Abstract

Tweetable abstract

Introduction

Montelukast is the most commonly prescribed leukotriene receptor antagonist (LTRA) to reduce airway inflammation in asthma and prevent exercise-induced bronchoconstriction. Montelukast has its own merits in being orally administered as once-daily dosing, and targets alternative inflammatory pathways to those targeted by inhaled corticosteroids (ICS), which may offer therapeutic benefits to specific groups of patients [1]. Therefore, the Global Initiative for Asthma guideline recommends LTRAs as an alternative or add-on controller to ICS in the management of persistent asthma, or as an alternative controller in initial asthma management in children aged ≤5 years [2].

The risk of neuropsychiatric events in montelukast users has long been an alarming issue for regulatory authorities. The issue first came under scrutiny in 2009, after the US Food and Drug Administration (FDA) issued a change in manufacturer labelling over the neuropsychiatric safety of montelukast [3], and contrasted with recommendations from the joint statement issued by the American Academy of Allergy, Asthma & Immunology and the American College of Allergy, Asthma & Immunology [4]. In March 2020, the FDA further issued a black box warning about the risk of mental health side-effects in montelukast users after reviewing case reports submitted to the FDA Adverse Event Reporting System [5]. The FDA also announced the initiation of an observational study using data under the Sentinel Initiative for the surveillance of medical products [6].

Recent work using data from electronic health record databases worldwide has reported possible relationships between montelukast and neuropsychiatric events, such as sleeping disorders [7], depression [6], dementia [8] and attention deficit/hyperactivity disorder (ADHD) [9]. The largest observational study to date initiated by the FDA Sentinel Initiative revealed a reduced risk of outpatient depressive disorders [6], yet paediatric Canadian studies identified increased risk of neuropsychiatric events [10, 11]. Two observational studies from Japan [8] and Norway [12] have also demonstrated possible paradoxical protective effects of montelukast in neurodegenerative diseases like dementia. However, the results are inconclusive to date owing to heterogeneity in study populations and outcomes. Therefore, we conducted this systematic review to describe the risk of neuropsychiatric events stratified by different psychiatric diagnoses in patients with asthma on montelukast. This review not only serves as an update on our 2018 systematic review [13], but also adds new perspectives on the overall safety or the possible therapeutic potential of montelukast in certain neuropsychiatric diseases in selected patient populations.

Methods

Search strategy

This systematic literature review was carried out from database inception to 7 September 2022. All studies were identified from databases, including PubMed, Cochrane Library and Embase, using Medical Subject Heading (MeSH) terms and keywords relevant to LTRAs and various neuropsychiatric events (figure 1). The list of keywords is included in supplementary material 1. Additional studies were screened from the bibliographies of the identified articles. Studies were screened and identified independently by authors CWHL and SP.

Open in a separate window

FIGURE 1

Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) flowchart of studies included in the systematic review. NE: neuropsychiatric event; LTRA: leukotriene receptor antagonist.

Data extraction

Data from studies were extracted using the PICO (population, intervention, comparison and outcomes) tool independently by CWHL and SP. Discrepancies in the inclusion of studies were discussed and resolved by the two authors. Study results were extracted by CWHL, and SP reviewed and provided additional details whenever necessary. The PICO tool is described in supplementary material 2. No discrepancies were found in the quality assessment process or described otherwise. This paper defined the collation of adverse drug event reports from pharmacovigilance databases worldwide as “pharmacovigilance studies”. Study characteristics of pharmacovigilance studies, case reports and case series are summarised in supplementary material 3–5. Information from case reports, including age, gender, the onset of adverse drug reactions, concomitant medications and prognosis, was extracted. The inclusion and exclusion criteria, exposure groups, study outcome(s), key results and the study periods for observational studies are reported and summarised in tables 1–4.

