Abstract
Background:
White matter hyperintensities (WMHs) are associated with late-life depression (LLD) and are considered a hallmark of MRI-defined vascular depression. However, their impact on depression recurrence in LLD is less well known.
Methods:
We investigated this relationship using data from a 2-year multi-site, longitudinal study, where baseline WMH volumes were obtained using 3 T FLAIR magnetic resonance imaging (MRI) from 145 participants, of which 102 had remitted LLD and 43 were control participants. We analyzed the effect of baseline WMH volume on LLD relapse over 2 years using regression and adjusting for total intracranial volume, age, sex, race, education, and study site. We performed survival analyses using a Cox proportional hazard model to determine whether baseline WMH volume was associated with time to relapse in LLD.
Results:
We found that participants with LLD had greater WMH volume at baseline than control participants, but not if accounting for peripheral cardiovascular disease. Participants with LLD who relapsed within 8 months of baseline had larger WMH volume than control participants but did not statistically differ from those that remained remitted; this effect was lost when expanding to participants that relapsed at any point in the 2-year study. WMH burden was not associated with time to relapse, suggesting greater WMH volumes are not indicative of faster relapse rates in LLD.
Conclusion:
Our results show that WMH burden may play a role in the early - but not the delayed - relapse of LLD, and they underscore the intricate dynamic of the biological markers underlying LLD treatment response and relapse.
Keywords: WMH, Small vessel disease, LLD, Depression, Late-life, Recurrence, Relapse
1. Introduction
Late-life depression (LLD) is a significant public health burden, associated with high disability, increased risk for cognitive decline and dementia, elevated suicide risk, and greater all-cause mortality (Taylor, 2014). Depression is a chronic, recurrent disorder with longitudinal studies and clinical maintenance trials finding relapse rates between 35 and 43 % over two years, and up to 57 % over four years (Andreescu et al., 2019; Deng et al., 2018; Reynolds III et al., 2006). Recurrence and relapse have been associated with several clinical and behavioral markers including higher number of previous episodes, severity of residual symptoms (Deng et al., 2018; Judd et al., 1998; Nierenberg et al., 2010; Nierenberg et al., 2003), residual anxiety and sleep disturbance (Andreescu et al., 2007; Deng et al., 2018; Reynolds III et al., 2006), and non-adherence with antidepressant treatment (Bockting et al., 2007; ten Doesschate et al., 2009). One of the most well-studied neurobiological markers of LLD are white matter hyperintensities (WMHs), considered a marker of small vessel brain disease. The vascular depression hypothesis (Alexopoulos et al., 1997) posits that vascular factors such as small vessel disease can cause, contribute to, or perpetuate symptoms of LLD.
The relationship between WMHs and LLD has been investigated in multiple studies, which consistently have found LLD is associated with greater WMH volumes (Firbank et al., 2004; Krishnan et al., 1988; Taylor et al., 2005; Wang et al., 2014) compared to never-depressed older adults (Chen et al., 2006; Herrmann et al., 2008; Wu et al., 2006), with some (Greenwald et al., 1998) but not all studies reporting on the importance on WMH lesions location in relationship to LLD (DeCarli et al., 2005). A greater WMH burden is often associated with a more chronic LLD disease course and poorer response to short-term antidepressants (Alexopoulos et al., 2005; Butters et al., 2004; Sheline et al., 2006). Other data suggests that WMHs in older adults accumulate at a faster rate in those with LLD than those with no prior history of depression (Chen et al., 2006; Taylor et al., 2003).
Few studies have focused on whether WMHs are associated with episode recurrence or an increased risk of relapse in LLD. Taylor et al. (2003) found that greater WMH burden was associated with poorer outcomes in LLD including greater risk of relapse, however this included both individuals who relapsed and those who did not remit. They found that WMH volume increased at a faster rate in the group that experienced relapse or did not remit compared to the group who were stably remitted (Khalaf et al., 2015; Taylor et al., 2003), which has been replicated (Chen et al., 2006). These previous works illustrate that the relationship between WMH volume and depression recurrence as an outcome of LLD is not well understood.
