|Home | About | Journals | Submit | Contact Us | Français|
While late-life depression (LLD) and amnestic mild cognitive impairment (aMCI), alone and in combination, is associated with an increased risk of incident Alzheimer’s disease (AD), the neurobiological mechanisms of this link are unclear. We examined the main and interactive effects of LLD and aMCI on the gray matter (GM) volumes in seventy-two physically healthy participants aged 60 and older. Participants were separated into normal controls, cognitively normal depressed, non-depressed aMCI, and depressed aMCI groups. Optimized voxel-based morphometry estimated GM volumes. The main and interactive effects of LLD and aMCI, and of depressive symptoms and episodic memory deficits on the GM volumes were analyzed. While decreased GM volumes in the mood regulating circuitry structures were associated with depression, GM atrophy in regions essential for various cognitive performance were related to aMCI. LLD-aMCI interactions were associated with widespread subcortical and cortical GM volume loss of brain structures implicated in AD. The interactions between episodic memory deficits and depressive symptom severity are associated with volume loss in right inferior frontal gyrus/anterior insula and left medial frontal gyrus clusters. Our findings suggest that the co-existence of these clinical phenotypes is a potential marker for higher risk of AD.
An estimated 35 million people worldwide are living with Alzheimer’s disease (AD) today, and this number is anticipated to rise to 113 million by year 2050 . While AD-disease modifying agents are currently in various phases of clinical trials, several of these have failed to show benefit. Modification of vascular, lifestyle, and psychosocial factors may prevent or reduce the incidence of AD in a significant proportion of individuals . Late-life depression (LLD), one such modifiable risk factor, affects a significant proportion of older adults, and is associated with poorer outcomes of co-morbid medical disorders, increased mortality risk and incident cognitive decline .
Late-life depression (LLD) is associated with increased risk of developing mild cognitive impairment (MCI) and AD [2, 4, 5]. In LLD, volume declines in the brain regions earliest affected by AD, including the frontal and medial temporal lobe (MTL) structures are seen, although these findings are not universal [6–15]. Circumscribed hippocampal abnormalities are reported in LLD, and depressed elderly with MTL volume loss may be at greater risk for development of dementia [16, 17]. Early evidence has also suggested a role of amyloid and neurofibrillary tangles in the pathophysiology of LLD [18, 19]. While these findings implicate neurodegeneration as a potential mechanistic link between depression and AD, we currently cannot predict who with LLD will progress to AD. The co-existence of LLD with persisting cognitive impairment, which often occurs in the elderly, accelerates the conversion to AD by neurobiological mechanisms that are unknown .
Amnestic mild cognitive impairment (aMCI), another common neuropsychiatric syndrome of the elderly, is a prodrome of AD in many individuals. However, while 10 to 15% of individuals with aMCI progress to AD annually, the conversion rates are highly variable and some revert back to normal . Deficits in multiple cognitive domains are robust predictors of aMCI conversion to AD . The severity of neurodegeneration in the MTL, insula and neocortical areas may separate aMCI converters from non-converters [23–25]. Moreover, the co-existence of depression and aMCI, is associated with greater progression to AD, relative to nondepressed aMCI, suggesting that their co-morbidity is a marker of disease severity . The differential GM volume effects by LLD and aMCI, when occurring alone or when they co-exist, are not fully understood.
We aimed to test the hypothesis that while LLD will be associated with GM volume reductions limited to the regions involved in emotional regulation, aMCI will be related to GM volume loss in areas involved in multiple cognitive functions. Furthermore, we hypothesized that the LLD-aMCI interactions will be associated with widespread GM volume loss in regions commonly involved in mild AD.
