Impaired Biophysical Integrity of Macromolecular Protein Pools in the Uncinate Circuit in Late-Life Depression

Abstract

Major depressive disorder is a common mood disorder in the elderly. Although the neuroanatomical abnormalities have been identified in patients with late-life depression (LLD), the precise biological basis of LLD remains largely unknown. The purpose of this study was to examine the biophysical integrity of macromolecular protein pools in the nodal regions of the “uncinate circuit”, a component of fronto-limbic circuitry that is connected by the uncinate fasciculus and is critical in the regulation of mood and emotions, using novel magnetization transfer (MT) imaging. Twenty-four patients with LLD and 27 non-depressed healthy control subjects (HCs) of comparable age, sex, and race were recruited from the communities of the greater Chicago Area. The nodal regions of the uncinate circuit, i.e., bilateral amygdala, hippocampus, and lateral and medial orbitofrontal cortices (OFCs), were examined. Compared with HCs, patients with LLD had significantly lower magnetization transfer ratio (MTR), a measure of the biophysical integrity of macromolecular protein pools, in bilateral amygdala and hippocampus. The lower MTR was negatively correlated with the depression score. Moreover, the MTR of these regions decreased with age and positively correlated with neuropsychological performance in the LLD group but not in the HC group. These findings suggest that LLD is associated with compromised biophysical integrity of macromolecular protein pools in nodal regions of the uncinate circuit, and that major depression may accentuate age-related attenuation of the biophysical integrity of macromolecular protein pools in this circuit. These findings provide important new insights into the neurobiological mechanisms of the pathophysiology of LLD.

Introduction

Major depressive disorder (MDD) is one of the most common mood disorders in the elderly. Late-life MDD (LLD) is associated with considerable psychosocial and medical morbidities, including frequent doctor visits and hospitalization, substance abuse, and suicide¹. In addition, MDD in the elderly differs from depression in non-elderly adults in several clinical and neurobiological domains, including the role of vascular and degenerative mechanisms in its pathophysiology, as well as associated cognitive changes and depression serving as a prodrome of early dementia of the Alzheimer type^{2, 3}. These observations, together with the imaging findings of more diffuse structural and physiological correlates of MDD in the elderly^{4, 5}, strongly suggest that depression in the elderly differs clinically and biologically from depression in non-elderly adults^{6, 7}.

The term “Limbic System”, which was believed as the brain’s center of emotions, derives from the concept of a limbic lobe articulated by the anatomist Broca⁸. The term was originally used to denote tissue surrounding the brain stem and beneath the neocortical mantle. It evolved over time and commonly includes the hypothalamus, amygdala, hippocampus, and the septal area and medial aspect of the prefrontal cortex in the vicinity of the corpus callosum^{8, 9}. Borders of the limbic system were later expanded to include basal surfaces of the cerebral hemispheres. The aforementioned structures can be divided architectonically into regions and subregions each with its distinct connectivity to other regions within and outside the limbic system^10-12. In this study, we proposed to examine a well-defined component of this complex brain network, the “uncinate circuit”, which has been consistently implicated in the regulation of emotions and is relevant to the pathophysiology of LLD. The uncinate circuit comprises the hippocampus, the amygdala, and the lateral and medial orbitofrontal regions, connected by the uncinate fasciculus.

The hippocampus, in addition to its most well-known role in memory, has an important role in stress responses and is volumetrically smaller in patients diagnosed with mood disorders^{13, 14}. Several mechanisms, including the impact of stress related steroids on the hippocampus, depression-related decrease in trophic factors, and possible vascular mechanisms, have been proposed as explanations for the smaller hippocampal volumes¹³. The amygdala has an important role in the regulation of emotions and processing emotional facial recognition^15-17. The amygdala extracts the affective components of stimuli, whereas the prefrontal cortex guides goal-directed behavior¹⁸. There is substantial evidence supporting a role of the amygdala in the pathophysiology of mood disorders^{19, 20}. The volumes of the amygdala are smaller in those patients with MDD who are either untreated or medication free for extended periods of time than controls^{21, 22}. Functional magnetic resonance imaging (MRI) studies demonstrate exaggerated activation of the amygdala in patients with MDD in response to both sad faces and sad words when compared with controls^{23, 24}. Lesions in the amygdala modify emotional responses including the ability to recognize facial expression and electrical stimulation of the amygdala evokes strong emotions including the recall of emotional events^{16, 20, 25, 26}. The orbitofrontal cortex (OFC) has a central role in regulating mood and related cognitive functions²⁷. Human MRI studies, including those from our laboratory, demonstrate smaller volumes of OFC in patients diagnosed with MDD compared with controls^{28, 29}. Postmortem studies revealed the reductions in glial cell counts, density, and markers, as well as alterations in glia-neuron ratios in brain tissue of patients with previously diagnosed mood disorders when compared with controls³⁰. Behaviorally, OFC dysfunction may interfere with assessments of rewarding stimuli and integration of the associated information into behavioral incentives in depression.