TABLE 1

Study characteristics of research studies on suicide-related outcomes

Study (publication year)	Study design	Study/inclusion period	Inclusion criteria	Exclusion criteria	Subjects (n)	End-points	Key findings
Paljarvi et al. [7] (2022)	Retrospective cohort study	2015–2019	For asthma group: 1. Asthma-related healthcare contact 2. Aged 15–64 years at index prescription of MTK 3. Alive for 12 months after index prescription	1. Dx of COPD 2. Dx of obstructive sleep apnoea 3. Dx of neoplasm 4. Pregnancy during study period	MTK users (36245) vs non-users (36245)	1. 12-month incident Dx for NEs, including bipolar disorders, depression, anxiety disorders, OCD and sleeping disorders 2. 12-month incident dispensed prescriptions for psychiatric medications	OR (95% CI) (propensity score-matched) for: 1. any non-fatal self-harm in 1 year: 1.06 (0.72–1.55)
Sansing-Foster et al. [6] (2020)	Retrospective cohort study	2000–2015	1. Aged >6 years with both medical and drug coverage for at least 183 days before medication initiation 2. Asthma Dx in any care setting 3. No Dx of COPD during 183 days prior to index date	1. Experienced outcome on cohort entry day 2. >45-day gap between consecutive enrolment periods for medical and drug coverage 3. Recipients of LTRA, LAMA and ICS 183 days prior to index date 4. Same-day dispensing for both MTK and ICS on index date	MTK (457377) vs ICS (457377)	1. Hospitalisation for depressive disorders defined as inpatient claims 2. Treated outpatient depressive disorders 3. Hospitalisation for self-harm 4. Modified self-harm (by E-codes)	HR (95% CI) for 1. self-harm: 0.92 (0.69–1.21) 2. modified self-harm: 0.81 (0.63–1.05)
Lu et al. [14] (2015)	ITS	2005–2010	1. Aged 5–64 years 2. Continuous health enrolment plan for past 12 months 3. Enrolled for health plan for current month 4. At least one outpatient or inpatient visit in the past year due to asthma diagnosis	History of: 1. COPD, cystic fibrosis, bronchiectasis 2. Pulmonary hypertension or embolism 3. Bronchopulmonary dysplasia 4. CHF	140000 (rolling cohort with asthma)	Monthly percentage of patients 1. dispensed an LTRA and non-LTRA 2. on mental health visits Quarterly percentage of patients 3. medically treated for suicide attempts	1 year after the FDA label change in 2009, suicide attempts: 1. did not change in adolescents 2. increased by 0.03 (95% CI 0.01–0.05) percentage points in young adults (aged 18–29 years) 3. increased by 0.01 (95% CI 0.00–0.01) percentage points in adults (aged 30–64 years)
Chen et al. [15] (2014)	Retrospective cohort study	2000–2008	1. Aged >10 years 2. Inpatient asthma Dx or ≥2 recorded outpatient asthma Dx 3. 1 year duration of asthma	1. Received asthma Dx from 1997–1999 2. Completed suicide (deaths within 2 weeks of self-harm)	cDDD 0–90: 165960 cDDD ≥90: 726	1. Self-harm (ICD-9 E950–E959, E980–E989)	1. Adjusted HR (95% CI) for asthma diagnosis: 1.70 (1.35–2.14) 2. Adjusted HR (95% CI) for MTK cDDD ≥90: 0.92 (0.29–2.91)
Schumock et al. [16] (2012)	Case–control	1997–2006	1. Aged 5–24 years 2. ≥1 prescription for asthma controller medication (ICS, LTRA, LABA, methylxanthines, immunomodulators, mast cell stabilisers, inhaled anticholinergics)	1. Received asthma controller medication >30 days before date of asthma diagnosis 2. Last enrolment date on or before index date 3. Not continuously enrolled for ≥6 months 4. Not continuously enrolled for ≥2 months and 20% or more of total months	Case: 344 Control: 3438	1. Suicide attempt (ICD-9 E950–E959)	Adjusted OR (95% CI) for 1. ever users of LTRA: 0.74 (0.46–1.20) 2. current users of LTRA: 0.70 (0.36–1.39)
Schumock et al. [17] (2011)	Ecological study	1999–2006	Poisson regression analysis of association between LTRA and suicide deaths at country level	–	249872 suicide deaths	1. Suicide deaths	1. Association between rate of MTK prescriptions dispensed and suicide rate: −0.0003, p=0.0217 2. Estimate rate multiplier for MTK: 0.9997 (95% CI 0.9994–0.9999)
Jick et al. [18] (2009)	Cohort	1998–2007	1. All patients receiving ≥1 MTK prescriptions	–	MTK (23500)	1. Rate of suicide completed 2. Person-year(s) at risk of suicide	1. Rate (95% CI) of suicide with asthma Dx: 1.02 (0.6–1.5) per 100000 person-years 2. At risk for suicide: 21050 person-years
Philip et al. [19] (2009)	Reviews of RCTs (116 trials#)	Up to 11 March 2008	1. All age ranges 2. Regardless of approval status for any indication completed by 11 March 2008 3. Early (Phase 1 and Phase 2a trials)/Late (Phase 2b/Phase 3)	1. Subjects who had clinically significant psychiatric disorders	MTK (20131) vs Placebo (9287) vs Active control (8346)	1. ADRs¶	1. No completed suicides reported in any study 2. 1 patient each in active-controlled and open-label studies

Open in a separate window

ADR: adverse drug reaction; cDDD: cumulative defined daily dose; CHF: congestive heart failure; Dx: diagnosis; FDA: US Food and Drug Administration; HR: hazard ratio; ICD-9: International Classification of Diseases, ninth revision; ICS: inhaled corticosteroid; ITS: interrupted time series; LAMA: long-acting muscarinic antagonist; LTRA: leukotriene receptor antagonist; MTK: montelukast; NE: neuropsychiatric event; OCD: obsessive compulsive disorder; OR: odds ratio; RCT: randomised controlled trial; vs: versus. ^#: controls include placebo and active control while trials included randomised, double-blind, placebo-controlled parallel-group trials; randomised, double-blind, placebo-controlled crossover trials; and open-label studies; ^¶: ADRs include completed suicide, suicide attempt, suicidal ideation and other terms suggesting self-injurious behaviour.

