Association of Aspirin Use with Reduced Risk of Developing Alzheimer’s Disease in Elderly Ischemic Stroke Patients: A Retrospective Cohort Study

Abstract

Background:

Currently there are no effective therapies to prevent or halt the development of Alzheimer’s disease (AD). Multiple risk factors are involved in AD, including ischemic stroke (IS). Aspirin is often prescribed following IS to prevent blood clot formation. Observational studies have shown inconsistent findings with respect to the relationship between aspirin use and the risk of AD.

Objective:

To investigate the relationship between aspirin therapy after IS and the new diagnosis of AD in elderly patients.

Methods:

This retrospective cohort study leveraged a large database that contains over 90 million electronic health records to compare the hazard rates of AD after IS in elderly patients prescribed aspirin versus those not prescribed aspirin after propensity-score matching for relevant confounders.

Results:

At 1, 3, and 5 years after first IS, elderly patients prescribed aspirin were less likely to develop AD than those not prescribed aspirin: Hazard Ratio = 0.78 [0.65,0.94], 0.81 [0.70,0.94], and 0.76 [0.70,0.92].

Conclusion:

Our findings suggest that aspirin use may prevent AD in patients with IS, a subpopulation at high risk of developing the disease.

INTRODUCTION

Alzheimer’s disease (AD) is the sixth leading cause of death in the United States; over 6.1 million Americans live with the disease as of 2017 [1]. It is the most prevalent dementia [2] and is characterized by neurodegeneration and declining cognition. Its etiology is largely unknown [3]. Multiple potentially modifiable risk factors are involved in AD including cardiovascular disease, cerebrovascular disease, metabolic and psychiatric factors, diet, lifestyle, and education [4]. Targeting these modifiable risk factors is a promising strategy that might prevent or delay up to 40% of dementias [5]. One area of particular interest is the link between AD and ischemic stroke (IS), defined as the abrupt cessation of blood flow to part of the brain due to a blockage that leads to neuronal damage and death [6]. IS can cause long-term cognitive impairment [2] or even vascular dementia, which is triggered by reduced blood flow to the brain [7]. Furthermore, anywhere from 7-41% of patients who experience at least one IS subsequently develop AD [8], and there is overlap between the symptoms and etiologies of vascular dementia and AD [7]. The Framingham study revealed that patients without dementia who experienced a stroke are twice as likely to develop dementia within the next ten years compared to patients without dementia who did not experience a stroke [9]. Several potential mechanisms could explain why IS increases the risk of dementia: it brings on ischemia, inflammation, excitotoxicity, and free radical production that damage neurons and lead to the degenerative changes seen in vascular dementia and AD [10].

Aspirin, or acetylsalicylic acid, is a widelyutilized drug that functions both as an antiplatelet and anti-inflammatory medication [11]. It binds to and irreversibly inhibits the enzyme cyclooxygenase 1, preventing formation of thromboxane A2 on platelets, which reduces platelet aggregation and blood clotting [12]. Consequently it is often prescribed to prevent IS, which results from abnormal and excessive clotting events [13]. It also acts as an anti-inflammatory agent by inhibiting the release of prostaglandins, molecules which are critical in mediating the immune system’s inflammatory response [11].

Multiple observational studies have examined the relationship between aspirin use and AD, with several finding that use before AD onset is associated with reduced incidence of the disease [14, 15], whereas use after AD onset has no effect on cognitive decline [16, 17]. Randomized controlled trials have failed to demonstrate any efficacy of aspirin or other non-steroidal anti-inflammatory drugs in preventing AD [18]. Importantly, however, half of these studies excluded patients with recent IS, and none have focused exclusively on patients with IS or other cerebrovascular events [18]. Considering that both ischemic and inflammatory processes are involved in the etiology of IS and AD, we hypothesize that patients who have experienced IS are less likely to subsequently develop AD if they are prescribed aspirin. We set out to elucidate this relationship through nation-wide retrospective patient cohort analyses using the TriNetX database. TriNetX contains de-identified patient-level data from millions of individuals throughout the country [19]. We have used the TriNetX database and its advanced built-in informatics and statistical tools for clinical studies on AD [20, 21] and other diseases [22-27]. We conducted a retrospective cohort study to compare the risk of developing AD in elderly patients taking aspirin versus those not taking aspirin after experiencing IS. To the best of our knowledge, this is the first observational study that has examined the relationship between aspirin use and AD specifically in IS patients. Our findings offer insights into personalized and targeted medical therapies for patients at risk for developing AD.

