Online platforms are an efficient means to detect early cognitive decline, but few studies have investigated the relationship between remotely collected subjective cognitive change and cognitive decline. We hypothesized that the Everyday Cognition Scale (ECog), a subjective change measure, predicts longitudinal change in cognition in Brain Health Registry (BHR), an online registry for neuroscience research. From the BHR database, we included participants aged 55+ who completed both the baseline ECog and repeated administrations of the CANTAB® Paired Associates Learning (PAL) test. Both self-reported ECog (Self-ECog) and study partner ECog (SP-ECog), and two PAL scores (first attempt memory score [FAMS] and total errors adjusted [TEA]) were assessed. We used multiple ECog scoring outputs, based on previously established cut-offs for likely impairment (Self-ECog positive [total score ≥1.31], SP-ECog positive [total score ≥1.36], and ECog consistent [any item≥3]). A linear mixed effects model was employed to assess the effect of baseline ECog on longitudinal change in PAL. Additionally, logistic regression models were used to assess the ability of ECog to identify ‘decliners’, who exhibited the worst PAL progression slopes corresponding to the fifth percentile and below. The study included a total of 16,683 participants, who were followed for 11.49±11.53 months. Both Self-ECog positive (estimate = -0.01, p <0.0019) and Self-ECog consistent (estimate = -0.008, p <0.0085) were significant predictors for longitudinal change in PAL FAMS after adjusting for age, gender, education, race, depression, family history of Alzheimer’s disease (AD), taking AD medication, and self-reported impairment. Those who were SP-ECog positive (Odds ratio [95% confidence interval] = 1.760 [1.143–2.667]) and SP-ECog consistent (2.021 [1.344–3.021]) had higher probability of being decliners based on PAL FAMS. Regarding the prediction of PAL TEA decliner, both Self-ECog consistent (1.248 [1.043–1.494]) and Self-ECog total (1.287 [1.054–1.557]) were associated with higher odds of being decliners. In the BHR’s unsupervised online setting, ECog demonstrated utility in predicting longitudinal progression in PAL scores, both in terms of continuous changes and when dichotomized as a decliner. Online, self-administered measures of subjective cognitive change, together with objective neuropsychological test results have great potential to identify individuals with cognitive impairments.

It is unclear to what extent genetic risk offsets the protective effects of better premorbid cognitive health on the risk of Alzheimer’s disease (AD). We tested for associations between measures of premorbid cognitive health, apolipoprotein (APOE) e4 ‘risk’ genotype, and their interaction, with risk of incident AD and age of diagnosis, in UK Biobank participants aged ≥55 years at baseline, adjusted for potential confounders. During follow-up, 3,505/252,340 (1.39%) participants received an incident diagnosis of AD. There were significant associations between better performance on each cognitive test with lower risk of incident AD, and later age at diagnosis. However, the benefit of better baseline cognitive scores on AD risk was significantly attenuated in APOE e4 carriers. These data demonstrate that the association between premorbid cognitive health and subsequent risk of AD is influenced by APOE e4 genotype. This has implications for risk stratification and targeted intervention.