Our sample comprised of seventy-two participants aged 60 and older (cognitively healthy, late-life depression [n=18]; amnestic mild cognitive impairment [n=17]; late-life depression with co-existing aMCI [n=12]; and cognitively normal, nondepressed controls [n=25]). All patients diagnosed as clinically significant depression and/or aMCI were recruited from the geriatric psychiatry and memory disorders clinics at the Medical College of Wisconsin. Control subjects were recruited from the community through local advertisements. All participants provided written informed consent and the study protocol were approved by the Medical College of Wisconsin Institutional Review board. During study visit 1, study participants received detailed clinical and neuropsychiatric assessments. The core neuropsychological battery administered to all participants included the (1) Mini-Mental State Examination (MMSE) (all subjects had to score ≥24); (2) Mattis Dementia Rating Scale-2 (MDRS-2) (age and education corrected MOANS scaled score of ≥5); (3) education adjusted Logical Memory II Delayed paragraph recall (LMII-DR) subscale from the Wechsler Memory Scale-Revised; (4) Physical Self Maintenance Scale/Instrumental Activities of Daily Living (PSMS/IADL); (5) 30-item Yesavage Geriatric Depression Scale (GDS); (6) Diagnostic assessment for Axis 1 disorders including the depression module from the Structured Clinical Interview for DSM IV; and (7) Hamilton Anxiety Scale (HAM-A). All participants had to score ≤4 on the modified Hachinski ischemic scale (HIS). Clinical assessment findings were reviewed during the weekly consensus conference attended by neurologists, neuropsychologists and a geriatric psychiatrist.
Inclusion criteria for cognitively normal subjects with LLD included a positive SCID for clinically significant depression (defined as having major or minor depression or dysthymic disorder). Seventeen participants met SCID criteria for major depression and one patient met criteria for minor depression. All participants had a GDS score of 10 or above, MMSE ≥26, PSMS ≤6 and IADL ≤9, score above the education-adjusted cutoff on the LMII-DR (Delayed recall score >8 for 16 or more years of education or score >4 for 8–15 years of education). Among the 16 out of the 18 cognitively healthy participants with LLD on antidepressants, 7 were on antidepressant monotherapy (selective serotonin re-uptake inhibitors (SSRI): N=2, serotonin norepinephrine reuptake inhibitors (SNRI): N=4, and bupropion: N=1); and 9 were on combination treatment.
Amnestic mild cognitive impairment (aMCI) (N=29) was operationally defined according to the established criteria: (1) subjective report of cognitive decline; (2) objective cognitive impairment that includes scoring 1.5 SD below on memory measures; (3) intact ADLs and relatively preserved IADLs; and (4) no dementia. For meeting criteria for objective cognitive impairment, participants had to score below the education-adjusted cutoff on the LMII-DR (i.e., ≤8 for 16 or more years of education, and ≤4 for 8–15 years of education), and score below 1.5 SD below the mean on one or more subscales (one of the impairment had to be memory) of the Repeatable Battery for the Assessment of Neuropsychological Status (RBANS), Behavioral Dyscontrol scale or the Boston Naming Test. Fourteen of the 29 aMCI subjects were on cognitive enhancing medications (cholinesterase inhibitors monotherapy: N= 9; memantine monotherapy: N=4; combined: N=1).
All participants with aMCI who endorsed clinically significant depression on SCID (N=12) were included in the aMCI with co-existing depression group (major depression (N=11). One aMCI subject who scored a 9 on the GDS met SCID criteria for dysthymic disorder and was also included in this group. Among the 10 out of the 12 aMCI subjects with LLD on antidepressants, 5 were on antidepressant monotherapy (selective serotonin re-uptake inhibitors (SSRI): N=2, serotonin norepinephrine reuptake inhibitors (SNRI): N=2, and bupropion: N=1); and 5 were on combination treatment.
Antidepressant doses had to be at stable doses for at least two weeks before study enrollment and until completion of the study procedures.
A sample of nondepressed, cognitively healthy participants (N=25) was recruited as the control group. Subjects in this group could not meet criteria for MCI or any Axis 1 psychiatric disorders or be on psychoactive medications.
Exclusion criteria included past or current history of concurrent Axis 1 psychiatric disorders, such as psychotic or bipolar disorders; alcohol or substance abuse/dependence during the past 5 years; active suicidality; MMSE scores <24; History of neurological disease including parkinson’s disease, dementia, multiple sclerosis, seizures, or stroke; Head injury with loss of consciousness; MRI contraindications; and unstable chronic medical conditions.
