Introduction

Parkinson’s disease (PD) is characterised by the pathological accumulation of alpha-synuclein positive Lewy bodies and neurites in the brain, with associated changes in the immune system both within the central nervous system (CNS) and the periphery. Beta-adrenoceptor modulation has been found to influence alpha-synuclein gene expression, with blockade leading to increased expression, and stimulation leading to decreased expression1. Beta-adrenoceptors are also present on immune cells, with stimulation related to immune-regulatory effects, and blockade related to immune activation2.

Epidemiological studies, including meta-analyses, have linked the use of beta-blockers and beta-agonists with an increased and decreased risk of PD, respectively1,3,4,5,6,7,8. However, some studies have found inconsistent results, and confounding factors and reverse causation may also contribute to these associations9,10,11,12,13.

Amongst those who have PD, relationships between beta-adrenoceptor modulating drugs and progression to major PD milestones such as postural instability and dementia, features which are relatively resistant to dopaminergic therapy and have important implications on quality of life and care14, have not been fully explored. Only one study has investigated these medications in relation to progression in PD and suggested that short-acting beta-agonist use was not associated with progression to dementia, but was associated with reduced time to death in PD, which was considered to reflect the effect of chronic obstructive pulmonary disease (COPD)/asthma severity in users of these medications6. Links between the use of these medications at the time of PD diagnosis and baseline clinical heterogeneity or subsequent motor progression to postural instability have not been investigated. This study sought to investigate these relationships using baseline and longitudinal data from the Parkinson’s Incidence Cohorts Collaboration (PICC)15, since beta-adrenoceptor modulation could potentially be a modifiable risk factor for PD progression. We hypothesised that beta-blockers would be associated with faster disease progression, while beta-agonists would display the opposite relationship.

Results

The overall PICC cohort consisted of 1107 patients with newly diagnosed idiopathic PD16,17,18,19,20,21. Demographic and baseline clinical details are shown in Table 1. The combined cohort had an average time from diagnosis to last visit of 5.5 (SD 3.1) years (Table 1).

Table 1 Demographic information for each cohort at baseline
Full size table

Beta-adrenoceptor drugs and baseline clinical heterogeneity

At the baseline visit, 195 (17.6%) of patients were on beta-blockers, 68 (6.1%) were taking beta-agonists, 4 (0.40%) were on both beta-agonists and beta-blockers, and 848 (76.6%) were not prescribed either of these drug classes (non-users). A detailed breakdown of specific drugs is provided in Supplementary Table S1.

Demographic and baseline clinical characteristics of groups stratified according to beta adrenoceptor drug use are summarised in Table 2. Beta-blocker users were significantly older, had a shorter disease duration at baseline, and higher baseline MDS-UPDRS motor and H&Y scores than non-users (Table 2). Frequency of cardiovascular disease and Cumulative Illness Rating Scale (CIRS) category scores were also higher in beta-blocker users than non-users, as anticipated given their typical clinical indications for use (Table 2). Beta-agonist users were significantly older and had fewer years of education compared to non-users (Table 2). The presence of autoimmune/inflammatory disease, percentage of ex/current smokers and CIRS category score were significantly higher in beta-agonist users compared to non-users, as anticipated given the typical clinical indications for these medications. (Table 2).

Table 2 Baseline clinical and demographic characteristics of PICC participants stratified by use of beta-agonists and beta-blockers
Full size table

Multivariate regression analysis was performed to assess the contribution of potential confounding factors to the observed group differences in baseline motor disease severity (Table 3). This analysis did not include the NYPUM and ParkWest cohorts due to the absence of CIRS data (Table 3). The results revealed that the associations between beta-blocker use and higher baseline MDS-UPDRS motor scores and H&Y stage were not statistically significant when accounting for relevant covariates (age at diagnosis, disease duration, comorbidities and study cohort) (Table 3).