In this study, we obtained baseline WMH volumes in a sample of older adults with remitted LLD, followed their clinical course longitudinally for up to 2 years to determine whether depression recurrence had occurred, and compared them to non-depressed older adults. We hypothesized that older adults with LLD would have larger cross-sectional WMH volumes at baseline compared to never-depressed older adults, and that those who experienced a depression relapse would have even greater WMH volume compared to those who do not. We also hypothesized that individuals with larger baseline WMH volumes would experience faster relapse rates than those with smaller WMH volumes.
2. Methods
2.1. Participants and study design
We recruited 145 participants into the REMBRANDT study (Recurrence markers, cognitive burden, and neurobiological homeostasis in late-life depression). The REMBRANDT study followed-up remitted LLD at three sites: University of Pittsburgh (PITT), Vanderbilt University (VUMC), and University of Illinois Chicago (UIC). We analyzed imaging data from 102 individuals with remitted LLD (LLD) and 43 never-depressed healthy controls (HC). Participants were recruited through registries, community outreach, clinical referrals, advertisements, and internet-based advertising. The study was approved by each of the three university’s Institutional Review Boards and participants provided written informed consent prior to study procedures. Recruitment is ongoing at all three sites.
Inclusion criteria for both LLD and non-depressed HC were being age 60 years or older; a Montreal Cognitive Assessment (MoCA) score of 24 or greater (Milani et al., 2018), or a MoCA-BLIND greater than or equal to 18 (MoCA Cognition, n.d.); and fluency in English. A diagnosis of mild cognitive impairment (MCI) was not exclusionary given its prevalence in LLD (Sheline et al., 2006). Exclusion criteria for both LLD and non-depressed participants were a history of a developmental disorder or IQ below 70; acute suicidal ideation within 3 months of study entry; acute grief (<1 month); current or past psychosis; and primary neurological disorders, including major neurocognitive disorders related to dementia, Parkinson’s Disease, stroke, epilepsy, etc. Peripheral neurological illness, such as essential tremor or regular headaches, were permissible.
Participants with LLD that entered through the treatment phase (i.e., were currently depressed) required a DSM5 diagnosis of major depressive disorder with a Montgomery-Asberg Depression Rating Scale (MADRS) (Montgomery and Åsberg, 1979) score greater or equal to 15. Participants with LLD that entered remitted required a DSM5 diagnosis of major depressive disorder in partial or full remission and a MADRS score of 10 or less within 4 months of starting the study. HC participants needed a MADRS score less than or equal to 8 and no history of psychiatric illness or psychotropic medication use. Past brief therapy for specific challenges (such as marital therapy or grief therapy) was allowed.
In the longitudinal phase, participants were followed for up to 2 years. During this time, their depression severity was assessed every 2 months using the MADRS (Montgomery and Åsberg, 1979). Relapse was considered when a participant’s MADRS score exceeded 10 in two consecutive assessments. For analytic purposes, we considered two time periods of relapse: (1) the participant first relapsed within 250 days (or roughly 8 months) from the baseline MRI (N = 21), chosen because all participants had completed at least 250 days in the study at the time of analysis, and (2) the participant first relapsed at any point during the 2-year study (N = 43). The analyses performed in this study were powered to detect statistically significant group differences for other measures and were not used to power this analysis. We note the terms recurrence and relapse are used interchangeably to describe any new depressive episode after treatment/admission to the study, regardless of timing.
2.2. Initial treatment phase (ITP)
Currently depressed LLD participants entered the study through an initial treatment phase (ITP), described by Taylor et al. (2024). They were treated for up to 28 weeks; if remission was not achieved, they were referred to clinical care and were withdrawn from the study. Participants entered the longitudinal phase of the study no sooner than 1 month and no later than 4 months following remission. In the ITP, participants were assessed at least every 4 weeks through clinical interview, MADRS, and clinical global impression (CGI) scale. The treatment plan consisted of an algorithm-guided treatment to facilitate remission in depressed patients. Patients began treatment using SSRIs, and treatments were adapted depending on the change MADRS after every 4 weeks. If there was substantial improvement (>50 % change in MADRS), patients would continue SSRI treatment until plateau or remission. A 25-50 % improvement in MADRS would lead to first-level augmentation using bupropion or mirtazapine. If there was <25 % improvement, patients would switch to a SNRI. For further details on the ITP, see the supplement. This protocol could be modified on the discretion of study clinicians (WDT, OA, CA) for given participants to accommodate past treatment history.
