Cognitive trajectories and incident dementia after a cardiovascular event in older adults

Cardiovascular disease (CVD) is a recognized risk factor for dementia. Here we determined the extent to which an incident CVD event modifies the trajectory of cognitive function and risk of dementia.


INTRODUCTION
Cardiovascular disease (CVD) and dementia are two co-existing agerelated chronic conditions that contribute to disease burden in older adults. 1 People with CVD contribute to increases in the cost burden of dementia. 2 Evidence suggests dementia in advanced age develops after a lifelong sequence of events, beginning with exposure to vascular risk factors in youth, followed by heart diseases. 3 A recent retrospective cohort study of predominantly younger adults found CVD diagnosis increased the risk of dementia by 17%. 4 However, despite previous reports that incident CVD accelerated cognitive decline, a large recent study of older adults (mean 73 years) found no increased risk of cognitive impairment among participants with CVD compared with no CVD. 5 Furthermore, the majority of past studies examined individual CVD events. There is good evidence that stroke, which causes direct end organ damage, is associated with an increased risk of cognitive decline and dementia [6][7][8][9] ; however, fewer studies have investigated the association with other CVDs. A 2018 review of cognitive decline after non-stroke CVDs (e.g., myocardial infarction [MI], heart failure [HF]) identified the need for more methodologically rigorous longitudinal studies on this topic. 10 Additionally, this review suggested that large cohort studies with a comprehensive assessment of cognitive and cardiovascular health at baseline were essential and highlighted that studies should exclude pre-existing cognitive impairment to determine the magnitude of decline secondary to each disease state. 10 Two reviews exploring the association between HF and dementia found significant associations, 11,12 but a recent review that included two additional large-scale studies did not find a statistically significant association. 13 However, this review considered the outcomes, dementia, cognitive impairment, and cognitive function separately and found HF was associated with reduced performance in cognitive tests. The association also varied for the different cognitive domains, emphasizing the importance of studying cognitive outcomes separately.
Therefore, the aim of this study was to determine the association between a CVD event and the incidence of dementia and cognitive trajectories after an event, in a large prospective cohort of initially healthy older individuals who were free of CVD or major cognitive impairment at baseline. Secondary aims included investigating the associations between two non-stroke CVDs, MI and HF, and the incidence of dementia and change in cognitive function over time.

Study population
Participants were from the ASPirin in Reducing Events in the Elderly (ASPREE) study, the full details of which were reported previously. 14 Briefly, this was a double-blind, randomized, placebo-controlled clinical trial investigating the effect of low-dose aspirin on disability-free survival in healthy older adults. CVD events and dementia were among prespecified endpoints, but the main trial results reported no signifi- to follow these participants annually.
The ASPREE study has multiple institutional review board approvals in the US and Australia. All the participants provided written informed consent before enrollment.

Cardiovascular disease events
Details of the CVD adjudication process have been published elsewhere. 15 Briefly, CVD was a composite of fatal coronary heart disease, nonfatal MI, fatal or nonfatal stroke, and hospitalization for heart failure (HHF). Health information was collected from hospitals/medical centers, treating physicians, death certificates, medical records, and hospital information obtained from family members. Committees whose members were blinded to the treatment group assignments used the information collected to adjudicate CVD events. 15 MI was defined based on the joint guidelines of the European Society of Cardiology and the American College of Cardiology, 18 and HHF was defined as any unplanned stay, overnight or longer, in a hospital or a similar facility, for which the principal reason for admission was HF. 15 All the HHF cases were the first incident of HHF.

Dementia
Full details describing the adjudication of dementia have been reported previously. 16  to measure psychomotor speed.

Covariates
Potential confounding factors considered were those known to be related to CVD and cognition and were measured at baseline.

Statistical analyses
The statistical analysis was performed in STATA version 17.0 (Stata Corp, College Station, Texas, USA). 24 Descriptive characteristics were presented according to CVD event. Incidence rates for dementia following a CVD event was compared with those without a CVD event.
Since death is a competing risk factor, we compared incidence rates after excluding participants who had died during the trial.
Linear mixed models were used to measure cognitive change over The effect size associated with the overall follow-up time variable represents the annual change in cognition scores for all participants over the follow-up period; the effect size of the time-varying variable represents the acute decline in cognition after a CVD event, 8 which was defined as the change in cognition in the time immediately after a CVD event and was estimated based on the first cognitive assessment administered after a CVD event as well as other cognition scores before and after a CVD event 8 ; lastly, the effect size for the time after CVD variable represents the annual change in cognition scores after a CVD event. The models included both random intercepts and slopes to account for participant-specific individual differences.
The unstructured correlation matrix was specified, and the model was fit using maximum likelihood. Sensitivity analyses were conducted by (1)

RESULTS
The study sample included 19 There were 922 participants with incident CVD in ASPREE ( Figure S1 and  . c This effect size shows the annual change in cognitive scores on each of the four tests after a CVD event.  25 Further, 14 (5.1%) were adjudicated as having dementia after MI (n = 277) and 9 (7.1%) after HHF (n = 127). Given these small numbers, no formal statistical analysis was performed for dementia risk after a CVD event.

Change in global cognition after CVD
The mean (SD) global cognition score at baseline was 93.4 (4.6) (  Figure 1 and Table 2).

