Psychopharmacology

, Volume 212, Issue 2, pp 119–129

Second-generation antipsychotics in dementia: beyond safety concerns. A clinical, systematic review of efficacy data from randomised controlled trials

review

DOI: 10.1007/s00213-010-1939-z

Cite this article as:
Gentile, S. Psychopharmacology (2010) 212: 119. doi:10.1007/s00213-010-1939-z

Abstract

Rationale

Antipsychotic drugs are widely used as a first-line pharmacological approach to treat dementia-related psychiatric symptoms. However, in this population of patients, such drugs have been associated with severe safety concerns. Hence, the aim of this review is to asses systematically the efficacy of second-generation antipsychotics (SGAs) in treating dementia-related neuropsychiatric symptoms in order to establish if the potential clinical benefits of such treatment outweigh the hypothesised risks related to pharmacological intervention.

Methods

The Cochrane Library, MEDLINE, EMBASE and PsycINFO were searched (from 1980 to June 22, 2010) using terms for included drugs (aripiprazole, clozapine, olanzapine, quetiapine, risperidone, ziprasidone, SGAs) and indications (elderly, dementia, Alzheimer's disease). Electronic database search were supplemented with hand search of references lists of electronically identified articles. Peer-reviewed, randomised, controlled trials published in English and investigating the efficacy of SGAs in patients with different forms of dementia were included in the review process. Information was drawn from the 30 articles that met the inclusion criteria.

Results

Nearly all reviewed studies suffer from methodological limitations too severe to draw definitive conclusions that may inform the decision-making process. Moreover, studies conducted with similar methodological design show conflicting efficacy results.

Conclusions

Because of their undemonstrated effectiveness, SGAs should be avoided in patients with dementia complicated by psychotic and/or behavioural symptoms. Hence, further researches are urgently needed to identify useful pharmacological strategies that can be used to improve the clinical condition of such patients and to reduce burden to caregivers when behavioural interventions are ineffective.

Keywords

Dementia Neuropsychiatric symptoms Second-generation antipsychotics Alzheimer Behavioural symptoms 

Introduction

Worldwide, nearly 25 million people have been diagnosed with some forms of dementia, and this number is expected to increase by five million each year (Ferri et al. 2005). Independent of the specific form of dementia, this devastating mental illness is often complicated by psychiatric and behavioural symptoms that contribute to patient and caregiver distress (Schulz et al. 1995), increased health-care costs (Murman et al. 2002), and may even lead to institutionalisation (Stern et al. 1997). In fact, delusions and/or hallucinations appear in 30–50% of these patients, and as many as 70% may develop agitated and/or aggressive behaviours (Craig et al. 2005).

Antipsychotic drugs are widely used as a first-line pharmacological approach to treat these psychiatric symptoms. In the US and Europe, up to 60% of people with dementia residing in care facilities are prescribed antipsychotics (McGrath and Jackson 1996; Lana et al. 2001; Bell et al. 2010). However, in this population of patients such drugs have been associated with an increased risk of severe adverse reactions, such as parkinsonism, oedema, chest infections and metabolic dysfunctions (Ballard and Waite 2006; Zheng et al. 2009); accelerated cognitive deterioration has also been reported (Ballard and Howard 2006; Kennedy et al. 2005), although not definitively confirmed (Livingston et al. 2007), as well as an increased risk of cerebrovascular events (especially for olanzapine (OLA) and risperidone (RIS; Wooltorton 2002; Wooltorton 2004), odds ratio 2.5–3 (Schneider et al. 2006a)) and mortality (odds ratio 1.5–1.7 (Schneider et al. 2005; Rossom et al. 2010)).

On the basis of these findings, the US Food and Drug Administration (FDA) issued a warning about a significant increase in mortality risk associated with second-generation antipsychotics (SGAs) use in patients with dementia (US FDA 2005). The overall message of the regulator warnings is that the risks associated with SGA use in elderly patients with dementia outweigh the benefits gained from behavioural control (Bullock 2005). Analogous risks have been reported with the use of first-generation antipsychotics (FGAs; Schneider et al. 1990; Wang et al. 2005). However, data which led to these warnings remain controversial (Ballard et al. 2009); in particular, a number of studies failed to demonstrate an association between both FGA and SGA treatments of dementia-related neuropsychiatric symptoms and an increased risk of stroke (Herrmann et al. 2004; Gill et al. 2005; Finkel et al. 2005; Liperoti et al. 2005; Liperoti et al. 2009).