2.3. Longitudinal phase design
HC and LLD participants entered the two-year longitudinal phase after the initial treatment phase. Given the goal of the REMBRANDT study to examine how cognitive performance and decline may both predict recurrence risk and be influenced by neurobiological processes contributing to depression recurrence, the study PIs in collaboration with the study neuropsychologist (Dr. Butters) decided on the 24-month follow-up interval to allow for capturing relevant changes in cognitive performance and decline in this population. Additionally, the results of a prior report on predictors of LLD recurrence support the use of a 24-month follow-up interval (Deng et al., 2018).
During the longitudinal phase, participants were contacted every 2-months plus additional visits as needed. LLD participants remained on the regimen that led to remission, but changes to the treatment plans were allowed if clinically necessary (e.g., changes in medications doses). We continued following participants after relapse until the completion of the two-year phase. Relapse was defined as having a MADRS score greater than or equal to 15 for a minimum of 2 weeks and meeting DSM5 diagnosis criteria for major depressive disorder.
2.4. Group definitions
We defined several groups including healthy controls (HC) and all remitted depressed patients in the study (LLD). We also defined those who remitted and did not relapse (REM) and those who first relapsed (REL) at any point in the study; further, we considered early vs. late REL to reflect a first relapse before or after 250 days (or roughly 8 months) from the baseline measurement, respectively. Early REL is notated REL (8 mo) while those that maintained remission during that 250-day period as REM (8 mo). Our study focuses only on a participant’s first relapse and does not consider remission or relapse after the first relapse.
2.5. Clinical assessments
We used a number of clinical assessments to collect data including the MINI for diagnostic assessments (Sheehan et al., 1998), MoCA for cognitive screening (Hawkins et al., 2014), the Cumulative Illness Rating Scale-Geriatric (CIRS-G) (Miller et al., 1992; Miller and Towers, 1991) for medical burden assessment, and the Five Factor Inventory to assess neuroticism (RR, 1992). Additionally, the MADRS was used to assess depression severity (Montgomery and Åsberg, 1979). Other assessments used in the study are detailed in Taylor et al. (2024) (Taylor et al., 2024).
2.6. MRI acquisition
Participants underwent a 3T MRI with a 32-channel head coil on Siemens Prisma (PITT), Philips Elition (VUMC), and GE Discovery MR750 (UIC). We used the Adolescent Brain Cognitive Development (ABCD) Study MRI Protocol (Casey et al., 2018). This protocol is listed in full detail for the one-hour scan in Taylor et al. (2024). A sagittal, whole-brain T1-weighted magnetization prepared rapid gradient echo (MPRAGE) was collected with repetition time (TR) = 2400 ms (2500 ms PITT), echo time (TE) = 2.9 ms, flip angle (FA) = 8 deg., field of view (FOV) = 256 × 256 (256 × 240 VUMC) with 176 slices (225 slices VUMC), 1 mm3 isotropic resolution, and GeneRalized Autocalibrating Partial Parallel Acquisition (GRAPPA) with acceleration factor of 2 (total time 5.63 to 6.65 min). An axial, whole-brain fluid attenuated inversion recovery (FLAIR) sequence was collected with repetition time (TR) = 4800 ms, echo time (TE) = 445 ms (113 ms UIC), inversion time (TI) = 1650 ms (1464 ms UIC), field of view (FOV) = 256 × 256 with 176 slices, 1 mm3 isotropic resolution, and GeneRalized Autocalibrating Partial Parallel Acquisition (GRAPPA) with acceleration factor of 3 (total time 5.96 to 6.5 min). Supports were included as needed.
Due to slight variations in imaging protocol and differences in the scanners used across sites, we conducted monthly scanning of ADNI (Alzheimer’s Disease Neuroimaging Initiative) structural phantom that included both the MPRAGE and FLAIR. We additionally computed the signal-to-noise ratio (SNR) from each scanner and found that they were within an expected tolerance. Lastly, we control for site in all analyses to reduce any potential differences in MRI, finding WMH volumes at baseline were not significantly different across the study sites.
