Exploring the risk of adverse drug events in combination with antiparkinsonics and antipsychotics—a two-decade real-world pharmacovigilance analysis based on the FAERS database

Abstract

Background

The combination of antiparkinsonics and antipsychotic drugs (AP) can improve the motor and mental symptoms of Parkinson’s disease (PD) and reduce the actual burden of chronic disease care. To explore the adverse drug events (ADEs) worthy of attention in this treatment management process, we conducted a real-world pharmacovigilance analysis based on the FDA Adverse Event Reporting System (FAERS) database.

Methods

The Standard pharmacotherapy for PD includes Levodopa/Carbidopa, Entacapone, Rasagiline, Pramipexole, Ropinirole, Rotigotine, Apomorphine, Amantadine, etc. Antipsychotic drug includes Quetiapine, Clozapine, and Pimavanserin. We collected the ADEs reports of FAERS that conformed to the combination regimens of anti-Parkinson’s drugs and AP during the 20-year period from the third quarter of 2004 to the second quarter of 2024. Disproportionate analysis and subgroup analysis were conducted through 5 algorithms, namely Ω shrinkage measure, additive model, multiplicative model, Combination risk ratio, and Chi-square. The time-to-onset (TTO) analysis was used to predict the variation of the risk size of ADEs occurrence over time. Finally, we explored the correlation between population characteristics and the occurrence of ADEs through Logistic regression.

Results

We collected a total of 6297 cases, including 38 316 ADEs records. The results of the disproportionate analysis show that the ADEs with the highest occurrence frequency include hallucination, general physical health deterioration, somnolence, stoma site discharge, urinary tract infection, memory impairment, etc. The TTO analysis results showed that the median TTO for all ADEs was 657.50 days, the median TTO for infection and inflammation was 716.00 days, and the median TTO for psychiatric symptoms was 823.00 days. All median TTOs conform to the early failure curve. The results of Logistic regression showed that gender was correlated with the occurrence of infection and inflammation, and the female population was more inclined to have important medical events related to infection and inflammation.

Conclusions

During the combined application of antiparkinsonics and AP, in addition to ADEs such as movement disorders and emerging mental symptoms, the risks of infection and inflammation should also be given key attention. Long-term follow-up should run through the entire process of disease diagnosis and treatment, and attention should be paid to the influence of drug dosage forms and dosages. The medication plan should be adjusted in a timely manner when ADEs occur.

Significance statement.

To explore the ADEs worthy of attention in the therapeutic management of the combination therapy of antiparkinsonics and antipsychotic drugs (AP), we conducted a real-world pharmacovigilance analysis based on the FDA Adverse Event Reporting System database. Adverse drug events (ADEs) reports that conformed to the antiparkinsonics and AP combination regimen during the 20-year period from the third quarter of 2004 to the second quarter of 2024 were collected. The analysis results suggest that in addition to ADEs such as movement disorders and emerging mental symptoms, the risks of infection and inflammation should also be given key attention. Long-term follow-up should run through the entire process of disease diagnosis and treatment, and attention should be paid to the influence of drug dosage forms and dosages. The medication plan should be adjusted in a timely manner when ADEs occur.

INTRODUCTION

Neurological disorders are the leading cause of disability and the second leading cause of death globally, with the prevalence of these disorders increasing over the past half century.¹ Parkinson’s disease (PD), one of the most common neurodegenerative disorders,² is a major contributor to neurological disability.³ It is estimated that 1 to 2 individuals per thousand are affected by PD,⁴ with advancing age recognized as an independent risk factor.⁵ The hallmark pathological feature of PD is the early and significant loss of dopaminergic neurons in the substantia nigra pars compacta, leading to dopamine deficiency and ultimately resulting in classical motor symptoms such as bradykinesia, rigidity, and tremor.^6,7 The heterogeneity in the mechanisms and symptoms of PD presents challenges for accurate diagnosis and effective treatment.⁸ Currently, standard treatment regimens primarily focus on managing motor symptoms, operating through the increase of dopamine levels in the brain or stimulation of corresponding dopamine receptors.² Levodopa remains the cornerstone of PD treatment, and its combination with carbidopa can minimize side effects while improving symptoms.⁶ Other alternatives for antiparkinsonian therapy include dopamine agonists, monoamine oxidase type B inhibitors, catechol-O- methyltransferase inhibitors, and additional medications such as amantadine.