Understanding pathways from genetic variation to cognitive impairment is critical for dementia prevention, risk stratification and the development of treatments. While genetic risk factors for dementia are known to associate with cross-sectional differences in biomarkers (e.g. lipids) in healthy people, potential influence over longitudinal trajectories is not understood.We leveraged genetic, general health and two-wave biomarker data from n = 17,817 UK Biobank participants. The outcomes were change in 26 common circulating blood biomarkers including inflammatory, cardiometabolic and lipid families. The presence of apolipoprotein (APOE) e4 ‘risk’ and e2 ‘protective’ alleles were tested separately versus ‘neutral’ e3e3 genotype, as were associations of non-APOE polygenic risk for Alzheimer’s disease. Biomarker change values were corrected for baseline levels, age, deprivation, sex, timepoint interval, smoking history, medication history, deprivation, genotyping chip and 10 genetic principal components (fully-adjusted).The average interval between assessments was 4.30 years (standard deviation; SD = 0.92). For e4 (versus e3e3), four associations were significant: accelerated change in total cholesterol, apolipoprotein b (ApoB) and low-density lipoprotein (LDL) each in the direction of poorer health (standardized β range = 0.021 SDs to 0.036 more change relative to e3e3), and c-reactive protein protectively (β = -0.059; all P < 0.001). For e2 allele presence, there were three significant associations: change in ApoB, total cholesterol and LDL in protective directions (β range = -0.057 to -0.090). There were no APOE genotypic interactions with baseline age, sex, or medication history, nor significant findings associated with non-APOE (Alzheimer’s disease) polygenic risk.APOE e genotype significantly modifies particularly lipid trajectories across time – most strongly ApoB levels. This adds nuance to lipids as a dementia risk factor, and, clinically, suggests more frequent lipid assessments in e4 carriers in that context. Our findings provide a plausible partial biological explanation for APOE’s progressive influence on neurocognitive health.

In Alzheimer’s disease research, subjective report of cognitive and functional decline from participant-study partner (SP) dyads is an efficient method of assessing cognitive impairment and risk of clinical progression. The extent to which discordance (disagreement) between self- and SP–report is associated with diagnosis of cognitive impairment is not known. We tested the hypothesis that discordance between baseline self- and SP-report Everyday Cognition Scale (ECog) scores was associated with greater probability of mild cognitive impairment (MCI) diagnosis. Dyads enrolled in the Alzheimer’s Disease Neuroimaging Initiative (ADNI) and the University of California, San Francisco Brain Health Registry (BHR), an online longitudinal aging-related research registry, completed an online adaptation of the 39-item ECog to assess subjective change across six cognitive domains. We derived four metrics of discordance between participant and SP ECog scores (dyadic discordance): Raw Score Difference, Absolute Score Difference, Overreport Score, and Underreport Score (Table 1). In the ADNI cohort, we fit a logistic regression model for each of the discordance metrics to independently evaluate their association with MCI diagnosis, after adjusting for dyad relationship and sociodemographic factors. Then, to further evaluate the predictive utility of these measures, we carried out a model selection procedure using cross-sectional data collected from ADNI dyads (N = 921; Table 1). Finally, we externally validated the model in a BHR cohort with clinically confirmed diagnoses (N = 279; Table 1). Higher Raw and Absolute Score Difference, greater Underreport scores, and lower Overreport scores were associated with greater probability of MCI in the ADNI cohort (Table 2). The model selection procedure identified a number of highly predictive variables, which were then included in a model that was externally validated in the BHR cohort. This model distinguished diagnostic groups in the BHR cohort with AUC = 0.892, Sensitivity = 0.61, Specificity = 0.95 (Figure 1) based on a restricted cubic spline regression model. Results indicate that ECog score discordance is associated with MCI diagnosis. The selected model showcased good predictive performance in the validation cohort, and highlights the potential utility of subjective dyadic discordance metrics to help identify older adults with MCI in diverse settings.

