Introduction

Delusions in Parkinson’s disease (PD) are associated with a significant impact on quality of life and an increased risk of carer burden, nursing home placement and hospitalisation (Aarsland et al. 1999). Although they are less common than hallucinations in PD, with an estimated prevalence of between 5 and 16%, delusions almost always require antipsychotic treatment (Kiziltan et al. 2007; Lee and Weintraub 2012; Factor et al. 2014). This treatment necessitates the delicate and complex balance between control of psychosis with parkinsonian side effects.

The current conceptualisation of PD psychosis, as defined by the National Institute of Neurological Disorders and Stroke and the National Institute of Mental Health, is a continuum where delusions occur through a progression of delusions and hallucinations rather than distinct symptom classes (Ffytche et al. 2017a; Ravina et al. 2007; Moskovitz et al. 1978). This progression occurs with cognitive decline and advancement of PD (Ffytche et al. 2017a). Isolated delusions, those without the co-occurrence of hallucinations, have been considered rare (Mack et al. 2012). However, this is not a consistent finding (Kiziltan et al. 2007; Stefanis et al. 2010). Additionally, specific PD subpopulations, such as those with dopamine dysregulation syndrome, have demonstrated elevated rates of isolated delusions (Cilia et al. 2014; Warren et al. 2017).

Identified risk factors for PD psychosis include severity and duration of PD, increasing age, cognitive impairment, REM sleep behaviour disorder, depression, and visual processing deficits (Aarsland et al. 1999; Ffytche et al. 2017b; Forsaa et al. 2010; Holroyd et al. 2001; Lee and Weintraub 2012; Mack et al. 2012; Pacchetti et al. 2005). The above model and risk factors are based upon research that focused solely on hallucinations or grouped delusions with hallucinations. The risk factors for isolated delusions are unknown.

We propose that the model for PD psychosis is incomplete and it is our hypothesis that delusions can occur without hallucinations or cognitive impairment. Additionally, we speculate that the content of delusions and the presence of perceptual phenomena vary with age and in certain sub-populations of PD, such as those with early onset PD. To further clarify the nature of delusions in PD, we conducted a systematic review of all published cases. We aimed to identify: 1) the spectrum of clinical presentations, 2) evaluation of known PD psychosis risk factors, 3) co-morbidities, and 4) effective treatment strategies.

Methods

The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) statement recommendations were followed (see Fig. 1 Moher et al. 2009). The study was pre-registered with PROSPERO: CRD 42017076792. Cases were identified by search of PubMed, Embase and PsychInfo (Inception to June 2017) with search terms: Parkinson’s, psychosis, delusions, paranoid or paranoia, Othello, jealousy, misidentification, Capgras. Search and screening was conducted independently by two authors (NW, ZH). Cases were accepted from any source including poster abstracts, case reports, cohort or case-control studies. However, if individual data for each case was not available in the cohort or case-control studies those cases were not included. References of selected articles were checked to identify other eligible studies. All languages were included.

Fig. 1
figure 1

PRISMA 2009 Flow Diagram

Full size image

Cases were included if there was documentation of both a delusion and idiopathic Parkinson’s disease. Cases with cognitive impairment or Parkinson’s disease dementia were included. However, with concern for potential diagnostic confusion, cases with a primary diagnosis of Lewy Body Dementia (DLB) or those with symptoms consistent with the international consensus guidelines for DLB were excluded (McKeith et al. 2005). Other exclusion criteria were: psychosis without delusions, presence of delirium, prior diagnoses of schizophrenia or other psychotic disorders. Case selection was made independently by two authors (NW, CO) with discrepancies discussed with all authors.

Data collected included: Demographics (gender, age), features of PD (age of PD diagnosis, levodopa equivalent daily dose (LEDD), dopamine agonists use, dyskinesia, visual processing deficits), features of psychosis (psychiatric history, age of delusion commencement, suspected trigger for delusion, type of delusion, hallucination presence and type, agitation), comorbid disorders (cognitive impairment, impulse control disorders, dopamine dysregulation syndrome, REM sleep behaviour disorder, anxiety, depression) as well as treatment use and outcome of psychosis management. Levodopa equivalent daily dose, if not supplied, was calculated using a standard formula (Tomlinson et al. 2010). Any mention of “dementia,” “severe cognitive impairment,” or a MMSE <24 was considered a ‘cognitive impairment’ (Folstein et al. 1975). If there was no documentation as to whether the variables were either present or not they were considered absent.

Data was analysed using descriptive statistics. Comparisons were made by Student’s t-test and Fisher’s exact test for continuous and nominal variables respectively. Effect sizes were calculated using Cohen’s d and effect-size correlation, rYl for the continuous variables, and odds ratios (OR) for the nominal variables (Cohen 1988). The Bonferroni correction was applied for multiple comparisons, giving p < 0.00217 for significance (Miller 1966).