TABLE 4

Study characteristics of research studies on neurodegenerative disease-related outcomes

Study (publication year)	Study design	Study period	Inclusion criteria	Exclusion criteria	Subjects (n)	End-points	Key findings
Ishikura et al. [8] (2021)	Retrospective cohort study	2006–2015	1. New diagnosis of bronchial asthma during study period (defined as ≥2 outpatient claims or ≥1 inpatient claims) 2. Aged ≥50 years at the time of diagnosis 3. ≥2 prescriptions of LTRA during study period	1. Zafirlukast users 2. Diagnosis of dementia before bronchial asthma diagnosis, before the first prescription or between the first and second prescription of LTRA 3. Enrolment period of <6 months 4. Diagnosis based on “Suspected” diagnostic codes	LTRA users (10471) vs non-users (10471)	1. Onset of dementia of all types identified by ICD-10 codes 2. Subtypes of dementia (Alzheimer's dementia and vascular dementia)	1. Adjusted HR (95% CI) for dementia: 0.42 (0.20–0.87) 2. Incidence rates of dementia in LTRA users: 0.52 cases per 1000 person-years
Xiong et al. [30] (2021)	Longitudinal observational study	2005–2021	1. Cognitively normal: normal cognition and did not use any medications for dementia 2. MCI: MCI diagnosis 3. AD: all-cause dementia diagnosis with AD as primary or contributing cause of cognitive impairment defined by NINCDS-ADRDA or NIA-AA criteria	1. Patients who did not complete the medication form 2. Patients who did not have baseline data to identify the variables used in propensity score matching	Number (Cognitively normal/MCI/AD): LTRA users (350/200/151) vs non-users (1050/600/453)	1. Immediate logical memory 2. Delayed logical memory 3. Psychomotor processing speed 4. Language test scores	Adjusted relative risk (95% CI) (cognitively normal/MCI/AD) for 1. Immediate memory: 0.992 (0.985–0.998)#/1.015 (0.991–1.039)/1.048 (0.972–1.130) 2. Delayed memory: 0.995 (0.987–1.002)/1.032 (0.993–1.074)/1.065 (0.932–1.217) Unstandardised coefficient (95% CI) (AD) for 1. Digit Symbol Substitution Test: 1.466 (0.253–2.678)# 2. Boston Naming Test: 0.529 (0.215–0.866)# 3. animal naming: 0.541 (0.215–0.866)# 4. vegetable naming: 0.309 (0.056–0.561)#
Grinde and Endahgl [12] (2017)	Retrospective cohort study	2004–2015	1. ≥2 prescriptions of MTKs or inhalation-type asthma medication 2. aged ≥60 years in 2014	–	MTK (23636) vs non-users (179837)	1. Use of dementia medicine¶,+ 2. Admission to nursing home 3. Death 4. Parkinson's medicine+ 5. Diabetes medicine+	Adjusted HR (95% CI) for 1. dementia medicine: 0.89 (0.81–0.98) 2. Parkinson's medicine: 1.06 (0.98–1.15) 3. diabetes medicine: 0.85 (0.80–0.90)

Open in a separate window

AD: Alzheimer’s disease; HR: hazard ratio; ICD-10: International Classification of Diseases, tenth revision; LTRA: leukotriene receptor antagonist; MCI: mild cognitive impairment; MTK: montelukast; NIA-AA: National Institute on Aging–Alzheimer's Association; NINCDS-ADRDA: National Institute of Neurological and Communicative Diseases and Stroke/Alzheimer's Disease and Related Disorders Association; vs: versus. ^#: significant after adjusting for a false discovery rate of 10%; ^¶: includes memantine, donepezil, rivastigmine or galantamine; ⁺: use of drugs defined under the Anatomical Therapeutic Chemical classification system.