Both clinical and preclinical studies demonstrate that these three brain regions have an important role in the neurocircuitry of emotions⁹. The OFC is connected to the amygdala and hippocampus through the uncinate fasciculus^{11, 31}. The uncinate fasciculus is a hook-shaped bundle that links the forward portions of the temporal lobe with the inferior frontal gyrus and lower surfaces of the frontal lobe³². It does this by arising lateral to the amygdala and hippocampus in the temporal lobe curving in an upward pathway behind the external capsule inward of the insular cortex and continuing up into the posterior part of the orbital gyrus. The medial OFC is connected to the hippocampus while the lateral OFC is the most heavily connected with the amygdala³³. The uncinate fasciculus has been implicated in LLD and displays lower fractional anisotropy (FA) in the patient group when compared with controls^{34, 35}. In addition to the aforementioned anatomical connections, there are multiple pathways and connections between regions and subregions of the three brain structures and other regions and networks in the brain. Nonetheless, anatomical and behavioral studies collectively provide the scientific rationale for considering this “uncinate circuit” in its entirety as a structural and functional unit that mediates stress and depression. In addition, our prior work and preliminary data from patients with LLD provided additional support for studying biophysical changes of the macromolecular protein pools in this circuit. In this study, we applied “magnetization transfer (MT) imaging”, a novel, supplementary MRI modality, to examine the biophysical integrity of macromolecular protein pools in the uncinate circuit in LLD.

Unlike diffusion-weighted MRI that measures water diffusion in brain tissue to probe microstructural properties of white matter (WM) such as the coherence of WM fibers, MT imaging exploits magnetization exchange between protons bound to macromolecules and free protons in tissue water to examine the biophysical integrity of macromolecular protein pools and their local microenvironments in the tissue (Fig. 1a). In the brain, bound protons are primarily present in myelin in WM and in cell membrane proteins and phospholipids in gray matter (GM). In contrast to free protons, bound protons are less mobile and possess a very short T₂ relaxation time. As a result, the MR signal from bound protons decays rapidly to noise levels before the start of data acquisition and therefore is undetectable by regular MRI. To solve this problem, an off-resonance prepulse is applied in MT imaging to selectively presaturate bound protons. Magnetization is then transferred from saturated bound protons to free protons through chemical exchange or direct dipolar coupling. This transfer of magnetization leads to a reduced MR signal from free protons (Fig. 1a). The contrast between images with and without the off-resonance presaturation pulse, defined as MT ratio (MTR), indirectly measures bound protons and, accordingly, informs the biophysical integrity of macromolecular protein pools and their local microenvironments³⁶. As demonstrated in MT studies of postmortem human brains^{37, 38}, lower MTR in WM is associated with demyelination and lower axonal density. Lower MTR in GM is believed to be linked to neuronal and synaptic loss, impaired cell membrane proteins and phospholipids^39-42, and Wallerian degeneration^{40, 42}. Although lacking specificity in its origins, MT imaging provides a new, innovative way to probe the integrity of macromolecular proteins and phospholipids in the brain.

Figure 1.

(a) Schematic diagram of the basic principle of magnetization transfer (MT) imaging. (b) Regions of interest (amygdala, hippocampus, lateral orbitofrontal cortex (OFC), and medial OFC) in the uncinate circuit for magnetization transfer ratio (MTR) analysis

There are a limited number of MT studies on LLD reported in the literature. Kumar et al.⁵ reported the finding that patients with LLD had significantly lower MTRs in genu and splenium of the corpus callosum, right caudate nucleus and putamen, and occipital WM than comparison controls. A more recent MT study was reported by Gunning-Dixon et al.⁴ who applied voxel-based analysis of MTR and found diffusive decrease of MTR in WM of the left hemisphere of the brains in patients with LLD. These MT studies were performed at a relatively low magnetic field strength (i.e., 1.5 T). This was mainly because high radio-frequency (RF) deposition associated with the short, rapidly repeated MT prepulses could easily reach the Food and Drug Administration (FDA)-approved specific absorption rate (SAR) limit, which constrained the clinical application of MT imaging to relatively low magnetic field strengths, e.g., 1.5 T. In this study, we used an improved pulsed MT technique, which enabled fast, high-resolution three-dimensional (3D) whole-brain MT acquisition at 3 T and maximized the WM/GM contrast within the FDA-prescribed SAR limits⁴³.