Abstract Backgrounds Digital, online assessments are efficient means to detect early cognitive decline, but few studies have investigated the relationship between remotely collected subjective cognitive change and cognitive decline. We hypothesized that the Everyday Cognition Scale (ECog), a subjective change measure, predicts longitudinal change in cognition in the Brain Health Registry (BHR), an online registry for neuroscience research. Methods This study included BHR participants aged 55 + who completed both the baseline ECog and repeated administrations of the CANTAB® Paired Associates Learning (PAL) visual learning and memory test. Both self-reported ECog (Self-ECog) and study partner-reported ECog (SP-ECog), and two PAL scores (first attempt memory score [FAMS] and total errors adjusted [TEA]) were assessed. We estimated associations between multiple ECog scoring outputs (ECog positive [same or above cut-off score], ECog consistent [report of consistent decline in any item], and total score) and longitudinal change in PAL. Additionally we assessed the ability of ECog to identify ‘decliners’, who exhibited the worst PAL progression slopes corresponding to the fifth percentile and below. Results Participants (n = 16,683) had an average age of 69.07 ± 7.34, 72.04% were female, and had an average of 16.66 ± 2.26 years of education. They were followed for an average of 2.52 ± 1.63 visits over a period of 11.49 ± 11.53 months. Both Self-ECog positive (estimate = -0.01, p < 0.001, R²m = 0.56) and Self-ECog consistent (estimate=-0.01, p = 0.002, R²m = 0.56) were associated with longitudinal change in PAL FAMS after adjusting demographics and clinical confounders. Those who were Self-ECog total (Odds ratio [95% confidence interval] = 1.390 [1.121–1.708]) and SP-ECog consistent (2.417 [1.591–3.655]) had higher probability of being decliners based on PAL FAMS. Conclusion In the BHR’s unsupervised online setting, baseline subjective change was feasible in predicting longitudinal decline in neuropsychological tests. Online, self-administered measures of subjective cognitive change might have a potential to predict objective subjective change and identify individuals with cognitive impairments.

Background The Everyday Cognition (ECog) 12-item scale, a functional decline measurement, can distinguish dementia from cognitively unimpaired (CU). Limited data compare ECog-12 performance by raters (self vs. informant) and scoring systems (average numeric vs. categorical grouping) to differentiate cognitive statuses.Objectives To evaluate the performance of ECog-12 in differentiation cognitive statuses.Design A cross-sectional diagnostic test study.Setting and Participants Data from the Alzheimer's Disease Neuroimaging Initiative (ADNI) study are analyzed. Participants were aged 55-90 years old divided into subgroups based on diagnostic criteria.Measurements We evaluated ECog-12 performance across different diagnostic groups, such as CU vs cognitive impairment (CI; mild cognitive impairment (MCI), and dementia), and the association between ECog-12 and CI. This procedure was repeated for self- and partner (informant)-reports. Additionally, types of ECog scores were also assessed, where an average ECog score was calculated (continuous numeric) as well as a categorical grouping ("any occasional declined" or "any consistently declined") based on item-level responses to ECog questions.Results ECog-12 cut-off scores of 1.36 (self-reported) and 1.45 (partner-reported) distinguish CU from CI with AUC 0.7 and 0.78, respectively. Adding a memory-concern question improved self-reported-ECog AUC to 0.79. Self- and partner-reported "consistently-declined" ECog-12 categorical grouping provided AUC 0.69 and 0.78. The study partner reported ECog-12 showed a greater association with CI than self-reported, with odds ratios of 35.45 and 8.79, respectively.Conclusion Study partner-reported ECog scores performed better than self-reported ECog-12 in differentiating cognitive statuses, and a higher study partner reported ECog score was a higher prognostic risk for CI. A memory concern question could enhance self-reported ECog-12 performance. This further emphasizes the need to obtain data from study partners for research and clinical practice.

