Drug interactions with dementia-related pathophysiological pathways worsen or prevent dementia

Abstract

Many risk factors are known to induce or precipitate dementia. Drugs acting via different mechanisms can modulate cognitive performance and exert either beneficial or deleterious effects on cognition through functional or neuropathological mechanisms. This review discusses the association between several classes of drugs and cognitive impairment and dementia risk. These drugs can be divided into drugs targeting CNS disorders (e.g., anticholinergic drugs, antiepileptics, antipsychotics, benzodiazepines, and opioids) and drugs targeting non-CNS disorders (e.g., antidiabetics, antihypertensives, proton pump inhibitors, and statins). Furthermore, we sought to highlight the pharmacological mechanisms underlying their possible detrimental or beneficial effects on cognition. Anticholinergic and antiepileptic drugs were excluded from this review because their effects on cognition are well known. Studies investigating benzodiazepines have revealed an increased risk of dementia. Conclusions on dementia risk or cognitive impairment regarding opioids and antipsychotic drugs are difficult to draw. These different classes appear to impair cognition not by a single clear mechanism of action specific to each class but by several relatively interdependent and interconnected mechanisms (e.g., impaired neurotransmission, neuroinflammation, neuronal death, oxidative stress, or interactions with dementia-related pathways). The dementia risk initially associated with the use of proton pump inhibitors might have been overestimated. In contrast, statins, antihypertensive medications, and antidiabetics could potentially decrease the risk of dementia and cognitive impairment by acting in ways opposite to the mechanisms cited above.

Linked Articles

This article is part of a themed section on Therapeutics for Dementia and Alzheimer's Disease: New Directions for Precision Medicine. To view the other articles in this section visit http://onlinelibrary.wiley.com/doi/10.1111/bph.v176.18/issuetoc

Abbreviations

ACE-Is

angiotensin-converting enzyme inhibitors

AD

Alzheimer's disease

AHMs

antihypertensive medications

Ang-II

angiotensin II

APs

antipsychotics

ARBs

angiotensin 1 receptor blockers

BDNF

brain-derived neurotrophic factors

BZDs

benzodiazepines

CCBs

calcium channel blockers

CF

cognitive function

DPP4I

dipeptidyl peptidase 4 inhibitor

GSK3ß

glycogen synthase kinase 3ß

MCI

mild cognitive impairment

MMT

methadone maintenance treatment

PPIs

proton pump inhibitors

RAS

renin angiotensin system

RCT

randomized control trial

TZD

thiazolidinediones

1 INTRODUCTION

Currently, 35.6 million people are living with dementia worldwide. Although recent studies have highlighted that the prevalence of dementia has remained stable over time (Y.-T. Wu et al., 2017), a better characterization of the risk factors of dementia, particularly the avoidable factors, is essential. Many risk factors, such as age, vascular risk factors, head trauma, and even educational level, are known to precipitate dementia. Drugs that act through different mechanisms mostly unrelated to their principal pharmacological properties could potentially modulate cognitive performance (Nevado-Holgado, Kim, Winchester, Gallacher, & Lovestone, 2016). On the one hand, many drugs known to decrease vascular risk factors, such as statins, antihypertensive medications (AHMs), and antidiabetics, could have a beneficial effect on cognition (Imfeld, Bodmer, Jick, & Meier, 2012; Lin et al., 2015; Yasar et al., 2016). On the other hand, many drugs are known to increase the risk of dementia (either degenerative dementia or vascular dementia) and cognitive decline. These drugs can generally be divided into drugs targeting CNS disorders such as benzodiazepines (BZDs) (Billioti de Gage, Pariente, & Bégaud, 2015) and opiates (Dublin et al., 2015) and drugs not targeting the CNS such as proton pump inhibitors (PPIs) (Tai et al., 2017). These drugs could act directly on cognitive function (CF) by modulating neurotransmission, ion channels, or receptor activity or indirectly via pathophysiological pathways related to vascular or neurodegenerative dementia by interacting with amyloid β production (Batchelor, Gilmartin, Kemp, Hopper, & Liew, 2017), τ hyperphosphorylation (Ramage et al., 2005), neuroinflammation, oxidative stress, endothelial function, or even cerebrovascular reactivity (Mendoza-Oliva, Zepeda, & Arias, 2014).

This article reviews and discusses the association between cognitive impairment and dementia risk and different classes of drugs. This article aims to highlight the pharmacological mechanisms underlying the possible detrimental or beneficial effects on cognition of the different drugs under discussion.

2 METHODS

2.1 Literature search strategy

A narrative review of observational and interventional studies published between 2007 and 2017 in English, French, or Spanish was conducted. The studies were identified by searching MEDLINE (via PubMed) using a combination of the following MeSH terms: Neurocognitive disorders/chemically induced OR Neurocognitive disorders/drug effect OR «cognitive disorders» [Title/Abstract] OR «cognitive impairment» [Title/Abstract] OR «dementia» [Title/Abstract]. The terms described above were coupled to the following different classes of drugs: drugs targeting the CNS disorders (BZDs, antidepressants drugs, antipsychotic [AP] drugs, antiepileptic drugs, and opiate drugs), drugs targeting vascular or metabolic risk factors (antihypertensive drugs and statins), and drugs targeting other pathways (PPIs). A complementary search was performed using the same terms on Google Scholar and ISI Web of Science to identify studies published within the last 4 years. These classes of drugs were included because they were noted in the French pharmacovigilance database as inducing minor or major neurocognitive disorders. Antihypertensive and antidiabetics medications were also included as they are widely used and have been the subject of numerous studies on their effects on cognition.

2.1.1 Primary outcome

In this review, we focused on the effect of medicines on minor and major neurocognitive disorders according to the definitions of the neurocognitive disorders provided by the Diagnostic and Statistical Manual of Mental Disorders, fifth Edition (DSM-V). The outcomes of interest included the incidence or prevalence of dementia, vascular dementia, Alzheimer's disease (AD), cognitive impairment, and the progression of cognitive decline.

2.1.2 Inclusion and exclusion criteria

Studies were included in the analysis if they were conducted in humans and examined the effect of different classes of drugs on cognition or dementia risk. We included non-observational studies that compared patients exposed or not exposed (without medication or treated with another medication) to a particular class of drug. We also included studies that evaluated cognition after withdrawal from a drug class of interest. Cross-sectional studies, randomized control trials (RCTs), meta-analyses, and systematic reviews were also included. Narrative reviews were included if they were relevant. We included studies involving patients regardless of their age or cognitive status at baseline. No attempt was made to identify unpublished reports. Opinion pieces, editorials, commentaries, news, and letters were excluded. Studies that compared adherence and non-adherence, studies investigating recreational use or substance abuse, studies comparing medication therapies and nonmedication therapies, and studies with outcomes that were not among the outcomes listed above were excluded. Animal studies were also excluded.