Notwithstanding these potential safety concerns, no decline in the use of this class of drugs has been observed in patients diagnosed with dementia (Leslie et al. 2009). In contrast, it has been reported that, in the US, at least $500 million (and, probably, substantially more) is spent each year on SGA medications for the treatment of dementia (Rosenhek et al. 2007). Hence, there is the need for a further, updated evaluation of the efficacy of such psychotropics currently prescribed in elderly care, to establish if the possible benefits on controlling psychotic and behavioural clinical manifestations prevail or not over the hypothesised risks related to pharmacological intervention.

Methods

Search strategy

The Cochrane Library (June 2010) MEDLINE (1966 to June 21, 2010), EMBASE (1980 to June 21, 2010), and PsycINFO (1985 to June 21, 2010) were searched using terms for included drugs (aripiprazole (ARI), clozapine, OLA, quetiapine (QUE), RIS, ziprasidone, SGAs) and indications (elderly, dementia, Alzheimer's disease (AD)). Electronic database search were supplemented with hand search of references lists of electronically identified articles, overall providing 303 articles.

Selection criteria

Peer-reviewed, randomised controlled trials (RCTs) published in English and investigating the efficacy of SGAs in patients with different forms of dementia were included in the review-process. Although considered for review, findings that emerged from post hoc analyses (Frank et al. 2004; Grossman and Okamoto 2004; Brodaty et al. 2005; Cummings et al. 2002; Rosenhek et al. 2007; Sultzer et al. 2008) were not included in the tabular data.

Data extraction

Information was drawn from the 30 articles that met the inclusion criteria.

Results

Studies in acutely agitated patients

Aripiprazole

A double-blind, placebo (PLA)-controlled, multicenter trial (Rappaport et al. 2009) evaluated the safety of three different doses of ARI in patients with dementia-related agitation. During the 4 h of the observation phase, the drug was well tolerated at all doses. Moreover, despite the analysis of efficacy outcomes being a secondary aim of the study, preliminary results evidenced a greater efficacy of ARI over PLA.

Olanzapine

Information from an analysis (Battaglia et al. 2003) of three RCTs suggested that OLA is effective and safe in treating acute agitation associated with dementia. Indeed, during the 2-h observational period, the rate of scores reaching calm states was faster during treatment with intramuscolar OLA than during PLA and active comparator treatments (lorazepam, LOR; haloperidol, HAL). Such results were also confirmed when OLA was administered intramuscularly (Meehan et al. 2002; Table 1).
Table 1

Studies in acutely agitated patients

Study/study's design and duration/sample size (Na)

Drug and doses (mg)

Efficacy evaluation/results

Main study limitations

Rappaport et al. 2009/DB and PC, 24 hs (N = 103)

ARI; 5, 10, 15 (fd)

ACES, CGI-I, CGI-S,PEC / ARI > PLA

• Statistical significance not reported

Battaglia et al. 2003/DB and PC, 24 hs (N = 71)

OLA; 2.5 (fd)

ACES, PANSS-EC / OLA > LOR = HAL > PLA

• Statistical significance not reported

Meehan et al. 2002/DB and PC, 2 hs (N = 137)

OLA; 2.5, 5 (fd)

ACES,BPRS, CGI-S, CMAI, PANSS-EC / OLA (5 mg) = LOR > PLA

• LOR used at a relatively low dose (1 mg)

ARI aripiprazole, OLA olanzapine, LOR lorazepam, HAL haloperidol, PLA placebo, ACES Agitation-Calmness Evaluation Scale, CGI-I Clinical Global Impressions-Improvement rating scales, CGI-S Clinical Global Impressions-Severity of Illness rating scales, CMAI Cohen-Mansfield Agitation Inventory, PANSS-EC PANSS-Excited Components, fd fixed dose, DB double-blind, PC placebo-controlled, hs hours

aN indicates the number of patients included in the SGA-treated group

Short-/medium-term studies: comparison with placebo

Aripiprazole

No differences in primary efficacy outcome measures between ARI and PLA were observed in two 10-week RCTs (De Deyn et al. 2005; Streim et al. 2008). ARI was superior to PLA merely in some secondary outcome measures. Partially conflicting results were reported in a randomised, double-blind, PLA-controlled, multicenter trial (Mintzer et al. 2007) that compared the effectiveness of ARI and PLA in institutionalised old patients (mean age: 80.2 years) diagnosed with Alzheimer's dementia. The drug showed greater efficacy than PLA in improving psychotic symptoms but only at dose of 10 mg/day.

