This retrospective cohort study was conducted using national claims data from the Health Insurance Review and Assessment Service (HIRA) database between 1 January 2008 and 31 December 2019. The Republic of Korea has a single-payer health insurance system that provides compulsory universal health insurance that covers the entire population of Korea (J. A. Kim et al., 2017). Healthcare providers must submit medical claims to the HIRA to receive an evaluation of the appropriateness of the reimbursement. The HIRA generates a research-ready database containing sociodemographic characteristics and clinical information. The database includes diagnoses, procedures, and prescription records from submitted claims after anonymizing all participants according to strict confidentiality guidelines from the medical care claims of all patients.

This study was approved by the Institutional Review Board (IRB) of Yonsei University (IRB number: 7001988-202004-HR-846-01E). The requirement for informed consent was waived because the data were publicly available and de-identified.

To evaluate the effect of RAS inhibitors on the development of PD in patients with ischemic heart disease (IHD), we enrolled patients aged ≥60 years who were diagnosed with angina or myocardial infarction during the identification period (1 January 2009, to 30 June 2009). Patients diagnosed with angina or myocardial infarction were identified based on the International Classification of Diseases 10^th Revision (ICD-10) diagnostic codes I20.0-I22.0. We designated a specific time point (1 July 2009) as the index date for all the study participants. All study participants were followed-up on the index date.

To identify new users of RAS inhibitors during the identification period, patients who had prescription records of RAS inhibitors prior to the identification period or who had been prescribed their first RAS inhibitors after the identification period were excluded. Patients who had died, had last claims records, or were previously diagnosed with parkinsonism (G20-G26), including PD before the index date, were also excluded. Given that PD may have a prodromal period (Postuma and Berg, 2016), a 1-year lag time was included to minimize the reverse causality of PD. The new users of RAS inhibitors during the identification period were classified into user groups. Patients who had never been prescribed RAS inhibitors during the study period were classified into a non-user group.

The RAS inhibitors used in this study included angiotensin-converting enzyme inhibitors (ACEIs) and AT₁ receptor blockers (ARBs), based on the Anatomic Therapeutic Chemical classification system of drugs by the World Health Organization (WHO) Collaborating Centre for Drug Statistic Methodology (World Health Organization, 2021). A detailed list of the RAS inhibitors is provided in Supplementary Table S1. To evaluate the cumulative effect of RAS inhibitors on the incidence of PD, we accumulated the doses and duration of RAS inhibitors during the follow-up period by collecting prescription information, including dosage, frequency, and days of supply.

The accumulated doses were calculated by multiplying the dosage, frequency, and number of supply days. These values were converted into the cumulative defined daily dose (cDDD) developed by the WHO (Methodology, 2021). The cumulative duration of RAS inhibitor use was calculated as the sum of the prescription days. The daily equivalent dosage, defined as the ratio of the cumulative dose to the cumulative exposure period, was calculated. To compare the preventive effect on PD, the patients were categorized into four groups according to their prescriptions: (1) use of non-BBB-crossing ACEIs, (2) use of BBB-crossing ACEIs, (3) use of non-BBB-crossing ARBs, and (4) use of BBB-crossing ARBs (Supplementary Table S1) (Alzahrani et al., 2020; C. K. ; Kim et al., 2015; Oka et al., 1988; Sink et al., 2009; Takai et al., 2004; Wharton et al., 2012; Yagi et al., 2013). These groups were based on the degree of BBB penetration and the type of RAS inhibitor prescribed.

Data on sociodemographic characteristics, including the patient’s age, sex, and insurance type (i.e., health insurance or medical aid), were collected during the identification period. Potential confounding factors affecting the development of PD, including comorbid diseases and concurrent medications, were recorded until the date of outcome occurrence or the end of the study period. These confounding factors were confirmed when patients had continuous claim records on diagnosis and prescription at least once per year until the last follow-up (J. Lee et al., 2018).