2.1.3 Study selection

First, the articles were screened by reading the titles and abstracts. Subsequently, the full paper was obtained for studies that appeared to have met the inclusion criteria or in cases where a decision could not be made based on the title and/or abstract alone. Then, the articles were screened by reading the full text and included if the criteria were met. Because the effects on the cognition of anticholinergic and antiepileptic drugs are well known, these two classes of drugs were not included in this narrative review.

2.1.4 Data collection

The data collected included the first author's name, publication year, study design, sample size, effect estimate, 95% CI, P value, and mean difference. The main outcome was divided into either dementia risk or CF (including cognitive decline or cognitive impairment). Each study was categorized according to its level of evidence from I to IV. High-quality RCTs were classified as I. Meta-analyses or systematic reviews including high-quality RCTs were classified as I. RCTs with a high risk of bias, clinical trials, prospective cohorts, and meta-analyses including observational and non-observational studies and open-label studies were classified as II. Retrospective cohort and case control studies were classified as III. Cross-sectional studies were classified as IV. Reviews were classified as “unclassifiable.”

2.1.5 Preclinical studies

Finally, preclinical studies that aimed to explain the possible pharmacological mechanisms of the drugs investigated were searched for in the PubMed and Google Scholar databases.

3 RESULTS

3.1 Study identification

Figure 1 shows the search and selection procedures. Three thousand two hundred and seventy-one records were identified in MEDLINE, and 783 records were identified through an additional search. One hundred and seventy-eight articles were included, 45 articles regarding anticholinergic and antiepileptic drugs were excluded. In total, 133 articles were analysed.

Figure 1

Literature search and selection. AHM, antihypertensive medication; AP, antipsychotics; BZD, benzodiazepines; PPI, proton pump inhibitors

3.2 Drugs targeting CNS disorders

3.2.1 Antipsychotics

Several studies have assessed the effect of APs on CF in healthy subjects and patients with dementia and even assessed the dementia risk among large cohorts of men with posttraumatic stress disorder (Table S1). Among healthy people, an RCT performed by Veselinović et al., 2013 showed deleterious effects of APs on processing speed and attention. Among patients with psychotic symptoms, the use of APs had heterogeneous effects on cognition (Nielsen et al., 2015). Indeed, an open-label study involving patients presenting with their first episode of schizophrenia who were taking APs showed a moderate improvement in cognitive test performance after 6 months of treatment with APs compared with their baseline performance (Davidson et al., 2009), and a meta-analysis highlighted that atypical APs (quetiapine, olanzapine, and risperidone) led to cognitive improvement in this patient population (Désaméricq et al., 2014).

The effects of APs on CF among patients with mild cognitive impairment (MCI) or dementia are unclear. Although a meta-analysis performed by Wolf, Leucht, and Pajonk (2017), an RCT performed by Ballard et al. (2008), and a review performed by El-Saifi, Moyle, Jones, and Tuffaha (2016) did not highlight a significant effect of APs in patients with dementia, an RCT (Vigen et al., 2011) and a cross-sectional study (Eggermont, de Vries, & Scherder, 2009) both found that APs could impair CF in AD and mild to moderate dementia patients respectively. Furthermore, a cross-sectional study found that patients with dementia took more APs on average than did patients without dementia (Lövheim, Karlsson, & Gustafson, 2008).

Finally, according to two cohort studies (Mawanda, Wallace, McCoy, & Abrams, 2017; Roughead et al., 2017), APs could increase the risk of dementia among patients with posttraumatic stress disorder. In conclusion, APs can improve cognition in schizophrenic patients, which is known to be altered, but the long-term use of APs could potentially increase the dementia risk.

The cognitive impairment induced by APs could be linked to the level of occupancy of dopamine D₂ receptors by a non-linear relationship (Veselinović et al., 2013). In addition, there is a negative correlation between the level of blockade of D₂ receptors by risperidone and the attentional score. The anticholinergic properties of APs could also play a major role in cognitive impairment, and the anticholinergic load among patients with schizophrenia is known to be associated with memory impairment (Nielsen et al., 2015). Terry and Mahadik (2007) proposed the hypothesis that APs could have deleterious effects on cognition by the following two mechanisms: First, by blocking the D₂ receptors on cholinergic neurons, APs lead to the release of ACh and excessive neuronal activity with high levels of intracellular calcium, leading to cell damage. Second, APs decrease the levels of nerve growth factors and brain-derived neurotrophic factors (BDNF). Indeed, these authors found that the chronic administration of risperidone diminished the expression of α7nAChR in memory-related brain regions.

3.2.2 Benzodiazepines

Two cohort studies (Bourgeois, Elseviers, Van Bortel, Petrovic, & Vander Stichele, 2015; Y. Zhang, Zhou, Meranus, Wang, & Kukull, 2016) and two case control studies (Høiseth et al., 2013; Lai, Yang, Tsai, Huang, & Chiou, 2014) evaluated the cognitive effect of BZDs. Bourgeois et al. (2015) and Y. Zhang et al. (2016) did not find any evidence that BZD users presented with a higher cognitive decline than non-users over time. Høiseth et al. (2013) showed no significant differences in cognitive test results of psychogeriatric patients between BZD users and non-users. However, a case control study performed by Lai et al. (2014) found an increase risk in cognitive impairment in diabetic patients taking BZDs (Table S2).

All other studies focused on the association between BZD and dementia risk. Eight case control studies (Billioti de Gage et al., 2014; Chan et al., 2017; Gomm et al., 2016; Imfeld, Bodmer, Jick, & Meier, 2015; Takada, Fujimoto, & Hosomi, 2016; Tapiainen et al., 2018; C.-S. Wu, Ting, Wang, Chang, & Lin, 2011; C.-S. Wu, Wang, Chang, & Lin, 2009) and four cohort studies (Billioti de Gage et al., 2012; Gallacher et al., 2012; Gray et al., 2016; Shash et al., 2016) showed a significant association between chronic BZD use and increased risk of dementia. In contrast, in their case control studies, Imfeld et al. (2015) showed that BZD use was not associated with an increased risk of AD or vascular dementia. Three meta-analyses (Islam et al., 2016; Penninkilampi & Eslick, 2018; Zhong, Wang, Zhang, & Zhao, 2015) suggested that BZD use was significantly associated with an increased risk of dementia, and a dose–response association was highlighted by Zhong et al. (2015). Furthermore, the final meta-analysis performed by Penninkilampi and Eslick (2018) only included studies that controlled for protopathic bias. Although the evidence supporting the idea that BZDs effectively increase the risk of dementia is becoming increasingly stronger, to date, no studies have shown a causal relationship.