During study visit 2, High-resolution Spoiled Gradient-Recalled Echo (SPGR) 3D axial images were acquired using a whole-body 3T Signa GE scanner (Waukesha, Wisconsin) with a standard transmits-receive head coil. The parameters were: TE/TR/TI of 4/10/450 ms, field of view (FOV) of 24*24 cm, flip angle of 12°, number of slices of 144, slice thickness of 1 mm, matrix size of 256×192 and voxel size of 0.938*0.938*1 mm3 (x, y, z). GDS was completed prior to MRI scans.
Optimized Voxel-based Morphometry (VBM) combining with Statistical Parametric Mapping (SPM8) software (http://www.fil.ion.ucl.ac.uk/spm) on a Matlab 7.5 (MathWorks, Natick, MA, http://www.mathworks.com/) platform were utilized to perform GM volume analysis . First, segmentation of the brain tissues was conducted in original space using the analysis of likelihood of each voxel of the brain images to be classified as GM, white matter (WM), or cerebrospinal fluid (CSF). Second, to create the study group-specific template, the high-resolution structural GM images for all participants were averaged after images were spatially normalized in the same stereotactic space based on the Montreal Neurological Institute template using 12 affine transformation and 16 nonlinear iterations. Third, the second segmentation was performed by using spatial normalization parameters for the native images based on study group-specific GM template, followed by interpolation of voxel size 2*2*2 mm3 using the trilinear interpolation method recommended in the SPM8 manual. This method, although time consuming, is more accurate than the Nearest Neighbor method. The 2-mm isotropic resolution was chosen because the down sampling has been found to provide results that are less dependent on the original position, scale and orientation of one dataset with respect to the other and to decrease the number of statistical tests by a factor of 8 . Fourth, the absolute GM volume was modulated for each subject by Jacobian determinants derived from the spatial normalization. Finally, all segmented gray matter images were smoothed using 8mm Gaussian kernel and transformed using logit transformation (logit(n) = 0.5 ln (n/(1 − n))) to make the data normally distributed for the further analyses.
Group comparisons for demographic information (age and education) were compared by analysis of variance (ANOVA), gender comparison was performed with χ2 test (p<0.05 to avoid false negative findings). Neuropsychological measurements were compared by analysis of covariance (ANCOVA), after controlling for age, gender and education. Neuropsychological and behavioral measures for group comparison were considered significant at p < 0.0045 (Bonferroni corrected, p<0.05/11≈0.0045), SPSS 18.0; SPSS Inc, Chicago, Illinois). The sources of the differences between the means of the four groups were examined by post hoc Fisher’s Least Significant Difference test for demographics; and post hoc Bonferroni-corrected tests were performed when appropriate for neuropsychological measures (after rejecting the null hypothesis that all group means are equal).
Volumetric measurements in the four study groups were tested for normal distribution and for homogeneity of variance. To find brain regions with significant group differences in gray matter volumes, 2×2 (depression × mild cognitive impairment) voxel-wise analysis of covariance (ANCOVA) was performed at p< 0.01 significance with false discovery rate (FDR) corrected for multiple comparisons across all subjects (number of contiguous voxels > 150). We employed the FDR and not the more conservative family wise error (FWE) procedure when performing correction for multiple comparisons. The FDR procedure controls for the proportion of false positives among the tests that show significant results. On the other hand, FWE correction has the tendency to wipe out both true and false positives when applied to the entire dataset that contain large number of voxels, thereby producing false negative findings (Type II error). Therefore, we believe that FDR is more appropriate than FWE for this preliminary VBM study [35, 36]. Age, education, gender, and intracranial volume (ICV) were included as covariates of no interest.
