Introduction

Early Parkinson’s disease (PD) research principally focused on evaluation and management of motor-based symptoms and impairment, with initial therapy directed towards mitigating motor-dominant symptomology. However, the spectrum of PD has evolved, with neurobehavioral features now recognized as major contributors to morbidity and mortality. Psychosis in PD is increasingly recognized as a distinct clinical manifestation associated with disease progression, dementia, and PD medications. Until recently, not only has its diagnosis been poorly codified, but also pathophysiologic understanding and effective symptom management have remained elusive.

Understanding of PD psychosis (PDP) has focused on its intrinsic pathology distinct from neurobehavioral complications of PD medications, with substantial focus directed towards complex therapeutic management involving both pharmacologic and non-pharmacologic approaches. Until recently, no Food and Drug Administration (FDA)-approved drugs had been available for PDP, although some atypical antipsychotics have found off-label use. Significant limitations are variably seen with these medications, including worsening of motor symptoms, sedation, and other systemic complications. The FDA also requires a “black box warning” when used in elderly patients secondary to increased morbidity risks. Recently, pimavanserin, a selective 5-HT2A receptor inverse agonist with minimal D2 receptor activity, gained FDA approval for treatment of PDP. We herein provide a brief review of PDP and will discuss the recent literature and findings regarding management of disease.

Definitions and Epidemiology

Definitions

Historical definitions have ranged widely from “perceptions without stimulus,” to Diagnostic and Statistical Manual of Statistical Disorders (DSM-IV-TR) standardization of “abnormal perceptions without a physical stimulus involving any sensory modality” [1]. As a response, the first standardized literature-based review of distinguishing criteria for defining PDP was proposed by a NIH-sponsored work group (NINDS-NIMH) in 2007 [2••]. The review provided definitions of psychotic symptoms as specific to PD, including delusions, hallucinations, a sense of presence, and illusions. These were incorporated into overall diagnostic criteria, both of which are outlined in Table 1.

Table 1 Definitions of behavioral phenomena and Diagnostic criteria for PDP, as proposed by NINDS/NIMH Group

The semiology of hallucinations in PD is more distinct than those secondary to schizophrenia. Visual hallucinations (VH) are by far the most common psychotic feature experienced in patients [1, 3•, 4]. They are generally well-formed people or animals, stereotyped, and reoccurring in low-stimulation environments, usually at home [2••]. Non-visual hallucinations are less frequent and are often experienced concurrently with VH rather than in isolation [5]. Insight may or may not be maintained. These fully formed VH differ from those in schizophrenia or delirium, wherein hallucinations may be mixed and insight frequently lost. Delusions in PDP are more commonly paranoid and related to infidelity or persecution. In contrast, religious, grandiose, and other delusion types are far less frequent, accounting for less than 10% [3•, 5, 6]. A sense of presence and visual illusions are referred to as “minor” hallucinations secondary to their decreased likelihood of being harmful or prolonged, although these may progress to clinically disabling hallucinations and delusions [7•, 8, 9]. The psychosis in Lewy body dementia very much resembles PDP, and there is debate whether these should be segregated.

Prevalence

Early characterizations of psychosis in PD lacked uniformity, not only determining prevalence, but also defining psychosis itself. Preliminary definitions of psychosis in PD commonly were inclusive of VH and delusions, with clear sensorium variably taken into context with disease progression. Thus, various studies reported prevalence of PDP in PD ranging from 6 to 40%, with differing study designs, population selection, and inclusion or exclusion of dementia, contributing to variability [10,11,12,13]. One longitudinal study showed that the majority of PD patients will eventually develop PDP [14••]. Overall, it is a very common feature.

Risk Factors

Multiple independent risk factors have been identified with the development of psychosis. Cognitive impairment and dementia have shown strong consistent associations with psychosis in longitudinal studies and are probably the most robust independent risk factor for development of psychotic symptoms [5, 8, 15,16,17]. Conversely, studies have shown that PD patients with earlier onset of VH may demonstrate more precipitous progression to dementia, with recurrent and persistent hallucinations precipitating loss of insight [18]. Older age has shown a correlation with PDP [1, 19], although the duration of PD may actually confer a stronger risk [4, 8]. Sleep disturbances have shown associations with psychosis in PD patients. Longitudinal studies have identified increased association for VH in patients with REM behavioral disorder (RBD) [14••], with more severe RBD showing correlation to psychotic symptoms in PD patients without dementia [20]. In general, more severe PD, especially midline features (balance and bulbar symptoms) and other non-motor features, associate with PDP.