Reactive astrocytes observed around amyloid plaques could release abnormal amounts of GABA (Pariente, de Gage, Moore, & Bégaud, 2016). Thus, it is conceivable that the negative cognitive effects of BZD could be strengthened by the presence of age-related amyloid plaques in non-demented patients (Pariente et al., 2016). Thus, patients with these pathological lesions could be at risk of experiencing more negative effects following BZD use. BZDs could precipitate these patients into the symptomatic phase of dementia (Pariente et al., 2016). Pathological lesions (induced by the hyperphosphorylation of τ proteins) could lead to an alternative network to counteract these lesions. By decreasing brain activation, BZDs could limit the capacity to create such networks and precipitate the development of dementia (Pariente et al., 2016). In addition, the chronic administration of BZDs in mice decreases the BDNF levels and inhibits hippocampal plasticity (Zhao et al., 2012). BZD could also induce neuronal death through a hypoxia mechanism caused by changes in the brain haemodynamic response (Abookasis, Shochat, Nesher, & Pinhasov, 2014).

3.2.3 Opioids

Two cross-sectional studies have shown that long-term opioid therapy induces cognitive impairment (Table S3) in patients with chronic back pain (Schiltenwolf et al., 2014) and patients with cancer at daily opioid doses of 400 mg or greater (Kurita et al., 2011). Nevertheless, a case control study was carried out a few years earlier by Kurita and de Mattos Pimenta (2008), and these authors did not find a clear association between opioid use and cognitive impairment among cancer pain patients. In their review, Højsted et al. (2012) concluded that opioid treatment induced no effect or worsening of CF in cancer pain patients and had no effect or improvement in chronic non-cancer pain patients. Soyka, Zingg, Koller, and Hennig-Fast (2010) performed a cross-sectional study investigating opioid substitution treatment and showed that patients receiving long-term methadone maintenance treatment (MMT; >6 months) had better performances in executive functions and visuo-construction than patients receiving short-term MMT. However, Wang, Wouldes, and Russell (2013) showed that MMT is associated with cognitive impairment if the daily dose is higher than 60 mg. One cohort study performed by Dublin et al. (2015) assessed the association between opioid use and dementia. These authors found an increased risk of dementia among patients with chronic consumption (>3 months) of at least 30 mg·day⁻¹ of opioids but failed to find an association with cognitive decline. All reviewed articles appear to suggest that cognitive impairment might be associated only with the highest doses and chronic exposure to opioids. In addition, treatment with opioids could lead to cognitive impairment that then fades over the duration of treatment (due to pharmacological tolerance).

Ramage et al. (2005) investigated the brains of young opiate abusers and found an excess of hyperphosphorylated τ neurofibrillary tangles and an increase in amyloid precursor protein. However, both studies did not find any increase in β-amyloid deposition. In addition, opioids could induce a neuroinflammatory state and microglial activation, which could predispose to neurodegeneration (Ramage et al., 2005). Furthermore, an in vitro study showed that morphine induces apoptosis in neurons and microglia activation (Dublin et al., 2015). Morphine could also lead to diminished hippocampal neurogenesis and altered LTP (Bao et al., 2007).

3.3 Drugs not targeting the CNS

3.3.1 Antidiabetics

According to a cohort study (Ng et al., 2014) and a review (Dumbrill & Moulton, 2018), metformin and dipeptidyl peptidase 4 inhibitors (DPP4I) could exert beneficial effects on cognition among type 2 diabetes patients with normal cognition (Table S4). In contrast, a systematic review performed by Areosa Sastre, Vernooij, González-Colaço Harmand, and Martínez (2017) did not find any evidence that treatment of type 2 diabetes delays or prevents cognitive impairment. The results of studies investigating the effect of antidiabetic drugs on cognition in AD or MCI patients showed that DPP4I, insulin, and thiazolidinediones (TZD) were effective in increasing cognition in patients with AD or MCI (cohort studies; Isik, Soysal, Yay, & Usarel, 2017; Plastino et al., 2010; Rizzo et al., 2014; RCT; Craft et al., 2012; systematic review; Cao et al., 2018, and meta-analysis; Liu, Wang, & Jia, 2015). Only one cohort study found that metformin use was associated with impaired cognitive performance in AD and MCI patients (Moore et al., 2013).

Several studies have investigated the association between dementia risk and antidiabetic medications. Metformin alone and combined oral therapy (metformin and sulfonylureas) decreased the risk of dementia compared to that among patients without these medications (Hsu, Wahlqvist, Lee, & Tsai, 2011) and patients taking TZD (Cheng et al., 2014) and could be more effective than sulfonylureas (Orkaby, Cho, Cormack, Gagnon, & Driver, 2017). A cohort study (Heneka, Fink, & Doblhammer, 2015) and a case control study (Chou, Ho, & Yang, 2017) have also shown the efficacy of TZD in decreasing dementia risk. A meta-analysis performed by Ye, Luo, Xiao, Yu, and Yi (2016) found that metformin and TZD reduced the dementia risk. In contrast, in a case control study, Imfeld et al. (2012) showed that the long-term use of metformin may be associated with an increased risk of AD.

Metformin could, among other effects, provide neuroprotection by inhibiting the opening of mitochondrial permeability transition pores (leading to neuronal cell death; Orkaby et al., 2017) and decreasing neuroinflammation (Orkaby et al., 2017). Vildagliptin, which is a DPP4I, showed anti-inflammatory, antioxidant effects and induced hippocampal neurogenesis (Isik et al., 2017). TZDs could inhibit the inflammatory response induced by amyloid peptides (Cheng et al., 2014). Finally, there is growing interest in intranasal insulin, which has shown numerous beneficial effects in MCI and AD patients (Cao et al., 2018).

3.3.2 Antihypertensive medications

The cohort studies that aimed to assess the effect of AHMs on cognition among cognitively normal patients found a decreased risk of cognitive impairment with calcium channel blockers (CCBs), β-blockers, angiotensin-converting enzyme inhibitors (ACE-Is), and diuretics (Gao et al., 2013; Gelber et al., 2013; R. Peters, Collerton, et al., 2015; Yasar, Zhou, Varadhan, & Carlson, 2008; Table 1). Demir, Gürol, Özyiğit, and Üresin (2016) performed a cross-sectional study and did not find a difference in the Mini Mental State Examination scores between renin angiotensin system (RAS) blocker users (angiotensin 1 receptors blockers [ARBs] and ACE-Is) and other AHM users. RAS blockers, which cross the blood–brain barrier, provide more cognitive benefit than those that do not cross the blood–brain barrier (Wharton et al., 2015). However, ARBs could be more effective than ACE-Is in improving cognition among patients with MCI (Hajjar et al., 2013). The beneficial effects of the acute use of ARBs (losartan) on cognition have been well established by Mechaeil, Gard, Jackson, and Rusted (2011) in an RCT involving healthy volunteers. However, the effects of the chronic use of AHM are not well highlighted in RCTs. Indeed, Anderson et al. (2011) and Wharton et al. (2012) failed to find that chronically blocking the RAS has a beneficial effect on cognition. Furthermore, the authors did not highlight a difference between ARBs (telmisartan) and ACE-Is (ramipril). Although the beneficial effects of AHM on cognition are difficult to prove in RCTs, AHMs do not appear to have a negative effect on cognition (Bratzke et al., 2016; Jongstra, Harrison, Quinn, & Richard, 2016; J. Peters, Booth, & Peters, 2015).