Comorbid diseases known as potential risk or protective factors for PD included hypertension, dyslipidemia, diabetes mellitus, end-stage renal disease, stroke, brain injury, dementia, chronic obstructive pulmonary disease (COPD), depression, gout, osteoarthritis, osteoporosis, and severe liver disease (Hu et al., 2020; Jafari et al., 2013; C. H. ; Li et al., 2015; Rugbjerg et al., 2010; Torsney et al., 2014; Tryc et al., 2013; Vale et al., 2018; Wang et al., 2018). Concurrent medications known to be potential risk or protective factors for PD include antipsychotics, antiemetics, calcium channel antagonists, antiepileptics, trimetazidine, dopamine depleters, mood stabilizers, beta-blockers, dihydropyridine calcium channel blockers, and statins (Ascherio and Schwarzschild, 2016; Jeong et al., 2019; S. ; Kim et al., 2020; López-Sendón et al., 2013; Mullapudi et al., 2016; Shin and Chung, 2012). The ICD-10 codes for comorbid diseases and details of the concurrent medications are listed in Supplementary Tables S2, 3.

The primary outcome measure was new-onset Parkinson’s disease during the follow-up period. New-onset PD was defined as an ICD-10 diagnostic code for PD (G20), accompanied by the PD registration code and prescription records for any PD treatment drug. These include L-dopa, carbidopa/benserazide, pramipexole, ropinirole, bromocriptine, selegiline, rasagiline, amantadine, entacapone, trihexyphenidyl, and benztropine. In Korea, the government established a registry called the Rare and Intractable Diseases Program, which registered patients with neurologist-confirmed PD (Park et al., 2019). To enhance the validity of PD diagnosis in this study, we used the PD registration code (V124) for the outcome definition. The date of study outcome was defined as the date of the first appearance of the PD diagnostic code for G20.

To adjust for selection bias, we used a propensity score (PS)-matched design. PS was estimated using multivariate logistic regression. The predictors of PS included sex, age, insurance type, follow-up duration, comorbid diseases, and concurrent medications. Patients who did and did not use RAS inhibitors were matched in a 1:1 ratio without replacement, using the greedy method. This was within a caliper that was 0.2 times the standard deviation of the logit PS. After PS matching, the balance diagnostics between RAS inhibitor users and non-users was assessed by evaluating the standardized difference between them. When the standardized difference was >0.1, it was regarded as a sign of imbalance (Austin, 2009). Furthermore, the similar distribution of PS between RAS inhibitor users and non-users after matching is confirmed graphically in Supplementary Figure S1.

Descriptive statistics were used to summarize the characteristics of the matched population. Between-group comparisons of the distribution of sociodemographic data, comorbid diseases, and concurrent medications between RAS inhibitor users and non-users were performed using Pearson’s chi-squared test and Student’s t-test. The incidence rate of PD per 1,000 person-years was calculated by dividing the number of outcome events by the total number of person-years at risk and multiplying the result by 1,000. The person-years at risk were accumulated for each patient from the index date to the outcome date, death, or 31 December 2019, whichever occurred first.

Multivariable Cox proportional hazard regression analyses adjusted for demographic characteristics, comorbidities, and concurrent medications were used to estimate the hazard ratio (HR) and 95% confidence interval (CI) for the risk of PD in relation to RAS inhibitor use. The proportional hazard assumption was confirmed by a statistical test of the correlation between Schoenfeld residuals and survival time, using a cumulative log(minus) log curve (Supplementary Figure S2). To investigate the cumulative dose-response and duration-response relationships, the risk of PD was investigated according to the cumulative dose, total duration, and daily equivalent dosage of RAS inhibitors, using a multivariate Cox regression model. Additionally, the adjusted hazard ratios (aHRs) of PD were estimated according to the type of RAS inhibitor (i.e., ACEIs and ARBs) used and the degree of BBB penetration. Subgroup analyses were performed to evaluate the effect of RAS inhibitors on PD risk according to patient sex and age. Age was divided into two groups based on the mean age of the study population.

Sensitivity analyses were conducted using three steps. First, the index date from 1 July 2009 was shifted to 1 July 2010 and 2011. As the identification period changed accordingly, drug exposure, outcomes, and confounding factors were re-assessed. Second, the lag time was extended for 3 and 5 years, resulting in a decrease in the follow-up period and the number of study participants. As the lag time was extended, captured PD events as outcomes during the lag time were excluded from the analysis. Finally, the operational definition of PD was changed, and the study outcome was re-evaluated according to the following new definitions. First, patients had a diagnostic code for PD accompanied by a PD registration code. Second, patients had a diagnostic code of PD and at least 60 days of prescription records for any PD treatment medications after the first diagnosis. All statistical analyses were performed using SAS statistical software (version 9.4; SAS Institute, Cary, NC, United States). p < 0.05 was considered significant.