TABLE 2

Study characteristics of research studies on depression-related outcomes

Study (publication year)	Study design	Study/inclusion period	Inclusion criteria	Exclusion criteria	Subjects (n)	End-points	Key findings
Paljarvi et al. [7] (2022)	Retrospective cohort study	2015–2019	For asthma group: 1. Asthma-related healthcare contact 2. Aged 15–64 years at index prescription of MTK 3. Alive for 12 months after index prescription	1. Dx of COPD 2. Dx of obstructive sleep apnoea 3. Dx of neoplasm 4. Pregnancy during study period	MTK users (36245) vs non-users (36245)	1. 12-month incident Dx for NEs, including bipolar disorders, depression, anxiety disorders, OCD and sleeping disorders 2. 12-month incident dispensed prescriptions for psychiatric medications	OR (95% CI) (propensity score-matched) for: 1. single episode of major depression in 1 year: 1.05 (0.96–1.15) 2. receipt of antidepressant: 1.16 (1.07–1.26)
Sansing-Foster et al. [6] (2020)	Retrospective cohort study	2000–2015	1. Aged >6 years with both medical and drug coverage for at least 183 days before medication initiation 2. Asthma Dx in any care setting 3. No Dx of COPD during 183 days prior to index date	1. Experienced outcome on cohort entry day 2. >45-day gap between consecutive enrolment periods for medical and drug coverage 3. Recipients of LTRA, LAMA and ICS 183 days prior to index date 4. Same-day dispensing for both MTK and ICS on index date	MTK (457377) vs ICS (457377)	1. Hospitalisation for depressive disorders defined as inpatient claims 2. Treated outpatient depressive disorders 3. Hospitalisation for self-harm 4. Modified self-harm (by E-codes)	HR (95% CI) for 1. hospitalisation for depressive disorders: 1.06 (0.90–1.24) 2. hospitalisation for depressive disorders in patients without a psychiatric history: 0.63 (0.37–1.07) 3. treated outpatient depressive disorders: 0.91 (0.89–0.93)
Ali et al. [20] (2015)	Case–control	1998–2009	1. Aged 1–17 years 2. Primary Dx of asthma 3. Health plan enrolment in the 12 months before and after asthma claim	1. Developmental disorders 2. Receiving long-term care 3. Pre-existing NE in 365 days before asthma claims	Case: 1920 Control: 5760	As post hoc analyses: 1. Anxiety disorders 2. Depressive disorders 3. Aggressive disorders	Adjusted OR (95% CI) for patients exposed to MTK in the past 365 days for: 1. depressive disorders: 1.01 (0.66–1.54)
Zhou et al. [21] (2013)	ITS	2003–2007	1. Aged ≤45 years 2. ≥2 asthma controller prescriptions in the study period 3. ≥1 medical and pharmacy claim 12 months before and after index date (1st prescription)	1. Aged ≥46 years (higher likelihood of COPD) 2. Other asthma controllers during 12 months before and up to 3 months after index date	MTK (232159) vs fluticasone (264704) vs LABA/ICS (89635)	1. Antidepressant dispensing rate (only medications as monotherapy are included)	Immediate change (95% CI) for antidepressant dispensing rates after index date: 1. Patient aged 1–11 years: MTK (0.14% (0.06–0.22%))# vs fluticasone (0.17% (−0.01–0.35%)) vs LABA/ICS (0.16% (−0.06–0.37%)) 2. Patient aged 12–17 years: MTK (0.80% (0.31–1.29%))# vs fluticasone (0.83% (−0.08–1.73%)) vs LABA/ICS (0.76% (0.22–1.30%))# 3. Patient aged 18–24 years: MTK (1.93% (1.55–2.32%))# vs fluticasone (1.72% (1.30–2.15%))# vs LABA/ICS (2.76% (2.35–3.17%))# 4. Patient aged 25–45 years: MTK (3.62% (3.15–4.09%))# vs fluticasone (2.38% (1.54–3.21%))# vs LABA/ICS (4.03% (3.34–4.73%))#
Holbrook and Harik-Khan [22] (2008)	Reviews of RCTs (3 trials¶)	+	Randomised, double-masked, controlled trials conducted by the ALA-ACRC including MTK as treatment arm	-	MTK (569) vs placebo/active control (900)	1. Emotional well-being§	1. No statistically significant changes in quality-of-life scores from all three trials for MTK arm

Open in a separate window

ALA-ACRC: American Lung Association Asthma Clinical Research Centers; Dx: diagnosis; HR: hazard ratio; ICS: inhaled corticosteroid; ITS: interrupted time series; LABA: long-acting β-agonist; LAMA: long-acting muscarinic antagonist; LTRA: leukotriene receptor antagonist; MTK: montelukast; NE: neuropsychiatric event; OCD: obsessive compulsive disorder; OR: odds ratio; RCT: randomised controlled trial; vs: versus. ^#: p<0.05; ^¶: controls included placebo and active control, plus randomised, double-blind, placebo-controlled parallel-group trials; ⁺: trials included were conducted by the ALA-ACRC; ^§: emotional well-being evaluated by the Juniper Mini Asthma Quality of Life emotional dimension scores.