Genetic PD types have not demonstrated clear association with psychosis; parkin, PTEN-induced putative kinase 1 (Pink1), α-synuclein (SCNA), and leucine-rich repeat kinase 2 (LRRK2) have not shown associations with PDP [21].

Caregiver Burden

The onset and progression of psychosis in PD affects not only the patient, but also family members and caregivers, as manifested by nursing home placement. An early case-control study in 1995 identified VH and delusions as the only significant risk factor for nursing home placement over 4 years [22], and a 2-year follow-up review found a 100% mortality rate among these nursing-home patients compared with 32% of community-based patients [23]. The results underscore the severity with which psychosis correlates with the disease state. The presence and involvement of a caregiver is associated with improved patient morbidity, mortality, and quality of life, although the stress and psychosocial burden associated with long-term care often leads to patient institutionalization [24•]. Caregivers themselves are impacted, with associated increased risks for chronic illnesses, depression, and mortality [25,26,27]. Recently, a palliative care approach has demonstrated increased merit in mediating this burden [28•].

Pathophysiology

Dopaminergic System

Early understanding of PDP linked dopaminergic systems to pathogenesis. Initial proposals hypothesized a pharmacologic “kindling” model of overstimulation and hypersensitivity of mesolimbic and mesocortical post-synaptic D2 receptors secondary to chronic levodopa treatment as disposing to psychosis [29]. However, no qualified relationship has been demonstrated between levodopa dose and PD patients with and without hallucinations, and hallucinations prior to levodopa exposure are observed [1, 8, 30]. While reduction of dopaminergic therapy remains a practical part of management of PDP, current understanding of PDP implicates multiple neurotransmitter systems. Furthermore, dopaminergic drugs can stimulate non-dopaminergic receptors.

Cholinergic System

In PD dementia (PDD), degeneration of cholinergic centers including the nucleus basalis of Meynert (NBM) and pedunculopontine nucleus (PPN) is observed; concurrent disruption of ascending cholinergic transmitter systems results in pronounced frontal cortical denervation causing impaired cognition and attention [31•, 32, 33]. In non-demented PD patients with VH, evidence has suggested that impaired cholinergic systems contribute to impaired visual processing related to pedunculopontine (PPN) degeneration, compared to those without VH [31•]. Pharmacologically, withdrawal of anticholinergic agents lessens hallucinations, and cholinesterase inhibitors may mildly improve psychosis features of PDD [34].

Serotoninergic System

Serotoninergic systems are also impaired in PD, with loss of neurons in the pontine median raphe nucleus and diminution of projections to the frontal and temporal cortices and putamen [35]. Serotonin content has been shown to be reduced to about 50% in the caudate and substantia nigra in PD patients [36]. 5-HT2A receptor profiles are also altered in PDP subjects. Compared to PD patients without psychosis, 5-HT2A receptor radio-ligand binding during position emission tomography (PET) in PD patients with VH has shown increased 5-HT2A binding in the ventral visual pathway including the bilateral infero-occipital gyrus, fusiform gyrus, and inferolateral temporal cortex [37•]. This could be interpreted as either reduced endogenous neuropeptide or relative upregulation of receptors in PDP patients. Postmortem tissue studies have shown increased 5-HT2A receptor levels in the inferolateral temporal cortex in PD patients with VH [38]. Visual input and processing in PDP patients may thus be altered through 5-HT2A receptor functioning.

Rating Scales

Assessment of PDP has been limited secondary to lack of a universally accepted and validated rating scales. Some scales were adopted from non-PD conditions and may not reflect PDP phenotype. As a result, in 2008, the Movement Disorders Society (MDS) Task Force on Rating Scales in PD provided critique and recommendations on the utility of available scales [39••]. Twelve scales were assessed for encapsulating PDP phenomena and clinical response over time.