Table 3 Multivariate regression models for baseline motor scores
Full size table

Beta-adrenoceptor drugs and motor and cognitive progression

There were no significant differences in follow-up duration for groups stratified by beta-blocker use/non-use and beta-agonist use/non-use (Table 2). Patients who were already at H&Y3 at baseline (n = 159) were excluded from the H&Y3 survival analyses. Following these exclusions, the higher baseline MDS-UPDRS motor score and H&Y stage in the beta-blocker use group and the higher age at diagnosis in the beta-agonist use group did not remain statistically significant (Supplementary Table S2).

The median time to reach H&Y3 was 3.52 (Interquartile Range (IQR) 1.77–5.49) years in beta-blocker users and 4.10 (IQR 2.22–6.04) years in beta-blocker non-users. Beta-blocker users progressed to H&Y3 significantly earlier than non-users (Kaplan-Meier analysis; Log Rank (Mantel–Cox), (χ2) = 9.286, df = 1; p = 0.002) (Fig. 1A). Cox regression analysis confirmed this significant relationship between beta-blocker use and increased progression to H&Y3 (Hazard Ratio (HR) = 1.538, p = 0.011), with the inclusion of relevant variables as potential confounders (age at diagnosis, disease duration, comorbidities (CIRS score, cardiovascular disease, diabetes, smoking status) and study cohort) (Table 4).

Fig. 1: Survival analyses of progression to H&Y3.
figure 1

Kaplan-Meier survival analyses of progression to H&Y3, following exclusion of those at H&Y ≥ 3 at baseline (n = 948). A Beta-blocker users and non-users. B Beta-agonist users and non-users (H&Y Hoehn and Yahr stage).

Full size image
Table 4 Cox regression models of progression to H&Y3
Full size table

For beta-agonist users, median time to reach H&Y3 was 4.57 (IQR 1.69–5.57) years, compared to 4.02 (IQR 2.11–5.89) years for non-users. There was no significant difference in rate of progression to H&Y3 between beta-agonist users and non-users (Log Rank (Mantel–Cox), (χ2) = 0.544, df = 1; p = 0.461) (Fig. 1B). This was confirmed by Cox regression analysis (HR = 0.772, p = 0.420) with relevant confounders (Table 4).

During the course of the study, 29.6% of patients developed dementia, with a mean duration from PD diagnosis to dementia onset of 4.50 (IQR 2.56–6.17) years amongst this group. Time to dementia did not differ in beta-blocker users versus non-users on Kaplan–Meier analysis (Log Rank (Mantel–Cox), (χ2) = 3.242, df = 1; p = 0.072) (Fig. 2A) or Cox regression (HR = 1.017, p = 0.926) with relevant baseline confounders (Table 5). Similarly, Kaplan–Meier analysis indicated that time to dementia did not differ in beta-agonist users compared to non-users (Log Rank (Mantel–Cox), (χ2) = 0.084, df = 1; p = 0.772) (Fig. 2B). This was confirmed on Cox regression analysis (HR = 1.091, p = 0.793) with relevant confounders (Table 5).

Fig. 2: Survival analyses of progression to dementia.
figure 2

Kaplan–Meier survival analyses of progression to dementia (n = 1107). A Beta-blocker users and non-users. B Beta-agonist users and non-users.

Full size image
Table 5 Cox regression models of progression to dementia
Full size table

Study cohort was included as a categorical covariate, with the largest cohort (PICNICS) as the reference category (Tables 3–5), in all multivariate regression analyses. (Use of each of the other cohorts as the reference category had no significant effect on the relationships and significance of the other covariates in the model (data not shown)). The inclusion of the study cohort as a covariate in regression analyses revealed heterogeneity in the effect size and significance of each cohort on the dependent variables (baseline motor severity and progression to H&Y3 and dementia). However, such adjustment for study cohort did not alter our primary finding of a significant relationship between beta blocker use and faster progression to H&Y3.