To investigate the main and interactive effects of the depressive symptoms and episodic memory scores on GM volume, a voxelwise multiple linear regression analysis was employed (3dRegAna, AFNI) between the GM volume of each subject and the GDS and LMII-DR scores in the four groups, respectively using the following equation as used in previous publications [37, 38]:
where GMi is the gray matter volume of ith voxel across group subjects, β0 is the intercept of the straight line fitting in the model; β1 and β2 are the main effects of the GDS and LM II-DR scores, β3 is the interactive effects of GDS and LMII-DR scores, β4, β5, β6, β7, and β8 are the effects of age, gender, education, group, and ICV as covariates of no interest in the above linear regression model. The cluster level was set at p< 0.01 with false discovery rate (FDR) corrected for multiple comparisons across all subjects (the cluster level threshold set at number of contiguous voxels > 150).
The demographic and neuropsychological characteristics are summarized in Table 1. While there were no education and gender differences between the four groups (p>0.05), cognitive healthy LLD group was significantly younger than the CN and nondepressed aMCI subjects (p=0.043).
Significant differences were found in MMSE, DRS-2 (attention, conceptualization, memory and total), LMII (immediate and delayed), HAM-A and GDS scores, after controlling for age, gender and education. Post-hoc analyses revealed the source of differences (Bonferroni p<0.0045). Depressed aMCI subjects had lower MMSE scores, relative to CN subjects. When compared to the CN and cognitively healthy subjects with LLD, the aMCI groups (depressed and nondepressed) performed poorly on DRS-2 memory and total, and LMII immediate and delayed memory recall scores. When compared to cognitively normal control and depressed groups, depressed aMCI subjects performed poorly on the DRS-2 conceptualization measure. The nondepressed aMCI subjects performed worse on the DRS-2 attention and conceptualization scores, relative to the nondepressed controls. Relative to the nondepressed groups, depressed (aMCI and cognitively healthy) subjects endorsed greater depressive and anxiety symptoms (Table 1).
The main effects of depression included GM volume loss in the bilateral dorsal cingulate cortex (dACC), orbitofrontal cortex (OFC) and ventromedial (vmPFC) prefrontal cortices; insula (Ins), posterior middle temporal gyrus (pMTG), middle occipital gyrus (MOG) and inferior frontal gyrus (IFG) on the left; and the right posterior cingulate cortex (PCC) and precuneus (Pcu) (Figure 1 and Table 2).
Main effects of mild cognitive impairment included GM volume loss in the bilateral amygdala/hippocampus (Amy/Hippo), vmPFC, ventrolateral PFC (vlPFC), insula, and anterior temporal pole (aTP); PCC/precuneus (Pcu) cluster, pMTG and retrosplenial cortex (RSC) on the right; and the left fusiform area (FFA); (Figure 1 and Table 3).
The main effects of depression and mild cognitive impairment were not associated with increased GM volumes in any brain regions.
The interactive effects of depression and mild cognitive impairment included GM volume loss in several brain regions including bilateral dorsomedial prefrontal cortex (DMPFC), dACC, vmPFC and cuneus; lentiform nucleus (LN) and superior frontal gyrus (SFG) on the left; and lingual gyrus (LG), Amy/Hippo and postcentral gyrus (PostCG) on the right (Figure 1 and Table 4) (p<0.01, FDR corrected).
After controlling for age, gender, education, ICV and individual group effects, depressive symptoms and episodic memory deficit interactions were seen in the right IFG/anterior insula (rIFG/aIns) and the left medial frontal gyrus (MeFG) clusters (p<0.01, FDR corrected) (Figure 2). There were no significant individual effects of depressive symptoms and episodic memory deficits on the GM volumes seen.
This is the first study to identify main and interactive relationships of LLD and aMCI on the GM volumes in older nondemented adults. While the main effects of LLD were associated with GM volume loss limited to the mood regulating circuitry structures, aMCI was associated with GM atrophy involving brain regions associated with multidomain cognitive functions. The interactions of LLD and aMCI detrimentally affect widespread GM regions implicated in early AD. Furthermore, the interactions of depressive symptoms with episodic memory deficits were associated with volume loss in the salience processing regions. Our findings shed light that the various subgroups of LLD, separated based on the degree of cognitive impairment, may have differential effects on the GM volumes, in nondemented older adults.