Table 1. Studies investigating AHMs

Main outcome	Authors	Year	Design	Level of evidence	Drug class	Patients (n, users/non-users, cases/control)	Main results	Measure
CF	J. Peters et al.	2015	Systematic review	I	CCBs	5,740	Limited evidence to conclude that nitrendipine may be associated with a reduced risk of dementia	NA
CF	Wharton et al.	2012	CT	II	ACE-Is (ramipril)	14	No difference in CF	NA
CF	Anderson et al.	2011	RCT	II	Telmisartan and ramipril	25,620	Two clinical trials, i.e., ONTARGET and TRANSCEND, performed among patients with a high vascular risk with established cardiovascular disease or diabetes without heart failure failed to find significant difference in cognitive dysfunction among telmisartan, ramipril, and placebo	Cognitive impairment telmisartan vs. ramipril OR; 0.90 [0.80, 1.01] Telmisartan vs. placebo OR; 0.97 [0.81, 1.17] Cognitive decline telmisartan vs. ramipril OR; 0.97 [0.89, 1.06] Telmisartan vs. placebo OR; 1.10 [0.95, 1.27]
CF	Bratzke et al.	2016	RCT	II	β-blockers, ACE-Is, ARBs, aldosterone inhibitors, and diuretics	612	β-blockers, ACE-Is, ARBs, aldosterone inhibitors, and diuretics were not related to cognitive impairment in a heart failure population	NA
CF	Mechaeil et al.	2011	RCT	II	Losartan	44	Losartan improved acute prospective memory vs. placebo in healthy young adults	P = 0.05
CF	Yasar et al.	2008	Cohort	II	ACE-Is and diuretics	326	Reduced incidence of impairment in global and domain-specific cognition in nondemented elderly women	ACE-Is MMSE HR; 0.19 [0.04, 0.77] Diuretics MMSE: HR; 0.17 [0.04, 0.56]
CF	Gelber et al.	2013	Cohort	II	β-blockers	2,285	β-blockers reduced the risk of cognitive impairment	IRR; 0.69 [0.50, 0.94]
CF	R. Peters et al.	2015	Cohort	II	Diuretics, ARBs, ACE-Is, CCBs, and β-blockers	238	CCB use diminished cognitive decline.	P = 0.03
CF	Demir et al.	2016	Cross-sectional	IV	RASB (ACE-Is and ARBs)	62	Nonsignificant differences in MMSE score between RASB group and non-RASB group	P = 0.09
CF	Yasar et al.	2016	Review	UN	ACE-Is, ARBs, diuretics, β-blockers and CCBs	NA	Observational studies and RCT were included. Limitations restrict the ability of the authors to draw any conclusions	NA
CF in MCI patients	Hajjar et al.	2013	RCT	II	Lisinopril, candesartan, or HCTZ	53	Candesartan may preferentially improve executive function in MCI patients compared to lisinopril or HCTZ	P = 0.008
CF in dementia patients	Gao et al.	2013	Cohort	II	CACE-Is	817	Small reduced rate of cognitive decline with CACE-Is compared to that with No CACE-Is	Qmci, P = 0.049
Dementia and CF	Chang-Quan et al.	2011	Meta-analysis	II	ARBs, diuretics, ACE-Is, β-blockers, and CCBs	32,658 with AHM/36,905 without AHM	Observational studies were included. AHM use could decrease the risk of the development of VaD and any dementia	VaD RR; 0.67 [0.52, 0.87] Dementia RR; 0.87 [0.77, 0.96]
Dementia and CF	Levi Marpillat et al.	2013	Meta-analysis	II	Diuretics, ARBs, ACE-Is, CCBs, and β-blockers	18,515	Observational and interventional studies included. AHM had benefits on overall cognition and reduced all-cause dementia	Overall cognition0.05 [0.02, 0.07] All-cause dementia0.84 [0.75, 0.93]
Dementia and CF	Parsons et al.	2016	Meta-analysis	II	DHP-CCBs, diuretics, ARBs, and β-blockers	30,776	RCTs were included. Nonsignificant reduction in CD or dementia	CD HR; 0.96 [0.87, 1.06] Dementia HR; 0.90 [0.76, 1.07]
Dementia and CF	Zhuang et al.	2016	Meta-analysis	II	ACE-Is and ARBs	54,678	Observational and interventional studies were included. ACE-Is and ARBs reduced the risk of VCI and VaD	VaD: RR; 0.78 [0.64, 0.93] VCI RR; 0.87 [0.75, 0.98]
Dementia and CF	Xu et al.	2017	Meta-analysis	II	NA	30,895	Prospective cohort studies were included. Decreased risk of dementia but not AD, nonsignificant decreased risk of cognitive impairment or cognitive decline	Dementia RR; 0.86 [0.75, 0.99] AD RR; 0.83 [0.64, 1.09] Cognitive impairment RR; 0.89 [0.57, 1.38] Cognitive decline RR; 1.11 [0.86, 1.43]
Dementia and CF	R. Peters et al.	2014	Systematic review	II	CCBs	NA	Observational and interventional studies included. No association between CCBs and AD risk	HR; 0.79 [0.53, 1.17] I^2 = 63%
Dementia and CF	Jongstra et al.	2016	Review	UN	NA	2,490	2 RCTs were included. No effects of withdrawing AHM on cognition or prevention of dementia	NA
Dementia and CF in patients with MCI	Wharton et al.	2015	CT	II	RAS blockers (ACE-Is and ARBs)	784	Rate of conversion from MCI to AD was significantly lower among the RAS users vs. non-users.RAS blocker medications that cross the BBB led to less cognitive decline than non-BBB-crossing RAS medications	P = 0.04
Dementia and CF	Rouch et al.	2015	Review	UN	CCBs, β-blockers, ACE-Is, and ARBs	1,346,176	Observational and interventional studies included. AHM, CCBs, and ARBs may be beneficial in preventing cognitive decline and dementia	NA
Dementia	van Middelaar et al.	2017	RCT	II	Diuretics, ARBs, ACE-Is, CCBs, and β-blockers	1,951	CCBs and ARBs decreased the risk of dementia	CCBs HR; 0.56 [0.36, 0.87] ARBs HR; 0.60 [0.37, 0.98]
Dementia (AD) in normal or MCI patients	Yasar et al.	2013	RCT	II	Diuretics, ARBs, ACE-Is, CCBs, and β-blockers	2,248	Diuretics, ARBs, ACE-Is, and β-blockers were associated with a reduced risk of AD in patients with normal cognition. Diuretics only were associated with a reduced risk of AD among MCI patients	Diuretics HR; 0.51 [0.31, 0.8] ARBs HR; 0.31 [0.14, 0.68] ACE-Is HR; 0.50 [0.29, 0.83] CCBs HR; 0.62 [0.35, 1.09] β-blockers HR; 0.58 [0.36, 0.93]
Dementia	Hussain et al.	2018	Meta-analysis	II	CCBs	75,239	Observational studies were included. Reduced risk of dementia with CCB users compared to non-users. DHP use also reduced risk of dementia	CCBs RR; 0.70 [0.58, 0.85] DHP RR; 0.56 [0.40, 0.78]
Dementia	Tully et al.	2016	Meta-analysis	II	Diuretics	52,599	Observational and interventional studies were included. Diuretics were associated with a reduced risk of dementia and AD	Dementia HR 0.83; [0.76, 0.91] AD HR 0.82; [0.71, 0.94]
Dementia	Chiu et al.	2014	Cohort	II	ARBs	24,531/24,531	Reduced risk of dementia, AD, and VaD among high vascular-risk patients	Dementia HR; 0.54 [0.51, 0.59] AD HR; 0.53 [0.43, 0.64] VaD HR; 0.63 [0.54, 0.73]
Dementia	Tully et al.	2016	Cohort	II	CCBs, β-blockers, α blockers, diuretics, ACE-Is, and ARBs	6,537	Loop diuretics reduced the dementia risk	HR; 0.45 [0.22, 0.93]
Dementia	Feldman et al.	2016	Cohort	III	CCB	15,664	Decreased risk of dementia with amlodipine	HR; 0.61 [0.49, 0.77]
Dementia	Hwang et al.	2016	Cohort	III	CCBs	18,423	CCBs reduced the risk of dementia, AD, and VAD	Dementia aHR; 0.81 [0.75, 0.87] AD aHR; 0.80 [0.72, 0.88] VaD aHR; 0.81 [0.70, 0.94]
Dementia in diabetes mellitus patients	Kuan et al.	2016	Cohort	III	ACE-Is and ARBs	2,377/2,377 ACE-Is,1,780/1,780 ARBs	In patients with type 2 diabetes mellitus, ACE-Is and ARBs decreased all-cause dementia risk	ACE-Is HR; 0.74 [0.56, 0.96] ARBs HR; 0.60 [0.37, 0.97]
Dementia	Davies et al.	2011	Case control	III	ACE-Is and ARBs	5,797/23,188	ARBs and ACE-Is reduced the risk of AD	ARBs OR; 0.47 [0.37, 0.58] ACE-Is OR; 0.76 [0.69, 0.84]
Dementia	Wagner et al.	2012	Case control	III	Diuretics, ARBs, ACE-Is, CCBs, and β-blockers	1,297/1,297	β-blocker use reduced the risk of dementia	β-blockers OR; 0.79 [0.61, 0.99]

