Association between CNS-active drugs and risk of Alzheimer’s and age-related neurodegenerative diseases

Objective As neuropsychiatric conditions can increase the risk of age-related neurodegenerative diseases (NDDs), the impact of CNS-active drugs on the risk of developing Alzheimer’s Disease (AD), non-AD dementia, Multiple Sclerosis (MS), Parkinson’s Disease (PD) and Amyotrophic Lateral Sclerosis (ALS) was investigated. Research design and methods A retrospective cohort analysis of a medical claims dataset over a 10 year span was conducted in patients aged 60 years or older. Participants were propensity score matched for comorbidity severity and demographic parameters. Relative risk (RR) ratios and 95% confidence intervals (CI) were determined for age-related NDDs. Cumulative hazard ratios and treatment duration were determined to assess the association between CNS-active drugs and NDDs at different ages and treatment duration intervals. Results In 309,128 patients who met inclusion criteria, exposure to CNS-active drugs was associated with a decreased risk of AD (0.86% vs 1.73%, RR: 0.50; 95% CI: 0.47-0.53; p <.0001) and all NDDs (3.13% vs 5.76%, RR: 0.54; 95% CI: 0.53-0.56; p <.0001). Analysis of impact of drug class on risk of AD indicated that antidepressant, sedative, anticonvulsant, and stimulant medications were associated with significantly reduced risk of AD whereas atypical antipsychotics were associated with increased AD risk. The greatest risk reduction for AD and NDDs occurred in patients aged 70 years or older with a protective effect only in patients with long-term therapy (>3 years). Furthermore, responders to these therapeutics were characterized by diagnosed obesity and higher prescriptions of anti-inflammatory drugs and menopausal hormonal therapy, compared to patients with a diagnosis of AD (non-responders). Addition of a second CNS-active drug was associated with greater reduction in AD risk compared to monotherapy, with the combination of a Z-drug and an SNRI associated with greatest AD risk reduction. Conclusion Collectively, these findings indicate that CNS-active drugs were associated with reduced risk of developing AD and other age-related NDDs. The exception was atypical antipsychotics, which increased risk. Potential use of combination therapy with atypical antipsychotics could mitigate the risk conferred by these drugs. Evidence from these analyses advance precision prevention strategies to reduce the risk of age-related NDDs in persons with neuropsychiatric disorders.

The retrospective cohort study reported herein investigated the association between CNS-active drugs, including antidepressants, sedatives, anticonvulsants, antipsychotics, and stimulants, and the incidence of AD and other NDDs including non-AD dementia, PD, MS, and ALS in patients 60 years of age or older.For this analysis, the risk of developing AD or other NDDs starting one year after exposure to CNS-active drugs was assessed using the Mariner USbased population insurance claims dataset.Pharmacologic interventions prescribed to treat neuropsychiatric disorders including antidepressants, sedatives, anticonvulsants, antipsychotics, and stimulants are categorized as CNS-activedrugs.These therapies modulate neurotransmitter signaling through multiple mechanisms that can impact neurodegenerative processes and potentially modulate the risk of developing AD and other NDDs (43)(44)(45)(46)(47)(48)(49)(50)(51).Secondary analyses to evaluate the impact of age and duration of treatment were conducted.Moreover, we identified and characterized patients who remained free of AD after drug exposure (responders) and those who developed AD (non-responders).
AD is a progressive disease characterized by impaired cognitive function that ultimately results in loss of autonomy and independent living (52-54).Currently, more than 6 million Americans are affected by AD which, along with other dementias, cost the nation $321 billion dollars in 2022 (54).Presently, select anti-amyloid therapies can slow progression of the disease at early stages (55, 56) but a cure for AD remains elusive.The preclinical phase of AD can begin two decades prior to diagnosis and represents a pivotal window for implementing preventative strategies that target risk factors.

Data source
The Mariner dataset used for this analysis contains insurance claims data within the United States, with a population primarily residing in the Southeastern region.The Mariner dataset contains patient demographic characteristics, prescription records, patient diagnosis, and procedure information organized under Current Procedural Terminology, International Classification of Diseases, Ninth Revision (ICD-9), and International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes (Supplementary Tables S1, S2), following the International Classification of Diseases.In September 2022, the Mariner database included 151 million patients with claims that went from 2010 to April 2021.
PearlDiver is a research software that facilitates interaction with individual commercial, state-based Medicaid, Medicare stand-alone prescription drug plan, group Medicare Advantage, and individual Medicare Advantage data sets (57).
This report follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.This study was approved by the University of Arizona Institutional Review Board.Requirements for informed consent were waived as the data were deidentified.