Of these, the Brief Psychiatric Rating Scale (BPRS) [40], Neuropsychiatric Inventory (NPI) [41], Positive and Negative Syndrome Scale (PANSS) [42], and Scale for Assessment of Positive and Negative Symptoms (SAPS) [43] met criteria for being “recommended,” while the Parkinson Psychosis Rating Scale (PPRS) [44], Parkinson Psychosis Questionnaire (PPQ) [45], Behavioral Pathology in Alzheimer’s disease Rating Scale (Behave-AD) [46], and Clinical Global Impression Scale (CGI) [47] were classified as “suggested.” Despite these evaluations, none of the above scales were considered to be comprehensive of psychotic symptoms in PDP, with the task force ultimately recommending the creation of a disease-specific scale. Newer scales have been published since that time. The SAPS-PD is a modified form of the SAPS scale used in schizophrenia and evaluates for hallucination- and delusion-type symptoms more typical for PDP; it was utilized among other rating scales in the initial studies on pimavanserin [48, 49]. Recently, Ondo et al. devised a novel scale with appreciable content validity and good inter- and intra-rater reliability, although it has yet to be validated against other scales [50]. Although scales are continuing to be developed, no universally validated and accepted tool exists.

Management

The decision on how aggressively to treat PDP depends on symptom severity and how it affects quality of life. Minor hallucinations may not need treatment, although they can quickly escalate in some cases. While pharmacologic management of PDP is best studied in well controlled trials, non-pharmacologic approaches remain an important initial intervention. The first step is to assess for any reversible metabolic, infectious, or other medical problems. Minor infections can often trigger or worsen PDP. A review of the patient’s active medications should be conducted, with attention first directed towards primary and secondarily psycho-modulatory medications distinct from PD medications. Antidepressants, sedatives, and narcotics, among others, may contribute to psychosis and should be reduced or discontinued if possible. Agents with anticholinergic properties, including tricyclic antidepressants, antispasmodics for bladder control, and over the counter antihistamines, warrant consideration for removal.

Following modification of non-PD based medications, attention can be directed towards adjustment of Parkinsonian medications, if plausible, based upon the patient’s motor symptomology. Previous literature has described an algorithmic approach of withdrawal of these agents, beginning with anticholinergics, followed by amantadine, monoamine oxidase B (MAO-B) inhibitors, dopamine agonists, catechol-O-methyltransferase (COMT) inhibitors, and finally levodopa [51•, 52•]. This approach serves the intention of first removing agents with greater anecdotal tendency to induce psychosis while minimizing compromise to motor function. Ultimately, any reduction in antiparkinsonian agents should be monitored for worsening motor disability. While “drug holidays” were initially implemented in times wherein levodopa-induced psychosis was the predominant proposed mechanism, this method of management was quickly abandoned upon recognition of the significant morbidity in which it resulted.

Pharmacologic Management

The introduction of pharmacologic agents for management of PDP is necessitated when both non-pharmacologic evaluations and dose adjustments of antiparkinsonian drugs do not sufficiently reduce psychoactive symptoms without worsening of motor symptoms. Earlier use of antipsychotic medications for mild hallucinations has been shown to reduce risk of subsequent deterioration [53]. However, all first-generation antipsychotics block dopamine-2 (D2) receptors, which will worsen PD motor features, especially gait and balance. This may occur within a single dose or be delayed [52•]. With the introduction of second generation antipsychotics, which possess relatively less affinity for D2 receptors, an alternative therapy showed promise to ameliorate psychosis without worsening motor function.