• Note. ACE-Is, angiotensin-converting enzyme inhibitors; AD, Alzheimer's disease; ARBs, angiotensin 1 receptor blockers; CACE-Is, centrally acting ACE-Is; CD, cognitive decline; CF, cognitive function; CT, clinical trial; DHP, dihydropyridine; HCTZ, hydrochlorothiazide; HR, hazard ratio; IRR, incidence risk ratio; NA, not available; MMSE, Mini Mental State Examination; qmci, quick mild cognitive impairment; OR, odds ratio; RAS, renin angiotensin system; RCT, randomized controlled trial; RR, relative risk; UN, unclassifiable; VaD, vascular dementia; VCI, vascular cognitive impairment.

All cohort studies and case control studies that have assessed the association between AHMs and dementia have shown a significant beneficial effect of ACE-Is, ARBs, CCBs, and β-blockers (Chiu et al., 2014; Davies, Kehoe, Ben-Shlomo, & Martin, 2011; Feldman et al., 2016; Hwang et al., 2016; Kuan et al., 2016; Tully et al., 2016; Wagner et al., 2012). ARBs and ACE-Is could also be effective in decreasing the dementia risk in patients with other co-morbidities, such as diabetes (Kuan et al., 2016). The reviews conducted by Rouch et al. (2015) and Yasar et al. (2016) concluded that AHMs could prevent vascular dementia and AD without highlighting a specific class of AHMs. All meta-analyses including observational and interventional studies found a benefit of AHMs on dementia risk (Chang-Quan et al., 2011; Hussain, Singh, Rahman, Habib, & Najmi, 2018; Tully, Hanon, Cosh, & Tzourio, 2016; Xu et al., 2017). However, Zhuang et al. (2016) found that central ACE-Is (which cross the blood–brain barrier) were associated with a reduced risk of dementia. Two RCTs (van Middelaar et al., 2017 and Yasar et al., 2013) showed that the use of diuretics, ARBs, and ACE-Is is associated with a reduced risk of dementia among patients with normal cognition. However, a systematic review (R. Peters, Booth, & Peters, 2014) and a meta-analysis including only RCTs (Parsons, Murad, Andersen, Mookadam, & Labonte, 2016) showed that there is no evidence that AHMs (with a focus on CCBs in R. Peters et al., 2014) could decrease the risk of dementia.

Taken together, these studies suggest that ARBs, ACE-Is, and CCBs appear to be more efficient than the other classes in preventing cognitive decline or dementia (with a preference given to ARBs; Levi Marpillat, Macquin-Mavier, Tropeano, Bachoud-Levi, & Maison, 2013; Rouch et al., 2015). However, although most observational studies have demonstrated a beneficial effect on cognition or a decrease in dementia risk, these results are difficult to highlight in RCTs.

By diminishing the intracellular calcium levels through the blockade of neuronal calcium channels, CCBs could prevent cognitive decline (Rouch et al., 2015) because an increase in cytosolic calcium could trigger the production of Aβ1-42 and induce neuronal death. CCBs also improve the regional cerebral blood flow and angiogenesis and decrease oxidative stress markers (R. Peters, Schuchman, Peters, Carlson, & Yasar, 2016).

