Predicting Cognitive Abilities and Treatment Outcome of Major Depressive Disorder in Elderly Patients Using Vortioxetine and Duloxetine Antidepressants

Lead Investigator: Dag Aarsland, King’s College London
Title of Proposal Research: Predicting Cognitive Abilities and Treatment Outcome of Major Depressive Disorder in Elderly Patients Using Vortioxetine and Duloxetine Antidepressants
Vivli Data Request: 8559
Funding Source: None
Potential Conflicts of Interest: None

Summary of the Proposed Research:

Depression affects up to 35% of the elderly population, and is often called late-life depression. In addition to lowered mood and other symptoms associated with depression, many people with late-life depression also experience subjective and/or observed cognitive impairment. Robust evidence suggests that while cognitive impairment for some people with late-life depression is reversible, there is a substantial proportion of those with impaired cognition that continues to worsen and can progress to dementia. In other words, late-life depression is a risk factor or an early stage of dementia. Evidence regarding the effect of antidepressant treatment on improving cognitive deficiency in late-life depression is mixed, indicating antidepressants may be associated with reduced dementia risk, no effect, or even increased risk of dementia. Previous studies found that vortioxetine, a novel antidepressant with a unique pharmacodynamic profile, had a positive short-term effect on cognition compared to placebo and to duloxetine, another anti-depressant. However, it is unknown whether cognitive ability is related to the effect of the drug or to the effect of improved mood and other depressive symptoms. This study aims to explore whether different baseline cognitive or other depression symptoms predict cognitive treatment outcome, and the association between remission of depression and cognitive improvement. We plan to utilize the data set from the previous study, and perform more detailed and advanced statistical analyses to address these questions. Findings of the study will lead to a deeper understanding of the effect of vortioxetine on cognition in people with late-life depression, and factors predicting cognitive outcome. Given the scale of the problem, this is a very important public health concern. In addition, this study will inform future research of whether antidepressants such as vortioxetine can be used for cognitive improvement and reduce the risk of developing dementia.

Statistical Analysis Plan:

The analysis will be performed using a a latent growth curve model (LGCM) approach where longitudinal data with several timepoints is available for the outcome variable. LGCM allows to examine the effect of predictor variables on the trajectory of the outcome variable over time. It assumes that patterns of observed change emerge as a consequence of underlying processes estimated as a series of growth parameters (Clapp et al., 2013). It can track the changes of phenomena over time with either linear or curvilinear trajectories (Burant, 2016). A stepped approach will be used for analysis following Bollen and Curran (Bollen & Curran, 2006). First, a baseline model will be estimated to explore specific type (e.g., linear, quadratic) of overall trajectory of expected recovery. Next, performance of each cognitive task over time will be added to the model. Their slopes and intercepts will be intercorrelated, whose effect size and significance will then be examined to test the hypotheses. Details of constructing the model may be subject to availability of longitudinal measures of cognition and other covariates. Model fit indexes such as comparative fit index (CFI), Tucker-Lewis index (TLI), and root-mean-square of approximation (RMSEA) will be monitored throughout the analysis. For research questions where outcome data is available for fewer than three timepoints, multiple regression analysis will be performed.

Missing data due to withdrawal will be deleted by observation, and only patients with completion of the entire trial process will be included into analysis (392 in total). However, in LGCM models, missing data may be handled using the Full Information Maximum Likelihood method which allows to make best use of all available data for incompletely observed variables ( The final decision on handling missingness for a particular model will be made upon data exploration and following the consultation of an expert statistician. Power calculation and secondary efficacy analyses will be in accordance with the initial study.

Requested Studies:

Randomised, Double-blind, Parallel-group, Placebo-controlled, Duloxetine-referenced, Fixed Dose Study Comparing the Efficacy and Safety of [Vortioxetine] Lu AA21004 in Acute Treatment of Major Depressive Disorder in Elderly Patients
Data Contributor: Lundbeck
Study ID: NCT00811252
Sponsor ID: 12541A

Public Disclosures: