With life expectancy and the percentage of elderly people rising in many developed and undeveloped countries around the globe, cognitive impairment and dementia are steadily gaining in societal importance. Depression and dementia constitute two of the most frequently seen medical conditions in geriatric patients. Approximately 30% to 40% of elderly patients with dementia will experience depression in the course of the disease.¹ Above and beyond the well-known parallels between neurocognitive symptoms in dementia and in acutely ill states of mood disorders (formerly termed "depressive pseudodementia"), evidence of a real association of mood and dementing disorders is accumulating. In most patients who have suffered a first affective episode, the course of illness is episodic or chronic. More than 90% of patients after a manic, and approximately 60% after a depressive episode will experience recurrence.² In 30% to 50% of patients, mild depressive symptoms and neurocognitive and psychosocial impairment persist into the remitted phases of the illness.³,⁴ Residual symptomatology and psychosocial impairment appear to be associated in a near-linear fashion.⁵ Residual symptoms lead to a higher utilization of the health care and social systems, a higher rate of recurrences, more hospitalizations and a higher rate of suicidal behavior.⁶ Persistent neurocognitive impairment is a major predicting factor for illness-related deterioration of quality of life (see Green et al.⁷ and Fujii et al.⁸ for patients with schizophrenia, and Depp et al.⁹ for patients with bipolar disorder).

The aim of the present review is to summarize the knowledge on cognitive impairment in acute and remitted phases of mood disorders and in dementia and the association of mood disorders and dementia. Additionally, we outline the neurocognitive profiles differentiating mood-associated and residual cognitive impairment from dementia and translate ideas on pathophysiological causation and progression into recommendations for patient management.

We identified relevant articles through a MEDLINE search using the key words: mood disorder, bipolar, depression, mania, dementia, cognitive impairment and cognitive function.

Up to half of all patients with mood disorders suffer from interepisodic residual symptoms in their remitted phases. Core affective or neurocognitive features of these residual symptoms have to be distinguished from treatment-emergent features, such as fatigue secondary to a side effect of medication or to other comorbid conditions. Of all patients with a score on the Hamilton Rating Scale for Depression (HAM-D) not exceeding 7, only about 20% reported being symptom-free. The most frequent residual symptoms were sleep disturbances, fatigue, and apathy, all tending to predate treatment.¹⁰ Sexual dysfunction, reported by approximately half of the patients with partial remission of mood disorders,¹¹ is coming more and more to the attention of clinicians. Thirty percent to 50% of fully remitted patients fail to attain premorbid levels of psychosocial functioning, possibly due to cognitive impairment.²,¹² The percentage of euthymic patients suffering from neurocognitive impairment, however, is not yet clear.¹³—¹⁵ Residual treatment-resistant features and neurocognitive impairment might be the leading causes of psychosocial impairment and lower quality of life.

Neurocognitive impairment in unipolar depressive and bipolar patients shows overlapping and distinct features. In a systematic review, Quraishi and Frangou¹⁶ differentiated data on neurocognitive impairment in bipolar patients seen in acute and/or euthymic phases and compared it with unipolar depressive patients. General intellectual function was largely preserved and impairment limited to acute episodes and performance scores. Attention showed abnormalities in acute and euthymic phases, with special focus on sustained attention and inhibitory control. Verbal memory was impaired in acute and euthymic phases, and visuospatial performance varied depending on the task. Executive function was impaired in all aspects in symptomatic patients.¹⁶ Martinez-Aran et al.¹⁷ compared euthymic and depressed bipolar patients in a cross-sectional study and showed that, after controlling for residual symptoms, the functional outcome in euthymic patients was related to verbal fluency. Balanzá-Martínez et al.¹⁸ reassessed neurocognitive functionig in euthymic bipolar patients 3 years after initial testing in a prospective follow-up study. They reported the same pattern and severity of neurocognitive deficits in the acute and euthymic phases.