The emergence of the angiotensin hypothesis and the implication of RAS in the pathogenesis of AD are currently broadly recognized (Kehoe, 2018). Thus, the modulation of RAS by ARBs and ACE-Is is noteworthy. ACE-Is also diminished ROS generation and enhanced CF (R. Peters et al., 2016) probably through the inhibition of cerebral ACE. However, according to (Fournier et al., 2009), ACE-Is only exert short-term protective effects, and chronic exposure to ACE-Is could lead to an increased Aβ42 brain burden and cognitive decline. Indeed, ACE mediates the conversion of Aβ1-42 to Aβ1-40. Thus, ACE-Is could inhibit the potential protective mechanism of ACE (Kehoe, 2018). Furthermore, ACE-Is are not harmless. A cohort study showed that compared to other AHMs, ACE-Is increased the risk of mortality in AD patients (Kehoe, Davies, Martin, & Ben-Shlomo, 2013).

Among the ARBs, losartan and telmisartan are the most studied (R. Peters et al., 2016). In a mouse model of AD, losartan normalized cerebrovascular reactivity (R. Peters et al., 2016), which is impaired in AD patients. These beneficial effects on memory have been well established, but the effects on the cerebral levels of Aβ1-42 remain unclear (R. Peters et al., 2016). However, losartan could counteract the deleterious effect induced by angiotensin II (Ang-II). Indeed, losartan can abolish the modulation of amyloid precursor protein secretase and increases in amyloid production (Zhu et al., 2011); reduce amyloid plaque formation and neuroinflammation (Danielyan et al., 2010); and reduce the increase in phospho-τ levels via a reduction in glycogen synthase kinase 3ß (GSK3ß; Tian, Zhu, Xie, & Shi, 2012). In addition, ARBs can down-regulate AT₁ receptors and up-regulate AT₄ receptors. By inhibiting the binding of Ang-II to the angiotensin 1 receptor, ARBs enhance the binding of Ang-II to AT₂ and AT₄ receptors. The activation of AT₂ and AT₄ receptors is well known to induce cognitive enhancement (Fournier et al., 2009). This mechanism of action could explain why ARBs might be more effective in preventing cognitive decline than ACE-Is, which decrease Ang-II production and thus decrease AT₂ and AT₄ receptor activation. In addition, ACE is overexpressed in the hippocampus of AD patients. This overexpression enhances the formation of Ang-II and decreases the release of ACh (Hwang et al., 2016). Thus, the diminished formation of Ang-II by ARBs could increase ACh release. Furthermore, a post-mortem study found that ARBs could reduce amyloid deposition in patients with and without AD compared to untreated patients and treated patients using other AHMs (Hajjar, Brown, Mack, & Chui, 2012).

Preclinical studies investigating β-blockers have found that carvedilol reestablished LTP and, thus, enhanced neuronal plasticity in a mouse model of AD (Kherada, Heimowitz, & Rosendorff, 2015). Carvedilol also attenuated the aggregation of Aß peptides in vivo. Propranolol was also able to decrease the Aß42 levels and hyperphosphorylation of τ and increase the phosphorylation of Akt and GSK3ß and BDNF production (Kherada et al., 2015).

3.3.3 Proton pump inhibitors

Recently, the association between PPI use and the risk of dementia or poorer CF in the overall population has raised much concern. Regarding PPI effects on cognition, a cross-sectional study found that PPI diminished cognitive performance (Nevado-Holgado et al., 2016), but this effect was not confirmed by two recent cohort studies (Lochhead et al., 2017; Wod et al., 2018; Table S5).

Regarding the dementia risk, initially, three cohort studies (Gomm et al., 2016; Haenisch et al., 2015; Tai et al., 2017) and one cross-sectional study (Herghelegiu, Prada, & Nacu, 2016) found a positive association between dementia risk and PPI use. However, this association was not found in more recent studies, including a case control study (Imfeld, Bodmer, Jick, & Meier, 2018) and a cohort study (Gray et al., 2018). One cross-sectional study (de Souto Barreto et al., 2013) and one case control study (Booker, Jacob, Rapp, Bohlken, & Kostev, 2016) even found a decreased risk of dementia with PPI use. A meta-analysis conducted by Wijarnpreecha, Thongprayoon, Panjawatanan, and Ungprasert (2016) failed to find a significant association between PPI use and an increased risk of dementia with a between-study heterogeneity of 99%. In conclusion, the evidence supporting an association between an increased risk of dementia and PPI use is actually quite limited (Batchelor et al., 2017) and might have been initially overestimated. These data are reassuring because PPIs are among the most widely used drug classes.

Although most epidemiological studies suggest that PPI use does not increase the dementia risk, in vivo and in vitro studies have shown that PPIs could interact with the pathophysiological pathway of AD. Indeed, lansoprazole increased amyloid β production (Aß40 and Aß42 oligomers) by enhancing ß-secretase activity and modulating γ-secretase (Batchelor et al., 2017). By inhibiting the vacuolar proton pump in microglial lysosomes, PPIs could basify lysosomes and thus hamper the degradation of fibrillar β-amyloid (Haenisch et al., 2015). Furthermore, chronic exposure to PPIs impairs endothelial function and accelerates human endothelial senescence by reducing telomere lengths (Batchelor et al., 2017). Finally, by decreasing the gastric pH, PPIs reduce the absorption of weak acids, such as cyanocobalamin, which might be a risk factor for dementia.

3.3.4 Statins

Studies investigating the effect of statins on cognition present discrepant results. In patients with normal cognition at baseline, two cohort studies found a beneficial effect of statins on cognition (Lilly, Mortensen, Frei, Pugh, & Mansi, 2014; Steenland, Zhao, Goldstein, & Levey, 2013; Table 2). However, a cross-sectional study (Benito-León, Louis, Vega, & Bermejo-Pareja, 2010), an RCT (Trompet et al., 2010), and two meta-analyses (Ott et al., 2015; Richardson et al., 2013) did not find a beneficial effect of statins on cognition among patients with normal cognition.

