cognitive dysfunction in older surgical patients. J Neurosurg Anesthesiol. 2022;34(2):248–50. pmid:33264229
- 14. Heinrich M, Sieg M, Kruppa J, Nürnberg P, Schreier PH, Heilmann-Heimbach S, et al. Association between genetic variants of the cholinergic system and postoperative delirium and cognitive dysfunction in elderly patients. BMC Med Genomics. 2021;14(1):248. pmid:34674705
- 15. Westphal S, Stoppe C, Gruenewald M, Bein B, Renner J, Cremer J, et al. Genome-wide association study of myocardial infarction, atrial fibrillation, acute stroke, acute kidney injury and delirium after cardiac surgery - a sub-analysis of the RIPHeart-Study. BMC Cardiovasc Disord. 2019;19(1):26. pmid:30678657
- 16. Bycroft C, Freeman C, Petkova D, Band G, Elliott LT, Sharp K, et al. The UK Biobank resource with deep phenotyping and genomic data. Nature. 2018;562(7726):203–9. pmid:30305743
- 17. Kurki MI, Karjalainen J, Palta P, Sipilä TP, Kristiansson K, Donner KM, et al. FinnGen provides genetic insights from a well-phenotyped isolated population. Nature. 2023;613(7944):508–18. pmid:36653562
- 18.
FinnGen. Genetic data. Available from: https://www.finngen.fi/en/genetic_data. Accessed 2025 September 1.
- 19.
UK Biobank. Data providers and dates of data availability. Available from: https://biobank.ndph.ox.ac.uk/showcase/exinfo.cgi?src=Data_providers_and_dates. Accessed 2025 September 1.
- 20. Abbott TEF, Fowler AJ, Dobbs TD, Harrison EM, Gillies MA, Pearse RM. Frequency of surgical treatment and related hospital procedures in the UK: a national ecological study using hospital episode statistics. Br J Anaesth. 2017;119(2):249–57. pmid:28854546
- 21.
Bupa. Schedule of procedures. Available from: https://codes.bupa.co.uk/procedures. Accessed 2025 September 1.
- 22. Evered L, Silbert B, Knopman DS, Scott DA, DeKosky ST, Rasmussen LS, et al. Recommendations for the nomenclature of cognitive change associated with anaesthesia and surgery-2018. Br J Anaesth. 2018;121(5):1005–12. pmid:30336844
- 23.
American College of Surgeons National Surgical Quality Improvement Programme. About the ACS risk calculator. Available from: https://riskcalculator.facs.org/RiskCalculator/about.html. Accessed 2026 January 7.
- 24. Mbatchou J, Barnard L, Backman J, Marcketta A, Kosmicki JA, Ziyatdinov A, et al. Computationally efficient whole-genome regression for quantitative and binary traits. Nat Genet. 2021;53(7):1097–103. pmid:34017140
- 25.