In recent decades, an increasing number of studies have highlighted that PD manifests numerous non-motor symptoms, including specific prodromal symptoms and additional symptoms that progress in the later stages of the disease. Psychiatric symptoms are a major component of the non-motor symptoms of PD, leading to adverse outcomes such as clinical disability. The range of Parkinson’s disease psychosis (PDP) includes, but is not limited to, anxiety, depression, hallucinations, delusions, and impulse control disorders.^9,10 A community study indicated that the prevalence of PDP exceeds 26%,¹¹ with the incidence of psychiatric disorders in PD patients reaching as high as 60%.¹² Cohort studies have demonstrated that psychiatric symptoms in individuals with Parkinson’s can increase the risk of long-term care by up to 3-fold and elevate mortality risk by nearly one-third.¹³ Conventional antiparkinsonian medications have limited efficacy in alleviating psychiatric symptoms, whereas long-term use of dopaminergic medications carries a risk of inducing psychiatric complications.⁶ As a result, antipsychotic medications (APs) have increasingly been incorporated into the clinical management of PD, typically for symptomatic relief, but their interactions and effects remain unclear, and their application is accompanied by ongoing doubts and limitations.¹⁴ Clozapine has been historically recognized as a recommended medication for PDP, while quetiapine has effectively become the first-line treatment due to its advantage of not requiring regular blood monitoring.^15-17 Not until 2016 did pimavanserin become the first drug specifically approved by the FDA for the treatment of PDP, showing a reduced mortality risk in clinical trials.^18-20 Furthermore, the use of APs may assist in the reduction of doses of dopamine receptor agonists and anticholinergic medications.

In the standard treatment of PD, adverse drug events (ADEs) can arise from various factors. The concomitant use of antiparkinsonian medications and APs may provide clinical benefits; however, it is also associated with a heightened risk of ADEs. Reported ADEs include cognitive impairment, drowsiness, extrapyramidal symptoms, and myocarditis.^21-23 Therefore, this study performs a pharmacovigilance analysis of the combination of these medications based on the FDA Adverse Event Reporting System (FAERS) database, examining ADEs including those that have not been widely reported or emphasized. The findings aim to furnish a pharmacovigilance report that supports the comprehensive clinical management of PD with psychiatric symptoms, thereby assisting healthcare practitioners in making informed clinical decisions and optimizing the management of the disease throughout its entirety, ultimately improving patient prognosis.

METHODS AND MATERIALS

Data Sources and Preprocessing

This study collected individual ADEs reports available in the FAERS database from the third quarter of 2004 to the second quarter of 2024, provided by healthcare professionals, consumers, and pharmaceutical companies. The dataset comprises 7 tables: demographic and administrative information of patients (DEMO), drug/biologic information (DRUG), adverse events (REAC), patient outcomes (OUTC), report sources (RPSR), drug therapy start and end dates (THER), and indications for use/diagnosis (INDI). Standard treatments for PD include Levodopa/Carbidopa, Entacapone, Rasagiline, Pramipexole, Ropinirole, Rotigotine, Apomorphine, and Amantadine. The APs analyzed in this study include Quetiapine, Clozapine, and Pimavanserin. We retrieved these FDA-approved drugs from the FAERS database to filter reports of ADEs associated with the concurrent use of antiparkinsonian and APs. Specific search terms can be found in Table S1. The process of retrieving and collecting individual ADEs reports adhered to the standards recommended by FAERS, with duplicate reports removed. Adverse drug events will be described at the levels of system organ class (SOC) and preferred term (PT) using medical terminology, referencing the Medical Dictionary for Regulatory Activities (version 27.0, https://www.meddra.org/). In the FAERS database, ADEs are classified into 4 categories: primary suspected (PS), second suspected (SS), concomitant (C), and interacting (I). To ensure the reliability of the analysis, this study exclusively collected reports of ADEs categorized as PS, while reports listed as SS, C, and I were excluded due to concerns about their relevance. The treatment outcomes of ADEs within the FAERS database fall into 7 categories: congenital anomaly (CA), death (DE), disability (DS), hospitalization—initial or prolonged (HO), life-threatening (LT), required intervention to prevent permanent impairment/damage, and other (OT). Among these categories, CA, DE, DS, HO, and LT are designated as Important Medical Events (IMEs), which will be the focal point of our study. For the sources of ADEs reports, we selected those submitted by healthcare professionals and consumers. Detailed processes for data retrieval, collection, and screening are illustrated in Figure 1. All statistical analyses and visualizations conducted in this study were performed using R software (v 4.3.2).

Figure 1.

Detailed technical roadmap of this study.

Descriptive Analysis

After filtering the ADEs reports related to the concurrent treatment of PD with psychiatric symptoms, we conducted a descriptive analysis of the reports. The results of this analysis include demographic information such as sex, age, and weight, reporter occupation (OCCP_COD), reporting country (REPORTER_ COUNTRY), treatment outcomes (OUTC_COD), and the year of reporting.

