Lack of clinically useful response predictors for treating aggression and agitation in Alzheimer's disease with citalopram

ABSTRACT FROM: Schneider LS, Frangakis, C, Drye LT, et al. Heterogeneity of treatment response to citalopram for patients with Alzheimer's disease with aggression or agitation: the CitAD randomized clinical trial. Am J Psychiatry 2016;173:465–72.

What is already known on this topic

Neuropsychiatric symptoms (NPS) associated with Alzheimer's disease (AD) include depression, apathy, hallucinations, delusions, wandering and anxiety. Of all the NPS, agitation and aggression are among the most problematic and are associated with increased risk of mortality, earlier institutionalisation, increased cost of care and markedly increased caregiver burden. Over two dozen randomised placebo-controlled trials with a variety of atypical antipsychotics have demonstrated modest efficacy for agitation and aggression, with and without psychotic symptoms. Unfortunately, the use of atypical antipsychotics in patients with dementia is associated with an increased risk of cerebrovascular adverse events and mortality.¹ As a result, clinical practice guidelines recommend their use only after environmental and behavioural interventions have been tried, and only when the behaviours represent a risk for harm to the patient and others.² Because of this unfavourable risk/benefit ratio, other pharmacological interventions have been investigated, though few have been studied in randomised placebo-controlled trials.

Methods of the study

This was a planned secondary subgroup analysis of the CitAD study published previously.³ CitAD was an investigator-initiated double-blind, randomised, placebo-controlled trial utilising generic citalopram up to 30 mg/day over 9 weeks, in patients with probable AD (MMSE 5–28) and significant agitation and/or aggression. The study randomised 186 patients with an average age of 78 and significant agitation and aggression. Prespecified primary outcomes demonstrated statistically significant benefit with citalopram compared to placebo. While the authors suggested that the effect size based on global impression was clinically relevant, the benefit was clearly modest, and ECG evidence of prolonged QTc interval, and possible cognitive worsening associated with citalopram, led to the conclusion that citalopram at 30 mg/day could not be generally recommended as a treatment option.

The study by Schneider et al attempted to explore the heterogeneity of response and identify clinically relevant predictors of response. The study utilised five pre-planned predictors, including residence (outpatient vs long-term care), psychosis, functional impairment severity, level of cognitive impairment and severity of agitation, as well as six other potential predictors, including age, gender and concomitant medication use. Univariate and then a two-stage multivariate interaction method were used for analyses.

What this paper adds

• While the results of the study revealed significant heterogeneity of response, there was no single covariate identified by the univariate analysis that predicted response to citalopram.

• The multivariate analyses identified two subgroups with large and opposite effects. About 20% of participants experienced robust effects with citalopram, about 10% showed significantly greater benefit with placebo, and about 70% showed minimal treatment effects overall.

• The patients who showed the most benefit with citalopram were typically outpatients, had milder cognitive impairment and had less severe agitation/aggression.

• The patients who showed greater benefit with placebo were more likely to be institutionalised, the youngest (ages 45–75) and the oldest (ages 83–92), more cognitively impaired, and had more severe agitation/aggression.

Limitations

• This was a secondary analysis, and the study was underpowered to determine predictors.

• The predictors identified were not readily applicable to clinical practice. For example, the patients that demonstrated the most robust response to citalopram were outpatients, had mild cognitive impairment, moderate to moderately severe agitation and were 76–82 years of age—a truly narrow range of clinical characteristics.

• Most of the sample (70%) demonstrated minimal response to citalopram and placebo.

• The groups with the most positive (and negative) responses were not necessarily consistent or reconcilable with a scientific rationale. For example, the patients in whom placebo was most effective resided in long-term care, and were the oldest (ages 83–92 years) or the youngest (ages 47–75 years).

What next in research

• The search for predictors of response in clinical trials is crucial, especially in conditions like NPS associated with AD where heterogeneity of clinical characteristics, types and severity of NPS, and treatment response is common.

• We need (1) a better understanding of the biological basis of NPS like agitation and aggression, which will allow us to develop biomarkers and help guide the development of new pharmacotherapies; (2) better measures of NPS to determine trial outcomes; (3) safer and more effective therapies for NPS in order to limit patient suffering and caregiver burden.

• Given the potential concern about QTc prolongation with citalopram, an obvious next step would be a trial with escitalopram, which has less tendency to prolong QTc.⁴

Do these results change your practices and why?

No, these results do not change my clinical practice and would be difficult to integrate into the daily management of AD patients. Conversely, the results of CitAD lend more support to the use of citalopram as an alternative to atypical antipsychotics for the treatment of significant agitation and aggression when pharmacotherapy is deemed necessary. In spite of prolonged QTc and possible slight cognitive worsening, these adverse effects are probably more acceptable to patients and caregivers compared with the increased risk of cerebrovascular events and mortality associated with antipsychotics. Concerns about prolonged QTc can be mitigated by monitoring treatment with ECGs.