Quetiapine in the treatment of psychosis in Parkinson’s disease

Abstract

Psychosis (delusions and/or hallucinations) is a common nonmotor feature of Parkinson’s disease (PD). Use of the older ‘typical’ antipsychotic drugs led to worsening of motor symptoms. The introduction of ‘atypical’ antipsychotics opened up a range of therapeutic options. These agents include clozapine, risperidone, olanzapine, aripiprazole and quetiapine. All have been used to treat psychosis in PD with varying success. Clozapine is the only drug with proven efficacy. We review the evidence for the efficacy of quetiapine. Eight open-label studies have assessed quetiapine use in 191 patients, with improvement in psychosis recorded in 152 (80%). In addition to the open-label studies, there have been two single-blind, randomized trials comparing quetiapine and clozapine. These studies suggest that quetiapine has similar efficacy to clozapine in controlling psychosis. Following the promising results of the open-label and clozapine comparison studies, five randomized, controlled trials (RCTs) have been performed to further establish the efficacy of quetiapine. Unfortunately, the results have been disappointing. The only positive placebo-controlled study excluded patients with delusions, which seem to be harder to treat than hallucinations. The four negative RCTs discussed seriously undermine the evidence from the open-label studies. The differences in design and interpretation of the RCTs emphasizes the need for further large, well-controlled trials, using strict inclusion criteria, appropriate psychosis rating scales, carer input and clinical significance. Currently, many physicians continue to cautiously offer a trial of low-dose quetiapine empirically. Clozapine should be considered in patients who can tolerate the required blood monitoring.

Background

Psychosis (used here to mean delusions and/or hallucinations) is a common nonmotor feature of Parkinson’s disease (PD). When hallucinations occur in the absence of delusions they are often not distressing and therefore may not require treatment. Delusions are often paranoid and persecutory, and are consequently more likely to result in subjective distress and behavioural disturbance. Prevalence estimates vary as a function of clinical setting. In a systematic review of clinic populations the prevalence of hallucinations alone was between 21% and 46% [Barnes and David, 2001]. The prevalence of delusions is about 5%, based on cross-sectional studies in clinic populations [Fenelon and Alves, 2010]. Definite risk factors for hallucinations include older age, longer disease duration, visual impairment, cognitive impairment and sleep disturbance [Barnes et al. 2003; Barnes and David, 2001]. Cognitive impairment is also a risk factor for delusions [Aarsland et al. 2001]. Psychosis indicates poor prognosis in PD [Aarsland et al. 1999; Goetz and Stebbins, 1995] and is often the most significant factor in the road to institutional care [Aarsland et al. 2000]. In addition carer burden and mortality are high [Gimenez-Roldan et al. 2003; Goetz et al. 2001; Aarsland et al. 2000]. Although the introduction of some anti-Parkinsonian drugs (dopamine agonists, amantadine, anticholinergics and levodopa), or an increase in dose, can be followed by the emergence of psychosis, the relationship to medications remains obscure. However, large prospective trials have shown clear evidence that the incidence of psychosis is higher in patients treated with dopamine agonists (when combined with levodopa) compared with those on levodopa monotherapy (9–17% versus 3–6%); this is also the common experience of all clinicians treating PD [Parkinson Study Group, 2000; Rascol et al. 2000]. A recent meta-analysis found that hallucinations were more frequent in early PD when patients were randomized to a dopamine agonist compared with placebo or levodopa [Stowe et al. 2008]. In the same study when ergot-derived versus nonergot-derived agonists were compared the increase in hallucinations was only seen for nonergot agonists. It is possible that this is an effect of different pharmacological properties of nonergot agonists, however it could be that hallucinations were underreported in the ergot studies, as these were generally performed earlier, prior to when nonmotor symptoms really came under scrutiny in PD research. In contrast to these studies, most cross-sectional surveys do not reveal different patterns of dopamine agonist use in hallucinators [Fenelon and Alves, 2010]. The reasons for this are currently unclear.