The current study was to expand on the previous MT studies on LLD but focused on the uncinate circuit using a region of interest (ROI)-based MTR analysis. Based on the literature, we hypothesized that MTR would be significantly lower in the nodal regions of the uncinate circuit in patients with LLD compared with healthy control subjects (HCs). We also hypothesized that MTR in the nodal regions of the uncinate circuit would be negatively correlated with depression score and positively correlated with neuropsychological performance. Examining the biophysical integrity of macromolecular protein pools in the uncinate circuit in patients with LLD may help to clarify the pathophysiology of depression in the elderly.

Materials and methods

Subjects

Data were collected as part of larger research studies investigating depression with and without diabetes at the University of Illinois at Chicago, approved by the University of Illinois at Chicago Institutional Review Board. All the subjects were recruited from the greater Chicago area through flyers, local advertisements, and relevant outpatient clinics, and written informed consent was obtained from all subjects. For the purpose of this report, we included 51 subjects in 2 groups: HCs (n=27) and patients with LLD (n=24) from our study samples. The subjects were age ≥ 60 years and passed the screen for current or history of neurological disorders (i.e., dementia, stroke, seizure, etc.), Axis I disorders (e.g., bipolar disorder) other than major depression, learning disability or attention deficit hyperactivity disorder, psychotropic medication use (including antidepressant drugs), current or history of substance abuse or dependence, or a Mini-Mental State Examination (MMSE) score < 24. The patients with LLD met the American Psychiatric Association’s diagnostic criteria (DSM-IV)⁴⁴ for major depression and required a score ≥ 15 on the 17-item Hamilton Depression Rating Scale (HDRS)⁴⁵, which was evaluated by a board-certified (AK) or board-eligible psychiatrist (OA). We should note that in this report all patients with LLD were free of antidepressant medication use, so the disease is studied in its untreated form. HCs denied a history of depressed mood, required a score of 8 or lower on the HDRS score, and were free of unstable medical conditions. All subjects were also administered the Center for Epidemiologic Studies Depression (CES-D) scale as an independent measure of depression severity⁴⁶. In addition, all participants were assessed for vascular risk by the Framingham Stroke Risk Profile score (FSRP)⁴⁷. The demographic and clinical measures across the subject groups are summarized in Table 1.

Table 1.

Demographic and clinical measures across subject groups

	HC (N = 27)		LLD (N = 24)		Statistics
	Mean	SD	Mean	SD	F	df	p
Age (years)	70.11	5.659	66.92	9.084	2.327	1,49	0.134
Sex	8M/19F	-	8M/16F	-	0.081	1	0.776
Race	9B/0H/2A/15W/0HP/1O		8B/2H/0A/13W/1HP/0O		6.046	5	0.302
Education (years)	15.93	2.645	15.79	2.604	0.033	1,49	0.856
WTAR (predicted full IQ)	105.61	12.869	108.00	12.985	0.402	1,45	0.529
MMSE	29.11	1.050	28.83	1.551	0.572	1,49	0.453
FSRP	10.70	4.046	10.09	4.111	0.284	1,48	0.596
Duration of current depressive symptoms (months)	-	-	18.22	23.75	-	-	-
CES-D	6.74	5.523	32.92	7.885	191.894	1,49	<0.0005
HDRS	1.44	1.625	18.96	3.368	579.357	1,49	<0.0005

HC, healthy control; LLD, patient with late-life depression

Sex: M=male, F=female; Race: B=black, H=hispanic, A=asian, W=white, HP=Hawaiian/Pacific Islander, O=other; Handedness: R=right, L=left, M=mixed; WTAR, Wechsler Test of Adult Reading; MMSE, Mini-Mental Status Exam; FSRP, Framingham Stroke Risk Profile; CES-D, Center for Epidemiologic Studies Depression scale; HDRS, Hamilton Depression Rating Scale