2.7. Structural processing and WMH segmentation
We conducted WMH segmentation using FLAIR sequences and the Lesion Segmentation Toolbox (Schmidt et al., 2012). These analyses were implemented through SPM12 using a 0.3 threshold, determined visually across a number of participants. Briefly, this toolbox co-registers the T1 image after bias correction and segments the white matter, gray matter, and cerebrospinal fluid. After an initial identification of WMH through outlier detection, a lesion-growing algorithm using a Markov random field extends and grows this lesion. This is used to estimate WMH volume. We additionally generated an intracranial volume mask by thresholding gray/white matter and cerebrospinal fluid probability maps by 0.1 and smoothing to generate a final mask. This was used to estimate intracranial volume.
2.8. Statistical analysis
We performed a cross-sectional analysis of baseline WMH volume to determine its effect on recurrence in LLD. Separate linear regression models were used to compare the average baseline WMH volume across groups (HC vs. LLD and HC vs. REL vs. REM), where we considered both early relapse and relapse at any point over the 2-year study. We adjusted for total intracranial volume (ICV) at baseline, demographic features including age at baseline, sex, race, education, study site. We note that education was considered using either True or False to indicate whether the participant had an advanced academic degree or not. We performed separate analyses including a measure of cardiovascular disease burden from the CIRS-G as a covariate (see supplement).
A sensitivity regression was performed to ensure WMH volumes were not influenced by baseline MADRS scores; linear regression models were created using the ‘lm’ function from the base ‘stats’ package in R (version 4.3.1). A Survival analysis was conducted using a Cox proportional hazards model, implemented using the ‘coxph’ function in the ‘survival’ package in R (4.3.1), to determine whether greater WMH burden led to earlier relapse times in individuals with remitted LLD.
Participants that were missing WMH and/or ICV data were excluded from all statistical analyses (N = 3); we note these individuals had remitted LLD and maintained stable remission for the duration of the study. An additional N = 6 participants did not have data on their educational backgrounds and were removed from the analyses when adjusting for demographic features. All 6 of these participants had LLD and 2 relapsed within 250 days of baseline; the remaining 4 maintained stable remission. Because our data was collected at multiple institutions with varied recruitment strategies, we performed sensitivity regression analyses using the participants that were treated to remission in the ITP (N = 58) to determine if our data is sensitive to the study site or screening MADRS scores (>10). We note that 39 of the participants from the ITP were recruited at VUMC, 8 were recruited at PITT, and 11 were recruited at UIC.
3. Results
The characteristics of our sample are reported in Table 1. We found that participants with LLD had greater baseline WMH volumes than HC [F(1,127) = 4.13, P = 0.04, Fig. 1a, Table 2]. This result did not hold when adjusting for cardiovascular disease burden (see supplement). We found early REL (8 mo) had greater WMH volume [F(1,126) = 4.51, P = 0.036, Fig. 1b, Table 3a] than HC; however, WMH volumes did not differ significantly between early REL (8 mo) and REM (8 mo) (Table 3b). We also conducted analyses using those that relapsed at any point in the study (REL). Neither individuals in stable remission (REM) nor REL differed in WMH volume from the HC group (Table 4a). Additionally, we found no differences in WMH volume between REL and REM (Table 4b).
Table 1.
Demographic characteristics of participants.
| Control (N = 43) | LLD (N = 102) | Overall (N = 145) |
|
|---|---|---|---|