Among the 537,116 patients diagnosed with IHD between January 2009 and June 2009, 31,114 RAS inhibitor users and 31,114 RAS inhibitor non-users were included in the analysis (Figure 1). The baseline characteristics of the study population are shown in Table 1. The mean age (±standard deviation) was 70.3 (±7.1) years, and males accounted for 44.4% of the cohort. The mean duration of follow-up was 7.6 and 7.7 years for RAS inhibitor users and non-users, respectively. After PS matching, the distribution between the two groups became similar for all variables, with standardized differences <0.1. This indicates that the differences between covariates were negligible.

A total of 1,086 new PD cases were identified from 477,255 person-years of observation in the main analysis. The overall incidence rate was 2.28 per 1,000 person-years. The median time to censor was 9.4 years (interquartile range, 6.1–9.5 years) in those censored with PD over 10 years of follow-up. The incidence rate of PD was significantly lower in RAS inhibitor users than in non-users (2.04 per 1,000 person-years vs. 2.51 per 1,000 person-years, p < 0.0001).

RAS inhibitor use was significantly associated with a reduced risk of PD after adjusting for confounders (aHR = 0.75; 95% CI = 0.66–0.85). Male sex and aged ≥65 years were associated with an increased risk of PD. Comorbidities of hypertension, dyslipidemia, diabetes mellitus, stroke, brain injury, dementia, COPD, depression, osteoarthritis, and osteoporosis were significantly associated with an increased risk of PD. Consistent use of medications including antipsychotics, antiemetics, calcium channel antagonists, antiepileptics, trimetazidine, and beta-blockers was significantly associated with an increased risk of PD. Statins were significantly associated with a decreased risk of PD (Table 2).

A significantly reduced risk of PD was observed with the use of ARBs (aHR = 0.74; 95% CI = 0.65–0.85). However, ACEI use was not significantly associated with the risk of PD (aHR = 0.91; 95% CI = 0.77–1.08). The risk of PD was significantly reduced with the use of BBB-crossing RAS inhibitors (aHR = 0.67; 95% CI = 0.57–0.78), but not with the use of non-BBB-crossing RAS inhibitors. Regarding the risk of PD by drug type in relation to the degree of BBB penetrability, only the use of BBB-crossing ARBs showed significant decreases in PD incidence (aHR = 0.62; 95% CI = 0.53–0.74) (Table 3).

Table 4 presents the PD risk breakdown by cumulative dose, duration, and daily equivalent dosage of BBB-crossing ARBs. In all cDDD groups, the use of BBB-crossing ARBs was significantly associated with a reduced risk of PD. Similar reduced risks of PD were observed in all groups at the daily equivalent dosage. Regarding cumulative duration, RAS inhibitor users who were exposed to ≥2 years of BBB-crossing ARBs showed a lower risk of PD than non-users, whereas those who were exposed to <2 years of BBB-crossing ARBs did not. The risk of developing PD tended to decrease with the increase in the cumulative dose and duration (P for trend = 0.03 and <0.01, respectively). Figure 2 shows the Kaplan-Meier curves of the cumulative hazard (1- survival rate) according to the cumulative duration of BBB-crossing ARBs. In the subgroup analyses, the aHRs of BBB-crossing ARBs were comparable in the populations aged<70 years (aHR = 0.65; 95% CI = 0.51–0.83) and ≥70 years (aHR = 0.58; 95% CI = 0.46–0.73). In the sex subgroup analysis, the aHR of BBB-crossing ARBs in the female population (aHR = 0.53; 95% CI = 0.43–0.66) was lower than that in the male population (aHR = 0.75; 95% CI = 0.58–0.96).

Sensitivity analyses for aHRs of RAS inhibitor use for PD are shown in Table 5. The PD risk associated with RAS inhibitor use remained lower than that the risk associated with RAS inhibitor non-use in all sensitivity analyses according to index date shifting, lengthening of the lagged period, and changing the outcome definitions.