Disproportionality Analysis of Adverse Drug Events

To identify ADEs related to the concurrent use of antiparkinsonian drugs and APs, we employed 5 algorithms—Ω shrinkage measure, additive model, multiplicative model, combination risk ratio, and Chi-square model, to conduct disproportionality analysis. A signal strength that exceeds the algorithm’s threshold indicates a potential correlation between drug use and ADEs. The Ω shrinkage measure is an algorithm that calculates the ratio of ADE reports associated with a combination of 2 drugs compared to the expected value ratio, with Ω₀₂₅ > 0 serving as the threshold for screening the pharmacovigilance signals related to this combination.²⁴ The Additive model exhibits sensitivity in detecting drug-drug interactions by employing a 2 × 2 contingency table; when the sum of additive effects (add) is greater than 0, it suggests a potential risk for drug interactions. The Multiplicative model can further refine the strength determination of signals identified by the Additive model, where a value of mul > 1 indicates an increased potential effect from combination therapy in comparison to monotherapy.²⁵ The Combination risk ratio model assesses risk based on the frequency of co-reported drug use, presenting results through the proportional reporting ratio (PRR). Proportional reporting ratio > 2, with χ² > 4, and a value of ≥ 3, indicates the presence of risk. Similarly, the Chi-square model, incorporating Chi-square statistics and Yates correction, uses a threshold of chi > 2 to signify potential drug interaction risks.²⁶Following the disproportionality analysis of ADEs reports at both the SOC and PT levels, we selected positive ADE signals that were based on the thresholds established by 5 algorithms described earlier. In alignment with Noguchi’s study, to mitigate the risk of overly conservative or optimistic algorithmic results, we selected ADE signals that satisfy the thresholds of 3 or more of the specified algorithms for a correlational analysis of combined drug therapy pharmacovigilance signals.²⁷

Subgroup Analysis

The FAERS database provides demographic data, including sex, age, and weight. To explore the differences in ADEs among populations who are on a 4-drug regimen, we conducted a subgroup analysis based on these demographic variables. This analysis was conducted based on the disproportionality analysis of the 4 algorithms, grouping the data by sex (male, female), age (< 60 years, ≥ 60 years), and weight (< 80 kg, ≥ 80 kg). Using the PRR and P-value obtained from the disproportionality analysis, we created volcano plots and forest plots to compare ADEs differences among the subgroups.

Probability Density Function of Time-to-Onset

To predict the changes in the risk of ADEs over time, we calculated the time-to-onset (TTO), defined as the time interval from the initiation of drug use to the occurrence of ADEs. After excluding cases with inaccurate TTO records, we computed the probability density function of TTO based on the Weibull distribution. The Weibull distribution is a probability distribution for continuous random variables that describes how risk can increase or decrease over time. The main parameters of the Weibull distribution are the shape parameter (α) and the scale parameter (β). Both α (α > 0) and β (β > 0) determine the probability density function; α controls the skewness, and β controls the width of the distribution, with the calculation formula detailed in Table S2. In this study, we focus on the probability density function of TTO described by the Weibull distribution with particular attention to the value of α. A larger α corresponds to a longer TTO, indicating that the failure curve is skewed to the right. When α < 1 and the 97.5% CI < 1, it signifies that the probability of ADEs occurrence decreases over time, indicating an early failure-type curve. When α = 1, it indicates that the probability of ADEs occurrence remains constant over time, representing a random failure-type curve. When α > 1 and the 97.5% CI > 1, it indicates that the probability of ADEs occurrence increases over time, signifying a degradation failure-type curve.

Logistic Regression

We conducted univariate and multivariate logistic regression analyses with sex (Male, Female), age (< 45 years, 45-65 years, ≥ 65 years), weight (< 70 kg, ≥ 70 kg), and duration of medication (< 1 year, 1-2 years, ≥ 2 years) as independent variables, and the occurrence of IMEs as the dependent variable to investigate the relationship between population characteristics and IMEs occurrence. Using logistic regression analysis, we calculated the odds ratios (ORs) for the occurrence of IMEs under different exposure conditions in various populations. Additionally, regarding reports of IMEs occurrence, we also conducted univariate and multivariate logistic regression analyses to further investigate the relationship between population characteristics and the occurrence of severe infections and inflammation.

RESULTS

Descriptive Analysis

We collected FAERS data spanning 20 years, from the third quarter of 2004 to the second quarter of 2024. A detailed screening process for ADEs reports is illustrated in Figure 1. Ultimately, we gathered 6297 cases, which included 38 316 related records. Examining annual report numbers (Figure 2), the growth of ADEs cases from 2004 to 2014 was slow; however, after 2014, particularly following the approval of pimobendan as a specific treatment for PD in 2016, the number of reported cases exhibited a rapid increase. The number of reported cases remained high and stable from 2019 to 2021, with 2022 marking the highest number of cases in the past 2 decades following the gradual resolution of the COVID-19 pandemic. Encouragingly, the number of annual case reports has significantly decreased in the subsequent period. Based on the case numbers from the first 2 quarters of 2024, it is projected that the growth rate of ADEs cases will gradually slow and stabilize starting from last year.

Figure 2.

The number of reports of adverse drug events (ADEs) each year from the third quarter of 2004 to the second quarter of 2024.