METHODS

Data sources

We accessed patient data through the cloudbased TriNetX Analytics network platform [28]. TriNetX contains nationwide, frequently updated, de-identified electronic health records from over 90 million unique patients across the United States covering diverse geographic locations, age groups, racial and ethnic groups, income levels, and insurance types. The built-in statistical functions within TriNetX perform statistical analyses on patient-level data. Because TriNetX reports population-level data and results without including protected health information, the MetroHealth System, Cleveland, Institutional Review Board has determined that any research using TriNetX is not Human Subject Research and is therefore exempt from review.

Study population

Our study population was restricted to persons over the age of 65 years, as most patients are diagnosed with AD between the ages of 65 and 85 [29]. We excluded persons with previous diagnosis of AD or vascular dementia or those with previous aspirin prescription. The exposure group (“Aspirin (+) cohort”) was comprised of patients who received an aspirin prescription within 3 months of their first IS, and the control group (“Aspirin (−) cohort”) was comprised of patients who did not receive an aspirin prescription within 5 years of their first IS (the period of study) (Fig. 1). There were further subgroups delineated to analyze outcomes by gender.

Fig. 1.

Flowchart of patient selection from TriNetX. IS, ischemic stroke; AD, Alzheimer’s disease.

Statistical analyses

All statistical analyses were conducted using the built-in functions on TriNetX. For each comparison, the Aspirin (+) and Aspirin (−) cohorts were propensity score matched for relevant variables that are associated with risk for developing AD: demographics (age, race, and sex [1]); preexisting medical conditions including cardiovascular [30], neurologic [31], and psychiatric [32] diagnoses; medication prescriptions that could influence these medical conditions; and blood pressure measurements. Propensity score matching involved 1 : 1 matching via nearest neighbor greedy matching algorithm, with a caliper of 0.25 standard deviations.

After balancing, a Kaplan-Meier survival analysis [33] was conducted to estimate the probability of developing AD at 1, 3, and 5 years after first IS in both the Aspirin (+) and Aspirin (−) cohorts (Fig. 2). In Kaplan-Meier analysis, patients who exit the cohort during the time interval of interest due to death or loss to follow-up are removed from the analysis by censoring [34] after the last entry in their electronic health record. Cox’s proportional hazards model was then used to compare the two cohorts with the proportional hazard assumption being tested with the generalized Schoenfeld approach. Hazard ratios (HR) and 95% confidence intervals (CI) were used to describe the relative hazard of rebound outcomes based on comparison of time to event rates. The same approach was utilized to generate HR and CI for mortality in the two cohorts at 1, 3, and 5 years after first IS. Separate subgroup analyses were performed in cohorts stratified by gender.

Fig. 2.

Retrospective cohort study design in TriNetX. AD, Alzheimer’s disease; IS, ischemic stroke.

RESULTS

Patient characteristics

The study population was comprised of 48,239 patients with IS age 65 and over, including 10,081 who were prescribed aspirin within 3 months of first IS and 38,158 who were not prescribed aspirin within 5 years of first IS (Fig. 1). Patients in the exposure group differed from those in the control group by racial makeup, comorbidities (including essential hypertension, diabetes, and atrial fibrillation), medication prescription (anticoagulants and platelet aggregation inhibitors), and blood pressure measurements (Table 1). After propensity-score matching, the two cohorts were well balanced, yielding 10,079 patients each in the overall Aspirin (+) and Aspirin (−) cohorts (Table 1), 4,558 patients each in the male Aspirin (+) and Aspirin (−) cohorts, and 5,497 patients each in the female Aspirin (+) and Aspirin (−) cohorts.

Table 1.

Patient characteristics before and after matching. Diagnostic and medication codes provided where applicable