| Age, Years | |||
| Mean (SD) | 66.6 (5.39) | 67.0 (4.55) | 66.9 (4.80) |
| Median [Min,Max] | 65.0 [60, 82] | 67.0 [60, 79] | 66.0 [60, 82] |
| Sex | |||
| Male (0) | 22 (51.2 %) | 28 (27.5 %) | 50 (34.5 %) |
| Female (1) | 21 (48.8 %) | 74 (72.5 %) | 95 (65.5 %) |
| Race | |||
| Non-white (0) | 12 (27.9 %) | 14 (13.7 %) | 26 (17.9 %) |
| White (1) | 31 (72.1 %) | 88 (86.3 %) | 119 (82.1 %) |
| Higher Education | |||
| False | 22 (51.2 %) | 60 (58.8 %) | 82 (56.6 %) |
| True | 21 (48.8 %) | 36 (35.3 %) | 57 (39.3 %) |
| Missing | 0 (0 %) | 6 (5.9 %) | 6 (4.1 %) |
| Study Site | 63 (61.8 %) | ||
| VUMC | 14 (32.6 %) | 77 (53.1 %) | |
| PITT | 14 (32.6 %) | 20 (19.6 %) | 34 (23.4 %) |
| UIC | 15 (34.9 %) | 19 (18.6 %) | 34 (23.4 %) |
| Baseline MADRS | |||
| Mean (SD) | 0.930 (1.50) | 5.53 (3.31) | 4.13 (3.58) |
| Median [Min, Max] | 0 [0, 6] | 5 [0, 18] | 3 [0, 18] |
| Missing | 0 (0 %) | 4 (3.9 %) | 4 (2.8 %) |
| Treatment Status | |||
| Control | 43 (100 %) | 0 (0 %) | 43 (29.7 %) |
| Entered Study Remitted | 0 (0 %) | 44 (43.1 %) | 44 (30.3 %) |
| Received Treatment | 0 (0 %) | 58 (56.9 %) | 58 (40.0 %) |
| Relapse Status | |||
| Did not relapse | 43 (100 %) | 59 (57.8 %) | 102 (70.3 %) |
| Relapsed in 2 years | 0 (0 %) | 43 (42.2 %) | 43 (29.7 %) |
| 8-month Relapse Status | |||
| Did not relapse | 43 (100 %) | 81 (79.4 %) | 124 (85.5 %) |
| Relapsed in 250 days | 0 (0 %) | 21 (20.6 %) | 21 (14.5 %) |
| Days to relapse | |||
| Mean (SD) | – | 286 (164) | – |
| Median [Min, Max] | – | 261 [69, 700] | – |
| Missing | 43 (100 %) | 59 (57.8 %) | – |
| WMH Volume, cm3 | |||
| Mean (SD) | 1.90 (3.87) | 3.40 (5.14) | 2.95 (4.83) |
| Median [Min, Max] | 0.548 [0, 20.0] | 0.923 [0, 21.1] | 0.844 [0, 21.1] |
| Missing | 0 (0 %) | 2 (2.0 %) | 2 (1.4 %) |
| ICV, mm3 | |||
| Mean (SD) | 1.5e6 (1.74e5) | 1.46e6 (1.6e5) | 1.47e6 (1.65e5) |
| Median [Min, Max] | 1.51e6 [1.18e6, 1.9e6] | 1.45e6 [1.13e6, 1.9e6] | 1.46e6 [1.13e6, 1.9] |
| Missing | 0 (0 %) | 3 (2.9 %) | 3 (2.1 %) |
| CIRS-G (H + V + E) | |||
| Mean (SD) | 1.32 (1.20) | 2.55 (1.71) | 2.22 (1.67) |
| Median [Min, Max] | 1 [0, 4] | 2 [0, 7] | 2 [0, 7] |
| Missing | 9 (20.9 %) | 10 (9.8 %) | 19 (13.1 %) |
Abbreviations: LLD, participants with remitted late-life depression; SD, standard deviation; VUMC, Vanderbilt University; PITT, University of Pittsburgh; UIC; University of Illinois Chicago; WMH, white matter hyperintensity; ICV, intracranial volume; CIRS-G (H + V + E), a sum of the heart, vascular, and endocrine/breast components of the Cumulative Illness Rating Scale-Geriatric.
Fig. 1.
Baseline WMH comparison by group. We report the mean (SD) WMH volume for each group with units cm3, where (a) HC: 1.9 (3.9); LLD: 3.4 (5.1); (b) REM (8 mo): 3.1 (5.0); REL (8 mo) 4.7 (5.5); (c) REM: 3.4 (5.2); REL 3.4 (5.0). Note HC is identical across figures.
Table 2.
Regression analysis comparing WMH volume between controls (HC) and those with remitted LLD (LLD) regardless of relapse status; bolded are the statistically significant variables at α = 0.05.
| Variable | B (SE) | β | t and p-value |
|---|---|---|---|
| ICV | 6.5e-06 (3.3e-06) | 2.2e-01 | 2.0, p = 0.05 |
| Age | 3.5e-01 (9.0e-02) | 3.4e-01 | 3.8, p < 0.001 |
| Sex – Female | −2.0e-01 (1.1) | −1.9e-02 | −0.2, p = 0.86 |
| Race – White | 1.2e-01 (1.2) | 8.9e-03 | 0.1, p = 0.92 |
| Higher Education | −6.7e-01 (8.5e-01) | −6.7e-02 | −0.8, p = 0.43 |
| Study Site (VUMC reference) | |||
| PITT | 8.4e-01 (1.0) | 7.3e-02 | 0.8, p = 0.41 |
| UIC | 4.6e-01 (1.2) | 3.9e-02 | 0.4, p = 0.69 |
| Group (HC reference) | |||
| LLD | 1.9 (9.3e-01) | 1.8e-01 | 2.0, p = 0.04 |
Abbreviations: B, linear coefficient; SE, standard error; β, standardized linear coefficient; ICV, intracranial volume; VUMC, Vanderbilt University; PITT, University of Pittsburgh; UIC; University of Illinois Chicago; HC, healthy control; LLD, participants with remitted late-life depression.