4 HCs did not have WTAR IQ and 1 LLD did not have FSRP in the database.

Neuropsychological tests

A neuropsychological battery was conducted on each subject across four domains: Verbal Memory (VerM) (California Verbal Learning Test–2nd Edition immediate total recall and long delay free recall, and Wechsler Memory Scale–3rd Edition (WMS-III) Logical Memory I and II); Visual Memory (VisM) (WMS-III Visual Reproduction I and II); Attention and Information Processing (AIP) (Stroop Color and Word trials, Trail Making Test A, and Wechsler Adult Intelligence Scale–3rd Edition (WAIS-III) Digit-Symbol Coding); Executive Function (EF) (Delis-Kaplan Executive Function System Category Switching, Trail Making Test B, Stroop Interference Score, WAIS-III Backwards Digit Span, and Self-Ordered Pointing Task Total Errors). Raw scores from the neuropsychological battery were standardized using healthy control sample means and SDs. Relevant scores were reversed so that higher scores consistently reflected better performance. Composite Z-scores were calculated for each domain. Cronbach’s alphas (α) were computed to assess how well the variables measured each latent construct. Values were considered good, indicating that each variable measured a unidimensional latent construct (VerM, α = 0.90; VisM, α = 0.84; AIP, α = 0.87; EF, α = 0.70). The investigator was blinded to the group allocation during the neuropsychological assessment and the following MRI data acquisition.

MT imaging data acquisition and processing

The MT imaging was performed on a Philips Achieva 3.0T X-Series MRI scanner (Philips Medical Systems, Best, The Netherlands) with an eight-channel phased-array (Philips’ SENSE-Head-8) head coil. Subjects were equipped with soft ear plugs, positioned comfortably in the head coil using custom-made foam pads to minimize head motion, and instructed to remain still. The MT images were acquired using a 3D spoiled gradient-echo sequence with the multi-shot echo-planar imaging (EPI) readout and the following parameters: repetition time (TR)/echo time (TE) = 64/15 ms, flip angle = 9°, field of view (FOV) = 24 cm, 67 axial slices, slice thickness/gap = 2.2 mm/no gap, EPI factor = 7, reconstructed voxel size = 0.83 × 0.83 × 2.2 mm³, with a non-selective five-lobed Sinc-Gauss off-resonance MT prepulse (B₁ / Δf / dur = 10.5μT / 1.5kHz / 24.5ms) optimized for maximum WM/GM contrast^{43, 48}. The image slices were parallel to the anterior commissure–posterior commissure line. Parallel imaging was utilized with a p-factor of 2.5. Before the MT scan, high-resolution 3D T₁-weighted magnetization prepared rapid acquisition gradient-echo (MPRAGE) images were acquired for image registration and segmentation between different image modalities in the postprocessing. The sequence parameters are: TR/TE = 8.4/3.9 ms, flip angle = 8°, FOV = 24 cm, 134 axial slices/no gap, voxel size = 0.83 × 0.83 × 1.1 mm³. In addition, T₂-weighted fluid-attenuated inversion recovery (FLAIR) images were also acquired using turbo spin echo sequence with the following sequence parameters: TR/TI/TE=11000/2800/68 ms, FOV = 24 cm, 67 axial slices without gap, and voxel size = 0.83 × 0.83 × 1.1 mm³ for delineation of hyperintensity areas.

T₁-weighted MPRAGE images, T₂-weighted FLAIR images, and the images without the off-resonance MT prepulse applied in the MT scan (M₀) were registered first, and the images with the off-resonance MT prepulse applied (M_s) were then registered to the co-registered M₀. MTR was calculated on a voxel-by-voxel basis with the co-registered M₀ and M_s and the equation MTR = (M₀-M_s)/M₀. The ROIs were placed on the nodal regions of the uncinate circuit⁴⁹, including amygdala, hippocampus, and lateral and medial OFC, in both the left and right hemispheres (see Fig. 1b). For all the ROIs, we used the software package FreeSurfer to automatically segment out the brain structures. Generation of the ROIs in the images and calculation of MTR in each ROI were performed using in-house developed programs.