From the demographic data of the 6297 collected ADEs cases (Table 1), after excluding missing data, the majority of the population characteristics reflect elderly males, with little variation across weight groups. Males accounted for 62% (n = 3798) and females for 38% (n = 2373). Additionally, minors represented 0.2% (n = 10), young adults 1.7% (n = 78), middle-aged individuals 18% (n = 792), and the elderly comprised 80% (n = 3579). Individuals weighing less than 70 kg accounted for 52% (n = 755), while those exceeding 70 kg made up 48% (n = 704). Nearly half (n = 3061) of the reports regarding ADEs cases were submitted by healthcare professionals (MDs, HPs, PHs, OTs), and about half (n = 3060) originated from major developed countries (US, JP, FR, GB, DE). Furthermore, the incidence of IMEs in patients receiving combined antiparkinsonian and antipsychotic therapy was high, with 81% (n = 4631) of ADEs cases experiencing IMEs. Therefore, given the annual report numbers and demographic data, the ADEs caused by the combination of antiparkinsonian drugs and APs should be carefully considered. It is crucial to remain vigilant in evaluating combined treatment regimens during clinical medication management to ensure comprehensive care for patients with PD.

Table 1.

Demographic characteristics of 6297 ADE cases

Characteristics	N = 6297a
Age	71 (10)
< 18 years	10 (0.2%)
18-45 years	78 (1.7%)
45-64 years	792 (18%)
≥ 65 years	3579 (80%)
Unknown	1838
Weight	70 (27)
< 70 kg	755 (52%)
≥ 70 kg	704 (48%)
Unknown	4838
Sex
Female	2373 (38%)
Male	3798 (62%)
Unknown	126
Reported person
Health professional physician (MD)	1250 (20%)
Non-healthcare professional consumer (CN)	3122 (50%)
Pharmacist (PH)	226 (3.7%)
Health professional (HP)	798 (13%)
Other health professional (OT)	787 (13%)
Unknown	114
Reported country
DE	1450 (23%)
US	1228 (20%)
NL	427 (6.8%)
CA	237 (3.8%)
ES	188 (3.0%)
FR	152 (2.4%)
IE	148 (2.4%)
GB	138 (2.2%)
IT	95 (1.5%)
JP	92 (1.5%)
AU	75 (1.2%)
BE	72 (1.2%)
Others	1958 (31.3%)
Unknown	37
Outcome
Hospitalization-initial or prolonged (HO)	2966 (52%)
Other (OT)	1079 (19%)
Death (DE)	1496 (26%)
Life-threatening (LT)	124 (2.2%)
Disability (DS)	45 (0.8%)
Required intervention to prevent permanent impairment/damage (RI)	8 (0.1%)
Unknown	579

^aMean (SD); n (%)

Signal Monitoring Data of ADEs at the SOC Level

Disproportionate analysis of ADEs at the SOC level indicates (Figure 3A, Table S3) that the most frequent occurrences are predominantly found in the following SOC categories: nervous system disorders (n = 7676), psychiatric disorders (n = 6073), general disorders and administration site conditions (n = 5257), injury, poisoning, and procedural complications (n = 4146), gastrointestinal disorders (n = 2350), infections and infestations (n = 1938), musculoskeletal and connective tissue disorders (n = 1902), investigations (n = 1575), and product issues (n = 1391). Based on the thresholds stipulated by the combination risk ratio algorithm used in the disproportionality analysis, only nervous system disorders, psychiatric disorders, product issues, and social circumstances met the criteria—PRR ≥ 2, χ² ≥ 4, a ≥ 3. The forest plot (Figure 3B) shows that the PRR values for these SOCs all exceed 2, with nervous system disorders (PRR = 2.36, 95% CI, 2.34-2.38),psychiatric disorders (PRR = 2.78, 95% CI, 2.76-2.80), product issues (PRR = 2.32, 95% CI, 2.27-2.37), and social circumstances (PRR = 2.40, 95% CI, 2.30-2.50). These results suggest that the combined use of antiparkinsonian drugs and APs is a risk factor for the occurrence of ADEs in these SOC categories.

Figure 3.

ADE signal monitoring results.