Clozapine

Clozapine was the first introduced “atypical” antipsychotic to be utilized for schizophrenia, although it was subsequently withdrawn from commercial use in Europe in 1977 following significant mortality secondary to agranulocytosis-related infections. It has minimal D2 affinity. Clozapine was first reported to improve PDP in a four-patient open-label pilot study with varying doses between 25 and 100 mg per day [54]. Multiple studies have since been published supporting the tolerability and efficacy of low dose clozapine for PDP [55,56,57,58,59]. Some reports also demonstrate improvement in motor features, especially tremor and dyskinesia [60, 61]. In a review of the earliest published studies assessing the efficacy of clozapine in PDP in over 300 patients, Factor et al. reported at least partial resolution of psychotic symptoms in greater than 80% of patients without worsening of motor symptoms, with doses ranging from 3.25 to 400 mg/day [13]. Agranulocytosis was reported in only one patient, and worsening of motor features in five, most of which were treated with doses greater than 100 mg/day. Although all initial reports were limited as open-label studies, results were sufficient to demonstrate improvement in symptoms with doses as low as 6.25 mg/day [62]. Of note, the first double-blind, placebo-controlled trial conversely showed exacerbation of parkinsonian symptoms with concurrent worsening of psychotic features in some patients with mean dose 170.8 mg/day (range 75–250 mg/day). The authors related adverse events to the substantially higher doses [63]. Two independent multi-center, double-blind, placebo-controlled trials subsequently evaluated lower doses of clozapine. The Parkinson Study Group US trial demonstrated significant improvement in psychosis outcomes over a 4-week study period assessed with CGI, BPRS, and SAPS rating scales, using doses 6.25–25 mg/day [64•]. The French Clozapine Parkinson Study Group used a comparative study protocol, although with CGI and PANSS scales, and reported similarly good efficacy outcomes compared to placebo, with dose range of 6.25–50 mg/day [65•]. Both studies were followed by a 12-week open-label extension with sustained benefit. Of note, the French study included a 4-week washout period which demonstrated recurrence of psychotic symptoms after clozapine weaning. Leukopenia was observed in one patient in the US group but resolved with clozapine withdrawal. Although the open-label period observed an increased rate of mortality (6/60 patients), cause of death was not attributed to study drug. The most common side effects include sedation, hypotension, and sialorrhea. The risk of agranulocytosis (1–2%) requires white blood count (WBC) monitoring and enrollment into special registries while on clozapine, which remains a limiting factor to greater use. In the USA, weekly WBC monitoring is required for the first 6 months, then bimonthly for 6 months, and then monthly indefinitely [66]. As was concluded in the 2002 review, the 2011 MDS report carried forward its previous designation that clozapine be rated as efficacious and clinically useful, with the American Academy of Neurology (ANN) Practice Parameter concluding clozapine should be considered for PDP with level B evidence, both noting the safety issues necessitating WBC monitoring [67, 68••, 69••].

Quetiapine

Quetiapine, a dibenzothiazepine derivative with similar structural characteristics to clozapine, shows comparatively low D2 receptor blocking activity and similar greater potency for 5HT2A receptor antagonism [70]. In schizophrenia trials, positive symptoms improved without motor side effects [71]. Multiple open-label studies have evaluated short- and long-term effectiveness of quetiapine in PDP, with doses ranging from 25 to 800 mg/day, identifying improvement in psychotic symptoms without significant worsening of extrapyramidal symptoms [72,73,74,75,76]. Better outcomes were achieved at lower doses, with motor symptoms more prone to worsen in patients with progressive dementia. Open-label comparison studies between quetiapine and clozapine show non-inferiority, with the additional benefit of lack of risk of agranulocytosis in quetiapine [77, 78]. However, four randomized controlled trials have all failed to show superiority over placebo, although the drug was well tolerated and did not worsen motor features [79,80,81,82]. Ondo et al. assessed quetiapine up to 200 mg/day in 31 PD patients with PDP and found no statistically significant difference in either psychosis measured by BPRS and the Baylor PD Hallucination Questionnaire. Motor examination was unchanged. The study may have been underpowered to achieve results, with authors further commenting that the short study period could potentially affect results [80•]. Rabey et al. similarly showed good tolerability and lack of motor worsening but did not demonstrate efficacy of quetiapine with a mean dose 120 mg/day in 58 patients measured by BPRS and CGI scores. Results were limited by a high dropout rate due to lack of efficacy [79]. The only double-blind randomized trial to show some efficacy of quetiapine found significant reductions in BPRS hallucination scores in 16 PD patients with VH, although the study was actually designed with changes in sleep architecture as the primary endpoint and was complicated by small sample size [81]. Based upon the inconsistent efficacy demonstrated, the updated MDS report concluded that there was “insufficient evidence” for use of quetiapine in treatment of PDP, although reported “acceptable risk” is without need for specialized monitoring regarding its safety profile [68••]. The AAN Practice Parameter reports that quetiapine may be considered for use with level C evidence [69••]. Sedation, potentially desirable, and orthostatic hypotension are the most common limiting side effects. Doses over 200 mg can start to impact motor features.