Olanzapine

A 10-week, randomised, double-blind trial (Mintzer et al. 2004) comparing the effectiveness of OLA vs PLA in patients with Alzheimer's disease provided doubtful efficacy data: indeed, no significant treatment effects were observed on primary measures of efficacy. Such effects became evident merely after a secondary comparison from the primary analysis repeated measure model pooling across all visits. Quite surprising findings were reported by Street et al. (2000). In fact, in this trial, OLA showed superior efficacy over PLA at low doses on primary outcome measures, but this difference disappeared when the drug was used at relatively high doses. In contrast, De Deyn et al. (2004) reported superior efficacy of OLA over PLA at the highest used dose (7.5 mg/day), at merely in secondary efficacy measures. In a 6-week, randomised, double-blind, PLA-controlled trial (Clark et al. 2001), performed on a specific subset of patients (those with no or minimal psychotic symptoms at baseline) who failed to meet the inclusion criteria for being included in larger trial (Street et al. 2000), the efficacy of the drug was more evident in preventing the onset of delusional symptoms than hallucination phenomena. The same parent study (Street et al. 2000) also led to identify retrospectively a subset of patients diagnosed with Lewy bodies dementia (LBD) in accordance with the International Work Group clinical criteria for LBD (McKeith et al. 1999). In this post hoc analysis, the drug failed to demonstrate definitive advantages in efficacy over PLA in patients with possible diagnosis of this specific form of dementia (Cummings et al. 2002).

Quetiapine

Data from small trials did not demonstrated any advantage of QUE over PLA in improving agitation or psychotic symptoms in patients with dementia and parkinsonism (Kurlan et al. 2007) or with specific diagnosis of Alzheimer's disease (Paleacu et al. 2008). Where the superiority was demonstrated, it merely emerged from a post hoc analysis of secondary outcome measures (Zhong et al. 2007).

Risperidone

A 12-week trial (Katz et al. 1999) suggested that RIS, at three different doses, might be more effective than PLA in improving psychotic and behavioural symptoms in institutionalised elderly patients with dementia. The most appropriate daily dose was 1 mg. RIS was also specifically effective in reducing aggressive behaviour (Brodaty et al. 2003); two post hoc analyses (Brodaty et al. 2005; Grossman and Okamoto, 2004) conducted on a specific subset of this population of patients confirmed the efficacy of the drug in improving psychiatric symptoms and global functioning in elderly patients with Alzheimer's or mixed dementia. The efficacy of the drug in improving dementia-related behavioural changes was also confirmed indirectly by measuring, through a secondary analysis of data obtained from a multicenter, double-blind, PLA-controlled parallel group study (Brodaty et al. 2003), the burden to nursing staff of older nursing-home residents (Frank et al. 2004). In contrast, a further, shorter trial (Mintzer et al. 2006) failed to demonstrate definitively the superiority of RIS over PLA in attenuating psychotic symptoms in patients with Alzheimer's disease (Table 2).
Table 2

Short-term studies: comparison with placebo

Study/study's design and duration/sample size (Na)

Drug and doses (mg)

Efficacy evaluation/results

Main study limitations

De Deyn et al. 2005/DB and PC, 10 ws (N = 106)

ARI, 5–15 (r)

POMs: NPI PSS / ARI = PLA

• Efficacy data coming by LOCF imputation strategy

SOMs: BPRS, CGI-S, NPIT / ARI > PLA

Streim et al. 2008/DB and PC, 10 ws (N = 131)

ARI, 2–15 (r)

POMs: NPI-NHP / ARI = PLA

• Efficacy data coming by LOCF imputation strategy

SOMs: ADCS-ADL, BPRS, CGI-S, CMAI, CSDD, NPI-NHT, TCDS / ARI > PLA

• Prohibited concomitant therapy: not specified

Mintzer et al. 2007/DB and PC, 10 ws (N = 366)

ARI; 2, 5, 10 (fd)

POMs: NPI PSS / ARI (10 mg) > PLA

• Efficacy data coming by LOCF imputation strategy

SOMs: BPRS, CGI-I, CGI-S, CMAI, NPIT / ARI > PLA (at different doses)

• Concomitant psychotropics allowed (TRA, ZPD, TMZ)

De Deyn et al. 2004/DB and PC, 10 ws (N = 520)

OLA; 1, 2.5, 5, 7.5 (fd)

POMs: NPI-NH / OLA = PLA

• Efficacy data coming by LOCF imputation strategy

SOMs: BPRS, CGI-C, CGI-S / OLA (7.5 mg) > PLA

• Concomitant psychotropics allowed (antidepressants, BDZs, ACIs)

Street et al. 2000/DB and PC, 6 ws (N = 159)

OLA; 5, 10, 15 (fd)

POMs: NPI-NH-CS; SOMs: BPRS, NPI-NH –t / OLA > PLA at and 10 mg; OLA = PLA at 15 mg

• Efficacy data coming by LOCF imputation strategy

• Concomitant psychotropics allowed (BDZs)

Clark et al. 2001/DB and PC, 6 ws (N = 121)