Table 3.
Regression analysis comparing WMH volume between controls (HC), relapsed within 8 months of baseline (REL 8 mo) and stably remitted within 8 months (REM 8 mo); bolded are the statistically significant variables at α = 0.05.
| Model | Variable | B (SE) | β | t and p- value |
|---|---|---|---|---|
| 3a | ICV | 6.5e-06 (3.3e-06) | 2.2e-01 | 2.0, p = 0.05 |
| Age | 3.4e-01 (9.0e-02) | 3.3e-01 | 3.8, p < 0.001 | |
| Sex – Female | −1.9e-01 (1.1) | −1.8e-02 | −0.2, p = 0.87 | |
| Race – White | 1.0e-01 (1.2) | 7.6e-03 | 0.1, p = 0.93 | |
| Study site (VUMC Reference) | ||||
| Higher Education | −6.0e-01 (8.5e-01) | −6.0e-02 | −0.7, p = 0.48 | |
| PITT | 9.0e-01 (1.0) | 7.8e-02 | 0.9, p = 0.38 | |
| UIC | 5.0e-01 (1.2) | 4.2e-02 | 0.4, p = 0.67 | |
| Group (HC reference) | ||||
| REM (8 mo) | 1.7 (9.5e-01) | 1.7e-01 | 1.7, p = 0.08 | |
| Early REL (8 mo) | 2.8 (1.3) | 2.0e-01 | 2.1, p = 0.04 | |
| 3b | ICV | 1.0e-05 (4.4e-06) | 3.1e-01 | 2.3, p = 0.02 |
| Age | 3.6e-01 (1.2e-01) | 3.1e-01 | 3.0, p < 0.01 | |
| Sex – Female | 4.8e-01 (1.5) | 4.1e-02 | 0.3, p = 0.75 | |
| Race – White | −7.519e-01 1.905e+00 | −4.6e-02 | −0.4, p = 0.69 | |
| Higher Education | −1.11 (1.02) | −1.0e-01 | −1.0, p = 0.32 | |
| Study site (VUMC Reference) | ||||
| PITT | 1.6 (1.3) | 1.2e-01 | 1.2, p = 0.23 | |
| UIC | −2.7e-01 (1.7) | −1.9e-02 | −0.2, p = 0.87 | |
| Group [REM (8 mo) reference] | ||||
| Early REL (8 mo) | 1.1 (1.3) | 8.3e-02 | 0.9, p = 0.40 |
Abbreviations: B, linear coefficient; SE, standard error; β, standardized linear coefficient; ICV, intracranial volume; VUMC, Vanderbilt University; PITT, University of Pittsburgh; UIC; University of Illinois Chicago; HC, healthy control; REM (8mo), LLD participants that maintained remission 250 days post-baseline; early REL (8 mo), LLD participants that relapsed within 250 days post-baseline.
Table 4.
Regression analysis comparing WMH volume between controls (HC), relapsed ever (REL) and stably remitted (REM); bolded are the statistically significant variables at α = 0.05.