The areas that showed GM atrophy in the depressed groups included structures of the mood regulating circuitry that are consistently reported in the literature. The neurodegeneration confined to these brain regions that are integral to emotional regulation provide important insights into the pathophysiology of LLD. Previously, volume loss in the OFC , dACC , prefrontal and temporal cortices, insula, and the PCC/precuneus [6, 9] has been previously described in LLD. Several studies including a recently published meta-analysis have revealed OFC to being one of the key brain regions that showed significant volume loss in LLD [8, 10, 40]. Interestingly, one postmortem study showed reduction in the glial cell number and density in the OFC. While loss of excitatory pyramidal neuronal density in layers III and V are seen in the dorsolateral prefrontal cortex, loss of layer II nonpyramidal inhibitory local circuit neurons are demonstrated in the rostral OFC . Smaller OFC volumes have been associated with functional disability in LLD . Volume reductions in the dACC, the cognitive subdivision of ACC, has been associated with poorer antidepressant treatment response in LLD [8, 39, 43]. Furthermore, longitudinal studies have also demonstrated volume declines in the prefrontal and temporal cortices in older adults with depressive symptoms . These structures are also key components of brain networks that are essential for episodic memory, executive function and attention-based cognitive performance. LLD often co-exist with varying levels of cognitive impairment in multiple memory and non-memory cognitive domains . While frontal lobe subregional volume loss has consistently been associated with executive dysfunction in LLD , the link of regional brain atrophy with impairment in other cognitive domains are not well understood. Future longitudinal studies need to examine the complex relationships between multiple cognitive domain deficits and brain volume changes in the depressed elderly.
Our findings of GM reductions in the amygdala-hippocampal area, temporal cortices, insula, fusiform gyrus and the PCC in the aMCI groups have been demonstrated in previous studies . GM atrophy in the amygdala-hippocampal area, thalamus and insula characterized patients with aMCI from cognitively healthy normals, and parietal and cingulate regional volume loss were considered pathognomonic of mild AD in one study . However, GM volume reductions in the inferior temporal cortices, MTL, fusiform gyrus and the PCC have also been previously reported in aMCI . aMCI patients who converted to AD had significantly greater GM atrophy in the insula, PCC, precuneus, amygdala, hippocampus, and frontal and temporal cortices, relative to the control group. MCI converters also showed more widespread GM volume reductions in the frontal, inferior parietal and temporal cortices, and the hippocampus than nonconverters [23, 24]. Recently published meta-analyses of structural neuroimaging studies have identified PCC, precuneus and MTL GM atrophy as antecedent biomarkers that can predict conversion of aMCI to AD [46, 47]. More importantly, the pattern of GM atrophic changes in MCI converters, especially in the temporal cortices, precuneus and the MTL, was similar to that seen in patients with mild AD, up to one year before the diagnosis of probable dementia.
To our knowledge, this the first study to demonstrate that the interactions between LLD and aMCI correlates with widespread GM volume reductions in the dorsal ACC, MTL, and prefrontal cortices, which are structures involved in early AD. Previously, greater fronto-parietal-temporal white matter atrophy in aMCI patients with subclinical depressive symptoms were reported than those without neuropsychiatric symptoms . However, the study by Lee et al. mostly included aMCI patients with subclinical depressive symptoms. In contrast, we conducted detailed neuropsychiatric assessments to categorically define LLD and aMCI in our study cohort, and focused our investigations to examine the effects of these neuropsychiatric syndromes on the GM volumes. We are intrigued by the findings of volume loss in the somatosensory and the visual cortices, which are brain regions that are only affected later in the course of AD. Interestingly, functional brain alterations in these brain regions have been reported in pre-clinical AD states, though these studies did not report on the GM volume differences in their study population . The GM atrophic patterns associated with LLD-aMCI interactions observed here illustrates that the co-morbidity of these behavioral phenotypes may be a marker of incipient AD.