OLA; 5, 10, 15 (fd)

POMs: BPRS, NPI-NH / OLA > PLA in improving delusional symptoms

• Efficacy data coming by LOCF imputation strategy

• Changes in BPRS score not shown

• Data coming from a specific subset of patients (those with no or minimal psychotic symptoms at baseline) who failed to meet the inclusion criteria for being included in the larger trial by Street et al. 2000

Kurlan et al. 2007/DB and PC, 10 ws (N = 20)

QUE, 25–300 (r)

POMs: ADCS-ADLQ, ADCS-CGI-S, BPRS / QUE = PLA

• Concomitant psychotropics allowed (ACIs)

• Small sample

Paleacu et al. 2008/DB and PC, 6 ws (N = 20)

QUE

150–300 (r)

POMs: NPI, CGI-I / QUE = PLA

• Concomitant psychotropics allowed (ACIs, ZPD)

• Small sample

Zhong et al. 2007/DB and PC, 10 ws (N = 241)

QUE, 100–200 (fd)

POMs: PANSS-EC / QUE = PLA

• Efficacy data coming by LOCF/OF-imputation strategy

SOMs: ADCS-CGI-S, CGI-C, CMAI, NPI-NH / QUE > PLA at 200 mg

• Concomitant psychotropics allowed (ACIs, BDZs, antidepressants)

Katz et al. 1999/DB and PC, 12 ws (N = 462)

RIS; 0.5, 1.0, 2.0 (fd)

POMs: BEHAVE-AD, CGI-C, CMAI / RIS > PLA

• Concomitant psychotropics allowed (LOR)

Brodaty et al. 2003/DB and PC, 12 ws (N = 167)

RIS, 0.5–2.0 (r)

POMs: BEHAVE-AD, CGI-C, CGI-S, CMAI / RIS > PLA

• Efficacy data coming by LOCF imputation strategy

• Concomitant psychotropics allowed (BDZs)

Mintzer et al. 2006/DB and PC, 8 ws (N = 235)

RIS; 1.0, 1.5 (fd)

POMs: BEHAVE-AD, CGI-C / RIS = PLA

• Concomitant psychotropics allowed (BDZs, ACIs)

ARI aripiprazole, OLA olanzapine, QUE quetiapine, RIS risperidone, LOR lorazepam, HAL haloperidol, TRA trazodone, ZPD zolpidem, TMZ temazepam, BDZs benzodiazepines, ACIs acetylcholinesterase inhibitors, PLA placebo, POMs primary outcome measures; SOMs secondary outcome measures, ADCS-ADLQ ADCS-Activities of Daily Living Questionnaire, ADCS-CGI-C ADSC Clinical Global Impression of Change, BEHAVE-AD Behavioural pathology in Alzheimer's Disease, BPRS Brief psychiatric Rating Scale, CGI-I Clinical Global Impressions-Improvement rating scales, CGI-S Clinical Global Impressions-Severity of Illness rating scales, CMAI Cohen-Mansfield Agitation Inventory, CSDD Cornell Scale for Depression in dementia, M-NCAS Modified Nursing Care Assessment Scale, NPI Neuropsychiatric Inventory, NPI PSS NPI Psychosis Subscale Score, NPI-NH NPI Nursing Home, NPI-NH-CS NPI-NH Core Symptoms, PANSS-EC Positive and Negative Syndrome Scale-Excitement Component, TCDS Total Caregivers Distress Scores, LOCF last observation carried forward, OC observed case, DB double-blind, PC placebo-controlled, ws weeks, fd fixed dose, r range, N/A data not available

aN indicates the number of patients included in the SGA-treated group.

Short-term studies: comparison with active comparators other than SGAs

Olanzapine

A 5-week trial (Verhey et al. 2008), which did not provide a PLA arm, demonstrated that both efficacy and safety of OLA and HAL in improving agitation in demented patients were similar.

Quetiapine

A small, 26-week trial (Ballard et al. 2005) was designed to assess the respective efficacy of QUE and rivastigmine (RIV) vs PLA as treatments of neuropsychiatric symptoms in institutionalised patients with dementia and to evaluate these treatments with respect to changes in cognitive performances. Neither QUE nor RIV were effective in improving agitation-related symptoms; moreover, compared with PLA, QUE was associated with a statistically significant worsening of cognitive functions. Such findings were replicated in a subsequent study (Tariot et al. 2006) that demonstrated that QUE was not more effective than PLA and HAL in improving psychotic symptoms in patients with dementia.