TABLE 3

Study characteristics of research studies on other neuropsychiatric outcomes

Study (publication year)	Study design	Study period	Inclusion criteria	Exclusion criteria	Subjects (n)	End-points	Key findings
Park et al. [23] (2022)	Self-controlled case series	2005–2007 and 2016–2018	1. Aged 3–30 years at the beginning of each study period 2. Dx of NE 3. Record of LTRA use (MTK or pranlukast) 4. Dx of asthma or allergic rhinitis	1. Patients who died during the study period 2. Patients who received LTRA or had a Dx of NE within 1 year before the observation period	17001#	1. Newly diagnosed NE during Obs period, categorised into psychotic, mood, anxiety, sleep, cognitive, movement and personality disorders	Incidence rate ratio (95% CI) (total/Obs period 1/Obs period 2) of NEs for: 1. all patients: 1.05/0.88/1.11¶ 2. patients started LTRA for 1–3 days: 0.68+/0.65¶/0.69+ 3. patients started LTRA for 4–7 days: 2.10+/1.35/2.36+ 4. patients started LTRA for 8–14 days: 1.60+/1.11/1.78+ Incidence rate ratio (95% CI) in patients with: 1. asthma: 1.00 (0.90–1.10) 2. allergic rhinitis only: 1.19 (1.01–1.39)¶
Paljarvi et al. [7] (2022)	Retrospective cohort study	2015–2019	For asthma group: 1. Asthma-related healthcare contact 2. Aged 15–64 years at index prescription of MTK 3. Alive for 12 months after index prescription	1. Dx of COPD 2. Dx of obstructive sleep apnoea 3. Dx of neoplasm 4. Pregnancy during study period	MTK users (36245) vs non-users (36245)	1. 12-month incident Dx for NEs, including bipolar disorders, depression, anxiety disorders, OCD and sleeping disorders 2. 12-month incident dispensed prescriptions for psychiatric medications	OR (95% CI) propensity score-matched) for: 1. any incident NE in 1 year: 1.11 (1.04–1.19) 2. any sleep problem: 1.13 (1.02–1.25) 3. insomnia: 1.13 (1.01–1.27) 4. receipt of sleep medication: 1.11 (0.99–1.25) 5. hypersomnia: 1.06 (0.78–1.44) 6. circadian rhythm disorder: 1.00 (0.67–1.48) 7. parasomnia: 1.04 (0.70–1.53) 8. anxiety or related disorder: 1.21 (1.05–1.20) 9. phobic anxiety: 1.13 (0.86–1.48) 10. generalised anxiety: 1.18 (1.05–1.33) 11. other anxiety: 1.11 (1.02–1.21)
Özata et al. [24] (2021)	Prospective cohort study	Nov 2017–Jun 2018	1. Age 1.5–5 years 2. Asthma diagnosed by an independent specialist physician based on the Global Initiative for Asthma guideline	1. Chronic diseases other than asthma 2. Psychiatric disease and a history of regular use of medication 3. Illiterate parents or non-Turkish speakers 4. Failure to attend follow-ups or not fully completing the CBCLs 5. ADHD and other psychiatric disorders	MTK users (50) vs ICS users (45)	1. CBCL scores for aged 1.5–5 years	Scores: ICS vs MTK in: 1. total CBCL score: 40 (IQR 27–52.5) vs 42.5 (IQR 29–61.5), p=0.34 2. internalisation score: 13 (IQR 8–18) vs 12.5 (IQR 9.75–21.5), p=0.41 3. externalisation score: 13 (IQR 7–17.5) vs 12.5 (IQR 7–20.25), p=0.40
Kang et al. [25] (2021)	Case–control	2003–2013	1. Newly diagnosed asthma during study period 2. Prescribed asthma disease controller 3. Aged >60 years	1. Case group: patients diagnosed with NEs before asthma diagnosis	Case: 31922 Control: 31922	1. NEs including mood disorder, sleep disorder, anxiety disorder, personality disorder, substance-related disorder, agitation, schizophrenia and self-harm	Adjusted OR (95% CI) for: 1. overall NEs: 1.67 (1.58–1.78) 2. sleep disorder: 1.54 (1.42–1.68) 3. mood disorder: 1.65 (1.51–1.81) 4. anxiety disorder: 1.63 (1.50–1.77)
Shim et al. [26] (2021)	Retrospective cohort study	2002–2015	1. Received health screening examinations between 2009 and 2010 2. Asthma diagnosis before the date of health screening 3. Aged 40–79 years	1. Use of LTRA prior to the date of health screening 2. NEs before the date of health screening	LTRA users (12168) vs non-users (49403)	1. NEs identified using ICD-10 codes§	Adjusted HR (95% CI) for: 1. LTRA use: 1.01 (0.83–1.23) Adjusted HR (95% CI) for LTRA use of: 2. <6 months: 1.01 (0.83–1.24) 3. 6–11 months: 0.81 (0.36–1.84) 4. 12–23 months: 1.37 (0.66–2.86) 5. ≥24 months: 0.71 (0.26–1.98)
Bayer et al. [27] (2020)	Prospective cohort study	Sep 2013–Mar 2014	1. Aged 3–18 years 2. Taking MTK for the first time due to asthma during study period	1. Experienced asthma attack when started MTK 2. Uncontrolled asthma symptoms 3. Existing chronic disease (diabetes, CKD) 4. Existing neuropsychiatric disease (epilepsy/ADHD)	125	1. Neuropsychiatric ADRs occurred 2. Effects of neuropsychiatric ADRs on patients’ QoL	1. NE in patients (62.4%) 2. Statistically significant elevations (p<0.001) of NEs post-MTK treatment in temperamental behaviour, nightmare and sleep disorders Child-reported QoL OR in: 3. preschool age: 2.66, p=0.048 4. school age: 5.95, p=0.027