Table 2. Studies investigating statins

Main outcome	Authors	Year	Design	Level of evidence	Patients (n, users/non-users, cases/control)	Main results	Measure [95% CI], P value, or MD
CF	Richardson et al.	2013	Systematic review	II	NA	Observational and interventional studies included.No adverse effects of statins on cognition	NA
CF	Trompet et al.	2010	RCT	II	5,804	Pravastatin treatment in old age did not affect cognitive decline	P > 0.05
CF	Lilly et al.	2014	Cohort	III	13,626	Decreased risk of cognitive disorders	aOR; 0.87 [0.83, 0.91]
CF	Benito-Leon et al.	2010	Cross-sectional	IV	137/411	No difference in cognitive function	P > 0.05
CF in patients with normal cognition and CF in MCI patients	Steenland et al.	2013	Cohort	II	1,244/2,363	Better attention among subjects with normal cognition at baseline	P < 0.006
CF in cognitively normal subjects and AD patients	Ott et al.	2015	Systematic review	I	27,643 included in MA	RCT were included. Statin therapy was not associated with cognitive impairments in cognitively normal subjects or AD subjects	No effect on cognition in cognitively normal subjects OR; 0.01 [−0.01, 0.03] No effect in AD subjects OR; −0.05 [−0.19, 0.1]
CF in AD patients	Padala et al.	2012	Open-label study	III	18	Improvement in cognition with discontinuation of statins and worsening with rechallenge among patients with AD	Improve MMSE scores after discontinuation (MD1.9 [3.0], P = 0.014) Decrease in MMSE scores after re-challenge (MD 1.9 [2.7], P = 0.007)
Dementia and CF	Swiger et al.	2013	Systematic review	I	NA	Inclusion of high-quality RCT and prospective cohorts. No effect of short-term use on cognition. Long-term use (>1 year) of statins reduced the risk of dementia	Long-term use HR; 0.71 [0.61, 0.82]
Dementia and CF	Power et al.	2015	Review	UN	NA	The protective effect of statins is due to reverse causation. No evidence of a causal preventative effect of late-life statin use on cognitive decline or dementia	NA
Dementia and CF	Mc Guiness et al.	2016	Review	UN	26,340	Two RCT included in this review (Prosper and HPS 2002). Statins given in late life to people at risk of vascular disease do not prevent cognitive decline or dementia	Prosper study OR; 1.0 [0.61, 1.65] HPS 2002 MMSE MD 0.06 [−0.04, 0.16]
Dementia	Song et al.	2013	Meta-analysis	II	57,020	Prospective cohort studies were included. Reduced risk of dementia	RR; 0.62 [0.43, 0.81]
Dementia	Wong et al.	2013	Meta-analysis	II	NA	Observational studies were included.Decreased risk of all types of dementia and AD	Dementia RR; 0.82 [0.69, 0.97] AD RR; 0.70 [0.60, 0.83]
Dementia	Macedo et al.	2014	Meta-analysis	II	27,622,899	Observational studies were included.Lower risk of dementia	OR; 0.7 [0.62, 0.87]
Dementia	X. Zhang et al.	2018	Meta-analysis	II	NA	Observational studies were included. Decreased risk of dementia (AD and non-AD dementia)	Dementia RR; 0.85 [0.80, 0.89] AD RR; 0.81[0.73, 0.89] Non-AD dementia RR; 0.81 [0.73, 0.99]
Dementia	Hendrie et al.	2015	Cohort	II	974	Decreased risk of incident dementia and AD	Dementia OR; 044 [0.21, 0.92] AD OR; 0.4 [0.18, 0.91]
Dementia	Hippisley-Cox and Coupland	2010	Cohort	II	2,121,786	Lower risk of dementia among women with simvastatin and atorvastatin	Simvastatin aHR; 0.88 [0.81, 0.96] Atorvastatin aHR; 0.84 [0.74, 0.96]
Dementia	C.-K. Wu et al.	2015	Cohort	II	15,200/42,469	Decreased risk of dementia associated with an increased dose of statin	≤88.4 mg HR; 0.83 [0.75, 0.92] ≤643.5 mg HR; 0.72 [0.64, 0.81] >643.5 mg HR; 0.39 [0.33, 0.45]
Dementia	Gnjidic et al.	2016	Cohort	II	2,056	No decreased risk of dementia	aOR; 0.66 [0.37, 1.19]
Dementia	Horsdal et al.	2009	Case control	III	11,039/110,340	Reduced risk of hospitalization for dementia	aOR; 0.67 [0.6, 0.75]
Dementia	Corrao et al.	2013	Case control	III	1,380/27,201	Reduced risk of dementia with statin use >7 months	7–24 months OR; 0.85 [0.74, 0.98] 25–48 months OR; 0.72 [0.61, 0.85] >48 months OR; 0.75 [0.61, 0.94]
Dementia	Chen et al.	2014	Case control	III	9,257/18,459	Dementia significantly decreased by 9% per year of treatment with statins	aOR; 0.91 [0.85, 0.97]
Dementia (AD)	Lin et al.	2015	Case control	III	719/719	Early statin use in mild to moderate AD patients leads to a lower risk of AD	OR; 0.85 [076, 0.95]
Dementia	Booker et al.	2016	Case control	III	11,956/11,956	Decreased risk of dementia	OR; 0.94 [0.90, 0.99]
Dementia	Chitnis et al.	2015	Cohort	III	8,062	No difference in the risk of dementia in an at-risk heart failure population	Current users HR; 0.93 [0.71, 1.21] Former users HR; 0.99 [0.79, 1.25]
Dementia	Chuang et al.	2015	Cohort	III	61,650/61,650	Decreased risk of dementia	HR; 0.92 [0.86, 0.98]
Dementia (AD) in patients with type 2 diabetes	Chen et al.	2014	Cohort	III	2,400/15,770	Decreased risk of AD among patients with type 2 diabetes	AD aHR; 0.48 [0.30, 0.76] Non-AD aHR; 1.07 [0.54, 2.12]
Dementia in patients with late onset depression	Yang et al.	2015	Cohort	III	1,844/1,844	Decreased risk of dementia among patients with late onset depression	HR; 0.67 [0.55, 0.83]
Treatment of dementia (AD and VaD)	Mc Guiness et al.	2014	Systematic review	I	1,154	4 RCT included. No beneficial effect of statins on ADAS-Cog or MMSE among patients with AD or VAD	ADAS-Cog MD 0.2; [−1.05, 0.52] MMSE MD-0.32 [−0.71, 0.06]

• Note. AD, Alzheimer's disease; ADAS-Cog, Alzheimer's disease Assessment Scale-cognitive; aHR, adjusted hazard ratio; aOR, adjusted odds ratio; CF, cognitive function; HR, hazard ratio; MA, meta-analysis; MCI, mild cognitive impairment; MD, mean difference; NA, not available; OR, odds ratio; PA, pooled analysis; RCT, randomized controlled trials; RR, relative risk; UN, unclassifiable; 95% CI, 95% confidence interval.

There are also discrepancies regarding their effects on cognition in patients with MCI or dementia. Indeed, Steenland et al. (2013) found a beneficial effect on cognition in patients with MCI. However, an open-label study performed by Padala et al. (2012) showed an improvement in cognition following their withdrawal in patients with AD or mixed dementia.