Mitchell R, Hemani G, Dudding T, Corbin L, Harrison S, Paternoster L. UK Biobank genetic data: MRC-IEU quality control, version 2. data.bris. 2019. https://doi.org/10.5523/bris.1ovaau5sxunp2cv8rcy88688v
- 26. Watanabe K, Taskesen E, van Bochoven A, Posthuma D. Functional mapping and annotation of genetic associations with FUMA. Nat Commun. 2017;8(1):1826. pmid:29184056
- 27. de Leeuw CA, Mooij JM, Heskes T, Posthuma D. MAGMA: generalized gene-set analysis of GWAS data. PLoS Comput Biol. 2015;11(4):e1004219. pmid:25885710
- 28. Gleason LJ, Schmitt EM, Kosar CM, Tabloski P, Saczynski JS, Robinson T, et al. Effect of delirium and other major complications on outcomes after elective surgery in older adults. JAMA Surg. 2015;150(12):1134–40. pmid:26352694
- 29. Werme J, van der Sluis S, Posthuma D, de Leeuw CA. An integrated framework for local genetic correlation analysis. Nat Genet. 2022;54(3):274–82. pmid:35288712
- 30. Lambert SA, Wingfield B, Gibson JT, Gil L, Ramachandran S, Yvon F, et al. Enhancing the polygenic score catalog with tools for score calculation and ancestry normalization. Nat Genet. 2024;56(10):1989–94. pmid:39327485
- 31. Ware EB, Faul JD, Mitchell CM, Bakulski KM. Considering the APOE locus in Alzheimer’s disease polygenic scores in the Health and Retirement Study: a longitudinal panel study. BMC Med Genomics. 2020;13(1):164. pmid:33143703
- 32. Menon DK, Rosand J. Finding a place for candidate gene studies in a genome-wide association study world. JAMA Netw Open. 2021;4(7):e2118594. pmid:34309672
- 33. Thedim M, Hu J, Maher M, Wiener-Kronish J, Saxena R, Vacas S. Perioperative polygenic and APOE-based genetic risk assessment for neurocognitive disorders: a biobank study. Br J Anaesth. 2025;:S0007-0912(25)00308-3. pmid:40562635
- 34. Denny JC, Bastarache L, Ritchie MD, Carroll RJ, Zink R, Mosley JD, et al. Systematic comparison of phenome-wide association study of electronic medical record data and genome-wide association study data. Nat Biotechnol. 2013;31(12):1102–10. pmid:24270849
- 35. Abelha FJ, Fernandes V, Botelho M, Santos P, Santos A, Machado JC, et al. Apolipoprotein E e4 allele does not increase the risk of early postoperative delirium after major surgery. J Anesth. 2012. pmid:22302107
- 36. Vasunilashorn S, Ngo L, Kosar CM, Fong TG, Jones RN, Inouye SK, et al. Does Apolipoprotein E genotype increase risk of postoperative delirium? Am J Geriatr Psychiatry. 2015;23(10):1029–37. pmid:26238230
- 37. Tagarakis GI, Tsolaki-Tagaraki F, Tsolaki M, Diegeler A, Tsilimingas NB, Papassotiropoulos A. The role of apolipoprotein E in cognitive decline and delirium after bypass heart operations. Am J Alzheimers Dis Other Demen. 2007;22(3):223–8. pmid:17606532
- 38. Leung JM, Sands LP, Wang Y, Poon A, Kwok P, Kane JP, et al. Apolipoprotein E e4 allele increases the risk of early postoperative delirium in older patients undergoing noncardiac surgery. Anesthesiology. 2007;107(3):406–11. pmid:17721242
- 39. Raptis V, Bhak Y, Cannings TI, MacLullich AMJ, Tenesa A. Dissecting the genetic and proteomic risk factors for delirium. Nat Aging. 2026;6(1):235–51. pmid:41286463
- 40. Yang LG, March ZM, Stephenson RA, Narayan PS. Apolipoprotein E in lipid metabolism and neurodegenerative disease. Trends Endocrinol Metab. 2023;34(8):430–45. pmid:37357100
- 41. Browndyke JN, Wright MC, Yang R, Syed A, Park J, Hall A, et al. Perioperative neurocognitive and functional neuroimaging trajectories in older APOE4 carriers compared with non-carriers: secondary analysis of a prospective cohort study. Br J Anaesth. 2021;127(6):917–28. pmid:34535274
- 42. Katznelson R, Djaiani G, Tait G, Wasowicz M, Sutherland AM, Styra R, et al. Hospital administrative database underestimates delirium rate after cardiac surgery. Can J Anaesth. 2010;57(10):898–902. pmid:20645040
- 43. Kim DH, Lee J, Kim CA, Huybrechts KF, Bateman BT, Patorno E, et al. Evaluation of algorithms to identify delirium in administrative claims and drug utilization database. Pharmacoepidemiol Drug Saf. 2017;26(8):945–53. pmid:28485014
- 44. Fry A, Littlejohns TJ, Sudlow C, Doherty N, Adamska L, Sprosen T, et al. Comparison of sociodemographic and health-related characteristics of UK biobank participants with those of the general population. Am J Epidemiol. 2017;186(9):1026–34. pmid:28641372