Vascular Depression: Recent Advances in Neuroimaging and Clinical Research

On February 25, at an afternoon symposium, 4 speakers addressed recent advances in
researching and modeling various factors associated with the diagnosis and outcome of geriatric depression.

Medical Comorbidity and Neuroanatomy in Geriatric Depression: Modeling the Relationships

“I will try to disentangle the role of medical burden, brain structural changes and cognitive performance in geriatric depression,” said Helen Lavretsky, MD, assistant professor of psychiatry, University of California, Los Angeles Medical School. “The task is very difficult to perform because these variables are closely related and they interact and co-exist very frequently,” Dr. Lavretsky said.

In addition to the close relationship and frequent interaction among variables, there are other research challenges involved in analyzing the multivariate components of geriatric depression. Challenges include the often small observed differences between depressed patients and normal controls, the need to control for age and gender, strict inclusion and exclusion criteria, and unclear cause-and-effect relationships. In addition, statistical analysis frequently assumes linear relationships between variables, which do not always exist (e.g., concerning the effect of vascular lesions on depression) .

An earlier model developed by Dr. Anand Kumar and colleagues indicated two alternative paths, via either high intensity brain lesions or frontal lobe atrophy, mediated by age and non-vascular and vascular medical burden. However, cognitive measures were not included in that model. The interaction of the individual risk factors, and the hierarchy of their effects on depression and depression outcomes were not addressed either (Neuropsychopharmacology 2002;26:229-236).

Dr. Lavretsky and colleagues used a statistical procedure called Classification and Regression Tree (CART) analysis to further explore the hierarchy of various predictors of depression, includingdemographic variables, measure ofmedical burden, MRI, and cognitive measures. “CART has the advantage of simultaneously examining all possible interactions among predictors and uncovering both linear and non-linear relationships among predictors,” said Dr. Lavretsky. CART presents a hierarchy of predictors in the form of adecision tree, by finding the best combination of predictors of an outcome.

In an initial study involving 81 patients, Mini-Mental State Examination (MMSE) score emerged as the best predictor of depression diagnosis. A large proportion of patients (those scoring below 27.5) was classified as depressed based on this factor alone. Other important predictors included frontal lobe volumes, lesion volumes, and medical burden (CIRS). In the analysis of a sub-sample with available cognitive test scores, verbal fluency test scores (a test of frontal executive function) was the best predictor of depression, followed by frontal lobe volumes and MMSE, which improved prediction to 100%. (CNS Spectrums, In press)

“CART can be used for modeling non-linear relationships among variables, measuring physical and mental health, as well as MRI and cognitive measures in depressed elderly subjects,” said Dr. Lavretsky. Measures of frontal lobe volumes and executive dysfunction and/or MMSE score combined are superior predictors of depression and need to be used together in the assessment of the outcomes. Replication of these results on independent samples is necessary to test the stability of the CART model.

Multifactorial Model of Geriatric Depression Outcome

“Geriatric depression is aheterogeneous disorder,” said David Steffens, MD, MHS, associate professor of psychiatry and medicine; head, Division of Geriatric Psychiatry, Duke University Medical Center, Durham, North Carolina. “Late-life depression is a very good model for what we were taught in medical school about bio-psychosocial illness,” said Dr. Steffens.

Geriatric depression can occur as a result of various stressors on a brain unequipped to handle them. Biological factors include dexamethasone non-suppression of cortisol, subcortical white-and gray-matter hyperintensities, decreased hippocampal volume, and decreased orbital frontal cortex volumes. “We’re talking a lot about biological variables but there is so much heterogeneity even within that,” said Dr. Steffens.

Various psychological factors, including response to loss, slow speed of processing, concentration and attention difficulties, frontal executive dysfunction, and difficulty processing negative feedback also affect the initial presentation and course of depression. Several social factors contribute as well, such as perceptions and availability of social support, marital status and the number of negative stressors. “People have a lifetime of experiences, and they find themselves in certain social situations that also affect the initial presentation and the course of depression,” said Dr. Steffens. In addition, clinical factors, such as dysthymia, cognitive impairment, activities of daily living, past response to treatment, and baseline depression severity are significant.