Clinical experience suggests that drug withdrawal or dose reduction sometimes can alleviate the psychosis, without the need for additional therapies. However, physicians treating PD psychosis sometimes find it can be impossible to reduce the dose of anti-Parkinsonian medication to a level leading to resolution of psychosis while retaining sufficient motor control for the patient [Cummings, 1999]. In this scenario, these patients therefore require maintenance of the anti-Parkinsonian drug least likely to aggravate psychosis (usually levodopa monotherapy) with the addition of specific antipsychotic therapy to treat the psychosis and to allow them to tolerate the required levels of dopaminergic treatment.

Previous attempts to control psychosis using the older ‘typical’ antipsychotic or neuroleptic drugs led to worsening of motor symptoms. The introduction of ‘atypical’ antipsychotics opened up new therapeutic options. Their characteristics were originally considered to include a lower incidence of extrapyramidal side effects in humans (compared with typical neuroleptics) [Kerwin, 1994]. However evidence from large randomized trials suggests that this difference is absent [Miller et al. 2008]. Recent literature focuses more upon specific pharmacological actions, and less upon categorization of an agent as ‘typical’ or ‘atypical’. Neuroleptic agents classified as ‘atypical’ include clozapine, risperidone, olanzapine, aripiprazole and quetiapine and have all been used to treat psychosis in PD with varying success [Friedman et al. 2006; Wolters and Berendse, 2001; Friedman and Factor, 2000]. Clozapine is the only drug with proven efficacy [French Clozapine Parkinson Study Group, 1999; Parkinson Study Group, 1999]. Olanzapine and risperidone have been shown to be effective, but are associated with significant motor worsening [Factor et al. 2002; Ondo et al. 2002]. Preliminary work with aripiprazole has also not been encouraging, with motor worsening occurring in 6 out of 14 patients in an open-label study [Friedman et al. 2006].

Quetiapine is a dibenzothiazepine compound which is similar in structure to clozapine. Quetiapine is widely used because it does not require blood monitoring. It exhibits strong antagonism of 5HT2 receptors and weak antagonism of D2 receptors. In this review the evidence for the efficacy of quetiapine in psychosis in idiopathic Parkinson’s disease is examined. A discussion on the role of cholinesterase inhibitors in PD psychosis and treatment of psychosis in other parkinsonian disorders is outside the scope of this review.

Search strategy and selection criteria

Suitable studies for this review were identified by searches of MEDLINE until January 2010 using the following terms:

1. Parkinson’s disease ‘or’ parkinsonism ‘or’ parkinsonian

2. Psychosis ‘or’ drug-induced psychosis ‘or’ hallucinations ‘or’ delusions

3. (i) ‘and’ (ii)

The key term ‘quetiapine’ was cross referenced with the results of (iii). The reference lists of all trial reports were examined. Only papers published in English were reviewed.

Open-label studies

Table 1 shows the details of open-label studies. The first reported open-label study with quetiapine in PD patients studied 10 patients [Evatt et al. 1996]. Significant improvements in both the Brief Psychiatric Rating Scale (BPRS) [Overall and Gorham, 1976] score by 31% and total Unified Parkinson’s Disease Rating Scale (UPDRS) [Fahn et al. 1987] by 30% were reported after 3 months. The same group reported a further study of 40 patients in which similar improvements in BPRS and UPDRS were seen [Juncos et al. 1999]. Another early open-label study with 11 patients reported a differential effect on hallucinations and delusions, with hallucinations improving in 6/11, but delusions improving in only 2/11 [Targum and Abbott, 2000]. This division in treatability between hallucinations and delusions was noted in another study which reported that when quetiapine doses were adjusted according to clinical need, doses required were higher for those patients experiencing delusions and hallucinations (265 ± 90 mg) than for hallucinations alone (110 ± 51 mg) [Mancini et al. 2004]. This study also found that patients with cognitive decline (Mini Mental State Examination [MMSE] score <22) had a poorer response to quetiapine than patients without cognitive decline (MMSE >23). Another study with 35 patients examined demented and nondemented PD patients with psychosis, and reported no significant improvement in BPRS score in either group [Prohorov et al. 2006]. However, an improvement in the Clinical Global Impression (CGI) [Guy, 1976] score was noted in 30% of patients in both groups. Three further open-label studies reported significant improvements in BPRS, by 33%, 26% and 30%, respectively, without deterioration in motor functioning as measured by the UPDRS motor subscale [Juncos et al. 2004; Brandstadter and Oertel, 2002; Fernandez et al. 1999].