A recent systematic review shows that in euthymic bipolar patients, both depressive and manic episodes in disease history are related negatively to neuropsychological functioning. Manic episodes were related to delayed verbal memory and executive functioning, whereas depressive episodes evidenced a more inconsistent influence on a broader range of neuropsychological functions. Verbal memory performance was negatively correlated to the number of previous manic episodes, duration of illness and hospital admission, all of which are highly intercorrelated variables. Unaffected first-degree relatives of bipolar patients showed subtle signs of impairment in several components of executive functioning and verbal memory, indicating that this might be a trait vulnerability factor for bipolar disorder.¹⁹

Mood disorders are associated with structural changes in the prefrontal cortex, including the cingulum, amygdala, thalamus, and, variably, the hippocampus,²⁰,²¹ although the data for bipolar patients are less consistent.²²,²³ In the prefrontal cortex, affected regions seem to be the dorsolateral and subgenual prefrontal cortex and the anterior cingulate cortex.²²,²⁴ Imaging techniques also aim at associating cognitive impairment with the functioning of specific brain regions. In a cross-sectional study, event-related functional magnetic resonance imaging (MRI) revealed dysfunction of the left ventral prefrontal cortex (VPFC) in manic, depressed, and euthymic bipolar patients during a Stroop test assessing executive functioning.²⁵

Mild cognitive impairment is defined by deficits in memory, normal daily activities, normal general cognitive functioning, abnormal memory function for age, and absence of dementia.²⁶,²⁷

Dementia is characterized in Alzheimer’s disease by the insidious onset of a slowly progressive deterioration in higher cognitive functions impairing daily activities. In vascular dementia, the initial course is more rapid or uneven. Secondary dementias occur in the context of endocrine disorders, encephalitis, or intoxications.

Essential characteristics of dementia include the development of multiple cognitive deficits including memory impairment and at least one of the following features: apraxia, aphasia, agnosia, or disturbance of executive functioning.

Compared with normal aging, dementia has a significant impact on everyday functioning. In the general population, memory impairment without dementia is present in 20%, mild cognitive impairment in 3.2%.²⁸ The prevalence of dementia rises exponentially from the sixth life decade onward to well above 40% in the cohort of people over 90 years of age. Approximately 15% to 20% of cases with mild cognitive impairment will develop dementia within 2 years, with longer observation time up to 70%.²⁹,³⁰ There is a scientific debate as to whether mild cognitive impairment is just early dementia or a separate entity,³¹ and some call for the integration of incorporating a measure of cognitive decline into the definition of mild cognitive impairment.³²,³³

First, as shown above, unipolar depressive as well as bipolar disorders seem to be accompanied by persistent neurocognitive impairment, even in interepisodic euthymic phases. Second, there are hints that mood disorder patients have a higher risk of mild cognitive impairment and dementia.¹⁴,³⁴,³⁵ In patients with a history of depression, a meta-analysis by Jorm et al.³⁴ reported relative risks for dementia of 1.87 (95% confidence interval 1.09 to 3.20) to 2.01 (95% confidence interval 1.16 to 3.50). In a Danish registry study with the prospectively collected data of 165,000 discharged patients, the risk of rehospitalization with a diagnosis of dementia was 4.2 for patients discharged with depression and 3.4 for patients discharged with mania in comparison to patients with other chronic diseases not affecting the CNS (RR given here for comparison to osteoarthritis, comparison to diabetes mellitus led to similar results).³⁵ The risk seems to increase with the number of episodes. After four episodes, the risk ratio was 2.05 (95% confidence interval 0.39; 10.68), and after five episodes or more, the ratio was 3.19 (95% confidence interval 0.69; 14.75) compared with the risk after the first episode.³⁶

On the other hand, in a 5-year prospective follow-up study including 766 elderly subjects, depression at baseline was predictive for the development of dementia.³⁷ In another prospective study including 114 anamnestic mild cognitive impairment patients, Modrego and Ferrandez³⁸ showed that the prevalence of depression at baseline increased the risk of dementia significantly (mean follow-up 3 years, hazard ratio 4:1; 95% confidence interval 2.4 to 6.9).