Risperidone

In a 12-week, double-blind trial which started after a 1-week, single-blind phase during which all previous psychotropic medications were discontinued (De Deyn et al. 1999), RIS was more effective than both HAL and PLA in reducing severity and frequency of behavioural symptoms in elderly demented patients. Such results were confirmed in a crossover study (Suh et al. 2004). However, in the study by Chan et al. (2001), impaired by the lack of a PLA arm, no differences in efficacy were found between such drugs. RIS seemed to be also more effective than RIV in the acute treatment of severe agitation (Holmes et al. 2007). In contrast, the drug failed to demonstrate specific advantages in efficacy over selective serotonin reuptake inhibitors (SSRIs), such as citalopram (CIT) (Pollock et al. 2007), or antiepileptic drugs, such as topiramate (TPM; Mowla and Pani 2010; Table 3).
Table 3

Short-term studies: comparison with active comparators others than SGAs

Study/study's design and duration/sample size (Na)

Drug and doses (mg)

Efficacy evaluation/results

Main study limitations

Verhey et al. 2008/DB and R, 5 ws (N = 30)

OLA, 2.5–7.5 (r)

POMs: CMAI; SOMs: CGI, NPI / OLA = HAL

• Lack of a PLA arm

• Concomitant psychotropics allowed (BDZs, antidepressants)

• Small sample

Ballard et al. 2005/DB, R and PC; 26 ws (N = 31)

QUE, 50–100 (r)

POMs: CMAI / QUE = RIV = PLA

• Efficacy data coming by LOCF imputation strategy

• Small sample

Tariot et al. 2006/DB, R and PC; 10 ws (N = 91)

QUE, 100–600 (r)

POMs: BPRS, CGI-S, MOSES / QUE = HAL = PLA

• Efficacy data coming by LOCF imputation strategy

• Concomitant psychotropics allowed (BDZs, ACIs)

• Data on HAL vs PLA not shown

De Deyn et al. 1999/DB and PC, 12 ws (N = 115)

RIS, 1–2 (r)

POMs: BEHAVE-AD, CGI, CMAI, FAST / RIS > HAL > PLA

• Concomitant psychotropics allowed (LOR)

• Severity of symptoms at baseline higher in the HAL- and PLA groups

Suh et al. 2004/DB, PC and CO; 18 ws (N = 60)

RIS, 0.5–1.5 (r)

POMs: BEHAVE-AD, CGI-C / RIS > HAL

• Lack of a PLA arm

• Concomitant psychotropics allowed (LOR)

Chan et al. 2001/DB and R, 12 ws (N = 29)

RIS, 0.5–2 (r)

POMs: BEHAVE-AD, CMAI, FAST / RIS = HAL

• Lack of a PLA arm

• Concomitant psychotropics allowed (BDZs, ACIs)

• Small sample

Holmes et al. 2007/DB and R, 6 ws (N = 12)

RIS, 1 (fd)

POMs: CMAI / RIS > RIV

• Small sample

• Data vs PLA not shown

• RIV used at low doses (6 mg/day)

Pollock et al. 2007/ R,C and OL; 12 ws (N = 50)

RIS, 0.5-2 (r)

POMs: CIRS-G, CSDD, NBRS, NPI, SIB / CIT > RIS

• Lack of a PLA arm

• Concomitant psychotropics allowed (ACIs)

Mowla and Pani, 2010/ DB and R, 6 ws (N = 20)

RIS, 1–2 (m)

CMAI, NPI / TPR = RIS

• Lack of a PLA arm

• Small sample

RIS risperidone, CIT citalopram, HAL haloperidol, OLA olanzapine, PLA placebo, BDZs benzodiazepines, ACIs acetylcholinesterase inhibitors, TPR topiramate, POMs primary outcome measures, SOMs secondary outcome measures, BEHAVE-AD Behavioural pathology in Alzheimer's Disease, CIRS-G Cumulative Illness Rating Scale-Geriatrics, CGI-I Clinical Global Impressions-Improvement rating scales, CMAI Cohen-Mansfield Agitation Inventory, CSDD Cornell Scale for Depression in dementia, FAST Functional Assessment Scale, NPI Neuropsychiatric Inventory, MOSES Multidimensional Observational Scale for Elderly Subjects, NBRS Neurobehavioral Rating Scale, NPI Neuropsychiatric Inventory, DB double-blind, PL placebo-controlled, R randomised, C controlled, OL open-label, CO crossover, ws weeks, r range, m mean

aN indicates the number of patients included in the SGA-treated group

Short-term studies: SGA head-to-head comparisons

Olanzapine vs risperidone

Both drugs demonstrated equivalent safety and effectiveness in two small trials (Fontaine et al. 2003; Mulsant et al. 2004). However, when head-to-head comparison also included a PLA arm and was performed on a larger sample (Deberdt et al. 2005), OLA and RIS did not demonstrate any superiority over PLA. In contrast, preliminary results (Gareri et al. 2004) had suggested that both OLA and RIS were more effective than specific FGAs, such as promazine.