Statistical analysis

Clinical and demographic measures were analyzed using univariate analysis of variance (ANOVA) for continuous variables and χ²-tests for categorical variables. Group differences in MTR were assessed using the linear mixed-effects model with diagnostic group as the between-group factor and hemisphere as the within-subject factor to deal with the correlated MTR data (from both the left and right hemispheres of the same subjects), controlling for age, sex, and race. Similarly, associations between the MTR and age, depression score, and neuropsychological task performance, respectively, were examined using the linear mixed-effects model analysis, controlling for age and/or other covariates such as sex or race. As MTR values are within a relatively small range (i.e., 0 - 1), to better demonstrate the association of MTR with individual variables, the MTR and independent variables were standardized (Z-scores) first before applying the linear mixed-effects model; therefore, we called the estimated coefficients from the above analysis of associations as “standardized coefficients”. Significance was set at p < 0.05. Correction for multiple comparisons was performed using the false discovery rate (FDR) approach^50-52 with the maximum acceptable FDR threshold set at 0.05. Post hoc power calculation was also performed to calculate the power using the current sample size and effect size. All statistical analyses were carried out using SPSS version 22 software (IBM Corporation, Chicago, IL, USA).

Results

Demographic and clinical measures

The demographic characteristics and clinical measures across subject groups, including age, sex, race, education, Wechsler Test of Adult Reading (WTAR) IQ, MMSE, FSRP score, and those related to depressive symptomatology, i.e., CES-D total score, HDRS score, and duration of current depressive symptoms, are summarized in Table 1. There were no significant differences in age, sex, race, education, WTAR IQ, MMSE, and FSRP between the two groups (p’s > 0.13). As expected, there were significant differences in depressive symptomatology measures between the two groups: the LLD group had significantly higher ratings on CES-D and HDRS than the HC group (p’s < 0.0005).

Regional MTR in uncinate circuit and depression

Linear mixed-effects model analysis (with diagnosis as the between-group factor and hemisphere as a within-group factor, controlling for age, sex, and race) revealed that MTR was significantly lower in amygdala (F = 10.816, df = 1,42, p = 0.002) and hippocampus (F = 5.880, df = 1,42, p = 0.020) in patients with LLD than HCs, but there were no significant group differences in MTR in lateral and medial OFCs (p’s > 0.16) (see Fig. 2 and Supplementary Table 1 for more details including the effect size). The significant group differences in amygdala and hippocampus survived correction for multiple comparisons. The post hoc power calculation revealed that the current sample size provided a power of 81% in detecting the group effect in amygdala while a power of 49% in hippocampus.

Figure 2.

Scatterplot of magnetization transfer ratio (MTR) in the (a) amygdala, (b) hippocampus, (c) lateral orbitofrontal cortex (OFC), and (d) medial OFC with CES-D score, adjusted for age, sex, and race

Further, amygdala MTR was negatively associated with HDRS (standardized coefficient = −0.374, standard error [SE] = 0.161, df = 20, p = 0.031) in the LLD group. In contrast, in the HC group, amygdala MTR was positively correlated with HDRS (standardized coefficient = 0.442, SE = 0.188, df = 23, p = 0.028) and hippocampus MTR had a trend to be positively correlated with HDRS (standardized coefficient = 0.317, SE = 0.183, df = 23, p = 0.096) (see Supplementary Table 2).

Regional MTR in uncinate circuit and age

Although there was no association between amygdala MTR and age or between hippocampus MTR and age in the HC group (amygdala: standardized coefficient = −0.085, SE = 0.245, df = 21, p = 0.731; hippocampus: standardized coefficient = −0.180, SE = 0.217, df = 21, p = 0.417), increasing age was associated with lower amygdala MTR and lower hippocampus MTR in the LLD group (amygdala: standardized coefficient = −0.587, SE = 0.135, df = 18, p < 0.0005; hippocampus: standardized coefficient = −0.606, SE = 0.168, df = 18, p = 0.002) (see Fig. 3 and Supplementary Table 3). The associations of amygdala MTR and age were close to be significantly different between the HC and LLD groups (standardized coefficient = 0.468, SE = 0.263, df = 41, p = 0.083) and the associations of hippocampus MTR and age were not significantly different between the HC and LLD groups (standardized coefficient = 0.419, SE = 0.265, df = 41, p = 0.122). Therefore, the patients with LLD showed a trend of accelerated aging in the uncinate circuit compared with HCs.

Figure 3.