Study design and variables
A subset of patients with non-melanoma skin cancer was selected for medical informatic analysis from the Mariner database as this medical condition is not associated with neurodegenerative disease and treatment for the condition does not require chemotherapy or surgical anesthesia for treatment which can impact risk of Alzheimer's and occurs with sufficient frequency in the general population to enable robust analyses.Patients younger than 60 years of age, with a history of neurosurgery or brain cancer, and with a history of NDD (including AD, dementia, MS, PD, and ALS) prior to the index date were excluded from this analysis.Participants were required to be continuously enrolled in the medical and pharmacy insurance database for a minimum of 6 months before and 3 years after the index date (Figure 1).The index date is described as the first drug prescription record for the treatment group, and a period of 6 months minimum after the first patient claim record for the control group.The treatment group was defined as patients with a medication charge for at least one CNS-active drug, while the control group included patients without any CNS-active drug medication charge throughout the duration of the study.The primary outcome of the study was defined as the incidence of AD, based on ICD-9 and ICD-10 codes, at least 1 year after the index date.The 1-year period is to remove potential unknown effects on NDDs prior to index date.Additionally, incidence of other NDDs including non-AD dementia, MS, PD and ALS was surveyed as exploratory endpoints.Medications considered in the study include FDA-approved antidepressants, sedatives, anticonvulsants, antipsychotics, and stimulants (Supplementary Table S3).These drugs were identified by drug codes and then grouped under categories according to their mechanism of action (Supplementary Table S1).Age in the treatment group was defined by the age of first CNS-active drug exposure.Following the analysis in (58) and (59) an assessment of comorbidities known to be associated with AD outcomes was conducted.The impact of CNS-active drug duration on the risk of AD was evaluated as described in (60), where analyses were conducted for durations of <1year, 1 to 3 years, 3 to 6 years, and >6 years.Cumulative hazard ratios were built using the propensity score matched population in the Bellwether-PearlDiver interface.The average follow-up time was 6.6 [2.4] years.The number of patients in each drug class and subclass, and the median adherence rate for each therapeutic are reported in Supplementary Table S3.A responder analysis was conducted including selected comorbidities known to be associated with AD and other age-related NDDs (61-64).Common co-therapies that have been previously identified as potential AD risk modifiers were also assessed (58)(59)(60)(65)(66)(67)(68).The drug groups included were antidiabetics, antihypertensives, anti-inflammatories, Selective Estrogen Receptor Modulators (SERMs), Androgen Deprivation Treatments (ADTs), statins and Menopausal Hormonal Treatment (MHT) (Supplementary Table S1).

Statistical analysis
Statistical analyses were conducted between May 1st and July 28th, 2022.Patient demographic statistics and incidence statistics were analyzed using unpaired and paired two-tailed t-tests or c 2 tests, as appropriate, to test the significance of the differences between continuous and categorical variables.In all analyses, a two-sided p < 0.05 was considered statistically significant.
To balance demographic and comorbidity characteristics between the treatment and control groups, propensity score matching was conducted as described in (58,59).A logistic regression was generated prior to propensity score matching to estimate the probability for each patient to receive CNS-active drug therapy given confounding variables like age, sex, region, Charlson Comorbidity Index (CCI) score, and comorbidities of interest.The confounding variables that showed statistically significant differences were used to propensity score match the treatment and control groups; these comprised all the demographic variables as well as comorbidities including Asthma, COPD, Cardiac Arrythmia, Cerebrovascular Disease, Chronic Artery Disease, Diabetes, Drug Abuse, Hypertension, Hypothyroidism, Liver Disease, Obesity, Osteoarthritis, Rheumatoid Arthritis, Tobacco Use, Acute Kidney Injury, and Capsulitis.