Hypotheses proposed to explain the neurodegenerative or at least neurocognitive impairment in the course of the disease include prefrontal and limbic damage by, for example, changes in the glucocorticoid hormone system and neurotransmitter pathways associated with glutamate.³⁴,³⁹—⁴³ Kempermann and Kronenberg⁴¹ suggest that cellular plasticity including neurogenesis might be severely and chronically disturbed in mood disorder patients. In these patients, the cognitive systems for coping with challenges of novelty and complexity might be continually working at their limits, potentially for a long time, without showing clinical symptoms. Naturally occurring cell death primarily affects newly generated neurons. In mood disorders, older neurons, for instance in the hippocampus, might die and the loss of older granule cell neurons might again affect regulation of adult neurogenesis. Environmental enrichment up-regulates adult hippocampal neurogenesis from the very low baseline levels in senescence. The impairment (as in mood disorders) could be due to a relative mismatch of functional challenge and possible plastic response.⁴¹ Adult hippocampal neurogenesis might be necessary to cope with complexity and novelty, and failure to cope might impair learning, learning being one way of coping with novelty. Rats with a tendency to be more stressed by exposure to new stimuli showed less adult hippocampal neurogenesis than did those with a relaxed response.⁴⁴ In mood disorders, especially in depressed states, patients show aversion to novelty and withdraw into routines; their stress resistance is reduced.

In a post-mortem study by Rapp et al. brains of Alzheimer’s patients with a lifetime history of depression showed significantly higher levels of plaque and tangle formation in the hippocampus than did the brains of patients without a history of depression.⁴⁵

Of the many models for the association of mood disorders and dementia, two are favored by several authors. One model suggests that mood disorders are a risk factor for earlier clinical manifestation of dementia. The second sees mood disorders as the cause of dementing states, for instance through neuronal loss via dysregulation of the glucocorticoid cascade.³⁶,⁴⁶ Manji et al.⁴⁷ suggest that impairment of neuroplasticity and cellular resilience may underlie the pathophysiology of mood disorders as such, and not only of neurocognitive impairment. Findings of discrete neurocognitive impairment in healthy co-twins discordant for unipolar depressive and bipolar disorders suggest that impairment of specific neurocognitive functions may, in some patients, be present before the onset of mood disorder and may consitute a trait factor or even an endophenotype.⁴⁸,⁴⁹

The differential diagnosis of mood disorders and dementia, especially in the early stages of disease, requires expertise on the part of the caring specialist. It remains difficult despite numerous efforts to delineate the two disorders.⁵⁰ Depression is frequently misdiagnosed as dementia and up to 32% of patients referred to special departments for dementia evaluation actually suffer from depression.⁵¹ On the other hand, depression often occurs in the early stages of dementing diseases. Early detection of dementia prepares the ground for delaying cognitive decline and gives the caregivers time to adjust to the new circumstances.

The three major conditions requiring differentiation are depicted in Figure 1. The conditions to be differentiated from (early) dementia are mood-associated cognitive impairment in acute episodes,(1) residual cognitive impairment in remitted patients with mood disorders (2) and, if possible, high risk patients with mood disorders who are expected to develop dementia in the future.(3)

In 1979, Wells proposed a widely quoted list of features suggested to differentiate depression with cognitive impairment (formerly termed depressive pseudodementia) from progressive dementia.⁵² Some features gained empirical support⁵⁰ and others lack predictive power (e.g., speed of progression). An update of the differential anamnestic, clinical, behavioral, and neuropsychological profiles of patients with present depression and (early) dementia is given in Table 1. Behavior at examination, deficit profile, and severity of the deficit add important information for the neuropsychological diagnosis.

The caring specialist’s first diagnostic step for patients presenting with cognitive impairment is to rule out neurocognitive decline due to normal aging. Functions affected by normal aging are mainly slowed recall of new information and problems with nonverbal memory (e.g., misplacing things). Usually, daily functioning is not grossly impaired.

It is essential to exclude the presence of comorbid conditions with secondary neurocognitive impairment by interviewing and physical examination. Additionally, a variety of medications can cause a nonprogressive cognitive impairment which resolves quickly after discontinuation.