Association between magnetization transfer ratio (MTR) and age in the (a) amygdala and (b) hippocampus, adjusted for sex and race

Neuropsychological tests

There were no significant group differences in the neuropsychological task performance in the four domains, EF, AIP, VerM, and VisM (F’s < 1.9, p’s > 0.17) (see Supplementary Table 4). We also performed the statistical analyses on component tasks in each domain, and there was no group difference in any component task either (F’s < 3.45, p’s > 0.07).

Regional MTR in uncinate circuit and neuropsychological performance

Amygdala MTR was positively correlated with neuropsychological task performance in the LLD group (p’s < 0.015) but not in the HC group (p’s > 0.30). Specifically, amygdala MTR in the LLD group was positively correlated with the composite Z-score of EF (standardized coefficient = 0.778, SE = 0.265, df = 18, p = 0.009), the composite Z-score of AIP (standardized coefficient = 0.489, SE = 0.176, df = 21, p = 0.011), and the composite Z-score of VisM (standardized coefficient = 0.526, SE = 0.194, df = 20, p = 0.013). There was no significant correlation between amygdala MTR and the composite Z-score of VerM in either group (p’s > 0.30). Hippocampus MTR tended to positively correlate with the composite Z-score of EF (standardized coefficient = 0.580, SE = 0.323, df = 18, p = 0.089) and the composite Z-score of VisM (standardized coefficient = 0.400, SE = 0.212, df = 20, p = 0.074) in the LLD group but not in the HC group (p’s > 0.27). There was no association between hippocampus MTR and the composite Z-score of AIP or VerM in either group (p’s > 0.43) (see Fig. 4 and Supplementary Table 5).

Figure 4.

Association between magnetization transfer ratios (MTR) and neuropsychological task performance (in composite Z-score) in three domains across regions of interest and subject groups: (a) amygdala, EF; (b) hippocampus, EF; (c) amygdala, AIP; (d) hippocampus, AIP; (e) amygdala, VisM; and (f) hippocampus, VisM, controlling for age

Discussion

There are two main findings of this study. First, patients with untreated LLD exhibited significantly lower MTR in the amygdala and hippocampus in the uncinate circuit when compared with non-depressed elderly control subjects, and lower amygdala MTR in patients with untreated LLD was associated with greater depression severity. Second, amygdala MTR and hippocampus MTR were negatively correlated with age and positively correlated with neuropsychological task performance across multiple domains in the LLD group but not in the HC group.

Using T₁-weighted structural MRI and volumetric and shape analysis techniques, previous studies have demonstrated reduced volume and local shrinkage of the amygdala^{19, 53, 54} and the hippocampus^{53, 55-58} in patients with LLD. In addition, previous MT studies on LLD have reported reduced MTR in the WM^{4, 59}. Aligning with these morphometric and WM MTR findings, the current MT imaging study focused on the nodal regions of the uncinate circuit that were implicated in emotion and memory in LLD. Specifically, we found significantly reduced MTR in the amygdala and the hippocampus in patients with untreated LLD. MT imaging and histopathology studies on postmortem human brain suggest that MTR in the WM correlates with both the axonal and myelin counts, with myelin being a primary component contributing to the reduced MTR. The origins of MTR changes in the GM are more complex and heterogeneous, and may reflect multiple neurobiological aberrations^39-42. Lower MTR in the GM are believed to link to impaired cell membrane proteins and phospholipids besides neuronal and synaptic loss^39-42. Wallerian degeneration, which is secondary to proximal and/or distal axonal damage, also has been implicated as a mechanism contributing to a lower MTR^{40, 42}. Exact biochemical and biophysical underpinnings of MTR in vivo are not fully understood. Nevertheless, MT imaging measures the magnetization exchange between macromolecules and water protons and can noninvasively inform the integrity of macromolecular proteins and phospholipids in the brain. Therefore, reduced MTRs observed in this study may reflect the compromised biophysical integrity of macromolecular protein pools in the uncinate circuit in patients with untreated LLD.