Discussion

This study demonstrates an association between the use of beta-blocker medications in early PD and subsequent motor progression in a large population-based incident PD cohort. Regular beta-blocker use was associated with more rapid progression to H&Y stage 3 over a mean of 5.5 years of follow-up. However, we observed no significant association between beta-agonist use and progression to H&Y3, and no significant relationship between beta-adrenoceptor drug use and progression to dementia.

Most patients on beta-blockers were prescribed these drugs for cardiovascular disease. This could potentially confound the relationship between beta blocker use and motor impairment, given that postural instability, gait dysfunction and impaired mobility in PD are also linked with increased cerebral white matter disease due to vascular risk factors and poor cardiovascular health22. Cardiovascular disease is also associated with older age, and this can itself be a risk factor for motor progression to H&Y316 and may also be linked with a more severe baseline motor subtype subsequently predictive of prognosis23. In keeping with this, we found that although baseline motor impairment was higher in beta-blocker users compared to non-users, this was confounded by age and comorbidity. However, in terms of axial motor progression, the association between time to H&Y3 and beta blocker use remained significant even accounting for age at diagnosis, baseline H&Y stage, cardiovascular disease, smoking, diabetes and comorbidity score in Cox regression analysis. This suggests that the association between beta-blockers and faster progression to H&Y3 in this cohort is not simply explained by the confounding effects of baseline H&Y stage, age and key comorbidity factors. However, data available in this study may not cover all possible aspects of cardiovascular co-morbidity (e.g. undiagnosed disease) and we cannot exclude the possibility that some residual confounding due to cardiovascular co-morbidity contributes towards the observed association.

Both beta-blockers and beta-agonists are known to cross the blood-brain-barrier (BBB) to different extents and may be mediating effects via central and/or peripheral mechanisms24,25. Brain concentrations of beta-blockers following oral intake have been found to be higher for the relatively lipophilic beta-blockers such as propranolol and metoprolol compared to atenolol24. However, separate analysis of those taking medium-high versus low BBB penetrating beta-blockers was not feasible due to inadequate numbers in each group and lack of data for fair dose equivalent comparisons.

Beta-agonist use at baseline was not associated with altered progression to H&Y3 in this cohort. However, the study may have been underpowered due to the relatively low number of individuals on beta agonist medications. Beta-agonists have been detected in the brain, with nasal administration leading to higher brain concentrations than intravenous administration26. It is possible to speculate that the lack of effect on progression to H&Y3 seen with beta-agonists in this study could be partially due to inadequate systemic and CNS concentrations achieved using inhaled beta-agonists. Thus, further studies investigating different routes of administration and brain pharmacokinetics will be important.

Whilst previous studies have not examined the relationship between beta-adrenoceptor drug use at baseline and subsequent motor progression, the prospective use of beta-blockers and beta-agonists has been associated with an increased and decreased PD risk, respectively1,3,4,5. However, further studies have suggested there may be confounding factors such as reverse causation, and association of the drugs with underlying diseases and risk factors (e.g. chronic obstructive lung disease and smoking) contributing towards these findings9,10,11,12. One study by Alexander et al. suggested a beneficial influence of the beta-agonist albuterol on some PD motor symptoms27, but no prior studies have assessed these aspects longitudinally. The evaluation of associations between beta-adrenoceptor use and subsequent clinical progression within PD (as in this study) is less likely to be influenced by reverse causation compared to studies assessing beta-adrenoceptor use and PD risk, thus also strengthening the current findings linking beta-blocker use with increased axial motor progression in PD.