BACKGROUND The Everyday Cognition scale (ECog-39) scores are associated with future cognitive decline. We investigated whether the 12-item ECog (ECog-12), which is being collected in Alzheimer's Disease Neuroimaging Initiative (ADNI)4, can predict progression. METHODS Baseline self (PT)- and study partner (SP)-ECog-12 data were extracted from the 39-item version collected in the ADNI. Weibull analysis examined the relationship between baseline ECog-12 and future clinical progression (change in Clinical Dementia Rating Sum of Boxes [CDR-SB] scores and diagnostic conversion). RESULTS Higher PT- and SP-ECog-12 scores were associated with faster CDR-SB worsening, with hazard ratios in cognitively unimpaired (CU) 3.34 and 9.61, mild cognitive impairment (MCI) 1.44 and 2.82, and dementia 0.93 and 1.82. They were associated with conversion from CU to MCI 3.01 and 6.24 and MCI to dementia 1.61 and 3.07. DISCUSSIONS P-ECog-12 provided a higher prognostic value for predicting clinical progression, so this can help identify and monitor patients at risk in research and health-care settings. Highlights The 12-item Everyday Cognition scale (ECog-12) data obtained from both raters increased diagnostic conversion risk from cognitively unimpaired to mild cognitive impairment (MCI) and from MCI to dementia. ECog-12, rated by study partners, was associated with an increased risk of Clinical Dementia Rating Sum of Boxes worsening in all diagnostic groups. Our results provide novel information about the specific scoring outputs and rater types (participant vs. study partner) of ECog-12 that can facilitate screening, prioritization, and longitudinal monitoring of the clinical progression of participants in Alzheimer's Disease Neuroimaging Initiative 4 and other Alzheimer's disease clinical studies, clinical trials, and in health-care settings.

INTRODUCTION:This study aimed to understand whether older adults' longitudinal completion of assessments in an online Alzheimer's disease and related dementias (ADRD)-related registry is influenced by self-reported medical conditions. METHODS:Brain Health Registry (BHR) is an online cognitive aging and ADRD-related research registry that includes longitudinal health and cognitive assessments. Using logistic regressions, we examined associations between longitudinal registry completion outcomes and self-reported (1) number of medical conditions and (2) eight defined medical condition groups (cardiovascular, metabolic, immune system, ADRD, current psychiatric, substance use/abuse, acquired, other specified conditions) in adults aged 55+ (N = 23,888). Longitudinal registry completion outcomes were assessed by the completion of the BHR initial questionnaire (first questionnaire participants see at each visit) at least twice and completion of a cognitive assessment (Cogstate Brief Battery) at least twice. Models included ethnocultural identity, education, age, and subjective memory concern as covariates. RESULTS:We found that the likelihood of longitudinally completing the initial questionnaire was negatively associated with reporting a diagnosis of ADRD and current psychiatric conditions but was positively associated with reporting substance use/abuse and acquired medical conditions. The likelihood of longitudinally completing the cognitive assessment task was negatively associated with number of reported medical conditions, as well as with reporting cardiovascular conditions, ADRD, and current psychiatric conditions. Previously identified associations between ethnocultural identity and longitudinal assessment completion in BHR remained after accounting for the presence of medical conditions. DISCUSSION:This post hoc analysis provides novel, initial evidence that older adults' completion of longitudinal assessments in an online registry is associated with the number and types of participant-reported medical conditions. Our findings can inform future efforts to make online studies with longitudinal health and cognitive assessments more usable for older adults with medical conditions. The results need to be interpreted with caution due to selection biases, and the under-inclusion of minoritized communities.

Scalable tools to efficiently identify individuals likely to have cognitive impairment (CI) are critical in the Alzheimer’s disease and related dementias field. The Everyday Cognition scale (ECog) and its short form (ECog12) assess subjective cognitive and functional changes and are useful in predicting CI. This study aimed to compare the ability of the online ECog and the in-clinic ECog in distinguishing between CI and cognitively unimpaired (CU) individuals, and to evaluate the effectiveness of the ECog12 compared to the full ECog in an online setting. Participants were recruited from the Brain Health Registry (BHR; online) and Alzheimer’s Disease Neuroimaging Initiative (ADNI; in-clinic) with available clinical diagnoses. Ability of ECog and ECog12 (Self- and study partner [SP]-ECog) to discriminate CI from CU were calculated using Receiver Operating Characteristic (ROC) curves. Area under the ROC curves (AUCs) between BHR and ADNI were compared using the DeLong test, as were AUCs between ECog12 and ECog in BHR. Both online and in-clinic ECog effectively discriminated CI from CU, with no significant differences in AUCs (BHR Self-ECog AUC = 0.722 vs. ADNI Self-ECog AUC = 0.769, DeLong P = .06; BHR SP-ECog AUC = 0.818 vs. ADNI SP-ECog AUC = 0.840, DeLong P = .50). Comparison between online ECog and ECog12 showed no significant differences in AUCs (Self-ECog AUC = 0.722 vs. Self-ECog12 AUC = 0.709, DeLong P = .18). Online ECog, including the short-form ECog12, is as valid as in-clinic ECog for identifying clinically diagnosed CI, offering a cost-effective and accessible screening tool for large-scale online studies for identifying potential candidates for disease-modifying therapy.