Of those studies that examined the relationships between the volume of amygdala or hippocampus with current depression severity in older adults, only a few suggested that smaller amygdala or hippocampus was associated with greater depressive severity^{60, 61}, whereas majority of the studies did not find such an association^{55, 56, 62, 63}. These inconsistencies may relate to heterogeneity of the studies, including different depressive scales, clinical settings, antidepressant treatments, age of onset, age range, and other possible characteristics. In contrast, existing literature showed more consistent association between amygdala or hippocampal volume and duration of the disease. Specifically, amygdala volume was negatively correlated with number of preceding depressive episodes but not with current disease severity in untreated mid- and late-life depression⁶³. Hippocampal volume was significantly correlated with total lifetime duration of depression^{55, 62, 64} but not with depression severity in LLD. The current study found that in patients with untreated LLD, lower amygdala MTR was associated with greater depression severity but not with duration of disease or age of onset (p’s > 0.5). These findings collectively suggest that volumetric changes of the amygdala and hippocampus may reflect an accumulative effect of depression and can serve as biomarkers for long-term, sustained impact of depression on brain structures, while lower amygdala MTR seems to represent short-term clinical characteristics and may potentially be used as a marker to monitor disease progression and evaluate treatment outcome.

Accelerated aging is implicated in depression. Depression increases risk of developing aging-related medical comorbidities in multiple organs including cardiovascular disease, diabetes, stroke, osteoporosis, vascular dementia, and Alzheimer's disease⁶⁵. At the cellular level, depression is associated with significantly shortened telomere^65-67, indicative of accelerated cellular aging. Distinct from depression in younger adults, LLD is associated with aging-related WM degeneration and GM atrophy in the brain. Specifically, within the WM, LLD is associated with greater burden of WM hyperintensities^{2, 68, 69} and compromised WM microstructural integrity^{34, 70-72}, presumably of a vascular origin⁷³. Within the GM, LLD is associated with significant atrophy in the frontal lobe, limbic lobes, and subcortical regions^{74, 75}. Relevant to this study, Hickie et al. found significant correlation between hippocampal volume and age in patients with LLD but not in control subjects⁵⁶. In line with their findings, we found amygdala MTR and hippocampus MTR were negatively associated with age in patients with untreated LLD, but not in control subjects. Further, lower MTR in the amygdala and the hippocampus was associated with poorer neuropsychological task performance in the LLD group, but not in the HC group. Our findings suggest that, accompanying pronounced GM loss with aging, biophysical integrity of macromolecular protein pools in these regions may decline faster with age in patients with untreated LLD, providing a supportive evidence for potentially accelerated aging of the brain in LLD.

Several limitations of this study should be considered. First, lower MTR has been reported across a spectrum of behavioral and neurological disorders as it represents underlying biophysical changes of the brain tissue that may occur in several disease states. While it may lack diagnostic specificity, it opens a window to study macromolecular proteins and phospholipids in the brain. Future studies using multimodal neuroimaging may help underpin the biological substrates underlying the impaired macromolecular protein pools in the uncinate circuit observed in the current study. Second, the present study has a modest sample size; therefore the results should be interpreted with caution. Further, with its relatively small sample size, this study does not have enough power to detect subtle biophysical changes of macromolecular protein pools in the medial and lateral OFC regions of the uncinate circuit in the patients with LLD. However, the post hoc power calculation showed that the current sample size still provided a power of 81% in detecting the group effect in amygdala. The relatively small sample size has also restricted our ability to uncover the association of MTR with specific cognitive processes. Thus, a composite Z-score approach was used to protect against multiple comparisons. Future studies with larger sample sizes can make these options more feasible. Third, a cross-sectional instead of longitudinal design was used in this study. The cross-sectional design is inherently more vulnerable to intersubject variance and cohort effects. Similarly, age-related change rather than aging-related change in MTR was reported in this study. Future studies with longitudinal data are needed to investigate aging effects on the regional MTR in the uncinate circuit in patients with LLD.

In summary, our findings are the first to demonstrate that the biophysical integrity of macromolecular protein pools in the amygdala and hippocampus of the uncinate circuit is compromised in patients with untreated LLD, when compared with non-depressed elderly controls. The measures of impaired macromolecular protein pools in the amygdala and hippocampus of the uncinate circuit were negatively correlated with age and positively correlated with neuropsychological task performance across multiple domains in the LLD group but not in the HC group, suggesting potentially accelerated aging of the uncinate circuit in untreated LLD. Given the heterogeneous nature of major depression and diverse biological abnormalities that have been reported in the late-life population, other brain regions and circuits may also play important roles in the pathophysiology of clinical depression in the elderly. Nevertheless, our data demonstrate significant biophysical compromise to the uncinate circuit, which may be a critical component to fronto-limbic dysfunction that is often invoked in the pathophysiology of mood and related behavioral disorders. Therefore, our findings have broad implications for the pathophysiology of mood disorders especially in late life.