Signal Monitoring Data of ADEs at the PT Level

The disproportionate analysis of ADEs at the PT level, excluding symptoms and causes closely related to PD, shows (Figure 3C, Table S4) that the most common PTs are hallucination (n = 802), general physical health deterioration (n = 622), somnolence (n = 348), stoma site discharge (n = 344), urinary tract infection (n = 338), and memory impairment (n = 309). The results of the heat map based on the thresholds established by 5 algorithms (Figure 3D, Table S4) reveal that among the 30 most frequent preferred terms (PTs), all except pneumonia, somnolence, pneumonia aspiration, and abnormal behavior meet the thresholds of at least 3 of the 5 algorithms, which indicates their status as risk factors for the occurrence of ADEs. Notably, just among the top 30 most frequently occurring PT items, were cognitive and mental disorders, including hallucination (n = 802, PRR = 17.24), hallucination visual (n = 226, PRR = 17.77), restlessness (n = 244, PRR = 10.46), aggression (n = 196, PRR = 6.12), delirium (n = 178, PRR = 8.42), and delusion (n = 171, PRR = 17.71), as well as neurological disorders, including memory impairment (n = 309, PRR = 3.46), hyperkinesia (n = 236, PRR = 223.06), and speech disorder (n = 215, PRR = 6.37), etc. The number of reports from both has exceeded 1000 (some demographic data are missing). In addition, stoma site symptoms include reaction (n = 86, PRR = 217.99), pain (n = 121, PRR = 128.87), and inflammation (n = 123, PRR = 295.65). Furthermore, infection (n = 98, PRR = 128.89), hypergranulation (n = 80, PRR = 214.9), hemorrhage (n = 85, PRR = 91.79), erythema (n = 256, PRR = 226.32), and discharge (n = 344, PRR = 197.36) also exceeded 1000 cases. In addition to the usual movement disorders, the results of the heat map (Figure 3D) suggest that infections in multiple body sites and diverse psychiatric and cognitive disorders present a significantly high risk of ADEs occurrence, which has not been adequately addressed in the past. In particular, urinary tract infection (n = 338, Ω₀₂₅ = 0.27, PRR = 17.77) and visual hallucination (n = 226, Ω₀₂₅ = 0.08, PRR = 3.12) satisfy the thresholds of all 5 algorithms, providing further evidence that infections and psychiatric disorders are high-risk factors for ADEs.

Subgroup Analysis

Based on demographic data concerning age, sex, and weight, we conducted a subgroup analysis based on disproportionality analysis. Note that this analysis only compares the frequency of occurrences between subgroups based on PT-level disproportionality analysis results and does not allow for correlation analysis. The results of the subgroup analysis based on age show (Figure 4A and B) that the elderly population (≥ 60 years) demonstrates a higher frequency of various neuropsychiatric disorders. Specifically, the following conditions were noted: neuroleptic malignant syndrome (PRR = 0.24, 95% CI, 0.13-0.41, P-value < .001), psychotic disorder (PRR = 0.39, 95% CI, 0.28-0.54, P-value < .001), paranoia (PRR = 0.3, 95% CI, 0.16-0.53, P-value < .001), mania (PRR = 0.1, 95% CI, 0.05-0.22, P-value < .001), emotional disorder (PRR = 0.21, 95% CI, 0.1-0.44, P-value < .001), impulse control disorder (PRR = 0.12, 95% CI, 0.06-0.23, P-value < .001). In contrast, the non-elderly population (< 60 years) has higher frequencies for infections and neurological disorders, including urinary tract infection (PRR = 2.68, 95% CI, 1.46-4.91, P-value = .001), pneumonia (PRR = 2.18, 95% CI, 1.27-3.74, P-value = .005), and confusional disorder (PRR = 2.53, 95% CI, 1.5-4.25, P-value < .001).

Figure 4.

The results of subgroup analyses based on age, gender, and weight, respectively.

Based on gender, the findings show (Figure 4C and D) that the male population exhibits higher incidences of gastrointestinal-related symptoms and lower respiratory tract infections, including bowel movement irregularity (PRR = 0.47, 95% CI, 0.23-0.96, P-value < .05), flatulence (PRR = 0.33, 95% CI, 0.13-0.88, P-value < .05), intestinal obstruction (PRR = 0.22, 95% CI, 0.07-0.73, P-value < .05), pneumonia aspiration (PRR = 0.36, 95% CI, 0.24-0.52, P-value < .001), and lower respiratory tract infection (PRR = 0.2, 95% CI, 0.07-0.57, P-value = .001). In contrast, female patients show higher rates of drug interactions, organ damage, circulatory disorders, and neuropsychiatric disorders, including drug interaction (PRR = 2.1, 95% CI, 1.38-3.19, P-value = .001), stoma site irritation (PRR = 2.58, 95% CI, 1.53-4.35, P-value < .001), acute kidney injury (PRR = 2.06, 95% CI, 1.17-3.65, P-value = .016), orthostatic hypotension (PRR = 2.21, 95% CI, 1.51-3.25, P-value < .001), circulatory hypotension (PRR = 3.21, 95% CI, 1.69-6.1, P-value < .001), altered state of consciousness (PRR = 2.68, 95% CI, 1.31-5.47, P-value < .01), PDP (PRR = 6.9, 95% CI, 3.47-13.74, P-value < .001), and acute psychosis (PRR = 3.42, 95% CI, 1.85-6.33, P-value < .001).

Based on weight, the results indicate (Figure 4E and F) that the obese population (≥ 80 kg) has higher frequencies of fungal infection (PRR = 12, 95% CI, 0.01-0.93, P-value < .05), stoma site pain (PRR = 0.42, 95% CI, 0.23-0.77, P-value < .01), and behavioral disorders (PRR = 0.11, 95% CI, 0.01-0.83, P-value < .05). In contrast, those with a weight below 80 kg exhibit higher frequencies of perceptual impairment and aggravated condition, including feeling abnormal (PRR = 3.46, 95% CI, 1.57-7.65, P-value = .002), paresthesia (PRR = 4.26, 95% CI, 1.42-12.75, P-value = .009), dysarthria (PRR = 3.91, 95% CI, 1.09-14.01, P-value < .05), psychotic disorder (PRR = 2.45, 95% CI, 1.17-5.15, P-value < .05), and condition aggravated (PRR = 2.21, 95% CI, 1.16-4.18, P-value = .019).