| Model | Variable | B (SE) | β | t and p-value |
|---|---|---|---|---|
| 4a | ICV | 6.5e-06 (3.3e-06) | 2.2e-01 | 2.0, p = 0.05 |
| Age | 3.4e-01 (9.0e-02) | 3.4e-01 | 3.8, p < 0.001 | |
| Sex – Female | −1.9e-01 (1.1) | −1.8e-02 | −0.2, p = 0.87 | |
| Race – White | 1.1e-01 (1.2) | 8.2e-03 | 0.1, p = 0.93 | |
| Higher Education | −6.7e-01 (8.5e-01) | −6.7e-02 | −0.8, p = 0.43 | |
| Study site (VUMC reference) | ||||
| PITT | 8.2e-01 (1.0) | 7.2e-02 | 0.8, p = 0.42 | |
| UIC | 4.4e-01 (1.2) | 3.7e-02 | 0.4, p = 0.71 | |
| Group (HC reference) | ||||
| REM | 1.9 (1.0) | 1.9e-01 | 1.9, p = 0.06 | |
| REL | 1.8 (1.1) | 1.7e-01 | 1.7, p = 0.10 | |
| 4b | ICV | 1.0e-05 (4.4e-06) | 3.2e-01 | 2.3, p = 0.02 |
| Age | 3.6e-01 (1.2e-01) | 3.1e-01 | 3.0, p < 0.01 | |
| Sex – Female | 4.6e-01 (1.5) | 3.9e-02 | 0.3, p = 0.76 | |
| Race – White | −7.8e-01 (1.9) | −4.8e-02 | −0.4, p = 0.69 | |
| Higher Education | −1.2 (1.1) | −1.1e-01 | −1.1, p = 0.28 | |
| Study site (VUMC reference) | ||||
| PITT | 1.5 (1.4) | 1.2e-01 | 1.1, p = 0.27 | |
| UIC | −3.9e-01 (1.7) | −2.7e-02 | −0.2, p = 0.82 | |
| Group (REM reference) | ||||
| REL | −1.4e-01 (1.1) | −1.4e-02 | −0.1, p = 0.89 |
Abbreviations: B, linear coefficient; SE, standard error; β, standardized linear coefficient; ICV, intracranial volume; VUMC, Vanderbilt University; PITT, University of Pittsburgh; UIC; University of Illinois Chicago; HC, healthy control; REM, LLD participants that maintained remission during the study course; REL, LLD participants that relapsed at any point during the study course.
To determine the sensitivity of our data to the initial screening MADRS and study site, we examined 58 remitted LLD participants who went through the ITP, of which N = 28 relapsed during the study (N = 17 relapsed within 250 days of baseline). There was no significant effect of study site, screening MADRS, age, sex, race, education, or ICV on WMH volume for those who received treatment prior to the start of the study (Table 5).
Table 5.
Sensitivity regression analysis using participants who were treated to remission for LLD in the ITP. We show results for both relapsed groups considered in this work; results show that WMH volume is not sensitive to any of the considered variables, including screening MADRS and study site.
| Model | Variable | B (SE) | β | t and p- value |
|---|---|---|---|---|
| 5a | ICV | 5.3e-06 (6.5e-06) | 1.8e-01 | 0.8, p = 0.42 |
| Age | 2.4e-01 (1.6e-01) | 2.3e-01 | 1.5, p = 0.14 | |
| Sex – Female | −1.9 (2.2) | −1.7e-01 | −0.9, p = 0.4 | |
| Race – White | −1.8 (2.7) | −1.1e-01 | −0.6, p = 0.52 | |
| Higher Education | −1.2 (1.7) | −1.1e-01 | −0.7, p = 0.49 | |
| Screening MADRS | −6.1e-02 (1.5e-01) | −6.5e-02 | −0.4, p = 0.69 | |
| Study site (VUMC reference) | ||||
| PITT | −6.5e-01 (2.0) | −4.8e-02 | −0.3, p = 0.74 | |
| UIC | −2.3 (2.2) | −1.7e-01 | −1.1, p = 0.29 | |
| Group [REM (8 mo) reference] | ||||
| Early REL (8 mo) | 1.61 (1.55) | 1.5e-01 | 1.0, p = 0.31 | |
| 5b | ICV | 6.2e-06 (6.7e-06) | 2.1e-01 | 0.9, p = 0.36 |
| Age | 2.5e-01 (1.6e-01) | 2.5e-01 | 1.6, p = 0.12 | |
| Sex – Female | −1.7 (2.4) | −1.6e-01 | −0.7, p = 0.47 | |
| Race – White | −1.7 (2.8) | −1.0e-01 | −0.6, p = 0.54 | |
| Higher Education | −1.3 (1.7) | −1.3e-01 | −0.8, p = 0.45 | |
| Screening MADRS | −4.2e-02 (1.6e-01) | −4.4e-02 | −0.3, p = 0.80 | |
| Study site (VUMC reference) | ||||
| PITT | −5.7e-01 (2.0) | −4.2e-02 | −0.3, p = 0.78 | |
| UIC | −2.4 (2.2) | −1.8e-01 | −1.1, p = 0.28 | |
| Group (REM reference) | ||||
| REL | 4.6e-01 (1.5) | 4.7e-02 | 0.3, p = 0.77 |
Abbreviations: B, linear coefficient; SE, standard error; β, standardized linear coefficient; ICV, intracranial volume; Screening MADRS, Montgomery-Asberg Depression Rating Scale used as an initial measurement of depression severity; VUMC, Vanderbilt University; PITT, University of Pittsburgh; UIC; University of Illinois Chicago; REM (8mo), LLD participants that maintained remission 250 days post-baseline; early REL (8 mo), LLD participants that relapsed within 250 days post-baseline; REM, LLD participants that maintained remission during the study course; REL, LLD participants that relapsed at any point during the study course.