Background Subjective cognitive decline (SCD) is defined as an individual’s perception of sustained cognitive decline compared to their normal state while still performing within boundaries for normal functioning. Demographic, psychosocial and medical factors have been linked to age-related cognitive decline, and Alzheimer’s dementia (AD). However, their relation to risk for SCD remains unclear. This study aims to identify demographic factors, psychosocial and cardiovascular health associated with SCD within the Brain Health Registry (BHR) online cohort. Methods Participants aged 55+ (N=27,596) in the BHR self-reported SCD measured using the Everyday Cognition Scale (ECog) and medical conditions, depressive symptoms, body mass index, quality of sleep, health, family history of AD, years of education, race, ethnicity and gender. Multivariable linear regression was used to examine whether SCD was associated with demographic, psychosocial, and medical conditions. Results We found that advanced age, depressive symptoms, poorer sleep quality and poorer quality of health were positively associated with more self-reported SCD in all models. No race or ethnicity differences were found in association with SCD. Males who reported alcohol and tobacco use or underweight BMI had higher ECog scores compared with females. Conclusion In addition to well-established risk factors for cognitive decline, such as age, our study consistently and robustly identified a strong association between psychosocial factors and self-reported cognitive decline in an online cohort. These findings provide further evidence that psychosocial health plays a pivotal role in comprehending the risk of SCD and early-stage cognitive ageing. Our findings emphasise the significance of psychosocial factors within the broader context of cardiovascular and demographic risk factors.

Background: Major depressive disorder (MDD) has increasing prevalence with age. Both objective measures of cognitive dysfunction and subjective report of cognitive difficulties related to MDD are often thought to worsen with increasing age. However, few studies have directly evaluated these characteristics across the adult lifespan.Methods: Participants included 23,594 adults completing objective and subjective measures of cognition on an online research registry. Linear regression including interactions of age group with depression was used to evaluate the association of self-reported MDD with measures of cognition in three age groups: 21-40 years; 41-60 years; 61+ years.Results: MDD (n = 2127) demonstrated poorer objective cognitive performance and greater subjective ratings of cognitive difficulties across all domains assessed compared to non-depressed individuals (ND; n = 21,467). Significant interactions of age group and MDD status with objective and subjective measures of cognition were observed for both middle age and older adults when compared to young adults but few significant differences between middle-aged and older adults were evident.Limitations: This study relied on self-report of MDD diagnosis, utilized remotely administered and unsupervised measures of cognition, and the sample was not diverse.Conclusions: The magnitude of association between MDD and cognitive correlates appears to plateau in middle age. Our results suggest that increased rates of dementia are not due to greater cognitive consequence of MDD in older adults and that age effects, and not greater effects of depression, may lead to increased diagnosis of MDD based on subjective report of cognitive symptoms.