Apart from a thorough anamnestic interview (not only with the affected patient but also with a family member or caregiver) and laboratory analyses, including genotyping, appropriate assessment of neurocognitive functioning is warranted. Although neurocognitive testing in affective patients leads to more valid results in stable euthymic states, screening tests for mild cognitive impairment have been developed that prove stable even with comorbid depression.⁵³

Screening tools for mild cognitive impairment and dementia are the commonly used Mini-Mental Status Test (MMST) and the DemTect,⁵⁴ the latter being more sensitive to mild cognitive impairment.⁵⁵,⁵⁶ Additionally, the Clock Drawing Test has been found effective for the detection of very early stages of cognitive impairment.⁵⁷ When early dementia is suspected, a neuropsychological evaluation is needed using well-standardized tests of cognition. Here, the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD)⁵⁸ test battery or tests from the Nuremberg Gerontopsychological Inventory (NAI)⁵⁹ are suggested. The results can help to validate the diagnosis and plan treatment. Neuropsychological testing is considered the most sensitive way to determine the presence, nature and severity of deficits.⁵⁰ Recent studies comparing elderly comparison subjects, elderly depressed patients and patients with progressive dementia suggest that the differentiation between the groups is fairly straightforward and facilitated by neuropsychological evaluation.⁶⁰

Jean et al.⁶¹ used the data of 44 depressed patients on the MMSE and the Dementia Rating Scale (DRS). At follow-up 7.5 years later, subjects who developed dementia had shown greater impairment on tasks measuring attention and memory at baseline than those who did not develop dementia. Future Alzheimer’s disease patients were differentiated on the basis of their difficulties on the MMSE-orientation subtest, whereas the patients with future non-Alzheimer’s dementia initially had more problems with executive functions and visuospatial abilities.⁶¹

Of patients with mild cognitive impairment, those that progressed to dementia during prospective follow-up were characterized by a higher severity of impairment on tests of episodic memory at baseline. Additionally, severity of impairment of semantic memory, attentional processing, and speed were predictive markers for progression.⁶²,⁶³ Swainson et al.⁶⁴ presented data showing that a visuospatial associative learning test (paired-associates learning [PAL] test, part of the Cambridge Neuropsychological Test Automated Battery) accurately distinguished patients with Alzheimer’s disease from depressed and comparison subjects.

Neuroimaging procedures are also being used increasingly for the differential diagnosis of dementing disorders. Techniques, such as magnetic resonance spectroscopy (MRS) and diffusion tensor imaging (DTI), advance the prospects of magnetic resonance imaging (MRI), single photon emission computed tomography (SPECT), and positron emission tomography (PET). CSF analysis assessing the concentration of tau protein (p-Tau231) can be used to differentiate geriatric patients with Alzheimer’s dementia from those with depression.⁶⁵

An appropriate, early pharmacological intervention in mood disorders is essential for achieving partial or full remission of symptomatology. Pharmacological approaches that specifically influence residual symptoms are studied only rarely. Though there is some evidence for decrease of residual fatigue, anxiety, or sexual dysfunction,⁶⁶—⁶⁸ little is known regarding neurocognitive impairment. Antidepressants and mood stabilizers exert major effects on the signaling pathways that regulate neuroplasticity and cell survival. Lithium, for example, stimulates cell proliferation in the subgranular zone and enhances expression of bcl-2, suggesting that its beneficial effect derives from interaction with hippocampal plasticity.⁶⁹

Even with appropriate pharmacological treatment, only a minority of patients reaches full symptom remission. Nonpharmacological interventions have long been an essential part of treating unipolar depression and are increasingly being studied in bipolar disorder. Cognitive behavior and interpersonal psychotherapy seem to reduce relapse risk in acute and subacute phases as well as shortly after remission.⁷⁰,⁷¹

Since monotherapy is in most cases not sufficient for sustained recovery, the combination of nonpharmacological with pharmacological treatment is recommended for greater efficiency.⁷²

Early, appropriate pharmacological as well as nonpharmacological treatment of mood disorders, including long-term prophylaxis, are essential in the prevention of persistent neurocognitive impairment in these patients. Recently, interest in potential neuroprotective interventions has also been rising. There is a growing literature on the neuroprotective effects of lithium which mainly stems from studies with cell cultures and animal research. In the treatment of chronic—but not acute—mood disorders, lithium appears to have robust neuroprotective effects against a variety of insults, including glutamatergic damage, ischemia, neurodegeneration, and oxidative stress. The effects of lithium include prevention of cellular damage and loss as well as, in some instances, reversal of damage after subsequent treatment with lithium.⁴⁷,⁷³ The effects may be shared by other mood-stabilizing agents, such as valproate, but the relevant literature is far less extensive than that available for lithium.⁷⁴,⁷⁵ The mechanisms for lithium’s neuroprotective effects appear to be diverse. Current models suggest that lithium modulates signal transduction pathways,⁷⁶,⁷⁷ inhibits glycogen synthase kinase-3β (GSK3β),⁷⁸,⁷⁹ and decreases N-methyl-D-aspartate (NMDA)-mediated excitotoxicity.⁸⁰,⁸¹

