Depressive symptoms and mild cognitive impairment in the elderly: an ominous combination

Both depression and mild cognitive impairment (MCI) are common conditions in older adults, yet, until recently, little was known about their individual and conjoint influence on development of neurodegenerative disorders such as Alzheimer’s disease (AD), nor the neurobiology underlying these associations. In the prior decade, lack of knowledge in this area spurred a call by NIMH for research to identify the dementia risk associated with depression and MCI. At the time, such dual-focused studies were hampered by two realities: poor characterization of cognitive performance and cognitive decline when depression was the primary focus of inquiry; and the fact that presence or history of depression was often an explicit exclusion criterion in mild cognitive impairment research. Fortunately, recent studies have been able to bridge this knowledge gap with longitudinal assessment of both depression and cognition.

The relationship between depression and cognitive disorders is complex. Epidemiological studies have long linked depression to development of Alzheimer’s disease. The question has been raised whether depression is a risk factor for development of dementia, or whether depression is a prodromal manifestation of dementing illness. More recent clinical studies have sought to test these hypotheses. For example, presence of depressive symptoms has been shown to increase conversion rates from MCI to AD [1] and dementia [2]. Using neuroimaging and genetic tools, other investigators have focused on understanding the neural basis of the association between depression and incident AD among non-demented depressed elderly.

In this month’s issue of Biological Psychiatry, the study by Lee et al [3] not only confirms the increased AD risk conferred by depression among patients with MCI, but it also examines longitudinal neuroimaging changes to investigate the underlying neurobiology of the relationship. The authors studied 243 MCI subjects from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) with two-year follow up, finding that those with persistent depressive symptoms had higher conversion to AD compared with those who remained without neuropsychiatric symptoms. MCI subjects with comorbid depressive symptoms showed more frontal, parietal and temporal white matter atrophy compared with the neuropsychiatrically asymptomatic patients.

Imaging studies such as Lee’s et al have the potential to inform us not only about brain circuitry in depressed demented patients, but in late life depression more broadly. Certainly, there have been numerous studies identifying neuroimaging changes in late life depression. The principal driver of much research in this area has been the vascular depression hypothesis, which states that cerebrovascular disease may predispose to, precipitate, or perpetuate some geriatric depressive syndromes. Structural imaging research found a greater burden of vascular disease in the white matter and subcortical gray matter among older depressives, and more recent diffusion tensor imaging studies have shown diminished white-matter structural integrity in late life depression, implicating an underlying frontostriatal disconnection syndrome. Beyond vascular depression, longitudinal volumetric neuroimaging studies have compared older non-demented adults with and without depression, linking depression with volume declines in the left frontal white matter [4] and the left hippocampus [5,6].

Based on these and other structural imaging findings (summarized in Table 1), a picture of the neurocircuitry of late life depression has emerged that includes prefrontal cortex, striatum and limbic structures. A rather parsimonious but illustrative circuit model of LLD is shown in Figure. 1. Projections from the orbitofrontal cortex and medial prefrontal cortex to the amygdala, limbic, and brainstem nuclei modulate endocrine, autonomic, and behavioral aspects of emotion. The model is derived from that of Phillips and colleagues [7], which focuses on dorsal and ventral systems, and is also informed by more recent work by Price and Drevets [8]. While the interactions among brain regions are over-simplified, they provide a sense of the complexity of the circuitry likely associated with geriatric depression. When considering cognitive decline and dementia in late life depression, obvious areas of focus include prefrontal cortex and hippocampus.

Table 1.

Recent neuroimaging findings in late life depression with relevance to cognitive decline and Alzheimer’s disease

Neuroimaging finding(s)	Study	Comment
Higher depression symptoms associated with smaller baseline frontal and temporal volumes, as well as greater decline in frontal white matter volume	Dotson et al, 2009 [4]	One of the first studies to examine depressive symptoms and regional brain volumes longitudinally. Sample size 110.
Baseline depression severity predicted a faster rate of decline in hippocampal volume, and incident depression was associated greater hippocampal volume decline	den Heijer et al 2011 [5]	Large (N = 514) population study of older adults that examined bidirectional hypotheses; small baseline hippocampal volume did not predict incident depression
Over 2 years, the depressed group showed a greater reduction in left hippocampal volume (normalized for total cerebral volume)	Steffens et al, 2011 [6]	Decline in hippocampal volume was also associated with subsequent decline in cognitive performance
Older depressed patients showed significant hippocampal shape alterations compared with normal elderly controls	Qiu et al 2009 [10]	Depressed patients with one APOE ε4 allele had more hippocampal shape alterations, a possible marker for AD risk
Older depressed adults with persistent cognitive impairment (PCI) showed baseline decreased activation in the dorsal anterior cingulate cortex, hippocampus, inferior frontal cortex, and insula (compared with older depressed patients without PCI)	Wang et al, 2011 [9]	Functional imaging study examining with two year clinical follow up. Depressed patients with PCI showed decreased activation in similar neural networks that have been associated with the development of Alzheimer disease among nondepressed individuals

Figure 1. Neural Circuit of Late Life Depression.