Early, efficient identification of Alzheimer’s disease (AD) risk is a critical need. Dyadic subjective report from a participant and an informant/study partner (SP) who knows the participant well, may be a reliable and accurate method for assessing AD risk. The Caregiver and Study Partner Portal within the Brain Health Registry (BHR; an online registry and cohort) is a scalable tool for remotely obtaining dyadic data. All BHR participants are asked to identify a potential SP, who is invited to separately enroll and consent. SPs complete a demographic survey and questionnaires pertaining to cognition, behavioral symptoms, activities of daily living, and the health of the participant. Participants and SPs both complete an online adaptation of the Everyday Cognition Scale (ECog) to assess subjective cognitive change. We analyzed demographic characteristics of the dyads enrolled in BHR and measured correlation between SP and self-report ECog scores. We used ADNI ECog data to estimate cut points likely to indicate cognitive impairment (1.43 for SP-report scores and 1.65 for self-report scores) and determined the number of BHR participants who are possibly impaired based on these cut points. Out of 93,134 BHR participants, 10,494 (11.3%) have an enrolled study partner (Figure 1). The mean SP-report ECog score is 1.35±0.50 and mean self-report ECog score is 1.46±0.47. SP and self-report ECog scores are correlated (r = 0.47, p < 0.001) (Figure 1). Out of 8,374 participants with available SP-report ECog scores, 2,194 (26.2%) are possibly impaired. Out of 10,018 participants with available self-report ECog scores, 2,384 (23.8%) of participants are possibly impaired (Figure 2). These findings demonstrate the feasibility of collecting online, dyad-report demographic and subjective cognitive change data from a large cohort of participant-SP dyads. The study failed to recruit an ethnoculturally and educationally diverse cohort of dyads. The positive correlation between self- and SP-report ECog scores suggests dyad concordance in subjective report of change. Future efforts will focus on improving study partner enrollment and BHR task completion, increasing dyad diversity, and investigating the ability of remote dyadic measures to track cognitive decline and clinical progression along the AD continuum.

BACKGROUND:In Alzheimer's disease (AD) research, subjective reports of cognitive and functional decline from participant-study partner dyads is an efficient method of assessing cognitive impairment and clinical progression. METHODS:Demographics and subjective cognitive/functional decline (Everyday Cognition Scale [ECog]) scores from dyads enrolled in the Brain Health Registry (BHR) Study Partner Portal were analyzed. Associations between dyad characteristics and both ECog scores and study engagement were investigated. RESULTS:A total of 10,494 BHR participants (mean age = 66.9 ± 12.16 standard deviations, 67.4% female) have enrolled study partners (mean age = 64.3 ± 14.3 standard deviations, 49.3% female), including 8987 dyads with a participant 55 years of age or older. Older and more educated study partners were more likely to complete tasks and return for follow-up. Twenty-five percent to 27% of older adult participants had self and study partner-report ECog scores indicating a possible cognitive impairment. DISCUSSION:The BHR Study Partner Portal is a unique digital tool for capturing dyadic data, with high impact applications in the clinical neuroscience and AD fields. Highlights The Brain Health Registry (BHR) Study Partner Portal is a novel, digital platform of >10,000 dyads. Collection of dyadic online subjective cognitive and functional data is feasible. The portal has good usability as evidenced by positive study partner feedback. The portal is a potential scalable strategy for cognitive impairment screening in older adults.

Recent research suggests genetic variation in the Klotho locus may modify the association between APOE ɛ4 and cognitive impairment. We tested for associations and interactions between these genotypes versus risk of dementia, cognitive abilities, and brain structure in older UK Biobank participants. Klotho status was indexed with rs9536314 heterozygosity (versus not), in unrelated people with versus without APOE ɛ4 genotype, corrected for various confounders. APOE ɛ4 associated with increased risk of dementia, worse cognitive abilities, and brain structure. Klotho was associated with better reasoning. There were no interactions; potentially suggesting an age- and pathology-dependent Klotho effect.