The potential mechanisms underlying the association between beta-blockers and increased motor progression in PD are complex and multifactorial. Increased axial motor impairment and progression to H&Y3 may be mediated by pathology outside the nigrostriatal tract, including cortical Lewy body pathology 28,29. Beta-adrenoceptor stimulation has been found to decrease expression of alpha-synuclein, and hence beta-adrenoreceptor blockade may be a risk factor for increased cortical alpha-synuclein accumulation and pathological aggregation1. Another potential mechanism of relevance is immune modulation, given that beta-adrenergic modulation has considerable effects on the immune system2. Both innate and adaptive immune cells have beta-adrenergic receptors and are likely to be modulated by beta-agonists and beta-blockers2,30. Generally, beta-adrenergic stimulation is associated with modulating overtly excessive immune responses and has demonstrated dopaminergic neuroprotective effects via inhibitory effects on microglia30. Consequently, beta-blockade would be expected to have opposing effects, leading to increased immune activation, which may be detrimental in the context of PD2. In addition, beta-agonists have also been shown to enhance brain extraction of levodopa, although this may be less relevant to relatively dopamine-resistant axial motor dysfunction25. However, these proposed underlying mechanisms, including alterations in alpha-synuclein expression and/or immune activation, remain speculative and require further investigation.

In keeping with a previous study6, beta-adrenoreceptor modulating drugs were not significantly associated with progression to PD dementia. One possible explanation is that the pathology of PD dementia is more multifactorial, and related to Alzheimer’s type pathology as well as alpha synuclein aggregates31, whereas the pathology of the axial motor features of PD may be more dependent on alpha-synuclein pathology.

A strength of our study is that it has used a large, comprehensive dataset comprising population-based, incident PD cohorts with an average 5.50-year prospective follow-up, with the ability to adjust for some of the potentially confounding influence of key comorbid conditions. However, it is important to acknowledge the limitations. Patients had different time periods of follow-up, although there were no significant differences between groups in the time from diagnosis to last visit (Table 2). Details of the duration and exact daily dose of beta-agonists and beta-blockers taken prior to PD onset and data on commencement or discontinuation of these drugs during the follow-up period were not available for all cohorts. Also, as with all drug studies, the accuracy of drug intake as per prescription could not be guaranteed, and all this limits the ability to confirm conclusions on the true causal influence of these drugs. Additionally, the relative effects of high and low BBB-penetrating beta-blockers/beta-agonists were not assessed due to low numbers of each drug type within the cohort and the lack of feasibility of fair dose comparisons. As previously discussed, the potential residual confounding of any uncaptured cardiovascular and other comorbid factors is also a possible limitation, while factors such as regular exercise, which may potentially have beneficial effects on disease progression32, are also not included as covariates in the analysis. Furthermore, regular use of some beta-blockers has been linked to dizziness, light-headedness and increased risk of falls, thus possibly confounding the relationship between beta-blockers and increased progression to postural instability/H&Y333. Formal assessment of orthostatic hypotension for inclusion as a potential confounder would be important in future studies.

While the inclusion of several different study cohorts from different parts of the UK and Europe is a strength, the inclusion of study cohort in the regression analyses indicated that there may be some heterogeneity among cohorts in terms of both disease severity at baseline and progression to H&Y3 and dementia (e.g. greater and less progression to H&Y3 in CamPaIGN and ICICLE cohorts respectively). Such variability may be due to additional population-level genetic and environmental factors unaccounted for in the data, and further characterisation of such additional relevant features would be valuable in future studies. A formal sensitivity analysis using a sub-cohort of the data was not performed due to the limited size of the individual cohorts and the absence of similar general comorbidity data to be used as a covariate in all cohorts. In addition, while details of ethnicity were not included in this study analysis, the cohorts in this study were mainly recruited from Caucasian majority populations, and further studies including cohorts from other countries and ethnicities will be important.

Overall, this study suggests that beta-adrenoceptor modulating drugs may be associated with heterogeneity of axial motor progression in PD. This is a motor feature which is considered to have a non-dopaminergic basis and thus more accurately reflects the underlying disease state than motor symptoms, which are dopa-responsive and confounded by PD medication use34. Further detailed longitudinal epidemiological cohort studies in even larger cohorts, with long-term follow-up and detailed data on medication type, dose, duration, confounding symptoms and comorbidities, will enable investigation into potential dose-response relationships and better correction for confounders. Such detailed studies, together with more experimental studies, will be required to confirm the current findings and, if confirmed, understand the mechanisms and significance of the associations seen.