Results
Of 3,182,960 patients, a subset of 855,153 met the inclusion and enrollment criteria (Figure 1).After adjusting based on age, sex, region, and selected comorbidities, a population of 309,128 patients was equally divided into treatment (mean age [SD]: 66.1 [4.0] years) and control groups (66.1 [4.0] years) (N:154,564).There were no significant differences (p>0.05) on age, sex, or region between adjusted treatment and control groups (Table 1).Although the CCI was not significantly different, patients on CNS-active drugs had significantly greater incidence of comorbidities (p < 0.0001) (Table 1).
The majority of patients in the treated group were exposed to either sedatives (38.86%) or antidepressants (33.28%),where the most prevalent drugs were Selective Serotonin Reuptake Inhibitors (SSRIs) (19.56%) and benzodiazepines (BDZPs) (26.73%), respectively (Supplementary Table S3).The median adherence for all drug subclasses was greater than 60% except for Z-drugs and BDZPs, which had a 43.34% and 15.33% median adherence, respectively (Supplementary Table S3).
Cumulative hazard ratios with 95% CI were generated from the propensity score matched population to evaluate the rate of disease conversion for AD and all NDDs combined.In patients aged 60 to 65 years old, there were no differences between patients with exposure to CNS-active drugs and the control group.However, as age increased, divergence between both groups was greater, with the non-exposure group showing increased risk of developing NDDs and AD compared to the treatment group.The greatest divergence occurred in the oldest group, in patients aged 75 to 79 years old (Figure 4).Interestingly, sex differences occurred in patients aged 70 years and older.Females with no exposure to CNS-active drugs had higher risk of developing NDDs and AD compared to males.In contrast, there were no differences in the rate of disease conversion between both sexes in the treated group (Figure 5).
The impact of duration of therapy on risk of developing AD and all NDDs combined was determined for patients with exposure to CNS-active drugs in the propensity score matched population (Table 3).Within the exposure group, the majority of patients received treatment for 6 years or longer (44.21%).In patients receiving CNS-active drugs for 3 years or more, CNS-active drugs were associated with a significant reduction in AD and NDD risk.Patients receiving CNS-active drugs for over 6 years exhibited the highest risk reduction in AD (RR: 0.22; 95% CI: 0.20 -0.25; p <0.0001) and NDD (RR: 0.28; 95% CI: 0.27 -0.30; p <0.0001).In contrast, when treatment duration was less than 3 years, no benefit on AD or NDD risk reduction was detected.A significant increase in AD risk occurred in patients exposed to CNS-active drugs for less than 1 year (RR: 1.23; 95% CI: 1.04-1.47;p = 0.02) as well as an increased risk for NDDs in patients receiving treatment from 1 to 3 years (RR: 1.09; 95% CI: 1.03-1.16;p = 0.0025) (Table 3).
A risk analysis was conducted to estimate the impact of multiple CNS-active drug combinations on the risk of developing AD (Figure 6).Drug groups with insufficient patient number were excluded from this analysis.Outcomes of this analysis indicated that addition of a second CNS-active drug was associated with greater reduction in AD risk compared to monotherapy.Specifically, for all antidepressants assessed (tricyclics, SSRIs and SNRIs), the combination of a BDZP, a Z-drug or a second-generation anticonvulsant was associated with greater reduction of AD risk compared to exposure to one antidepressant alone.BDZP and second-generation anticonvulsants exerted greater reduction in AD risk in all drug combinations explored, except when combined with SSRIs.Similarly, Z-drugs showed decreased AD risk in all combinations with the exception of tricyclic antidepressants.Interestingly, atypical antipsychotics were associated with increased AD risk, which was mitigated by the addition of a tricyclic, SNRI, Zdrug or second-generation anticonvulsant, with Z-drugs providing the greatest mitigation of AD risk.

A B
Relative risk of developing Alzheimer's disease (AD) in patients with exposure to different classes of CNS drugs (A).Sex differences in relative risk of developing Alzheimer's disease (AD) in patients with exposure to different classes of CNS drugs (B) (females appear in red, males appear in blue).CI, confidence interval; RR, relative risk.
Addition of a Z-drug to any CNS-active drug was associated with the greatest reduction of AD risk for every drug class assessed, with the combination of a Z-drug and an SNRI exerting greatest AD risk reduction (Figure 6).