Huang et al. [9] (2020)	Retrospective cohort study	1 Jan 1997–31 Dec 2013	1. Aged ≤12 years 2. ≥1 claim of inpatient admission or ≥3 claims of ambulatory visit 3. New-onset asthma diagnosed during study period	1. ADHD diagnosis before asthma diagnosis or MTK treatment 2. Patients with follow-up periods <6 months	MTK (12806) vs non-MTK users (12806)	1. Occurrence of ADHD using ICD-9 codes (314.X)	1. Adjusted HR (95% CI) for MTK on association with ADHD: 1.04 (0.93–1.17) 2. Prolonged use of MTK (>90 days) did not increase the risk of ADHD when compared with ≤90 days (1.08, 95% CI 0.95–1.23) and >90 days (1.00, 95% CI 0.86–1.15)
Glockler-Lauf et al. [10] (2019)	Case–control	2004–2015	1. Aged 5–18 years with physician-diagnosed asthma in study period 2. Asthma maintenance medication prescribed	1. No valid health card number/residence code 2. Existing surgery, psychiatric disorders or non-pharmacologically treated asthma (<1 record of asthma maintenance drug)	Case: 898 Control: 3497	1. First NEƒ following physician-diagnosed asthma	1. Adjusted OR (95% CI) for MTK use: 1.91 (1.15–3.18) 2. 42.4% of NE in 90 days 3. Most prevalent presenting complaint for first NE: anxiety (436 out of 898, 48.6%); sleep disturbance (234 out of 898, 26.1%)
Benard et al. [11] (2017)	Retrospective cohort	2011–2016	1. Aged 1–17 years 2. Physician-confirmed asthma 3. A clinic visit during MTK initiation as monotherapy or adjunct therapy to ICS/ICS+LABA	Without MTK prescription or maintenance asthma medications, or denied taking medications of interest	MTK (84) vs ICS monotherapy (84)	1. Incidence of parent-reported neuropsychiatric ADR leading to drug cessation 2. Characteristics of parent-reported NE ADRs	1. Relative risk (95% CI) for ADR reported by parents: 12.0 (1.6–90.2) 2. Incidence (95% CI) of MTK cessation: 12% (7–21%)
Ali et al. [20] (2015)	Case–control	1998–2009	1. Aged 1–17 years 2. Primary Dx of asthma 3. Health plan enrolment in the 12 months before and after asthma claim	1. Developmental disorders 2. Receiving long-term care 3. Pre-existing NE in 365 days before asthma claims	Case: 1920 Control: 5760	1. ND (composite end-point) 2. Psychiatric disorder diagnosis 3. NE diagnosis 4. Psychotropic medication receipt As post hoc analyses: 1. Anxiety disorders 2. Depressive disorders 3. Aggressive disorders	Adjusted OR (95% CI) for 1. ND with MTK exposure in the year prior to index date: 1.01 (0.88–1.14) 2. ND with high chronic cumulative dose of MTK: 0.64 (0.50–0.82) Psychiatric disorder diagnosis with high chronic cumulative dose of MTK: 0.64 (0.49–0.83) 3. Anxiety disorders for patients exposed to MTK in the year prior to index date: 0.95 (0.66–1.37) 4. Aggressive disorders for patients exposed to MTK in the year prior to index date: 0.92 (0.58–1.45)
Lu et al. [14] (2015)	ITS	2005–2010	1. Aged 5–64 years 2. Continuous health enrolment plan for past 12 months 3. Enrolled for health plan for current month 4. At least one outpatient or inpatient visit in the past year due to asthma diagnosis	History of: 1. COPD, cystic fibrosis, bronchiectasis 2. pulmonary hypertension or embolism 3. bronchopulmonary dysplasia 4. CHF	140000 (rolling cohort with asthma)	Monthly percentage of patients 1. dispensed an LTRA and non-LTRA 2. on mental health visits 3. medically treated for suicide attempts (quarterly percentage)	1 year after the FDA label change in 2009, mental health visits 1. increased by 0.25 (95% CI 0.01–0.49) percentage points in adolescents 2. increased by 1.00 (95% CI 1.00–1.00) percentage points among young adults (aged 18–29 years) 3. increased by 0.61 (95% CI 0.31–0.91) percentage points in adults (aged 30–64 years)
Bisgaard et al. [28] (2009)	Reviews of RCTs (9 trials##)	1995–2004	1. Aged 6 months to 14 years 2. Asthmatic patients without significant comorbidities	–	MTK (1999) vs placebo (873) vs active control (379)¶¶	1. All ADRs	1. One case of NE in open-label extension studies, in which the patient had background attention deficit disorder and depression, later withdrawn due to asthma exacerbation 2. One case of epilepsy in long-term double-blind studies, leading to discontinuation
Philip et al. [29] (2009)	Reviews of RCTs (46 trials++)	Up to 25 April 2008§§	1. Randomised, double-blind, placebo-controlled trials 2. Parallel-group or crossover trials 3. Multiple dose administration 4. Inclusive of ≥20 patients per treatment group of all ages 5. Patients aged ≥3 months 6. Completed by 25 April 2008	Patients with clinically significant psychiatric illnesses at baseline (in trials)	MTK (11673) vs Placebo (8827) vs Active control (4724)	1. BRAEs	OR (95% CI) for 1. patients with BRAEs: 1.12 (0.93–1.36) 2. BRAE leading to discontinuation: 0.52 (0.17–1.51)