Depression is often experienced differently in older adults than in younger individuals, who frequently exhibit a low mood or crying. Symptoms in older people tend to include social withdrawal, apathetic picture, decreased interest in normal activities, cognitive changes, psychomotor agitation or retardation, and generalized anxiety. Those with vascular depression present with symptoms such as major depression and apathetic picture. They tend to have an older age of onset (or change in character of symptoms if early onset), presence of non-CNS vascular disease, and documented brain damage.

Biological, social and psychological factors all play roles in predicting bad outcome in geriatric depression. Predictors include subcortical white and gray matter lesions, lower perceived social support, number of negative stressors, frontal executive dysfunction, functional impairment, and prior depression history. Possible correlates include older age and greater severity of baseline depression.

Dr. Steffens and colleagues developed a hierarchical linear model to study the effect over time of a given factor on depression outcome. Demographic, social, clinical and MRI variables were included. Depression outcome was measured by a change in rating using the Montgomery-Åsberg Depression Rating Scale (MADRS). The model illustrates a MADRS trajectory over time for each subject and then combines individual results to assess the importance of a given predictor variable.

Patients were treated using a staged treatment algorithm for somatic treatment, including both medication and ECT. Results indicated that MADRS declined precipitously in the first few months of treatment, and then declined gradually, to remission levels, after four years. Lower MADRS scores over time were related to a variety of functional, biological, demographic, as well as social support factors. Most significant were the following: passage of time (likely a treatment effect), lower baseline MADRS scores, male gender, higher perceptions of one’s social support during baseline depressive episode, lower baseline gray matter hyperintensity score, and lower IADL impairment.

Several hypotheses emerged from the results. “The good news is it looks like geriatric depression is treatable,” said Dr. Steffens. “But asking about social support, particularly perceptions of one’s social support, is important.” In addition, subcortical gray matter lesions may have more impact on expression and course of depression than deep white matter lesions, and IADL impairment may reflect white matter lesions that accompany depression.

Future studies are needed, using more sophisticated models with longitudinal variables, to explore the role of psychological factors like executive dysfunction and spirituality, as well as other MRI variables (orbital frontal cortex, hippocampus.) Among other things, it is expected that further research will help to better define the role of non-pharmacological interventions (e.g., improving perceptions of support through interactions with families).

MRI Subcortical Hyperintensities and Response to
Sertraline and Citalopram in Geriatric Depressed Outpatients

“It’s unclear, based on prior studies, to what degree the severity of subcortical hyperintensities (SH) predicts treatment response in geriatric depression,” said Stephen Salloway, MD, Director of the Neurology and Memory Disorders Program, Butler Hospital, Providence, Rhode Island; and associate professor of clinical neurosciences and psychiatry and human behavior, Brown University Medical School. Most existing data attempting to explore this relationship comes from poorly controlled trials of severely depressed inpatients.

Based on existing literature, Dr. Salloway hypothesized that people with high SH scores would not respond as well to antidepressant treatment as people with low scores, and would present with more cardiovascular risk factors, especially hypertension. In a test of these hypotheses, Dr. Salloway and colleagues studied differences in sertraline treatment response between geriatric outpatients with high versus low levels of SH (determined using standardized MRI). “It’s the first study to evaluate antidepressant treatment response in a placebo-controlled trial of depressed geriatric outpatients, using a standardized MRI, “ said Dr. Salloway.

Contrary to the hypothesis, patients treated with sertraline did not differ significantly from those given placebo after eight weeks with regard to Hamilton Depression Scale change score and clinical global impression (CGI) scores. Additionally, no difference in five cardiovascular risk factors, including hypertension, was observed between the groups. “At least from this data there’s no difference in treatment response in geriatric outpatients with depression. High SH doesn’t seem to attenuate the response to either drug or placebo. “These are novel data examining the relationship between SH and antidepressant treatment response in geriatric depression,” said Dr. Salloway.

Study limitations include small sample size, fairly restricted range of (mainly low) SH severity, and use of a visual versus a quantitative volumetric MRI rating scale. Future research is indicated, including larger, controlled trials that utilize quantitative MRI, less stringent medical inclusion criteria, and older
participants.