Background and purpose: Previous studies testing associations between polygenic risk for late-onset Alzheimer’s disease (LOAD-PGR) and brain magnetic resonance imaging (MRI) measures have been limited by small samples and inconsistent consideration of potential confounders. This study investigates whether higher LOAD-PGR is associated with differences in structural brain imaging and cognitive values in a relatively large sample of non-demented, generally healthy adults (UK Biobank). Method: Summary statistics were used to create PGR scores for n=32,790 participants using LDpred. Outcomes included 12 structural MRI volumes and 6 concurrent cognitive measures. Models were adjusted for age, sex, body mass index, genotyping chip, 8 principal components, lifetime smoking, apolipoprotein (APOE) e4 genotype and socioeconomic deprivation. We tested for statistical interactions between APOE e4 allele dose and LOAD-PGR vs. all outcomes. Results: In fully adjusted models, LOAD-PGR was associated with worse fluid intelligence (standardised beta [β] = -0.080 per LOAD-PGR standard deviation, p = 0.002), matrix completion (β = -0.102, p = 0.003), smaller left hippocampal total (β = -0.118, p = 0.002) and body (β = -0.069, p = 0.002) volumes, but not other hippocampal subdivisions. There were no significant APOE x LOAD-PGR score interactions for any outcomes in fully adjusted models. Discussion: This is the largest study to date investigating LOAD-PGR and non-demented structural brain MRI and cognition phenotypes. LOAD-PGR was associated with smaller hippocampal volumes and aspects of cognitive ability in healthy adults, and could supplement APOE status in risk stratification of cognitive impairment/LOAD.

Objective: Atherosclerosis is the underlying cause of most cardiovascular disease, but mechanisms underlying atherosclerosis are incompletely understood. Ultrasound measurement of the carotid intima-media thickness (cIMT) can be used to measure vascular remodeling, which is indicative of atherosclerosis. Genome-wide association studies have identified many genetic loci associated with cIMT, but heterogeneity of measurements collected by many small cohorts have been a major limitation in these efforts. Here, we conducted genome-wide association analyses in UKB (UK Biobank; N=22 179), the largest single study with consistent cIMT measurements. Approach and Results: We used BOLT-LMM software to run linear regression of cIMT in UKB, adjusted for age, sex, and genotyping chip. In white British participants, we identified 5 novel loci associated with cIMT and replicated most previously reported loci. In the first sex-specific analyses of cIMT, we identified a locus on chromosome 5, associated with cIMT in women only and highlight VCAN as a good candidate gene at this locus. Genetic correlations with body mass index and glucometabolic traits were also observed. Two loci influenced risk of ischemic heart disease. ConclusionS: These findings replicate previously reported associations, highlight novel biology, and provide new directions for investigating the sex differences observed in cardiovascular disease presentation and progression.

Importance: Recent research has suggested that genetic variation in the Klotho (KL) locus 29 may modify the association between apolipoprotein e ( APOE ) e4 genotype and cognitive 30 impairment. 31 Objective: Large-scale testing for associations and interactions between KL and APOE 32 genotypes vs. risk of dementia (n=1,570 cases), cognitive abilities (n=174,513) and brain 33 structure (n = 13,158) in older (60+ years) participants. 34 Design, setting and participants: Cross-sectional and prospective data (UK Biobank). 35 Main outcomes and measures: KL status was indexed with heterozygosity of the rs9536314 36 polymorphism (vs. not), in unrelated people with vs. without APOE e4 genotype, using 37 regression and interaction tests. We assessed non-demented cognitive scores (processing 38 speed; reasoning; memory; executive function), multiple structural brain imaging, and clinical 39 dementia outcomes. All tests were corrected for age, sex, assessment centre, eight principal 40 components for population stratification, genotypic array, smoking history, deprivation, and 41 self-reported medication history. 42 Results: APOE e4 presence (vs. not) was associated with increased risk of dementia, worse 43 cognitive abilities and brain structure differences. KL heterozygosity was associated with 44 less frontal lobe grey matter. There were no significant APOE/KL interactions for cognitive, 45 dementia or brain imaging measures (all P>0.05). 46 Conclusions and relevance: We found no evidence of APOE/KL interactions on cognitive, 47 dementia or brain imaging outcomes. This could be due to some degree of cognitive test 48 imprecision, generally preserved participant health potentially due to relatively young age, 49 type-1 error in prior studies, or indicative of a significant age-dependent KL effect only in the 50 context of marked AD pathology. 51