In general, this work is supportive of previous studies suggesting that beta-adrenoceptor-stimulating drugs warrant further investigation as candidates for drug repurposing trials in PD. Additionally, considering the beneficial use of beta-blockers for the symptomatic control of tremor in PD, as well as for comorbid conditions such as cardiovascular disease and migraine35, further thorough investigation and clarification of any potential negative association between beta-blockers and PD axial motor progression will be important to help prescribers navigate the relative benefits and risks of these drugs for people with PD.

Methods

Participants and data collection

Data was utilised from PICC, a collaboration of 6 population-based PD cohorts in Northern Europe15. Each study obtained demographic and clinical data as near as feasible to the disease diagnosis (baseline (first) visit; mean (standard deviation) disease duration 0.19 (0.32) years; range 0–2.35 years) and during prospective follow-up15. Details have been previously published (Cambridgeshire Incidence of PD from General Practitioner to Neurologist (CamPaIGN))16; Incidence of Cognitive Impairment in Cohorts with Longitudinal Evaluation-PD (ICICLE-PD)17,36; New Parkinson Patient in Umeå (NYPUM)18, The Norwegian ParkWest study (ParkWest)19, Parkinsonism: Incidence, Cognition and Non-motor heterogeneity in Cambridgeshire (PICNICS)20 and Parkinsonism Incidence in North-East Scotland (PINE)21.

In brief, patients were diagnosed with idiopathic PD using UK PD Brain Bank criteria without family history as an exclusion criterion. Patients were followed up at regular intervals and in this study, data from follow-up visits performed every 12, 18 or 24 months were used. All patients had a confirmed clinical diagnosis of idiopathic PD at their last clinic visit or autopsy.

Standard protocol approvals, registrations and patient consents

All studies had ethical approval from their regional Ethics Committees and written informed consent was obtained from all participants. CamPaIGN- Cambridgeshire 3 Research Ethics Committee (08/H0306/26); ICICLE Newcastle- Newcastle and North Tyneside 1 Research Ethics Committee (08/H0906/147); NYPUM- The Regional Ethics Committee of Umeå University, (Um dnr 03-387); Park West- Regional Committee for Medical and Health Research Ethics, Western Norway (REK 131/04); PICNICS- Essex Ethics Committee (REC 07/H0302/138); PINE- Multi-centre Research Ethics Committee for Scotland A (REC05/MRE00/94).

Demographic and clinical assessments

Data acquisition for each study cohort has been previously described in detail15 and included demographic and clinical data. Home visits and/or telephone follow-up to help ascertain key outcomes, as feasible, were offered to minimise attrition bias.

Motor symptoms were evaluated using the Hoehn and Yahr (H&Y) scale and the Unified Parkinson’s Disease Rating Scale (UPDRS) Part III (CamPaIGN, NYPUM, ParkWest, PINE) or the Movement Disorders Society-UPDRS (MDS-UPDRS) Part III (ICICLE-PD, PICNICS). UPDRS-III scores were converted into MDS-UPDRS-III as previously described37. Global cognitive function was assessed using the Mini-Mental State Examination (MMSE)38. Dementia was diagnosed according to Diagnostic and Statistical Manual of Mental Disorders, 4th Edition (PINE, CamPaIGN) or Movement Disorder Society PD dementia criteria39 (PICNICS, ICICLE-PD, NYPUM, ParkWest)15, using appropriate neuropsychological assessments to determine decreased global cognitive ability and impairment in more than one domain and clinical interviews to determine functional impairment and to exclude features making the diagnosis uncertain. Levodopa equivalent daily dose (LEDD) was calculated using published methods40.