Results

A total of 633 non-duplicated articles were identified in the databases. Following full text screening, 119 studies with 184 cases of delusions in PD were identified, with publication date ranges from 1976 to 2017 (see Fig. 1 and online supplementary material). There were 114 cases from individual case reports, 63 from case series, and 7 from case control studies.

There were 121 males (65.8%) and 63 females. The mean age of onset of PD was 51 years (SD = 11.8) and duration of PD was 9.9 years (SD 6.1), with 91 cases (49.5%) classified as early onset PD (onset before 50 years: Pagano et al. 2016). Dyskinesia or significant motor fluctuation was reported in 57 cases (31%). The Mean LEDD was 1116.7 mg (SD = 1022.3). Dopamine agonists were used in 115 cases (62.5%). Cognitive impairment was present in 33 cases (17.9%) but was not specifically excluded in 70 cases.

Delusions were predominantly paranoid in nature (152 cases, 82.6%) with themes of persecution (117 cases) and delusional jealousy (44 cases). Misidentification syndromes also occurred (21 cases, 11.4%), with Capgras (17 cases), reduplication (3 cases) and Fregoli (1 case). Other delusional themes included: grandiosity (7 cases), delusions of reference (7 cases), infestation (6 cases), nihilism (3 cases), guilt (2 cases), somatic concerns (2 cases) and religion (1 case). There was more than one delusional type in 28 cases, and unspecified in six. The onset of delusions was reported to be related to a change in dopaminergic treatment in 53 cases (28.8%) and secondary to deep brain stimulation in 10 (5.4%). With increasing age, the frequency of misidentification syndromes increased, delusional jealousy decreased, and persecutory delusions remained stable (see Fig. 2). Of the 91 cases with early onset PD, only one clearly had a misidentification syndrome (Islam et al. 2015).

Fig. 2
figure 2

Frequency of delusion type across age-range of cases

Full size image

Hallucinations were reported in 92 cases (50%): visual (45 cases), auditory (14 cases), both visual and auditory (24 cases), tactile (3 cases), olfactory (1 case), and unspecified (7 cases). Minor hallucinations, such as presence phenomena (2 cases) and illusions (1 case) were noted rarely. The presence or absence of hallucinations was not recorded in 21 cases. When documented, anxiety (43/59 cases) and depression (47/70 cases) were commonly comorbid, although in many of these cases it was unclear if this disturbance was historical or concurrent. Significant agitation or aggression was reported in 61 cases. Impulse control disorders were noted in 45 cases (24.5%), most frequently hyper-sexuality (29 cases), pathological gambling (11 cases) or compulsive shopping (8 cases), less frequently punding (4 cases) and binge eating (3 cases). Dopamine dysregulation syndrome was noted in 37 cases (20.1%). The presence of REM sleep behaviour was documented in only nine cases.

When comparing the group of isolated delusions (n = 87) with those with both delusions and hallucinations (n = 97) there were some statistically significant differences (see Table 1). The cases with isolated delusions were more likely to be younger, with earlier onset of PD. Impulse control disorders I (OR = 5.90) and dopamine dysregulation syndrome (OR = 3.35) were more common in those with isolated delusions. Cognitive impairment was seen less frequently in the cases with isolated delusions (OR = 0.25) than in those with both delusions and hallucinations. There was no statistically significant difference between groups for anxiety, agitation or depression.

Table 1 Cases of delusions in Parkinson’s disease
Full size table

Since there was a significant difference between groups for age of onset, and the average age of onset for all cases was close to the defined limit for early-onset PD; a second comparison was performed, comparing all early-onset cases with all others, using the same variables (see Table 2).

Table 2 Comparison of groups by age of onset
Full size table

In this comparison, the age of onset data were unavailable for 8 cases. Cases of early onset PD represented >50% of the sample. The main findings of note were significantly higher frequency of impulse control disorders (OR = 7.07) and dopamine dysregulation syndrome (OR = 3.99) in the early onset cases. There was a lower rate of cognitive impairment in the early onset cases (9.1% compared to 26.7%, OR 0.27) but this result did not reach statistical significance (p = 0.0028).

Most cases either improved (93 cases, 50.5%) or resolved (35 cases, 19%) with treatment, however, many (49.7%) required more than one treatment strategy. Using anti-psychotics (124 cases) or reducing dopamine replacement therapy (102 cases) were the primary treatment options. The most commonly used anti-psychotics were clozapine (61 cases), quetiapine (59 cases), olanzapine (18 cases) and aripiprazole (7 cases). Associated anti-psychotic induced motor impairment was frequently noted and there was one death related to quetiapine induced neuroleptic malignant syndrome (Schattner et al. 2016). Electro-convulsive therapy was used in 11 cases and cholinesterase inhibitors in three. Deep brain stimulation was used successfully as a treatment in three cases. In the majority of cases it was unclear which treatment strategy lead to improvement or resolution of psychosis.