Discussion
This study sought to explore the impact of CNS-active medications in an aging population on the development of AD and other age-related NDDs using health claims records.Outcomes of this retrospective analysis indicated that use of CNS-active drugs was associated with significant risk reduction of AD, non-AD dementia, MS, PD, and ALS.Additional analyses revealed that antidepressant, sedative, anticonvulsant, and stimulant medications were associated with a decreased risk of AD, while antipsychotics, specifically atypical antipsychotics, were associated with increased risk of developing AD.Additionally, sedatives were associated with a greater reduction in AD risk in women compared to men, and antipsychotic treatment was associated with increased risk of AD only in males.
Age and duration of therapy emerged as two main modulators of AD risk associated with CNS-active drugs, where AD risk reduction was apparent after 3 years of drug exposure and reduction of AD risk was greater with increasing age.During early stages of AD pathology, a misdiagnosis of a neuropsychiatric disorder can occur as a consequence of the common symptomatology of both diseases (69, 70).Thus, the increased AD risk observed in patients receiving CNS-active drugs for less than 3 years may be the result of AD pathology already established in this group of patients, but not yet diagnosed.These findings suggest that accounting for treatment duration and examining patients in the appropriate age range are key elements of the association between Relative risk of developing Alzheimer's disease (AD) in patients with exposure to different subclasses of CNS drugs.CI, confidence interval; RR, relative risk.

FIGURE 4
Hazard ratio curves for risk of developing Alzheimer's disease (AD) and neurodegenerative diseases (NDDs) combined in propensity score matched patients.AD, Alzheimer's disease; NDD, neurodegenerative disease.CNS-active drugs and incidence of NDDs.This could also explain disparities in the literature where studies that considered treatment duration indicated that antidepressant and sedative drugs were associated with reduced incidence of dementia (71)(72)(73), while studies that did not account for duration of treatment found an increased risk associated with these drugs (74-78).
Risk analysis evaluating sex differences did not reveal significant differences between males and females on AD risk.However, when NDDs were combined a slight but significant difference emerged where females exhibited greater risk reduction compared to males.Consistent with these findings, results from the cumulative hazard analysis indicated a reduction in rate of disease conversion for both sexes in the treated group, with females exhibiting a greater magnitude of divergence between the treated and untreated groups.Within the treated group, females and males exhibited a comparable trajectory of disease conversion suggesting that CNSactive drugs activate mechanisms that reduce risk of NDD that are independent of biological sex.
Results from these analyses indicated that antidepressants, including tricyclics and tetracyclics, SSRIs, SNRIs and SSRI/ Serotonin partial agonists, were associated with a significant risk reduction of developing AD.Mechanistically, antidepressants exert multiple actions in the brain including modifications of synaptic transmission, reduced inflammation, and increased neurogenesis that could ameliorate the pathological changes of depression that could increase the risk of developing AD (43,45,(79)(80)(81)(82)(83).Moreover, antidepressants can impact amyloid plaque development (47,84).The majority of these beneficial effects are thought to be regulated by increases in serotonin and norepinephrine signaling (85).Hazard ratio curves for risk of developing Alzheimer's disease (AD) and neurodegenerative diseases (NDDs) combined in propensity score matched patients, females versus males.AD, Alzheimer's disease; NDD, neurodegenerative disease.Sedatives used for the management of insomnia including Zdrugs and BDZPs (86) were associated with reduced risk of developing AD and was greater in females relative to males.Zdrugs were associated with greater risk reduction compared to BDZPs.Both Z-drugs and BDZPs could protect against AD by improving sleep, thereby reducing inflammation, tau aggregation, and amyloid beta accumulation (12,50,87).Notably, Z-drugs bind with greater specificity to the GABA-A chloride channel receptor than BDZPs and promote slow-wave sleep, crucial for Ab clearance and memory consolidation, which could provide an additional benefit in AD risk reduction (88) compared to BDZPs.Interestingly, sedative drugs, especially BDZPs, are highly lipophilic which could induce longer duration of action in women, who have larger volumes of adipose tissue than men (89).Of note, median adherence for both sedative drug classes assessed, specifically for BDZPs, was lower than 50%, which could impact the results on the association between AD risk and treatment with sedative drugs.
Anticonvulsants were associated with reduced AD risk, despite previous studies reporting contradictory findings on cognition and AD risk (90-93).Data contained herein provide new evidence for potential long-term benefits of anticonvulsants.Neuronal hyperexcitability that occurs in the early stages of AD (94) could potentially be counteracted by anticonvulsants as observed in preclinical studies (46,48).
Atypical antipsychotics were associated with a significant increase in risk of developing AD.Notably, an association with increased risk of AD was only significant in males.Multiple analyses suggest that certain antipsychotic drugs can impair memory and cognitive function with short-term treatment (44, 95, 96) whereas their long-term impact on AD risk was uncertain (97,98).Findings reported herein provide evidence for an adverse impact on AD risk.Relative risk of developing Alzheimer's disease (AD) in patients receiving different combinations of CNS-active drugs.BDZP, benzodiazepines; CI, confidence interval; SNRI, serotonin and norepinephrine reuptake inhibitors; SSRI, selective serotonin reuptake inhibitors.Antipsychotic antagonistic effects on neurotransmitters like acetylcholine and serotonin, crucial for memory and learning, could contribute to AD vulnerability (99, 100).Interestingly, women's psychotic symptoms respond to lower doses of antipsychotics compared to men (101,102), which could influence the modulation of AD risk differently in females compared to males.Interestingly, among all CNS-active drugs stimulants were associated with the greatest AD risk reduction.Although the association between stimulants and AD remains poorly understood, evidence indicates that low-dose methamphetamine may promote neuroprotection via sAPP alpha production (51), neuronal differentiation, and synaptic plasticity (49) that likely involves norepinephrine and dopamine mechanisms.