Variable	Before Matching			After Matching
	Exposure Group (n = 10,081)	Control Group (n = 38,158)	Standard Mean Difference	Exposure Group (n = 10,079)	Control Group (n = 10,079)	Standard Mean Difference
Demographics
Age at index event	72.8	72.8	0.01	72.8	72.7	0.01
Race (White)	74.7	66.9	0.15*	74.7	74.8	0.01
Race (Black)	17.8	19.1	0.01	17.8	17.9	0.01
Race (Unknown)	7.5	13.9	0.20*	7.5	7.3	<0.01
Sex (Male)	45.3	44.6	0.01	45.3	45.9	0.01
Diagnoses (%)
Essential hypertension (I10)	67.3	76.1	0.19*	67.3	64.5	0.06
Diabetes (E08-E13)	34.0	41.0	0.14*	34.0	32.1	0.03
Pure hypercholesterolemia (E78.0)	27.0	32.1	0.12*	27.0	25.5	0.04
Ischemic heart diseases (I20-I25)	23.7	34.8	0.24*	23.7	21.7	0.03
Obesity (E66.9)	19.6	21.3	0.06	19.7	18.1	0.03
Atrial fibrillation (I48)	12.6	19.7	0.21*	12.6	11.4	0.02
Acute myocardial infarction (I21)	3.6	5.6	0.09	3.6	3.2	<0.01
Other anxiety disorders (F41)	24.7	28.2	0.08	24.7	24.0	0.03
Depressive episode (F32)	22.5	26.6	0.12*	22.5	21.7	0.01
Nicotine dependence (F17)	13.4	14.5	0.02	13.4	12.4	0.02
Reaction to severe stress; adjustment disorders (F43)	6.4	7.7	0.07	6.4	6.6	0.02
Major depressive disorder, recurrent (F33)	6.0	8.8	0.12*	6.0	5.6	0.01
Alcohol related disorders (F10)	3.0	3.3	0.04	3.0	2.9	0.01
Conductive and sensorineural hearing loss (H90)	8.3	10.6	0.09	8.3	7.8	0.04
Other and unspecified hearing loss (H91)	7.7	7.9	0.02	7.7	7.5	<0.01
Injuries to the head (S00-S09)	15.5	18.7	0.10*	15.5	14.2	0.01
Other cerebrovascular disease (I67)	10.2	19.2	0.29*	10.2	9.8	0.03
Occlusion and stenosis of precerebral arteries, not resulting in cerebral infarction (I65)	9.9	13.6	0.14*	9.9	9.1	0.02
Transient cerebral ischemic attacks (G45)	6.6	11.2	0.21*	6.6	6.5	0.01
Occlusion and stenosis of cerebral arteries, not resulting in cerebral infarction (I66)	0.8	1.6	0.11*	0.8	0.8	0.01
Medications (%)
Diuretics (CV700)	42.4	42.2	0.04	42.4	39.5	0.03
Angiotensin-converting enzyme inhibitors (CV800)	30.2	30.2	0.01	30.2	27.7	0.02
Calcium channel blockers (CV200)	30.0	31.8	0.05	30.0	28.0	0.04
Antihypertensives, other (CV490)	9.5	10.7	0.01	9.5	8.6	<0.01
Angiotensin II inhibitors (CV805)	19.8	19.9	0.02	19.8	18.1	0.03
Atorvastatin (83367)	20.2	20.6	0.03	20.2	18.7	0.04
Simvastatin (36567)	17.4	17.8	0.04	17.4	16.5	0.01
Pravastatin (42463)	9.2	8.5	0.01	9.2	8.6	0.02
Rosuvastatin (301542)	6.8	6.3	0.01	6.8	5.9	0.01
Lovastatin (6472)	3.3	3.4	<0.01	3.3	3.1	0.01
Anticoagulants (BL110)	23.1	28.6	0.13*	23.1	22.0	0.02
Platelet aggregation inhibitors (BL117)	4.0	8.2	0.17*	4.0	3.7	0.01
Labs
Blood pressure, systolic
<140	38.0	24.3	0.27*	38.0	36.4	0.04
140-160	34.0	20.4	0.28*	34.0	32.5	0.05
>160	25.3	14.2	0.26*	25.3	23.6	0.05
Blood pressure, diastolic
<90	42.5	27.1	0.30*	42.5	40.8	0.04
90-100	21.2	12.4	0.22*	21.2	20.0	0.04
>100	11.3	6.6	0.15*	11.3	10.6	0.02

^*Standard Mean Difference greater than 0.1, a threshold used to declare imbalance.

Aspirin use is associated with decreased risk of AD among patients with IS

At 1, 3, and 5 years after first IS, patients in the Aspirin (+) cohort were at a consistently reduced risk of developing AD compared to the matched control group: HR = 0.78, 95% CI: [0.65,0.94], 0.81, 95% CI: [0.70,0.94], and 0.76, 95% CI: [0.70,0.92], respectively (Fig. 3). Decreased risk for developing AD was observed in both male and female subgroups, though not significant at the 1-year follow-up timepoint for males or the 3-year follow-up timepoint for both genders, possibly due to the limited sample sizes (Fig. 3). In addition to a reduced risk of AD, patients in the overall Aspirin (+) cohort experienced a significantly reduced mortality at 1, 3, and 5 years: HR = 0.87, 95% CI: [0.80,0.95], 0.86, 95% CI: [0.81,0.92], and 0.85, 95% CI: [0.80, 0.91], respectively (Fig. 4).