Medication and comorbidity data categorisation

Medication history and comorbidity information were collected at the baseline visit based on available details from referral letters, prescriptions and patient interviews. In some cohorts, the patients’ general practitioners were contacted to obtain missing details (ICICLE-PD). Beta-adrenoceptor modulating drug categories were: beta-agonist users (oral or inhaled beta-agonist use) and non-users; beta-blocker users and non-users. Details of the exact daily dose of each drug and prior duration of use were not available for all studies.

Comorbidity burden was quantified by categorising the available medical history data based on the scoring guidelines from the CIRS and using the total number of comorbidity categories affected (CIRS category score)41. Due to the nature of the most common clinical indications for beta-agonists (immune mediated reversible airway obstruction including asthma and COPD) and beta-blockers (ischaemic heart disease, hypertension, cardiac arrhythmias), the presence or absence of the comorbid disease categories encompassing these conditions (autoimmune/inflammatory diseases and cardiovascular diseases respectively) were also separately recorded in cohorts where data was available.

Assessment of outcomes

Motor progression was assessed using time from diagnosis to reaching postural instability, as measured by a score of 3 or higher on the H&Y scale. The H&Y stage is particularly reflective of axial motor dysfunction (including postural instability and gait dysfunction) and reaching H&Y3 is a key motor milestone in PD progression14. The date of reaching H&Y3 was calculated as the midpoint between the first visit at and beyond which the H&Y status was consistently recorded as ≥3, and the preceding visit, based on the available follow-up data. For the purposes of this study, ‘consistent’ recording of the H&Y stage as ≥3 was defined as recording of the H&Y stage as ≥3 for all future visits until the end of follow-up. Censoring was at the time of the last assessment.

Cognitive progression was assessed by measuring the time from diagnosis to the fulfilment of criteria for PD dementia39. The date of dementia was calculated as the midpoint between the visit at which dementia was first diagnosed and the preceding visit16. Where patients were lost to follow-up in the research clinic, dementia diagnoses were determined from clinical records or death certificates, with date of dementia diagnosis being determined by expert consensus of experienced clinical researchers on a case-by-case basis using the midpoint between the date that dementia was first recorded and the preceding clinic visit or research appointment, as appropriate.

Time (years) from PD diagnosis to dementia was calculated and used for survival analysis. Censoring was at the time of the last assessment.

Statistical analysis

Data was analysed using IBM SPSS Statistics (Version 27). Survival plots were generated using GraphPad Prism version 9.

Baseline differences in demographic and clinical (including motor and cognitive) variables between beta-adrenoceptor drug users and non-users groups were evaluated using non-parametric Kruskal–Wallis tests for continuous variables (due to lack of normal distribution indicated by Shapiro–Wilk tests) and Chi-squared tests for dichotomous variables. Regression analyses were used to confirm any observed associations between the drug categories and key baseline cognitive and motor variables, adjusting for relevant covariates. Such analysis only included patients with all the covariates available for analysis.

Associations between beta-agonist users versus non-users and beta-blocker users versus non-users and time to reach H&Y3 or PD dementia were assessed using Kaplan-Meier survival analyses. Pairwise log-rank comparisons were conducted to determine if groups differed in survival distribution. Cox regression analyses were performed to control for relevant covariates. The event per variable ratio was >10 in each analysis, indicating suitability for survival analysis and Cox regression. The proportional hazards assumption was checked by visual inspection of Kaplan–Meier curves and Schoenfeld Residual plots for each covariate. Significance threshold was considered to be p < 0.05, and the available data was estimated to have 80% power to detect an HR of <0.57 and >1.75.

In all multivariate regression analyses, relevant covariates included variables that were significantly different (p < 0.05) between drug user/non-user groups (Table 2) and other relevant variables associated with the drug category or the outcome measure. Study cohort was included as a categorical covariate, with the largest cohort (PICNICS) as the reference category (Tables 3–5). Covariates included in each regression analysis are detailed in the ‘Results’ tables.