Quetiapine vs risperidone

An 8-week trial (Rainer et al. 2007) suggested that QUE and RIS were equally effective in treating behavioural and psychological symptoms of dementia.

Olanzapine vs quetiapine and risperidone

Until now, one of the largest and best designed trials on the effectiveness of SGAs in patients with dementia complicated by behavioural and psychiatric symptoms is the Clinical Antipsychotic Trials of Intervention Effectiveness-Alzheimer's Disease (CATIE-AD; Schneider et al. 2006b). The trial, conducted at 45 sites in the US, was designed to randomly assign under double-blind conditions more than 400 patients to OLA, RIS, QUE or PLA in a 2:2:2:3 ratio. The median time to discontinuation for any reason was not significantly different among the SGA and the PLA groups, whereas the time to the discontinuation of treatment for adverse reactions favoured PLA. However, when information from the CATIE-AD study were analysed to asses specifically the efficacy of SGAs in improving psychiatric and behavioural symptoms (Sultzer et al. 2008), such data suggested statistically significant beneficial effects with both OLA and RIS, whereas QUE did not demonstrate any superiority over PLA. Conversely, none of these three SGA drugs induced appreciable improvements on functional abilities, quality of life or caregiving time needed (Rosenhek et al. 2007; Table 4).
Table 4

Short-/medium-term studies: SGA head-to-head comparison

Study/sample size (Na)

Drug and doses (mg)

Efficacy evaluation/results

Main study limitations

Fontaine et al. 2003/DB and P, 2 ws (N = 39)

OLA 2.5–10, RIS 0.5–2 (r)

POMs: CGI-I, NPI; SOMs: BEHAVE-AD, MOSES, PGDRS, QUALID / OLA = RIS

• Lack of a PLA arm

• Concomitant psychotropics allowed (LOR)

• Small sample

Mulsant et al. 2004/DB and R, 6 ws (N = 85)

OLA 2.5–10, RIS 0.25–1.5 (r)

POMs: UKU; SOMs: NPI, TMT, WMS / OLA = RIS

• Lack of a PLA arm

• Concomitant psychotropics allowed (ACIs, BDZs)

Deberdt et al. 2005/ DB and PC, 10 ws (N = 400)

OLA 2.5–10, RIS 0.5–2 (r)

POMs: CGI-S, NPI; SOMs: BPRS, CMAI / OLA = RIS = PLA

• Efficacy data coming by LOCF imputation strategy

• Concomitant psychotropics allowed (BDZs)

Gareri et al. 2004/DB, 8 ws (N = 40)

OLA 5, RIS 1 (fd)

POMs: NPI/ /OLA = RIS > PMZ

• Lack of a PLA arm

• Concomitant psychotropics allowed (LOR)

• Small sample

Rainer et al. 2007/R and RB, 8 ws (N = 72)

QUE 50–400, RIS 0.5–4 (r)

POMs: NPI; SOMs: AKT, BPRS, CMAI, CSDD, PDS / QUE = RIS

• Lack of a PLA arm

• Efficacy data coming by LOCF imputation strategy

• Concomitant psychotropics allowed (BDZs)

Schneider et al. 2006b/DB and PC, 36 ws (N = 278)

OLA 2.5–17.5, QUE 20–200, RIS 0.5–2 (r)

POMs: TtD; SOMs: CGI-C / OLA = QUE = RIS = PLA

• Concomitant psychotropics allowed

• QUE and RIS was used at relatively low doses compared with OLA

RIS risperidone, QUE quetiapine, OLA olanzapine, LOR lorazepam, BDZs benzodiazepines, PLA placebo, PMZ promazine, POMs primary outcome measures, SOMs secondary outcome measures, LOCF last observation carried forward, ADAS-Cog Alzheimer's Disease Assessment Scale Cognitive subscales, ADLS Activity of Daily Living Scale, ADR-QoL Alzheimer's Disease Related Quality of Life, AKT Age-adjusted Concentration Test, BEHAVE-AD Behavioural Pathology in Alzheimer's Disease, BPRS Brief Psychiatric Rating Scale, CAS Caregiver Activity Survey, CGI-I Clinical Global Impressions-Improvement rating scales, CGI-C Clinical Global Impressions of Change, CGI-S Clinical Global Impressions-Improvement Psychosis, CMAI Cohen-Mansfield Agitation Inventory, CSDD Cornell Scale for Depression in Dementia; HUIM3 Health Utilities Index Mark 3, MOSES Multidimensional Observational Scale for Elderly Subjects, NPI Neuropsychiatric Inventory, DS Dependence Scale, PDS Progressive Deterioration Scale, PGDRS Psychogeriatric Dependency Rating Scale, QUALID Quality of Life in late-Stage Dementia Scale, TtD Time to discontinuation for any reason, UKU Udvalg for Kliniske Undersogelser, WMS Wechsler Memory Scale, TMT Trail Making Test, R randomised, RB rater blinded, DB double-blind, PC placebo-controlled, P parallel, fd fixed doses, r range