Time-to-Onset Analysis

To examine the temporal changes in ADEs risks associated with the combined use of antiparkinsonian drugs and APs, we calculated the TTO for cases and computed the probability density function of TTO based on the Weibull distribution. A total of 2004 cases had complete TTO records, with the onset times of all cases calculated in 6-month intervals for TTO. The histogram (Figure 5A) indicates that approximately half of the 2004 cases experienced ADEs within 22 months of treatment, with a median TTO of 657.50 days. In the probability density function (Table 2), the shape parameter (α = 0.78, 97.5% CI, 0.75-0.80) and scale parameter (β = 807.79, 97.5% CI, 760.10-855.48) indicate an early failure-type distribution, suggesting that the probability of ADEs occurrence decreases over time.

Figure 5.

The frequency distribution of ADE occurrence based on Time-to-onset.

Table 2.

Parameters for calculating the probability density function of time-to-onset based on the Weibull distribution

Group	Number	TTO (days)		Weibull distribution				Failure type
				Shape parameter		Scale parameter
		Median (IQR)	Min-Max	α	97.5% CI	β	97.5% CI
Total	2004	657.50 (159.50-1405.00)	1-6375	0.78	0.75-0.80	807.79	760.10-855.48	Early failure
Infection and inflammation	800	716.00 (159.50-1454.25)	1-5353	0.75	0.71-0.79	822.83	743.45-902.21	Early failure
Psychiatric symptoms	1035	823.00 (192-1546)	1-5673	0.82	0.79-0.86	955.98	881.70-1030.26	Early failure
Fibroma	20	1113.50 (1073.25-1340.75)	82-2653	1.97	1.29-2.65	1409.31	1082.45-1736.17	Degradation failure

Abbreviation: TTO, time-to-onset.

We focused on 3 significant ADEs, including infection and inflammation, psychiatric symptoms, and fibroma.

A total of 800 cases reported the TTO for infections and inflammation. The histogram (Figure 5B) indicates that the median TTO for all 800 cases is 716.00 days. In the probability density function (Table 2), the shape parameter (α = 0.75, 97.5% CI, 0.71-0.79) and scale parameter (β = 822.83, 97.5% CI, 743.45-902.21) indicate an early failure-type distribution, suggesting that the probability of ADEs related to infections and inflammation decreases over time.

A total of 1035 cases reported the TTO for psychiatric symptoms. The histogram (Figure 5C) shows that the median TTO for all 1035 cases is 823.00 days. In the probability density function (Table 2), the shape parameter (α = 0.82, 97.5% CI, 0.79-0.86) and scale parameter (β = 955.98, 97.5% CI, 881.70-1030.26) indicate an early failure-type distribution, suggesting that the probability of ADEs related to psychiatric symptoms decreases over time following combined medication.

A total of 20 cases reported the TTO for fibroma. The histogram (Figure 5D) shows that the median TTO for all 20 cases is 1113.50 days. In the probability density function (Table 2), the shape parameter (α = 1.97, 97.5% CI, 1.29-2.65) and scale parameter (β = 1409.31, 97.5% CI, 1082.45-1736.17) indicate a degradation failure-type distribution.

Logistic Regression Analysis

The occurrence of IMEs results in poor outcomes for patients. To explore the relationship between population characteristics and the occurrence of IMEs, we conducted univariate and multivariate logistic regression analyses. Age (< 45 years, 45-64 years, ≥ 65 years), weight (< 70 kg, ≥ 70 kg), gender (female, male), and duration of medication (< 1 year, 1-2 years, ≥ 2 years) were treated as independent variables, while all IMEs served as the dependent variable for logistic regression. Both univariate and multivariate logistic regression analyses (Figure 6A and B) indicated that a duration of medication exceeding 1 year is significantly associated with all IMEs. The results of the univariate logistic regression (Figure 6A) showed that, compared to a duration of less than 1 year, a duration of 1-2 years (OR = 2.35, 95% CI, 1.20-4.96, P-value = .017) and more than 2 years (OR = 2.07, 95% CI, 1.30-3.36, P-value = .003) are risk factors for all IMEs. Consistent with the univariate logistic regression results, the multivariate logistic regression results (Figure 6B) also indicated that a duration of 1-2 years (OR = 2.33, 95% CI, 1.19-4.94, P-value = .019) and more than 2 years (OR = 2.11, 95% CI, 1.32-3.44, P-value = .002) are risk factors for all IMEs.

Figure 6.