Olanzapine

Olanzapine is a thienobenzodiazepine with structurally similar characteristics to clozapine, with higher affinity for 5-HT2A receptors than D2 receptors. However, olanzapine does possess much greater D2 receptor affinity than clozapine or quetiapine [83]. It has shown antipsychotic efficacy and tolerability in not only schizophrenia and bipolar disorder [84, 85], but also Alzheimer’s disease, and in an open-label study with dementia with Lewy bodies without major side effects [86, 87]. Wolters et al. first studied olanzapine in 15 non-demented PD patients with psychosis and found clinically significant reductions in psychosis over 2 months as measured by the BPRS score, with a mean dose of 6.5 mg/day [63]. This open-label study showed good tolerance of olanzapine and did not show motor deterioration. However, a number of succeeding open-label studies, retrospective reviews, and case reports have reported clinically significant worsening of motor function [88, 89]. Three RCTs have been performed confirming these findings [90, 91]. Breier et al. conducted two simultaneous placebo-controlled studies, one in the USA and the other in Europe, identifying improvement in both olanzapine and placebo groups across multiple psychosis rating scales, but no significant treatment-arm difference. Motor function worsened significantly in the olanzapine group compared to placebo. Ondo et al. conducted a smaller placebo-controlled study with similar results. Motor worsening usually began gradually after several weeks, and most impacted gait, posture, and balance. The MDS report labeled olanzapine “unlikely efficacious” with an “unacceptable risk” safety profile secondary to extrapyramidal deterioration [68••], with level B evidence of lack of efficacy based upon the above studies. The AAN Practice Parameter concluded that olanzapine should not be routinely considered for PDP [69••].

Pimavanserin

The implication that 5-HT2A receptor stimulation is associated with psychosis is corroborated by evidence demonstrating 5-HT2A receptor blockage as important to the antipsychotic mechanism of atypical antipsychotics. Pimavanserin was subsequently developed as a novel very specific 5-HT2A/C receptor inverse agonist with no activity on dopamine or other G-protein coupled receptors [92]. Initial studies showed tolerability and safety in non-demented PD patients with doses of either 25–100 mg/day. The half-life of 55–60 h suggests that steady state may not be achieved until 10–14 days after initiation, potentially delaying clinical effect [49]. Four RCTs have been conducted comparing pimavanserin to placebo, but only two of four were published. A phase II RCT of 60 patients confirmed safety and tolerability of doses between 25 and 60 mg/day without worsening of motor symptoms; treatment-arm patients demonstrated a trend towards improvement in psychosis, but statistical significance was not achieved across all measured assessments, possibly due to a small sample size, shortened evaluation period, and non-specific assessment measures (SAPS) [93••]. Two unpublished phase IIb/III trials showed large placebo responses and failed to meet their primary efficacy point [94•]. Consequently, a larger phase III multi-center RCT was conducted with 199 patients comparing pimavanserin 40 mg/day to placebo over 6 weeks. The study design was modified from the previous phase II trials to minimize placebo response. This study employed a placebo run-in prior to randomization in which brief psychosocial therapy was offered, and a modified version of the SAPS scale with focus on hallucinations and delusions (SAPS-PD) was administered by central raters via video interview. The SAPS-PD and CGI scores significantly improved with good tolerance and no decline of motor dysfunction [95••]. Both sensory hallucinations and delusions improved equally. Secondary measures assessing sleep and cognition also improved. In 2016, pimavanserin became the first FDA-approved designated antipsychotic for use in PDP. A meta-analysis of all four RCTs assessed a total of 680 patients with PDP, confirming significant efficacy in reduction in psychosis as measured by the SAPS hallucinations and delusions scales (SAP-H and SAP-D, respectively) [96].