Table 1.

Open-label studies.

Study	Study type	Number of patients	Duration	Drop out	Mean dose (unless stated)	Diagnostic criteria for PD	Baseline psychosis	Standard rating scales used	Effect on psychosis	Effect on Parkinsonian symptoms (UPDRS)	Effect on cognition (MMSE unless stated)
Evatt et al. [1996]	Open label	10	3 months	None	Median 50 mg/day (25–150 mg/day)	Not stated	Not stated	BPRS, UPDRS	Significant improvement in BPRS (baseline vs. 3-month treatment scores = 37.2 t 11 vs. 25.8 t 11, p < 0.04) (30%)	Significant improvement in total UPDRS (68.1 ± 11 at baseline vs. 47.6 ± 18 at 3 months, p < 0.004)	Not assessed
Targum et al. [2000]	Open label	11	52 weeks	2/11 lack of efficacy 4/11 adverse events	Up to 300 mg	Not stated	Score >3 on BPRS hallucinations/delusions scale	BPRS, CGI, UPDRS	Hallucinations markedly improved in 6 patients. Delusions improved in 2 patients only. Total BPRS scores changed from 38.92 ± 16.07 at baseline to 35.46 ± 11.98 at the last observation (p = not significant).	No change in first 12 weeks. Subsequent deterioration due to disease progression.	Not assessed
Juncos et al. [1999]	Open label	40	12 weeks	Not stated	Between 25 and 800 mg	Not stated	Not stated	BPRS, CGI, UPDRS	Significant improvement in BPRS and CGI (mean BPRS scores improved by 12.5 ± 2 (baseline = 34.7; p < 0.001) (36%)	Improvement over 12 weeks	Not assessed
Fernandez et al. [1999]	Open label	24	4 weeks	3/24 due to AEs	40.6 mg	¾ cardinal features of PD	DSM-IV drug-induced psychosis	BPRS, UPDRS, MMSE	Mean BPRS score substantially improved from 32.6 at baseline to 22.8 (p < 0.024) after 4 weeks (30%)	No significant difference on UPDRS motor subscale (baseline 42.2; 4 weeks 44.9, 6% increase)	No significant differences on MMSE
Brandstadter and Oertl [2002]	Open label	7	3 months	1/7 due to AEs	164 mg	Not stated	Not stated	BPRS, UPDRS	Mean BPRS score substantially improved from 34.3 ± 4.27 at baseline to 25.4 ± 1.27 (p < 0.001) after 14 days (26%)	No evidence of change on UPDRS motor subscale.	Not assessed
Prohorov et al. [2006]	Open label	35	24 weeks	8/35 due to lack of efficacy 3/35 due to AEs	150 mg (demented group) 76 mg (nondemented)	Not stated	Presence of hallucinations or delusions, which are significantly affecting lives.	BPRS, MMSE, CGI, UPDRS	No significant change in BPRS in either group, 30% improvement in CGI	Mild worsening in demented group (Baseline 40.4 ± 10.2, Follow up 45.7 ± 14.5; 13% increase)	Mild worsening in demented group
Mancini et al. [2004]	Open label	35	12 months	3/35 due to lack of efficacy	185 mg	UK brain bank	‘disabling mental disorders’	PPRS, MMSE, UPDRS	Significant improvement in total and partial PPRS when compared to baseline	No significant difference on UPDRS	No significant differences on MMSE
Juncos et al. [2004]	Open label	29	6 months	1/29 due to lack of efficacy 3/29 due to AEs	Median 62.5 mg (Range 12.5–400 mg)	Not stated	‘severe enough to warrant treatment’	BPRS, NPI, UPDRS, CGI	BPRS mean improvement from baseline of 33% at week 24 (p < 0.003) Total NPI change not significant. NPI subtotal change (delusions and hallucinations) significant.	No significant changes in UPDRS motor subscale. (Baseline 27.8 ± 1.7 vs. final 26.3 ± 1.7; 5% increase)	No significant change in MMSE. Improvement in immediate story memory and delayed story memory
Reddy et al. [2002]	Notes review	43	10 months	4/43 due to AEs	54 mg	Not stated	‘Psychosis leading to substantial communication and social problems’	UPDRS, MMSE	35/43 improvement in psychosis	Not significant across group. Baseline 32 ± 14, final 33 ± 16 (3% increase)	Not assessed
Fernandez et al. [2003]	Notes review	106	15 months	19/106 due to lack of efficacy 9/106 due to AEs	60 mg	¾ cardinal features of PD	DSM-IV drug-induced psychosis	UPDRS	87/106 improvement in psychosis	Mild worsening UPDRS-motor score (45 vs. 48; p = 0.004) was seen (7% increase))	Not assessed