Open in a separate window

ADHD: attention deficit hyperactivity disorder; ADR: adverse drug reaction; BRAE: behaviour-related adverse experience; CBCL: Child Behavior Checklist; CHF: congestive heart failure; CKD: chronic kidney disease; Dx: diagnosis; FDA: US Food and Drug Administration; HR: hazard ratio; ICD-9/10: International Classification of Diseases, ninth/tenth revision; ICS: inhaled corticosteroids; IQR: interquartile range; ITS: interrupted time series; LABA: long-acting β-agonist; LTRA: leukotriene receptor antagonist; Obs: observation period; OCD: obsessive compulsive disorder; MTK: montelukast; ND: neuropsychiatric disturbance; NE: neuropsychiatric event; OR: odds ratio; QoL: quality of life; RCT: randomised controlled trial; vs: versus. ^#: including 4300 patients enrolled in observation period 1 (1 January 2005–31 December 2007) and 12701 patients enrolled in observation period 2 (1 January 2016–31 December 2018); study also included 6484 (38.14%) patients with only diagnoses for allergic rhinitis; ^¶: p<0.05; ⁺: p<0.005; ^§: NEs include F00–F09 organic, including symptomatic mental disorders; F10–F19 mental and behavioural disorders due to psychoactive substance use; F20–F29 mood disorders; F40–F49 neurotic, stress-related and somatoform disorders; F50–F59 behavioural syndromes associated with physiological disturbances and physical factors; F99 unspecified mental disorder; X60–X84 suicide attempt; and R45.8 suicide ideation; ^ƒ: NEs include substance-related disorders, schizophrenia, anxiety, sleep disturbance, mood and personality disorders, and agitation; NEs with outpatient visits were excluded; ^##: controls included placebo and active control, plus randomised, double-blind, placebo-controlled parallel-group trials and open-label studies; ^¶¶: only including participants in trials enrolling asthma patients; ⁺⁺: controls included placebo and active control, plus randomised, double-blind, placebo-controlled parallel-group trials and open-label studies and randomised, double-blind, placebo-controlled crossover trials; ^§§: including 30 trials on asthma, 12 on seasonal allergic rhinitis, two on perennial allergic rhinitis, one on respiratory syncytial virus-induced bronchiolitis and one on migraine headaches.

Results

Discussion

This systematic review did not reveal significant associations between montelukast use and suicide-related events. While the evidence on depression remains conflicting, older adults may be particularly susceptible to specific neuropsychiatric events such as anxiety and sleeping disorders. Future studies are required to confirm the associations between montelukast and neurodegenerative diseases.

Montelukast was not found to be harmful in suicide-related events in reviews of RCTs and observational studies using ICD codes to define study outcomes [6, 15, 16]. In clinical trials, suicidal events might be too rare to be detected given a rate of 1.02 cases per 100000 person-years [18], resulting in nonsignificant associations observed in subsequent reviews [19, 28]. Besides, the sensitivity of ICD codes in capturing suicide could be somewhat compromised in observational studies owing to a conservative approach adopted by clinicians when classifying deaths as suicides [53]. The findings may also be susceptible to residual confounding because most studies investigated the associations of montelukast and suicidal events with administrative data, which may not capture nonclinical characteristics, such as socioeconomic factors, in the study cohorts. Despite these methodological limitations, the consistency of findings in research studies suggests that montelukast poses no additional risk for suicide-related events at a population-wide level.

However, the possibility of montelukast inducing suicide-related events in certain individuals should not be excluded, because current methodologies struggle to address differences in baseline characteristics at an individual level. Asthma diagnosis itself might be a contributor to suicidal events, as revealed by a Taiwanese cohort study showing self-harm to be associated with asthma (adjusted HR 1.70, 95% CI 1.35–2.14) [15]. Other clinical diagnoses may also contribute to the risk of suicide, the magnitude of which remains largely unknown. Statistical adjustments for these diagnoses in observational studies would be useful in accounting for the differences in baseline characteristics. The findings, however, imply associations of montelukast with suicide-related events in the whole study population, rather than the safety of the drug in every single individual involved. Montelukast may still be harmful in individuals with specific combinations of baseline comorbidities associated with suicide whom the current study designs failed to capture. Developing risk-scoring tools for suicide risk and matching patients based on the risk scores may be useful in reflecting individual differences on the baseline risk of suicide as we aim to further unmask the relationship between montelukast and the risk of suicidal events on an individual level.

Despite the identification of adverse drug reactions on depression by pharmacovigilance databases, these associations were not found in RCTs and observational studies. Asthma may be associated with depression [54, 55], and results from observational studies did not show an increased risk of depression after adjusting for asthma severity [6, 16]. Besides, comparable increases in antidepressant prescription rates observed in comparison groups [21, 32] may suggest that the reports of depression identified from pharmacovigilance databases could be attributed to both poor asthma control and the use of LTRAs, rather than either factor alone. While montelukast was associated with incident antidepressant prescriptions [21, 32], the outcome definition may be prone to classification bias because antidepressants may be indicated for other disorders such as neuropathic pain [7]. In addition, the follow-up periods in previous RCTs of up to 24 weeks [22, 29] may be too short to detect the associations between montelukast and depression on long-term use. Future research studies should look to address these methodological limitations in terms of adjustments for baseline asthma severity, outcome definitions and duration of follow-up.