HPA Axis- Hypothalamic-Pituitary-Adrenal Axis

We are now starting to witness the emergence of functional magnetic resonance imaging (fMRI) as a tool to understand the relationship between late life depression and cognitive decline. Wang et al [9] examined 23 older depressed patients using a functional imaging within-scanner target detection task to assess activation in key brain areas. Patients were followed clinically with depression and cognitive assessments for two years. Those with persistent cognitive impairment (PCI) at two-year follow-up had significantly decreased baseline activation in several brain regions, including the dorsal anterior cingulate cortex, hippocampus, inferior frontal cortex and insula, compared with non-PCI depressed patients. It is interesting that in prior fMRI studies of MCI and AD (without depression), cognitively impaired patients showed decreased activation in the medial temporal lobe (hippocampus and fusiform gyrus) and cingulate cortex. Future functional imaging studies, including fMRI and positron electron tomography imaging for cerebral blood flow and for amyloid labeling, should further our understanding of the relationship between depression and incident dementia.

What is the salient take-home message from studies like Lee et al, which examine neuroimaging links between depression and cognitive impairment? From a clinical standpoint, identification of depression as marker for increased risk of conversion from depression to dementia among patients with MCI is key for those caring for individuals with depression or cognitive impairment, as careful screening for and follow-up on both conditions becomes relevant. Having acknowledged the clinical imperative, one must then ask the simple question: why depression? What are the biological features of depression that it should affect the dementia conversion rate? Certainly one of those features is presence and severity of cerebrovascular disease. The relationship between vascular disease and development of AD has received considerable attention recently. Previously, our group reported that change in white matter hyperintensity volume was significantly associated with development of dementia, although we found that this association was greater among patients who developed non-Alzheimer dementias than among those with incident AD.

Lee et al found changes in frontal, parietal and temporal white matter among depressed MCI patients who converted to AD. Obviously, frontal and temporal white matter changes are consistent with our model (Figure 1) of late life depression. So do structural changes in late life depression accelerate the pathophysiological changes in AD? There is support for this hypothesis not only from the present study, but also in the observation that depression has been associated with increased hypercortisolemia, and some have advocated a depression-as-stress model of depression-related glucocorticoid neurotoxicity, particularly for vulnerable structures such as the hippocampus. Thus, depression-associated hippocampal neuronal loss may negatively and synergistically affect the underlying neurodegenerative processes in AD, leading to earlier clinical expression of disease.

Another plausible hypothesis to consider is that depression may be marker of more advanced structural change associated with neurodegeneration. The longitudinal course of MCI can be variable; depression-associated brain changes may indicate an underlying progression of disease, heralding further cognitive decline and development of AD. This view presupposes the intersection of pathology in common pathways for depression and dementia.

Is there a unique neurocircuitry for depression in dementia, or more specifically for AD-associated depression? Possible support for an affirmative answer to this question may be found in the consensus among clinicians and researchers that there is a separable clinical entity of depression of AD, with a proposed criteria set that requires only three symptoms of syndromal major depression. Certainly, apathy is common in AD, and there has been considerable discussion about the clinical overlap between apathy and any unique depressive syndrome of AD. Frontal lobe dysfunction is an obvious area of intersection in apathy, depression and dementia. Given similar phenomenology of depression of AD with subsyndromal depression, as well as available evidence linking frontal lobe dysfunction to a variety of geriatric neuropsychiatric syndromes, it becomes hard to argue that we have identified a unique brain circuit for AD-associated depression. Rather, one might assert that affected areas of the brain in AD include those that underlie depression, and as the neurodegenerative process hits these regions, one consequence is development of depressive symptoms.

Future pathological studies of well characterized patients with depression and cognitive impairment, including dementia, will help inform us more about the specific neuroanatomy and functional changes unique to depression among demented patients.