PD, Parkinson’s disease; BPRS, Brief Psychiatric Rating Scale; CGI, Clinical Global Impression; UPDRS, Unified Parkinson’s Disease Rating Scale; MMSE, Mini Mental State Examination; NPI, Neuropsychiatric Inventory; AE, adverse event, DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, 4th Edition.

In addition to these prospective open-label studies there have been two large retrospective notes reviews. Reddy and colleagues reported a review of 43 patients treated with open-label quetiapine [Reddy et al. 2002]. In their study 35 patients showed an improvement in psychosis, rated by clinical interview and carer report. When patients with and without dementia were compared, improvements in psychosis occurred in similar numbers in each group. However, significantly more patients in the demented group experienced a worsening of motor symptoms (p < 0.02, Fisher’s exact test). The UPDRS motor score for the nondemented group showed a decrease (improvement) of 2 points, while the demented group showed an increase (worsening) of 2.7 points.

Fernandez and colleagues performed a retrospective analysis of 106 quetiapine treated patients, 87 of whom had PD [Fernandez et al. 2003]; 35% had complete remission of their psychosis, 47% had partial improvement while 18% had poor to no response. A mild but statistically significant worsening in UPDRS-motor scores of 7% was seen in the whole cohort by 15 months. Interestingly when quetiapine responders were compared with nonresponders, more nonresponders were noted to be significantly more demented, delusional or had experienced ‘threatening’ psychosis.

Taken together the above prospective open-label studies assessed quetiapine use in 191 patients, with improvement in psychosis recorded in 152 (80%). The magnitude of the improvement ranged from 26% to 36% and was reported to be clinically meaningful in all eight reports.

Clozapine comparison studies

In addition to the open-label studies, there have been two single-blind randomized trials comparing quetiapine and clozapine (Table 2). Morgante and colleagues performed a 12-week randomized trial in 45 patients comparing quetiapine and clozapine [Morgante et al. 2004]. No reduction of Parkinsonian medication was performed prior to enrolment. Prescription was open label, but the raters were blinded. A significant improvement of 25–30% in both treatment groups, as assessed by the BPRS and the CGI, was reported. There was no motor worsening in either treatment group, as measured by the UPDRS motor subscale. Merims and colleagues reported a 22-week randomized trial of 27 PD patients [Merims et al. 2006]. Selegeline, amantadine and anticholinergic drugs were withdrawn for at least 2 weeks prior to entry into the trial. No reduction of dopaminergic medication was performed. Psychosis was rated by a blinded rater using the Neuropsychiatric Inventory (NPI) [Cummings et al. 1994] and CGI. Eighty-one per cent of patients reached a clinically ‘satisfactory’ control of psychotic symptoms. Both groups showed improvement over time as assessed by the CGI (p < 0.001), with no significant difference between the groups. No significant changes were found in the NPI-rated severity of hallucinations or delusions over the trial in both treatment arms. In the clozapine-treated group, the NPI hallucination frequency score was reduced significantly by 1.5 points after the second month compared with baseline (p = 0.0015). In addition the NPI delusion frequency score was reduced by 2 points compared with baseline after 3 months (p = 0.0149). In the quetiapine-treated group, these changes did not reach significance. The results of these studies suggest that quetiapine shows similar efficacy to clozapine in reducing dopaminergic psychosis in PD patients, with a possible superior effect of clozapine on the frequency of hallucinations and delusions.