Other Antipsychotics and Drug Classes

A number of other atypical antipsychotics and non-antipsychotic class drugs have been evaluated for psychotic symptoms in PD, although because of variable efficacy and extent of motor worsening, they are not recommended for standard use.

Risperidone is another antipsychotic structurally distinct from clozapine, with combined 5-HT2 and D2 receptor antagonism. Although generally labeled as an atypical antipsychotic, it carries a dopamine receptor binding profile similar to typical antipsychotics [97]. Initial open-label small studies in PDP identified efficacy at low doses, although subsequent reviews and follow-up studies demonstrated clinically significant worsening of motor symptoms in a dose-dependent manner [98, 99].

Ziprasidone is an atypical antipsychotic with a high 5-HT2A/D2 receptor-antagonism affinity ratio, initially suggestive of less risk for extrapyramidal side effects [100]. No double-blind RCT has been performed in PDP patients. However, psychiatric-based review of ziprasidone use in schizophrenia has demonstrated worsening motor function [101]. Smaller case series in PDP have not shown as significant results [102]. Larger randomized, placebo-controlled studies have not been conducted.

Aripiprazole showed a promising pharmacologic profile of D2 and 5-HT1 receptor “modulation” with mixed agonist/antagonist properties and a high 5-HT2A/D2 receptor affinity ratio [103]. Several case reports and open-label trials have shown mixed efficacy in improving psychosis but were complicated by a high risk of motor complications [104,105,106]. Plans for controlled trials were abandoned.

Melperone is another atypical antipsychotic with higher 5-HT2A receptor affinity used in management of schizophrenia. An open-label study was published in 1996 showing improvement in BPRS scores in 30 patients with PDP, without worsening of motor symptoms [107]. Results of an unpublished double-blind, placebo-controlled trial showed lack of efficacy as measured by the SAPS scale, although no significant adverse effects were noted [108].

Cholinesterase inhibitors have been evaluated for efficacy in demented PD patients, some of whom had psychosis. In these trials, psychotic symptoms were predominantly measured as a secondary outcome or as an adverse event. While reports have not demonstrated compromise of motor functioning, results have shown possible reduction for VH but not delusions. In this regard, rivastigmine has shown more consistent improvement [109, 110] compared with donepezil [111].

SEP-363856, a 5-HT1A antagonist and trace amine-associated 1 (TAAR1) receptor antagonist, is currently being studied for PDP, as well as schizophrenia.

Conclusion

Psychosis in PD is an increasingly recognized morbid complication of disease progression that will eventual impact the majority of patients. Characterizations of problematic psychotic features are now well defined and fairly unique, mostly visual hallucinations and persecution delusions. Pathophysiological studies reveal complex mechanisms and implicate dysfunction of multiple neurotransmitter systems, particularly in relation to visual processing, but exact pathophysiology is not known.

While initial management may be directed to non-pharmacologic approaches, drug therapy is often ultimately warranted. Targeting of the serotoninergic system has shown benefit for psychotic features, with minimization of dopaminergic antagonism a desired quality in novel therapeutics. Clozapine has demonstrated efficacy in well-designed, randomized, placebo-controlled studies and is deemed efficacious at lower doses. Despite this, frequent blood-count monitoring due to risk of agranulocytosis has limited its use. Quetiapine has not shown benefit in controlled trials but has been frequently used secondary to more benign adverse effects and ease of use. Both have minimal risk of worsening motor features secondary to dopamine receptor antagonism.

Pimavanserin benefits from a highly selective 5-HT2A/2C receptor activity. It has demonstrated efficacy in randomized, placebo-controlled trials without worsening of motor function and is currently the only FDA approved drug for PDP. However, its long-term efficacy remains to be determined.

While new therapeutics and targets continue to be investigated, a more complete understanding of PDP pathology is needed to further refine drug targets. Furthermore, trials evaluating current and novel drugs are limited secondary to a lack of a validated rating scale. Ultimately, investigation into novel agents will require exploration of not only selective receptor targets, but also a unified approach to the clinical evaluation of PDP itself.