Results from a second placebo-controlled trial, conducted by Dr. Steven Roose and colleagues at Columbia University, using citalopram in 111 depressed geriatric outpatients, age 75 and over, demonstrated that older citalopram subjects had significantly more SH than younger subjects in the sertraline study. As in the sertraline study, there was no effect of SH on treatment response.

In conclusion, age appears to be a risk factor for developing SH, but here may not be an association between SH and antidepressant treatment response in geriatric depressed outpatients.

Executive Dysfunction in Geriatric Depression

“The intention of the vascular depression hypothesis has been to function as an intellectual platform for studies of brain disturbances that lead to late-life depression,” said George S. Alexopoulos, MD, professor of psychiatry, Weill Medical College of Medicine; and professor, Weill Graduate School of Medical Sciences, Cornell University, New York City. Recently, the vascular depression hypothesis has lead to the exploration of cognitive functions in depression.

Prior to the vascular hypothesis, cognitive impairment was often viewed either as a by-product or a concurrent, irrelevant product of depression. Based on this hypothesis, many now believe that cognitive impairment and depression are both meaningful clinical expressions of related brain abnormalities, resulting from vascular and other causes. “It is the same brain disturbance that generates the depressive syndrome and the cognitive syndrome,” said Dr. Alexopoulos. Recent neuropsychological studies conducted by Dr. Alexopoulos and colleagues demonstrate that executive dysfunction is common in geriatric depression, and improves as depression improves, but never normalizes.

Several studies have explored neurobiological correlates of depression. Functional neuroimaging studies have shown that in depression, dorsal neocortical areas of the brain are hypometabolic, while ventral limbic structures are hypermetabolic. Other studies indicate normalization of the metabolism of these areas during remission. In addition, clinical research demonstrates that impairment in tasks of initiation, perseveration, and response inhibition are associated with poor treatment response, low likelihood of remission and high probability for relapse.

Based on these findings, Dr. Alexopoulos and colleagues proposed that mechanisms that lead to executive impairment and the mechanisms that perpetuate affective symptoms in depressed elderly patients are related. Neuroimaging and neuropsychology tests were conducted to dissect the cognitive manifestations of depression and further explore this relationship.

In one study, Dr. Alexopoulos and colleagues used diffusion tensor imaging to examine the integrity of white matter at pre-selected regions. In a small number of subjects, they observed a relationship between integrity of white matter in certain localized regions of the brain (e.g., those lateral to the anterior cingulate) and executive dysfunction, poor antidepressant response, and disability in geriatric depression. Larger scale studies are necessary to confirm these findings.

In a second study, Dr. Alexopoulos and colleagues performed a battery of neuropsychology tests to further dissect the cognitive dysfunction of depression and aging. Results indicate that functions such as selective attention and sustained attention were abnormal in depressives compared to controls, regardless of age. In contrast, inhibitory control and focused effort seem preferentially impaired in patients who had geriatric depression. Studies are needed to determine whether brain abnormalities in inhibitory control and sustained effort are associated with the course of geriatric depression.

Dr. Alexopoulos and colleagues used a cognitive neuroscience approach to test simple tasks associated with identifiable brain circuitry. Reaction times were measured after subjects were exposed to alerting, orienting, and conflict inducing stimuli. Results demonstrate an association between attention scores in the conflict system and persistence of depressive symptomatology in people who received antidepressant treatment with citalopram. In contrast, impairment in the orienting and alerting systems had no association with antidepressant response.

In conclusion, vascular and age-related abnormalities may result in both depressive symptoms and specific cognitive abnormalities. “These cognitive abnormalities can be mapped and catalogued and we can have a ‘phenomenology’ of cognitive impairment inlate onset depression that would further characterize the syndrome,” said Dr. Alexopoulos. In addition, microstructure abnormalities that can now be identified with reasonable precision at pre-selected regions, may have an impact on the course of depression. Finally, cognitive neuroscience paradigms, with or without neuroimaging, may be used to identify specific aspects of the circuitry dysfunction associated with the course of geriatric depression.