Table 2.

Clozapine comparison studies.

Authors	Study type	Number of patients	Duration	Drop out	Dose	Diagnostic criteria for PD	Baseline psychosis	Standard rating scales used	Effect on psychosis	Effect on Parkinsonian symptoms (UPDRS)	Effect on cognition (MMSE unless stated)
Morgante et al. [2004]	Rater-blinded, vs. clozapine	45	12 weeks	Clozapine 3/23 due to AEs Quetiapine 2/22 due to AEs	Clozapine 26 mg Quetiapine 91 mg	Criteria defined by Gelb et al. [1999]	Psychosis duration >3 weeks, BPRS >3 on hallucinations/ delusions subscores	BPRS, CGI, UPDRS	Significant improvement from baseline in BPRS in both groups	No significant change from baseline on UPDRS-motor	Not assessed
Merims et al. [2006]	Rater-blinded, vs. clozapine	27	26 weeks	Clozapine 7/14 3 due to lack of efficacy, 3 due to neutropenia, 1 due to poor compliance Quetiapine 4/13 (all 4 due to lack of efficacy	Not stated	Not stated	‘clinically significant’	NPI, CGI, UPDRS	Significant improvement in CGI score for both groups. NPI score only significant for hallucinations item in clozapine group.	No significant change from baseline on total UPDRS	Not assessed

PD, Parkinson’s disease; BPRS, Brief Psychiatric Rating Scale; CGI, Clinical Global Impression; UPDRS, Unified Parkinson’s Disease Rating Scale; MMSE, Mini Mental State Examination; NPI, Neuropsychiatric Inventory.

Randomized controlled trials

Following the promising results of the open-label and clozapine comparison studies, several RCTs have been performed to further establish the efficacy of quetiapine (Table 3). Unfortunately, the results have been disappointing. To date there have been one positive and four negative RCTs. The first reported was a 12-week double-blind RCT with 31 subjects [Ondo et al. 2005]. Psychosis was rated using the BPRS and the Baylor PD Hallucination Questionnaire, an in-house scale. Twenty-six subjects completed the study. No significant group differences were found in the total BPRS score or on the hallucinations question. On the Baylor PD Hallucination Questionnaire a trend towards improvement was noted, but this did not achieve statistical significance (p = 0.19). UPDRS scores did not change significantly relative to placebo. The authors hypothesize that the study may have been inadequately powered to detect a treatment effect using the more traditional BPRS.

Table 3.

Randomized controlled trials.