Fig. 3.

Hazard ratio of AD in propensity-score matched Aspirin (+) and Aspirin (−) cohorts at 1, 3, and 5 years after first IS. AD, Alzheimer’s disease; HR, hazard ratio; CI, confidence interval.

Fig. 4.

Hazard ratio of death in propensity-score matched Aspirin (+) and Aspirin (−) cohorts at 1, 3, and 5 years after first IS. HR, hazard ratio; CI, confidence interval.

DISCUSSION

In our study, elderly patients who have experienced an IS and have been prescribed aspirin display a significantly lower rate of AD diagnosis compared to matched patients who have not been prescribed aspirin. These reduced risks were consistent at three follow-up timepoints (1, 3, and 5 years after first IS). Previous studies have observed higher all-cause mortality among healthy older adults who received daily aspirin compared to those who received placebo [35]. Interestingly, our study shows that in patients who have experienced IS (a decidedly less healthy population), mortality is significantly lower in the Aspirin (+) cohort compared to the Aspirin (−) cohort. Although the risk for AD was also lower in the Aspirin (+) cohort, this is unlikely to be the only mechanism responsible for the lower death rate since the number of deaths were higher than the number of AD diagnoses; for example, there were 1,043 deaths but only 213 AD diagnoses at the 1-year timepoint for the Aspirin (+) cohort. This suggests that other factors contributed to the lower death rates in patients prescribed aspirin after first IS, and that the interplay between aspirin and mortality likely depends on population characteristics. Further research is needed to understand the factors underlying the observed relationship between lower risk of mortality and aspirin prescription in patients who have experienced IS.

While aspirin is given after IS to prevent blood clot formation, it also displays anti-inflammatory effects, even at the low dose that is commonly prescribed to IS patients [36]. Inflammation is a critical (albeit poorly understood) component of AD. Patients with AD display elevated levels of pro-inflammatory cytokines in the blood, including interleukin-6, tumor necrosis factor-receptor 1, and C reactive protein [37]. In response to the protein misfolding that occurs in AD, pro-inflammatory signals are triggered to alter the function of the endoplasmic reticulum in central nervous system cells [38]. In addition, the activation of microglia is a well-characterized phenomenon in AD, which results in the release of pro-inflammatory factors [39]. Perhaps aspirin attenuates the inflammation implicated in AD pathogenesis, as inflammation is a hallmark of the cellular damage inflicted after IS [40, 41]. Indeed, some of our previous work reveals that tumor necrosis factor inhibitors, another class of anti-inflammatories, are associated with reduced incidence of AD in patients with rheumatoid arthritis, which is characterized by high levels of inflammation [42, 43]. Further research is needed to understand how aspirin’s thrombolytic and anti-inflammatory properties separately or jointly contribute to its association with reduced AD incidence in IS patients.

This analysis has several important limitations. First, the patients included on the TriNetX platform represent those who had medical encounters with healthcare organizations in the TriNetX network, so the patient electronic health record data may not be representative of or generalizable to the entire elderly American population with IS. Second, due to the retrospective and longitudinal nature of this study, our findings indicate association, not causation: while patients in the Aspirin (+) cohort display a lower rate of AD, there may be other factors in this group not captured in the analysis that could explain this effect. In addition, the data available on TriNetX do not capture the compliance or length of aspirin therapy in the Aspirin (+) cohort. Incorporation of this information into future studies would allow for better understanding of how length of aspirin prescription correlates with AD risk. Finally, analyses of aspirin dosage (for example, 81 mg or 325 mg prescriptions versus placebo) were not included due to limited sample sizes, as fewer than 15% of aspirin prescriptions on TriNetX include dosage information. Future studies should gauge the relative AD risk in patients prescribed low- or high-dose aspirin to determine if a dose-dependent risk reduction exists.

While previous studies have examined the benefits of aspirin and other non-steroidal anti-inflammatory drugs for the prevention of AD, we believe this is the first study that describes the association between aspirin and AD specifically in elderly patients who have experienced IS. Since patients who have experienced IS are at increased risk of developing AD, aspirin therapy may serve as an affordable and effective method to prevent or delay the onset of AD. It will be especially important to balance the potential benefits of aspirin for AD prevention with the risks (for example, the increased risk of adverse bleeding events). Tailoring therapies for at-risk subpopulations is becoming increasingly important in the era of precision medicine, and hopefully carefully targeted real-life trials can gauge the risk-benefit equation for aspirin or other therapies to prevent the development of AD in patients most at risk.