At present, little is known about the potential neuroprotective effects of lithium treatment on patients. Recent evidence suggests that the neuropsychological impairment experienced by bipolar patients may be related more to the disease itself than to drug treatment.⁸² In fact, research indicates that chronic lithium treatment may correct some of the previously reported neurocognitive abnormalities in these patients.⁴⁷,⁸³ Englesman et al.⁸⁴ reported stable memory performance over a 6-year period in 18 bipolar patients. Letendre et al.⁸⁵ reported preliminary results from eight patients suffering from HIV-related neuropsychological impairment who responded to a 12-week lithium treatment. Moore et al.⁸⁶ were able to show that a 4-week lithium treatment increased gray matter volume throughout the brain in bipolar patients. Unfortunately, no region-specific imaging data, imaging data on long-term treatment with lithium, or data on treatment of older patients are available.

Magnetic resonance spectroscopic (1 hour-MRS) analysis revealed that after 4 weeks of lithium treatment, the levels of N-acetylaspartate derivates (NAA, a potential marker for neuronal density and viability) were increased by about 5% in the brain.⁸⁷,⁸⁸ The high correlation between lithium-induced NAA increase and regional voxel gray matter content indicated that the NAA increase was occurring primarily in CNS gray matter.⁸⁷ It is feasible that these MRS changes are related to the neuropsychological alterations of bipolar patients, since Li et al.⁸⁹ showed a significant correlation between abnormal 1 hour-MRS spectra and neuropsychological deficits in patients suffering from epilepsy.

Studies are underway to evaluate whether long-term lithium treatment in bipolar patients can affect neurocognitive impairment. The hypothesis of a lower prevalence of mild cognitive impairment or dementia in lithium-treated patients with mood disorders still has to be evaluated.

One could also envision that continuous neurocognitive training in unipolar depressive and bipolar patients could help to protect cognitive functioning from declining in the course of disease. To date, however, there are no data available to verify this hypothesis.

The study types found in the literature search are for the most part cross-sectional and case-controlled, and are therefore characterized by a high risk of bias. Additionally, inference on causality cannot be drawn from cross-sectional studies and only with caution from case-control studies. While registry data may include information from several time points, the evidence levels which emerge are usually inferior to those obtained in individual prospective observation. Prospective studies usually have a short follow-up, with the consequent problem that outcomes manifest only after longer periods of time, especially with low numbers of participants. Another important problem is the difficulty in differentiating circumscribed study populations (e.g., late onset depression and dementia). Additionally, covariates influencing cognition, such as comorbidity and medication, can only be monitored effectively with adequate numbers of participants and careful study design. Difficulties add up in meta-analyses of heterogeneous studies, affecting usefulness of the results.

In differential diagnosis, assessment of neurocognitive impairment profiles complements information from disease history and comorbidity, physical examination and other diagnostic tools. With growing evidence that mood disorders increase the risk of dementia, the early identification of high risk individuals has become an important issue. Cognitive domains such as attention, verbal, semantic and visuospatial memory are affected in remitted patients with mood disorders as well in elderly subjects progressing from mild cognitive impairment to dementia. To increase the body of evidence, intelligent study designs focusing on subject definition, application of adequate cognitive tests, and thorough evaluation of disease history, comorbidity, and medication are required. In terms of management, early patient education, adequate treatment, and use of possible preventive strategies will become increasingly necessary as knowledge of risk factors and effectiveness of special interventions grows.

1=Differential diagnosis mood-associated cognitive impairment versus early dementia; 2=residual cognitive impairment in remitted patients with mood disorders; 3=identification of high risk patients with mood disorders for development of dementia

Thanks to Catherine Aubel for her help in editing the manuscript.