	Study type	Number of patients	Duration	Drop out	Dose	Diagnostic criteria for PD	Baseline psychosis	Standard rating scales used	Effect on psychosis	Effect on Parkinsonian symptoms (UPDRS)	Effect on cognition (MMSE unless stated)
Ondo et al. [2005]	Double blind, placebo controlled	31	12 weeks	Placebo 4/21 (2 due lack of efficacy, 2 due to AEs) Active 2/10 (due to unrelated illness)	Up to 200 mg	Not stated	‘subjectively problematic’ hallucinations BPRS 11 ± 5	Baylor PD Hallucination Questionnaire, BPRS, UPDRS	BPRS total and hallucination question; no significant change vs. placebo.  Baylor PD hallucination questionnaire: trend toward improvement but not significant (p = 0.19)	UPDRS on Motor scores did not change significantly compared with placebo	No significant differences vs. placebo
Rabey et al. [2007]	Double blind, placebo controlled	58	12 weeks	Placebo 11/28 (lack of efficacy in 9, death in 1, noncompliance in 1) Active 15/30 (lack of efficacy in 10, AE in 3, noncompliance in 1, somatic problems in 1)	119.2 mg	Not stated	‘psychotic PD’: severity not stated	BPRS, MMSE, CGI, UPDRS	No significant differences between groups on BPRS	No significant difference on UPDRS-motor. Quetipine group baseline 37 ± 9.6; 12 weeks 39.2 ± 9.8. (6% increase)	No significant differences vs. placebo
Kurlan et al. [2007]	Double blind, placebo controlled	40 (9 PD)	10 weeks	Placebo 7/20 Active 3/20 (all withdrew consent, not secondary to AEs)	120 mg	Sum of ratings for tremor at rest, rigidity, bradykinesia and gait items of UPDRS ≥2	DSM-IV drug- induced psychosis	BPRS, NPI, MMSE, CGI, UPDRS	No significant improvement in BPRS but large placebo effect	No significant worsening vs. placebo	No significant differences vs. placebo
Shotbolt et al. [2009]	Double blind, placebo controlled	24	12 weeks	Placebo 3/13 (due to AEs) Active 3/11 (due to AEs)	Quetiapine 72.7 mg Placebo 96.2 mg	UK Brain Bank	>3/7 on BPRS psychosis items	BPRS, NPI, Baylor PD hallucination scale. UPDRS	No significant effect on BPRS or NPI. No significant effect on time to dropout.	No significant difference vs. placebo in UPDRS- motor	Not reported
Fernandez et al. [2009]	Double blind, placebo controlled	16	8 weeks	Placebo 1/8 Active 4/8 (2 due to AEs, 2 due to lack of efficacy)	58.3 mg	UK Brain Bank	Persistent VH >1 month. NOTE: delusions excluded	BPRS, UPDRS, CGI	Improvement on hallucination subscale of BPRS (p = 0.02) and on CGI.	UPDRS-motor scores improved in quetiapine arm compared to placebo (5.74 [SD = 6.84] vs. 2.83 [SD = 7.46]; p = 0.10).	Not reported

PD, Parkinson’s disease; BPRS, Brief Psychiatric Rating Scale; CGI, Clinical Global Impression; UPDRS, Unified Parkinson’s Disease Rating Scale; MMSE, Mini Mental State Examination; NPI, Neuropsychiatric Inventory; AE, adverse event, DSM-IV, Diagnostic and Statistical Manual of Mental Disorders, 4th Edition; VH, visual hallucination.

Rabey and colleagues reported a 3-month double-blind RCT with 58 subjects [Rabey et al. 2007]. Psychosis was rated using the BPRS. No significant group differences were found for any of the outcome measures. Fifteen patients dropped out in the quetiapine group (50%; lack of response in 10). In the placebo group 11 patients dropped out (40%; lack of response in 9). The authors suggest that the high dropout rate may have had an influence on the result. However the dropout rate due to lack of response was similar in both placebo and treatment groups. One further RCT incorporated the projected high dropout rate into the study design, using time to dropout as the primary outcome measure [Shotbolt et al. 2009]. If quetiapine is effective this would be longer in the treatment group, as dropout due to lack of response is reduced. However no difference in time to dropout was found. Secondary outcome measures (BPRS, NPI, Baylor PD hallucination questionnaire, UPDRS) were also analysed at 6 weeks. No significant group differences were found.

The fourth RCT examined 40 patients with behavioural disorders, dementia and Parkinsonism, only nine of whom had a diagnosis of PD [Kurlan et al. 2007]. No significant improvements in psychosis as measured by the BPRS were recorded. There was no motor worsening. The negative finding was attributed to a large placebo effect, lack of power and possibly inadequate dosage (mean 120 mg/day).

The only positive RCT study to date was also the smallest, only recruiting 16 patients, 5 of which discontinued prematurely [Fernandez et al. 2009]. Also of note is that patients with delusions were excluded. A significant improvement on the hallucination subscale of the BPRS with quetiapine was reported (38% versus 1%), with no evidence of motor worsening. However, there was no between arm difference in change in total BPRS score.