The neuroanatomical correlates of the depressive symptom-episodic memory deficit interactions have not been previously reported. We found that the interaction between greater depressive symptoms (i.e., higher GDS scores) and higher episodic memory deficits (i.e., lower LM II–DR scores) were associated with decreased GM volumes in the right aIns/IFG and left MeFG clusters. aIns, a crucial hub that is engaged in the information transfer across multiple brain networks is implicated in salience processing. Moreover, our results underline the complex role of the right aIns/IFG in the higher-order social, emotional and cognitive processes including cognitive control, judgment and decision-making, empathy, and experiences of positive and negative feelings . The involvement of the superior aspect of the MeFG in the severity of depressive symptom-memory deficit interactions is intriguing, though GM reductions in this region have been previously reported in LLD, in MCI converters and in AD patients [6, 23]. The role of this region in mood regulation and cognitive functions should be a focus of future investigations.
Various pathophysiologic mechanisms are proposed to explain the link of LLD with neurodegeneration and development of AD. LLD, at least in some, may share common pathophysiologic mechanisms with AD. Cortical amyloid and neurofibrillary tangle accumulation comparable to that seen in mild AD is reported in LLD [18, 19]. Also, LLD of early onset, recurrent and of longer duration may lead to chronic exposure of toxic cortisol levels, resulting in GM volume loss . Subclinical cerebrovascular disease that is associated with LLD may damage white matter tract integrity and accelerate cognitive decline in some patients . LLD may accelerate neuronal injury through confluence of various mechanisms thus increasing the risk of AD.
This study is not without limitations. First, the cross-sectional study design precludes us from assuming causality and from capturing the temporal sequence of the associations between LLD and GM volume reductions. Second, while small sample size in the co-morbid group is a limitation, we found detrimental effects of depression and MCI interactions on the GM volumes across the whole sample. Our findings need to be replicated in future studies with larger sample sizes where similar analyses are conducted using FDR as well as more stringent FWE correction procedures. Third, an ideal study design would have been to include antidepressant and cholinesterase inhibitor naïve participants to eliminate the potential confound of medication exposure, though the ethical aspects restricted us from performing such a study. It is noteworthy that majority of the depressed participants (cognitively normal with LLD: N=9 and aMCI with co-existing LLD: N=10) were nonresponders to at least one therapeutic antidepressant trial (i.e., at adequate dosages for at least 6 weeks). Both increased and decreased GM volumes are associated with antidepressant exposure in older adults [53, 54]. Interestingly, remitters with treatment resistant depression show GM volume increases than nonremitters in younger depressed adults .
Our results provide preliminary evidence that LLD with and without amnestic MCI has differential effects on GM volumes in the elderly. Since the interactive effects of LLD and aMCI are associated with GM loss in regions that are early targets of AD neuropathology, the co-existence of these clinical phenotypes may be a potential marker of AD. Neuroimaging biomarkers may provide greater degree of accuracy in predicting those with LLD who is at greater risk of future progression to AD.
This work was supported by Alzheimer’s Association New Investigator NIRG-11-204070 (JSG), Advancing Healthier Wisconsin Endowment for Research to MCW (JSG), Extendicare Foundation (JSG), National Alliance for Research in Schizophrenia and Depression (NARSAD) New Investigator, 1UL1RR031973 from the Clinical and Translational Science Award program of the National Center for Research Resources, NIH grant R01 AD20279 (SJL), and the National Natural Science of Foundation of China (81171021, 91132727, 81171323 (CX)). The authors thank Ms. Carrie O’Connor for editorial assistance, Mr. Douglas Ward for assistance with statistical analysis, and Ms. Judi Zaferos-Pylant and Mr. Yu Liu for MRI technical support.
The authors disclose no potential conflicts of interest, financial or otherwise, related directly or indirectly to this work. All authors have made substantial intellectual contribution to this manuscript in one or more of the following areas: design or conceptualization of the study, analysis or interpretation of the data, or drafting and revising the manuscript. All authors have given final approval of this manuscript.
This work was presented as a poster at the Alzheimer’s Imaging Consortium session of the Alzheimer’s Association International Conference, Vancouver, British Columbia, Canada, July 14-19, 2012.