Discussion

The gender distribution in this study reflects published PD cohorts. The age of onset of PD is lower than would be expected in a clinical cohort of PD, reflecting the high proportion of early onset PD cases in this series (Pringsheim et al. 2014). Excessive dopamine replacement therapy was not used in most cases, although psychosis often followed a change or increase in dopaminergic medications. Delusions were primarily paranoid in nature with common themes such as delusional jealousy perhaps reflecting stage of life and environment. Misidentification syndromes, which are uncommon in primary psychiatric populations, occurred more frequently with older age and cognitive impairment (Cummings 1985). Hallucinations only occurred in around half of the cases and were again more common with age and cognitive impairment. Interestingly, a similar picture of difference in perceptual phenomena and type of delusions with increased age and cognitive impairment is seen in Alzheimer’s disease (Ismail et al. 2011).

Those with delusional PD psychosis appear more likely to have impulse control disorders, dopamine dysregulation syndrome and possibly mood and anxiety disturbance, compared to published cohorts of PD (Barone et al. 2009; Vazquez et al. 1993; Cilia et al. 2014; Weintraub et al. 2010). The association with co-morbid disorders was further increased when comparing cases of isolated delusions to those who had both delusions and hallucinations. The group with isolated delusions had lower age of onset and was less likely to have cognitive impairment. These findings were similar when cases were compared by age of onset, perhaps suggesting that the underlying differences between the groups may be influenced by age of onset.

Cognitive impairment has been considered a necessary component in the development of PD psychosis (Ffytche et al. 2017a). The rate of impairment was higher in cases with combined hallucinations and delusions than in cases with isolated delusions. We speculate that cognitive impairment may be less important for delusion formation than for hallucinations in PD psychosis.

A hypothesis derived from schizophrenia research suggests that delusion formation occurs through the aberrant signalling of salience to environmental stimuli and their mental representations, mediated by sensitised meso-limbic dopaminergic neurons (Kapur 2003). Chronic use of dopaminergic medication has been shown to sensitise meso-limbic neurons (Moskovitz et al. 1978). This sensitisation may be particularly important for those with early onset PD, in whom disease progression is slower with a greater subsequent sparing of the ventral tegmental area and the projections to the ventral striatum (Wickremaratchi et al. 2009; Pagano et al. 2016; Kish et al. 1988). A relative and sensitised mesolimbic predominance of dopaminergic neurons is also thought to be associated with the development of impulse control disorders, dopamine dysregulation syndrome and mood or anxiety fluctuations (Brewer and Potenza 2008; Eskow Jaunarajs et al. 2011; Linazasoro 2009).

Distorted ‘top down’ cognitive explanations of this aberrant salience may also be required for delusion formation (Poletti and Bonuccelli 2013). Mood and anxiety disturbance, as seen frequently in cases of isolated delusions, may alter or impair the cognitive explanation (Morgante et al. 2012). There is a suggestion that those with PD, especially with impulse control disorders, have an increased rate of ‘jumping to conclusions,’ a cognitive characteristic that may lead to early acceptance of delusional explanations (Djamshidian et al. 2012).

It is important to note that the above are hypotheses that require further development and correlation with clinical data. Furthermore, there are a number of limitations inherent to the methodology of this study that should be considered. Case reports and case series are subject to reporting and publication bias. Inevitably, potentially useful data are omitted from reports, such as ethnicity, presence of REM sleep behaviour disorder and visual processing deficits, that were rarely mentioned. Available data may be incomplete, such as details regarding perceptual phenomena or detailed assessment of cognitive function. For this analysis, continuous variables (e.g., cognitive impairment) were dichotomised to permit analysis in the absence of more complete data. Similarly, omission of a reported symptom (e.g., impulse control disorder, hallucination) was taken as equivalent to definite absence for the purpose of analysis. These simplifications were necessary given the source material, but potentially limit clinically application of our findings. The aim of this study was primarily descriptive, given the existing knowledge in this area is limited.

This is the first systematic review of delusions in PD psychosis and our findings suggest that the current conceptualisation of PD psychosis does not account for all cases and perhaps is an over simplification. There are possible implications for screening, diagnosis and treatment. Prospective studies, with careful examination of psychiatric phenomena, and assessment of other relevant risk factors, such as cognitive impairment, REM sleep behaviour disturbance and processing deficits may help guide clinicians in targeting treatment and other interventions. In particular, advanced imaging modalities including functional MRI (fMRI), diffusion tensor imaging (DTI) and single-photon emission computed tomography (SPECT) have recently been applied to evaluation of the meso-limbic circuit affected in Parkinson’s disease (Sharman et al. 2013; Son et al. 2016). Combining these techniques with specific examination of psychosis in early onset PD may allow expansion and testing of hypotheses generated by this study.