Discussion

In summary, four out of five of the placebo-controlled studies found that quetiapine was not effective in PD psychosis, while all of the studies against clozapine and the open-label studies found that it was. The only positive placebo-controlled study excluded patients with delusions, which seem to be harder to treat than hallucinations. This is corroborated by the open-label study of Mancini and colleagues, who found that higher doses of quetiapine were required in patients with delusions [Mancini et al. 2004]. This was also reported in the notes review of Fernandez and colleagues [Fernandez et al. 2003].

It is possible that the varied associations and aetiologies of psychosis have played a part in the conflicting results from these studies. These studies have differed in the selection of patients with drug-driven psychosis, psychosis associated with cognitive impairment, other Parkinsonian disorders, hallucinations with and without delusions and patient numbers. Several studies found that patients with cognitive decline had a poorer response to quetiapine [Mancini et al. 2004; Fernandez et al. 2003]. Ondo and colleagues suggest this as a reason why their RCT was negative [Ondo et al. 2005]. One study showed a neurocognitive improvement in patients treated with quetiapine [Juncos et al. 2004]. The authors attribute this to a reduction in the cognitive dysfunction associated with psychotic symptoms. This has not been reported elsewhere in the PD–quetiapine–psychosis literature, although if true, it could indicate that management of the primary cognitive dysfunction could lead to a secondary improvement in psychosis. Indeed, some evidence to this effect is available from the rivastigmine literature, where significant benefits of rivastigmine over placebo in change in NPI scores have been reported [Emre et al. 2004]. Trial duration may be an important factor in establishing efficacy. All of the negative placebo-controlled studies were relatively short (10–12 weeks). The open-label studies and one of the quetiapine versus clozapine were considerably longer (6–12 months). It may be that this prolonged treatment period is required before significant improvements are seen in this patient group.

Caution should be exercised when interpreting the data from the quetiapine versus clozapine trial as patients were unblinded. In addition this comparator study design does not provide information of possible placebo effect size or spontaneous remission. Although these studies show no significant differences between quetiapine and clozapine, quetiapine is always less effective. If a placebo arm had been included it is possible that there would also have been no difference between quetiapine and placebo. The blinded raters were aware of the study design, which may have introduced bias of effect sizes in both groups. However, in the Morgante and colleagues study the decrease in BPRS score in the quetiapine group (28.6%) was similar to that seen in the Parkinson Study Group trial, a placebo-controlled trial with clozapine, suggesting that the results are reliable [Morgante et al. 2004; Parkinson Study Group, 1999]. One other point is that the quetiapine versus clozapine trials tended to recruit more severely psychotic patients (such patients would be excluded from placebo-controlled trials, because of the urgent need for treatment). For example, the baseline BPRS for the Ondo and colleagues quetiapine study was 11 ± 5, compared with 15.5 ± 3.4 for the quetiapine group in the Morgante and colleagues study (higher scores indicate more severe psychosis) [Ondo et al. 2005; Morgante et al. 2004]. It may be that severely psychotic patients show a better/more clinically meaningful response to treatment, or regression to the mean, accounting for the positive results.

It is notable that there is no widely accepted validated scale specifically designed for use in PD psychosis. The most commonly used scales were the NPI and BPRS. The NPI uses information from the caregiver. Thus, it may be more reliable as patients with PD psychosis may minimize their symptoms. Both BPRS and NPI are composite measures of psychiatric symptoms. In PD psychosis studies the hallucinations and delusions subscales are of principal interest. Some of the studies performed post hoc analyses of these subscales in isolation. Fernandez and colleagues only achieved a positive result when they looked at the BPRS hallucination question only, after the total BPRS changes were not significant [Fernandez et al. 2009]. Juncos and colleagues found that change in NPI scores only reached significance by using this method [Juncos et al. 2004]. However, this approach was also used in the Ondo and colleagues trial, but did not yield a significant result [Ondo et al. 2005].