The results of univariate and multivariate logistic regression based on age, gender, weight, and medication duration, respectively.

Furthermore, we analyzed the occurrence of IMEs related to infections and inflammation as the dependent variable and examined its correlation with independent variables (gender, weight, duration of medication). Specifically, since the majority of the patient population is elderly, age was not included as one of the independent variables. The results of the univariate logistic regression (Figure 6C) indicated that, compared to females, males (OR = 0.19, 95% CI, 0.03-0.83, P-value = .043) serve as a protective factor against IMEs related to infections and inflammation. Consistent with the results of univariate logistic regression, the multivariate logistic regression results (Figure 6D) indicated that compared to females, males (OR = 0.13, 95% CI, 0.02-0.73, P-value = .029) also act as a protective factor against IMEs associated with infections and inflammation. Overall, females are at a higher risk of experiencing IMEs related to infections and inflammation when receiving combined treatment with antiparkinsonian drugs and APs.

DISCUSSION

Currently, the pathogenesis of PD remains incompletely understood. The prevailing view is that the loss of dopaminergic neurons is a key pathological mechanism involved. Additionally, the accumulation of misfolded proteins is a major characteristic of neurodegenerative diseases, which in the case of PD manifests as the formation of Lewy bodies observed in the brain, spinal cord, and peripheral nervous system.²⁸ The formation of Lewy bodies tends to follow a stereotypical progression, and some studies suggest that the progression of Lewy body pathology corresponds to the staging of PD and specific symptoms.²⁹ The treatment of PD is limited to symptom relief and improvement, specifically addressing classical motor symptoms such as bradykinesia, rigidity, and resting tremor. The main therapeutic strategies involve increasing dopamine concentrations in the brain or enhancing the activation of dopamine receptors. Changes in brain dopamine levels can easily lead to new psychiatric symptoms or disorders, and late-stage PD is also associated with a high risk of comorbid psychiatric illnesses.^30,31 The benefits of using APs in patients with PD arise partly from managing the psychiatric symptoms that occur, and there is a growing consensus on the effectiveness of medications like clozapine in controlling tremor symptoms in Parkinson’s patients.^32,33 Reports indicate that one in 4 PD patients take APs within 5 years,³⁴ and despite only a one-year increase in disease duration, this proportion exceeds half.³⁵ However, while polypharmacy offers therapeutic advantages, it also carries a higher risk of adverse reactions and poor outcomes. Therefore, investigating the current status of ADEs associated with the combined use of antiparkinsonian drugs and APs in a real-world context has profound implications for enhancing and ensuring comprehensive disease management for PD, especially PDP.

In this study, based on the FAERS database, we conducted a real-world pharmacovigilance analysis using 20 years of data on the combination therapy of antiparkinsonian drugs and APs. The research indicates that, despite the use of APs, psychiatric symptoms remain one of the most common ADEs in the treatment of PD. This finding aligns with previous clinical observations and is not surprising, as there is currently no guaranteed safe and effective treatment for psychiatric symptoms associated with PD. Furthermore, the research process revealed that psychiatric symptoms might also lead to IMEs, which warrants attention and management. Hallucinations and delusions are the most common symptoms of PDP,^9,10,36 frequently observed in the late stages of PD, and they are likely induced or exacerbated by antiparkinsonian medications or psychoactive drugs.^17,37

This study also found a significant number of ADEs reports related to infection and inflammation following the use of antiparkinsonian drugs in conjunction with APs, which does not seem to have been previously noted in past studies. We speculate that the emergence of these ADEs may be due to several possible reasons. First, APs are known to potentially interfere with immune system function,³⁸ which may be amplified in PD patients due to various mechanisms and is associated with systemic inflammation. Possible mechanisms include drug-drug interactions and physiological changes associated with aging in patients.^39,40 Second, both antiparkinsonian drugs and APs may affect the composition of gut microbiota, and dysbiosis can alter metabolic responses in the gut, leading to systemic inflammation that spreads from the gut to peripheral blood and may cross the blood-brain barrier to affect the central nervous system.^41,42 On the other hand, changes in gut microbiota composition can influence the production of neurotransmitters such as dopamine, potentially creating new challenges for treatment plans.⁴³ Third, the formulation and administration method of antiparkinsonian drugs may be critical factors in inducing inflammation. For example, the enteral administration of levodopa/carbidopa combination formulations appears to have a higher susceptibility to inflammation compared to traditional oral medications. Our research also identified numerous reports related to infections and inflammation that were closely associated with lesions in the gastrointestinal tract or stoma sites. Additionally, we speculate that the combined use of these 2 classes of drugs may lead to potential metabolic interactions that could affect drug clearance rates and levels of inflammatory mediators, although high-quality evidence for this is still lacking. Moreover, there is evidence suggesting that the immune system plays a significant role in the onset and progression of PD, with autoimmune processes in PD patients potentially continuing to induce neuronal degeneration or inflammation.⁴⁴ Furthermore, APs may indirectly increase inflammatory responses by activating specific immune pathways.