Seven cohort studies (Chen et al., 2014; Chuang, Lin, Lin, Sung, & Kao, 2015; Hendrie et al., 2015; Hippisley-Cox & Coupland, 2010; Li et al., 2010; C.-K. Wu et al., 2015; Yang et al., 2015) and five case control studies (Booker et al., 2016; Chen, Liu, Chen, & Wu, 2014; Corrao et al., 2013; Horsdal et al., 2009; Lin et al., 2015) found that statin use was associated with a reduced risk of dementia. According to these studies, beginning treatment with high doses of lipophilic statins before an early old age (<80 years old) and taking these statins for a long time might reduce the risk of dementia. However, two cohort studies (Chitnis et al., 2015; Gnjidic et al., 2016) did not find any differences in the risk of dementia among statin users and non-users. All meta-analyses involving observational studies (Macedo et al., 2014; Song, Nie, Xu, Zhang, & Wu, 2013; Swiger, Manalac, Blumenthal, Blaha, & Martin, 2013; Wong, Lin, Boudreau, & Devine, 2013; X. Zhang, Wen, & Zhang, 2018) found a decreased risk of dementia and AD. However, these meta-analyses demonstrated several limitations, such as high heterogeneity across studies. According to Power, Weuve, Sharrett, Blacker, and Gottesman (2015) and McGuinness, Craig, Bullock, and Passmore (2016), the data supporting a causal preventive effect of statins on cognitive decline or dementia if given late in life are actually insufficient, and statins are not effective in treating dementia (McGuinness, Craig, Bullock, Malouf, & Passmore, 2014).

The precognitive effect of statins could be due to pleiotropic effects (Mendoza-Oliva et al., 2014). By decreasing the prenylation of Rho GTPases, statins could up-regulate the production of endothelial NOS (Hamel, Royea, Ongali, & Tong, 2016) and thus restore neurovascular coupling and cerebral blood flow, which are impaired in AD patients. Their effects on CF could also be due to their actions on amyloid plaques. Statins attenuated the accumulation of amyloid β deposits (Mendoza-Oliva et al., 2014). Statins could also exert anti-inflammatory effects by reducing the levels of isoprenyl intermediates (Mendoza-Oliva et al., 2014) and increasing neurogenesis and neuronal survival (Hamel et al., 2016) through the up-regulation of α7nAChR-cascading PI3K-Akt (Hamel et al., 2016). Statins increase the phosphorylation of Akt and GSK3ß, leading to enhancement in LTP (Hamel et al., 2016). However, in vitro studies investigating neuroblastoma cells have highlighted the potentially deleterious effects of statins, such as an increase in ROS or the induction of apoptosis in these cells through a mitochondrial pathway. An in vivo study has also shown that simvastatin could alter LTP (Mendoza-Oliva et al., 2014).

4 DISCUSSION

This review aimed to summarize the recent literature on drug use and their effects on cognition. Studies were included regardless of their benefit or deleterious effect on cognition. Most studies investigating BZDs suggested that these drugs increased the risk of dementia. Firm conclusions on dementia risk and cognitive impairment associated with opioids and APs are difficult to draw. However, these different classes appeared to impair cognition not by an unambiguous mechanism of action specific to each class but by several relatively interdependent mechanisms (Figure 2) that somewhat differ across the different classes investigated (e.g., impaired neurotransmission, neuroinflammation, neuronal death, oxidative stress, and interactions with dementia-related pathways). In contrast, by acting in an opposite way to the mechanisms cited above (Figure 3), statins, AHM, and antidiabetics could potentially decrease the risk of dementia. The dementia risk initially associated with the use of PPIs might have been overestimated. However, this review was not exhaustive, and other drugs not mentioned here, such as corticosteroids, histamine H₂ receptor antagonists, antineoplastic agents, or even some cardiovascular agents, might induce deleterious effects on cognition.

Figure 2

Potential mechanisms of drug-induced cognitive disorders. ACE, angiotensin-converting enzyme; ACE-Is, angiotensin-converting enzyme inhibitors; AchIs, anticholinergics; AEDs, antiepileptic drugs; APs, antipsychotics; APP, amyloid protein precursor; BZD, benzodiazepines; PPI, proton pump inhibitors; VPA, valproic acid

Figure 3

Potential mechanisms of drugs preventing cognitive disorders. ACE-Is, angiotensin-converting enzyme inhibitors; ARBs, angiotensin 1 receptor blockers; AT1R, AT2R, and AT4R, angiotensin receptors 1, 2, and 4; BBs, β blockers; CCBs, calcium channel blockers; DPP4I, dipeptidyl peptidase 4 inhibitor; TZD, thiazolidinediones

This review also aimed to raise awareness among physicians, of drug-induced neurocognitive effects. Before prescribing medication, physicians should be aware of the drug's cognitive safety. Importantly, many drugs exhibit anticholinergic properties, and a high anticholinergic load is associated with cognitive impairment. Furthermore, the rational and restricted use of BZDs and PPIs, which are widely prescribed, should be encouraged given their many adverse effects.

Further research should focus on assessing the cognitive safety of drugs. For instance, a better characterization of the effects of APs and opioids on cognition could provide more information about their effectiveness and safety to physicians, allowing them to make informed decisions and integrate the drugs' cognitive safety into the overall benefit/risk ratio. Future research should also focus on assessing the properties of drugs that cross the blood–brain barrier and thus potentially exhibit CNS action, for example, central ACE-Is. A better characterization of the implications of the RAS system in dementia is also of great interest, and Phase II trials are already underway in patients with mild to moderate AD. Indeed, the RADAR trial (ISRCTN: 93682878, EudraCT:2012-003641-15) aims to assess the efficacy of losartan in reducing AD pathology (Kehoe et al., 2018) and the SARTAN-AD trial aims to compare the efficacy of perindopril and telmisartan (https://clinicaltrials.gov/ct2/show/NCT02085265) to slow down the progression of AD.

Finally, assessing the effects of each drug on cognition should be a priority during the development of drugs from preclinical to clinical studies by using different paradigms of neurodegenerative diseases and different reliable biomarkers, including functional MRI, PET, or even EEG. After marketing approval, cognitive safety should be monitored by relying on epidemiological cohort or case control studies and the pharmacovigilance system.

4.1 Nomenclature of targets and ligands

Key protein targets and ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY (Harding et al., 2018), and are permanently archived in the Concise Guide to PHARMACOLOGY 2017/18 (Alexander, Christopoulos et al., 2017; Alexander, Fabbro et al., 2017; Alexander, Peters, et al., 2017).

CONFLICT OF INTEREST

The authors have no conflicts of interest to declare, except for R.B., who has participated on the boards of Lundbeck, Otsuka Pharmaceutical, and Novartis via medical conferences and the publication of articles.