Other psychosis scales include the Parkinson Psychosis Rating Scale and the Baylor PD Hallucination Questionnaire [Ondo et al. 2005; Friedberg et al. 1998]. While these were designed for PD psychosis they have not been validated independently. This makes interpretation of the studies using these scales more difficult. A semistructured interview, the North-East Visual Hallucinations Interview (NEVHI), has been developed to assess visual hallucinations (not delusions) [Mosimann et al. 2008]. Preliminary validity and reliability measures were promising. This may be helpful in future treatment studies, particularly if it becomes accepted as a ‘gold standard’.

In terms of safety, reported side effects included sedation, confusion and postural hypotension. Other recognized side effects such as weight gain, drooling, diabetes and prolonged QT interval were generally not reported for these studies, possibly because the doses used for the treatment of PD psychosis are typically smaller than those used in schizophrenia, or because of the short duration of the PD–psychosis trials. There was no clear evidence for worsening of Parkinsonism at the low doses used, with only mild changes in UPDRS III (the motor component of the scale) ranging from 3% to 7%. This reflects findings from the schizophrenia literature, where even at higher doses (up to 750 mg) the frequency of extrapyramidal side effects, including induction of Parkinsonism, was low and not different to placebo [Arvanitis and Miller, 1997; Small et al. 1997; Borison et al. 1996]. Excluding lack of efficacy, trial drop-out rates due to other side effects were comparatively low (ranging from 8% to 27%). Lack of efficacy was a more common reason for drop out. There were no deaths in any of the prospective trials, although in general recent concern has been raised on the use of other atypical neuroleptics (risperidone and olanzapine) in patients with other disorders which co-exist with cognitive impairment. A meta-analysis by the Committee on Safety of Medicines found a threefold increase in the risk of stroke for risperidone when used to treat behavioural disturbance in older patients with dementia. There was a similar risk with olanzapine [Committee on Safety of Medicines, 2004]. Another large meta-analysis showed that death occurred more often among patients randomized to drugs; the odds ratio was 1.54 (95% confidence interval [CI], 1.06–2.23; p = 0.02) [Schneider et al. 2005]. It is unclear whether this risk is equally applicable to PD patients, without dementia, who are being considered for treatment with quetiapine for hallucinations/delusions, rather than for behavioural disturbances, but it seems sensible to exercise caution and consider this as a potential side effect when assessing the risks and benefits of using quetiapine in this clinical setting.

Summary

The four negative RCTs seriously bring into question the early promise shown by quetiapine in open-label studies. The differences in design and interpretation of the RCTs emphasises the need for further large well controlled trials, using strict inclusion criteria, appropriate psychosis rating scales, carer input and clinical significance. Duration of treatment may be significant. The open-label trials were longer than the RCTs. Longer RCTs are needed (>3 months duration). However, in practice it would be difficult for patients with psychosis to remain on placebo for long periods, and so novel trial designs or primary endpoints may be appropriate in this scenario. Currently, many physicians believe that quetiapine is effective in treating PD psychosis, and continue to cautiously offer a trial of low dose quetiapine empirically, in the absence of clear evidence for a proven alternative (with the exception of clozapine). In addition it is notable that the American Academy of Neurology's task force on PD treatment recommends quetiapine as a treatment option after clozapine has been considered [Miyasaki et al. 2006]. However, it should be remembered that in many European countries the prescription of quetiapine for PD psychosis is off-label, with possible economic consequences. The standard management of reducing anti-parkinsonian drugs as much as possible should be employed, and the evidence for the use of cholinesterase inhibitors to manage psychosis should be also considered. Clozapine use should be considered if patients can tolerate the required blood monitoring.

Funding

This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors.

Conflict of interest statement

All of the authors were involved in a randomized controlled trial of quetiapine in Parkinson’s disease psychosis, which is now complete. This was a noncommercial trial funded by the Parkinson’s Disease Society. Dr Michael Samuel has received honoraria for lectures/educational material from UCB, GSK, Medtronic and Orion. He has received unrestricted educational grants from Britannia, Solvay, GSK, Teva, Ipsen, Boehringer-Ingelheim and Medtronic. He has received funding for educational trips from Teva, Ipsen, Pfizer, Medtronic, UCB and Boehringer-Ingelheim.