The onset of PD is often insidious, with a progressively developing course. Treatment typically begins only after symptoms become apparent and tends to continue for a long duration. Logistic regression analysis indicates that the duration of combination therapy is associated with the risk of experiencing IMEs and worsened prognosis. Specifically, combination therapy lasting longer than 1 year, especially data from those on treatment for over 2 years, not only suggests poor control of the symptoms but also signals an unfavorable prognosis. Time-to-onset analysis indicates that the median TTO for experiencing ADEs after combination therapy is approximately 22 months. Among all IMEs, we especially focus on those related to infection and inflammation. Univariate and multivariate logistic analyses targeting infection and inflammation-related IMEs show that women have a higher risk compared to men. Although initially the risk of PD in men is twice that in women,^45,46 subsequent studies have shown that disease progression is faster in women, with a higher risk of adverse outcomes such as mortality.⁴⁷ Genomic features reveal some sex-specific aspects of PD pathology, with upregulated genes in female dopaminergic neurons affecting signaling pathways and neuronal maturation.⁴⁸ Analysis of the ratio of the 2 types of dopamine receptors between sexes also shows significant differences,⁴⁹ which may correlate with women’s susceptibility to ADEs. A medication duration exceeding 1 year is a risk factor for all IMEs, though there are no significant differences in infection and inflammation-related IMEs across different drug duration groups. This indicates that long-term medication for chronic diseases can indeed lead to many adverse consequences, necessitating the development of more refined management strategies. It also serves as a reminder to all clinical decision-makers, patients, and external caregivers that vigilance against infections and inflammation must not be relaxed throughout the comprehensive management of chronic diseases.

As this study is an observational study based on real-world data, it still has certain limitations, primarily due to missing data records and numerous unknown biases that cannot be completely avoided through statistical methods. This could lead to significant irreducible errors in the data. Furthermore, factors such as the dosage of drugs, the duration of therapy, and the timing of increasing or decreasing medications in combination therapy play a decisive role in the occurrence of ADEs, but relevant data in the FAERS database are missing. Therefore, we are unable to analyze the roles played by variations in drug doses and types in the occurrence of these severe ADEs. Consequently, we hope that more comprehensive real-world studies and large-sample multicenter randomized controlled trials can address the shortcomings of the current research and provide more reliable conclusions.

CONCLUSION

This study, based on the FAERS database, conducted a 20-year pharmacovigilance analysis of the combination therapy of antiparkinsonian drugs and APs. The results indicate that in the comprehensive management process of the combination therapy of antiparkinsonian drugs and APs, in addition to ADEs such as motor disorders and newly emerging psychiatric symptoms, special attention should also be paid to the long-term risks of infection and inflammation associated with combined medication. We believe that long-term follow-up is necessary and should be integrated throughout the duration of the combination therapy, as it helps clinicians timely adjust medication plans and alleviate the actual burden on family members or caregivers. Furthermore, this issue serves as a wake-up call for current PD treatment strategies; in the future development of antiparkinsonian drugs, it is crucial to focus not only on controlling symptoms and eliminating underlying causes but also on reducing medication-related ADEs that diminish quality of life. This study provides a pharmacovigilance analysis report on combination therapy for PDP, which can help guide clinicians to improve decision-making and optimize management throughout the disease course.

Author Contributions

Junyi Wang (Conceptualization [lead], Data curation [lead], Formal analysis [lead], Methodology [lead], Resources [lead], Software [lead], Validation [lead], Visualization [lead], Writing—original draft [lead], Writing—review & editing [equal]), Sen Lin (Conceptualization [lead], Data curation [lead], Formal analysis [lead], Investigation [lead], Methodology [lead], Resources [lead], Software [lead], Validation [equal], Visualization [lead], Writing—review & editing [equal]), Chen Bai (Software [supporting], Visualization [supporting]), Huimin Zhang (Resources [supporting], Software [supporting], Visualization [supporting]), Haoqi Liu (Resources [supporting], Validation [supporting], Visualization [supporting], Writing—original draft [supporting]), Min Wang (Data curation [supporting], Resources [supporting], Software [supporting], Writing—original draft [supporting]), and Rongjuan Guo (Funding acquisition [lead], Investigation [equal], Project administration [equal], Supervision [equal], Writing—review & editing [equal])

Funding

This study was supported by the Beijing Municipal Health Commission, the Depression Chinese Medicine and Western Medicine collaborative research project(Grant No. 2023BJSZDYNJBXTGG-014). The funding agency was not involved in any aspect of the study design, data collection, data analysis, or manuscript writing.

Conflicts of interest

None declared.

Data availability

The data are available in the FDA Adverse Event Reporting System (https://www.nber.org/research/data/fda-adverse-event-reporting-system).

Ethics approval and consent to participate

The data utilized anonymized data from an open-access database, eliminating the need for ethical scrutiny.

Consent for publication

Not applicable.