Cortes
Responder analyses indicated that non-responders were older and predominantly female compared to patients who did not develop AD which could contribute to the greater incidence of AD in women (54).Additionally, non-responders were characterized by an overall higher incidence of comorbidities than responders, specifically cardiovascular comorbidities, whereas responders had a higher incidence of obesity.Consistent with these findings, previous reports indicate that weight loss at midlife is a risk factor for AD, and that high body mass index in late life could be protective against AD (103-105).Interestingly, responders were also characterized by a greater exposure to antiinflammatory drugs and MHT which have both been associated with decreased risk of developing AD (60,65,106).

Limitations
Several limitations of this study are due to the Mariner database which does not include values for laboratory tests or APOE genotype information or socioeconomic factors like race, ethnicity, or level of education or lifestyle modifications or psychological therapies which are recommended for several of the neuropsychiatric disorders assessed.Although the groups were propensity score matched, these factors could have an impact on AD development which is not accounted for in the analyses.Further, there was no information regarding efficacy of CNSactive drugs to control the neuropsychiatric disorder or whether therapies were changed during the course of treatment.
The average median adherence for all drugs considered in the study was 71.88%, which is higher than the estimated adherence to chronic medications in the general population (50%) 107.Because all main analyses were conducted with all drugs grouped together, adherence was sufficient to provide reliable results.However, low adherence rate would specifically impact results derived from drug classes with low median adherence.Both sedative drug classes had an adherence rate lower than 50%, with 15.33% for benzodiazepines and 43.34% for Z-drugs.The low adherence rate would underestimate the magnitude of impact of these drugs on NDD risk.

Conclusion
Outcomes of analyses reported herein indicate that CNS-active drugs targeting neuropsychiatric disorders including depression, insomnia, epilepsy, schizophrenia, and ADHD can significantly modify the risk of developing AD.Antidepressants, sedatives, anticonvulsants, and stimulants, were all associated with significant reduction in the risk of AD, especially following long-term treatment.In contrast, antipsychotics, specifically atypical antipsychotics, were associated with a significant increased risk of developing AD, which could be mitigated when combined with other CNS-active drugs such as SNRIs.Women and men exhibited distinct response profiles to sedative and antipsychotic drugs, where sedatives were associated with a greater reduction in AD risk in women compared to men whereas antipsychotic treatment was associated with increased risk in AD only in males.Notably, non-responders to these therapeutics were characterized by a high incidence of cardiovascular disorders.Collectively, the data indicate that long-term management of neuropsychiatric disorders using appropriate pharmacologic treatment can significantly reduce risk of AD and other NDDs.

TABLE 1
Baseline characteristics for unadjusted and propensity score-matched patients with or without exposure to CNS active drugs.

TABLE 2
Relative risk of AD development in propensity score matched patients after exposure to different classes of CNS-active drugs.

TABLE 3
Impact of CNS drug therapy duration on risk of Alzheimer's disease (AD) and neurodegenerative diseases (NDDs) combined.

TABLE 4
Demographic characteristics, comorbidities and cotreatments in patients who develop Alzheimer's disease (AD) (non-responders) and those who remain free of AD (responders) after treatment with CNS drugs compared to the untreated population.