Interestingly, the risk for depression was reduced in montelukast users in observational studies [6, 7], which could be attributed to the immunomodulatory effects of montelukast. M1 macrophage-mediated inflammation was associated with depression, and many antidepressants were shown to possess immunomodulatory effects by reducing plasma interleukin-1β and interleukin-6 levels [56]. By lowering serum proinflammatory tumour necrosis factor-α and interleukin-6, montelukast may offer benefits in depression [57]. LTRAs have also shown antidepressant activities by boosting the proliferation of neural progenitors and neurogenesis in mice [58]. However, corticosteroids, a drug class known to be associated with depression, were often used as a comparison arm in observational studies [6]. Therefore, the findings cannot exclude the possibility that montelukast induces depression, but to a lesser extent than ICS. Besides, the sole inclusion of psychiatric outcomes resulting in healthcare claims may fail to capture mild cases of neuropsychiatric events related to montelukast or ICS use [6]. The protective effects reported in a recent cohort study, but only before propensity score matching [7], may suggest undiscovered confounding.

Emerging evidence from pharmacovigilance databases and observational studies indicated adverse relationships between montelukast and anxiety in older adults, but not in children. First, it is worth noting that cytochrome P450 enzymes [35] and the additive neurotoxicity of eicosapentaenoic acid, a precursor for the biosynthesis of leukotrienes [39], may be alternative culprits, as pointed out by another review of case reports [59]. Whether genetic polymorphisms in cytochrome P450 enzymes may affect the associations remains unknown. While asthma control is associated with anxiety [55], the possible involvement of leukotriene pathways in inducing anxiety-like behaviour was also highlighted [60]. Knockout of the 5-lipooxygenase activating protein, a key protein involved in eicosanoid metabolism, resulted in lower expression of transcription factor cFOS and induced anxiety in mice [60]. More importantly, the decline of cFOS was associated with age-dependent anxiety phenotypes [60], which may explain the adverse associations identified particularly in elderly patients. Using real-world data, two large population-based studies enrolling over 50000 adults [7, 25] revealed associations of montelukast with the risk of anxiety, and the findings should be robust.

Montelukast was consistently associated with sleeping disorders in adults and the elderly in studies with different designs. However, these associations were called into question when patients without asthma were enrolled [23], and when subtypes of sleeping disorders were defined as study outcomes. The fact that montelukast was used off-label for patients with obstructive sleep apnoea [61] further complicated the interpretation of findings when various types of sleeping disorders were analysed as a composite outcome. Therefore, stratification of patient populations and outcomes of sleeping disorders are crucial in future studies.

Based on data from Europe [12] and Asia [8, 30], montelukast appears to be associated with a lower risk of developing neurodegenerative diseases in the elderly. Statistically significant associations were found using dementia medications [12] and new-onset dementia [8] as outcome definitions. However, some studies [12, 30] were not designed to evaluate the safety of montelukast in developing neurodegenerative diseases, but rather to explore the therapeutic potential of montelukast in these outcomes. Therefore, the study populations may not be representative of patients with asthma in general. The Japanese cohort study on dementia only included new users of LTRAs to avoid results being distorted by the previous use of medications [8], yet the effect sizes were small using cognitive test scores as the outcome [30]. Montelukast was found to modulate neuroinflammation in animal models when used at high doses [62]. A recent case series also demonstrated the potential of high-dose montelukast in improving memory [63]. Future work shall continue to focus on establishing clinical evidence and the optimal dose-response relationship of montelukast in the treatment of neurodegenerative diseases. The safety of montelukast for neurodegenerative diseases in general asthma patients remains to be established.

Evidence from previous literature on the relationship between montelukast use and neuropsychiatric events in children was inconclusive. To date, case reports on neuropsychiatric events related to montelukast use in paediatric subjects are more common [10, 11, 16, 20, 28, 31, 41, 43–45, 64]. While this could be attributed to the fact that montelukast is more commonly prescribed for children [13], physiological alterations in brain development and the hypothalamic–pituitary–adrenal axis during adolescence may cause dysregulations in neurotransmitters and stress hormones, which could also act as possible contributors to the neuropsychiatric events observed [65].

The three observational studies enrolling paediatric patients [10, 11, 27] revealed statistically significant increases in the risks of developing various neuropsychiatric events. However, small cohort sizes [11], narrow scopes of definitions of neuropsychiatric events [10] and over-representation of paediatric subjects from low-income families [10] limited the generalisability of the results. While molecular models suggest that neurogenesis declines in adulthood and offers protective effects against molecules entering the brain via the blood–brain barrier [66], emerging evidence has suggested otherwise because montelukast was found to be associated with various neuropsychiatric events in the elderly, rather than in children. The heterogeneity in study outcomes across paediatric observational studies also implies further work is required to confirm the neuropsychiatric safety of montelukast in children.

Conclusion

Montelukast has not been shown to increase the risk of suicide attempts in patients with asthma in observational studies. The evidence on depression was inconsistent, and possible protective effects are yet to be unmasked. However, current evidence revealed possible associations between montelukast and anxiety, as well as specific subtypes of sleeping disorders. The signals from observational studies were alarming for older patients, although the associations could not be confirmed in children because the study outcomes in this patient group were heterogeneous. The stratification of specific subtypes of psychiatric disorders and age is warranted in future studies to reveal how age and the types of psychiatric diagnoses may affect the relationship.

Supplementary material

Acknowledgements

References