Change in delayed recall for episodic memory after CVD
At baseline, the mean (SD) score for delayed recall for episodic memory was 7.7 (2.8) (Table 1) Table 2).

Change in processing speed after CVD
The mean (SD) processing speed score at baseline was 36.7 (10.1) (  (Figure 1 and Table 2).

Change in cognitive function after MI or HHF
For MI (n = 355), the results were largely consistent with all CVD; however, the effect sizes were smaller for processing speed (acute change:  (Table 4).

Sensitivity analyses
For the sensitivity analyses, (1) after excluding 1052 participants who died, the results (Tables S6, S7, and S8) were largely similar to the TA B L E 4 HHF and change in cognitive performance over time (n = 18,363).

Global cognition (3MS) Delayed recall (HVLT-R) Processing speed (SDMT) Verbal fluency (COWAT)
Cognitive main study results; however, there was a significant acute increase in global cognition after an event; (2) for CVD other than stroke (excluding 404 participants with stroke), the results (Table S9) were consistent with the primary findings, though the increase in acute global cognition after CVD became significant; (3) after excluding participants who were adjudicated as having dementia (n = 872), the results for CVD (Table S10), MI (Table S11), and HHF (Table S12) remained largely consistent with the main results, but here again the acute increase in global cognition was significant; (4) lastly, when the trial randomization (treatment) groups were added as a potential confounding factor, this was not significant, nor did it change the results from the linear mixed models.

DISCUSSION
In this large prospective cohort of individuals predominantly 70 years and older without a prior CVD event or major cognitive impairments at baseline, the difference in incidence rates of dementia after a CVD event and those without a CVD event was not significant; however, after excluding participants who had died during the trial, dementia incidence rates were higher after incident CVD than in those without cognition, verbal memory, and temporal orientation scores after a diagnosis of coronary heart disease, which is consistent with our findings. 27 In that study, however, they found no associations with short-term cognition, but they didn't assess processing speed, which was found to be significant in our study.
Exploring multiple cognitive domains such as global cognition, delayed recall, processing speed, and verbal fluency helps in further understanding the impact of CVD events on cognition. Interestingly, while we showed that CVD was followed by the strongest acute drop in processing speed, over time there was a faster decline in global cognition, episodic memory, and verbal fluency. Impairment in episodic memory is common in the early stages of dementia, and impairment in verbal fluency is also associated with most brain disorders such as dementia or traumatic brain injury. It is not entirely clear from our study why there was an acute change in processing speed and a more gradual decline in global cognition, delayed recall, and verbal fluency. A possible hypothesis is that processing speed could be linked to higher-order cognitive processes. 28,29 Further research is required to understand this difference in the short-and long-term impacts of CVD events on the different cognitive domains, which will inform individ- This study highlighted the need for both short-and long-term monitoring of cognition in people after incident CVD. However, there is no consensus on the best tools that can be used for cognitive assess-  30 and HF, 31 another study suggested that MoCA might not be sensitive enough to detect subtle cognitive changes in the early stages and recommended that cognition in individuals at different stages of the CVD continuum be assessed through a complete neuropsychological evaluation using cognitive assessments that target memory and executive functions and verbal fluency. 32 Good cognitive performance supports adherence to medication, self-care, and the ability to retain disease-related knowledge, 33 which could also help prevent recurrence of another incident CVD, further emphasizing the importance of preventing CVD events in order to maintain cognitive function and independence in older adults.
A major strength of this study was the prospective follow-up of a large cohort of individuals without any CVD events or major cognitive impairment at baseline, with regular cognitive assessment by trained staff. These ensure high-quality data and instills confidence that the events and assessments have been recorded accurately. All CVD events and dementia were adjudicated by expert committees of clinicians, and the exact timing of these events was recorded. With the availability of the CVD diagnosis date, we were able to use a time-varying covariate, which allowed us to control for the immortal time bias. 34 Moreover, the ASPREE study systematically collected data on a wide range of risk factors for both CVD and cognition, which allowed us to adjust for multiple covariates.
One limitation of this study was the short follow-up (median of 6.4 years) in a healthy population. As a consequence, there were few dementia cases following CVD, which restricted our ability to perform further analysis. A longer follow-up might have provided a better estimate of the potential long-term impact on cognitive trajectories.
Because they relate to healthy subjects, the findings might not be broadly applicable to all individuals. A second limitation is that we did not correct for multiple comparisons, and the results should be interpreted with caution because there is the potential for false positive associations. Third, although CVD adjudication was done robustly, there were no details about the event. Given that CVDs, like HF, are chronic conditions, severity of CVD, treatment, and so forth might impact on cognition. Lastly, we considered four cognitive domains, but other domains, for example working memory, were not included, and our test of verbal fluency (COWAT-F) examined only one subdomain, which is prone to learning effects.
In conclusion, after a CVD event, the incidence of dementia is higher than among participants without a CVD event. Additionally, incident CVD events, MI and HHF, were associated with an acute drop in processing speed and a greater decline in global cognition. CVD and HHF were also associated with a longer-term decline in delayed recall and verbal fluency over time. The main clinical implications of this study are the need for both short-and long-term monitoring of cognition among people who have had a CVD event. Further research is required to understand the mechanisms for cognitive change following any CVD event.

ACKNOWLEDGMENTS
This study was mainly supported by grants from the National Insti-