aN indicates the number of patients included in the SGA-treated group

Discussion

General considerations

The efficacy of SGAs in improving behavioural and psychiatric symptoms of dementia is still a matter of debate in medical literature. Two systematic reviews concluded that the evidence for both OLA and RIS supports their effectiveness compared to PLA (Carson et al. 2006; Sink et al. 2005). However, such conclusions were drawn from a number of studies significantly lower than that considered in this article. Conversely, expert consensus guidelines recommended RIS as first-line agent and OLA and QUE as second-line options (Alexopoulos et al. 2004), whereas conclusions from a meta-analytic approach suggested that ARI and RIS were the only medications that had shown statistically significant effects sizes on symptom rating scales (Schneider et al. 2006a). However, nearly all these studies, independent of the drugs that were promoted, reported small-effect sizes (Lee et al. 2004).

To further attenuate the enthusiasm about the efficacy of treatment with SGAs of dementia-related psychotic and behavioural symptoms, it must be stressed that a growing body of evidence seems to demonstrate that patients who discontinue such medications are not more likely to show symptomatologic worsening than those who continue pharmacological regimen (Ruths et al. 2008; Ballard et al. 2004; Ballard et al. 2008).

Moreover, it is also interesting to note that studies comparing SGAs to benzodiazepines and/or antihistaminics are anecdotal (Meehan et al. 2002) or completely lacking, despite both classes of drugs being widely used as calming agents in different psychiatric disorders characterised by agitated and/or aggressive behaviours (Doody et al. 2001).

Also, many trials employed the last observation carried forward (LOCF) imputation strategy to account for drop-outs. The LOCF methodology may lead to the overestimation of the effectiveness of medications, since this methodology allows to include in the statistical analysis patients who prematurely discontinue the trials after the first observation due to lack of efficacy (Gentile 2007, 2009).

Last but not least, several reviewed trials (De Deyn et al. 1999; Chan et al. 2001; Meehan et al. 2002; Battaglia et al. 2003; Brodaty et al. 2003; Fontaine et al. 2003; Frank et al. 2004; Gareri et al. 2004; Suh et al. 2004; Deberdt et al. 2005; Tariot et al. 2006; Kurlan et al. 2007; Pollock et al. 2007; Rainer et al. 2007; Zhong et al. 2007; Verhey et al. 2008) drew pooled conclusions from patients diagnosed with different forms of dementia, without providing any results stratified for diagnosis.

Specific considerations

Studies on acutely agitated patients

Among the three available trials, two involving OLA and one ARI (Meehan et al. 2002; Battaglia et al. 2003; Rappaport et al. 2009), two did not report the statistical significance of the behavioural changes that were recorded (Battaglia et al. 2003; Rappaport et al. 2009). In the only trial that provided this information, the active comparator (LOR) was used at daily doses not higher than 1 mg (Meehan et al. 2002).

Short-term studies vs PLA

Three trials are available on ARI, overall involving more than 600 patients (De Deyn et al. 2005; Mintzer et al. 2007; Streim et al. 2008). Data from such studies, on some occasions impaired by methodological limitations such as the permitted use of concomitant psychotropic medications (Mintzer et al. 2007), provided controversial findings: indeed, in one of these trials (Streim et al. 2008), ARI did not demonstrate any superiority over PLA in primary outcome measures. The same conclusions can be applied to the three OLA-trials involving 800 patients: in fact, in two of these studies, the drug was found be more effective than PLA, but merely in secondary outcome measures (Clark et al. 2001; De Deyn et al. 2004).

Among the three trials available on QUE, overall including less than 300 patients (Kurlan et al. 2007; Zhong et al. 2007; Paleacu et al. 2008), two demonstrated that the efficacy of the drug was similar to that shown by PLA (Kurlan et al. 2007; Paleacu et al. 2008)

Three of the four available trials (Katz et al. 1999; Brodaty et al. 2003; Frank et al. 2004) that provided original data on the use of RIS in treating dementia-related psychotic and behavioural symptoms (overall involving 664 patients), although suffering from the same limitations shown by studies on the other SGAs, were concordant in suggesting that RIS, used at a wide range of daily dose (0.5–7.0 mg), was more effective than PLA. However, the trial by Mintzer et al. (2006), conducted on a relatively large sample and which specifically selected patients with AD suffering from delusions or hallucinations, showed no significant clinical effects of two fixed daily doses (1.0 or 1.5 mg) of RIS.

Short-term studies vs active comparators others than SGAs

Five trials (De Deyn et al. 1999; Chan et al. 2001; Suh et al. 2004; Tariot et al. 2006; Verhey et al. 2008) compared the effectiveness of three different SGAs (OLA, QUE, RIS) vs HAL, but only two (De Deyn et al. 1999; Tariot et al. 2006) also provided a PLA arm. The only trial on OLA, both trials on QUE and one of the three trials on RIS found that HAL was as effective as the specific SGA studied. It is also interesting to note that the superiority of HAL over PLA in improving neuropsychiatric symptoms of dementia is still far from being definitively established (Teri et al. 2000). When the comparison was made with acetylcholinesterase inhibitors, QUE was no more effective than RIV (Ballard et al. 2005). Conversely, RIS demonstrated a statistically significant superiority on RIV which, however, had been used at dosages lower than the recommended ones (Holmes et al. 2007), but not on TPM (Mowla and Pani, 2010). However, it should be noted that acetylcholinesterase inhibitors, such as donezepil, seem to be ineffective in controlling Alzheimer-related behavioural disturbances, and even when used at recommended doses. (Howard et al. 2007)

The only available trial vs SSRIs (CIT), impaired by the lack of a PLA arm, also demonstrated that this antidepressant was more effective than RIS (Pollock et al. 2007).

SGA head-to-head comparison

Six studies, overall involving less than 1,000 patients (Fontaine et al. 2003; Gareri et al. 2004; Mulsant et al. 2004; Deberdt et al. 2005; Schneider et al. 2006b; Rainer et al. 2007) provided original data. Of note, four of the six original trials (Fontaine et al. 2003; Mulsant et al. 2004; Gareri et al. 2004; Rainer et al. 2007) omitted to provide a PLA arm. The two studies comparing OLA and RIS (Deberdt et al. 2005) or OLA, QUE and RIS (Schneider et al. 2006b) vs PLA demonstrated no advantages either between the SGAs that were investigated or between the active compounds and PLA. Only a post hoc analysis suggested a similar efficacy of both OLA and RIS (Sultzer et al. 2008), and significantly higher than that shown by QUE and PLA. However, this superiority did not induce clinically relevant changes in the quality of life of patients with dementia (Sultzer et al. 2008).

Conclusions

Reviewed studies show discouraging findings. Indeed, nearly all researches focused on investigating the efficacy of SGAs in treating dementia-related neuropsychiatric symptoms and which provided reassuring information suffer from methodological limitations too severe to draw definitive conclusions that may inform the decision-making process. Beyond the frequent use of the LOCF imputation strategy, such limitations also include: the unavailability of the statistical significance of results, especially in studies conducted on acutely agitated patients; the lack of a PLA arm (this limitation impairs the findings of nine of the 15 studies that compared SGAs with active comparators, either other SGAs or different classes of drugs); the propensity to allow the use of concomitant medications. In addition, when the efficacy of SGAs was compared with that of active comparators, such as LOR and RIV, both drugs were used at relatively low doses. Moreover, studies conducted with similar methodological design show conflicting results (Brodaty et al. 2003; Mintzer et al. 2007).

In the light of this background, we are forced to conclude that SGAs should be avoided in patients with dementia complicated by psychotic and/or behavioural symptoms. Unfortunately, the same (and even more severe) concerns must be extended to FGAs (Liperoti et al. 2009). Hence, further researches are urgently needed to identify effective pharmacological strategies that can be used to improve the clinical condition of such patients and to reduce burden to caregivers when behavioural interventions are ineffective.

Acknowledgments

Dr. Gentile acknowledges Medical Information Service (Annamaria Desiati, Lucia Fantini, Luana Salvati), a free-access service of Eli Lilly SpA, Sesto Fiorentino, Italy, for the kind and professional support in providing full-text articles. Dr. Gentile also acknowledges Mr. Richard Stuart Harrison for his precious editorial assistance.

Disclosure of interests

Dr. Gentile has received speaker bureau/consultant honoraria by Boehringer Ingelheim and Eli Lilly Italia SpA. During the last 5 years, Dr. Gentile has also received travel funds by Bristol-Myers Squibb, Eli Lilly, Lundbeck and Novartis. No financial support for this study was received. No potential conflicts of interest are directly related to the subject of this paper.

Copyright information

© Springer-Verlag 2010

Authors and Affiliations

  1. 1.Department of Mental Health, Mental Health Center n. 63ASL SalernoCava de’ TirreniItaly

Personalised recommendations