To better visualize these effects, we plotted WMH volumes using a log-scale and plotted four groups: HC, REM (those who stayed remitted throughout the observation period), early REL (those who relapsed before 250 days), and late REL (those who relapsed after 250 days). The early REL, as indicated by our results, showed the largest WMH volumes (Fig. 2). The mean and standard deviation of WMH volumes per group are as follows, in units of cm3: HC – 1.9 (3.9); REM – 3.4 (5.2); Early REL – 4.7 (5.5); Late Rel – 2.1 (4.3).
Fig. 2.
4-way group comparison of baseline WMH volume. We compare the log-transformed WMH volumes from HC, REM (those who stayed remitted through the observation period), late REL (those who relapsed after 8 months), and early REL (those who relapsed before 8 months). The early REL as indicated in Fig. 1b (middle) had the highest WMH.
Using a Cox proportional hazards model, we performed survival analyses to examine the effect of WMH volume on time to relapse for LLD participants, finding WMH has no statistically significant effect on time to relapse (Fig. 3). None of the other factors we considered (i.e., ICV, age, sex, race, education, study site) had a significant association with time to relapse (not reported).
Fig. 3.
Differences in time to relapse based on median split in WMH volume. We illustrate the probability of staying in remission with LLD over time (N = 93; 9 excluded due to missing data) used in the survival analysis. While there is some divergence in remission probability between those with low vs. high WMH volume, WMH volume was not associated with faster relapse times.
4. Discussion
In this study, we replicated previous research that has found individuals with LLD have greater WMH volume compared to never-depressed controls. Our results related to depression relapse were mixed. As seen in Fig. 2, the early relapsers had significantly greater WMH volume than controls but not stable remitters and later relapsers. There was no significant difference between stable remitters and later relapsers. Additionally, WMH volume did not predict earlier times to relapse in LLD.
Our finding that participants with remitted LLD had a greater WMH burden than the never-depressed, healthy control participants is consistent with the vascular depression hypothesis (Szymkowicz et al., 2023; Taylor et al., 2013) and replicates the results from multiple prior studies (Firbank et al., 2004; Krishnan et al., 1988; Taylor et al., 2005; Wang et al., 2014). Taylor et al. (2003) similarly found no baseline differences in WMH volume between patients with LLD that experienced poor outcomes (i.e., patients that did not remit or had relapsed) and those that achieved remission, but suggested that WMHs accumulated faster longitudinally in the poor outcome group. These results have been replicated (Chen et al., 2006; Khalaf et al., 2015), but also show there was no difference in WMH accumulation between the control group and the overall depressed group regardless of outcome (Chen et al., 2006).
The difference in significance between the findings that associate WMH volume with LLD relapse within roughly 8 months, but not when we include LLD participants who relapsed later than 8 months, point toward a potential “accelerant” effect of WMHs on relapse in LLD. We may speculate that vascular disease, as measured through WMH burden, compromises the stability of treatment response relatively early after patients achieve remission. This observation could have implications on clinical guidelines, suggesting that patients with LLD and a high burden of WMHs may require more intensive monitoring of mood during follow-up in the first year after achieving remission to possibly circumvent a relapse.
While WMHs have been associated with LLD onset, severity and treatment response (Herrmann et al., 2008; Taylor et al., 2013), few studies have examined their impact on recurrence in individuals with remitted LLD. The focus of most research that targets relapse in LLD is on clinical factors including number of previous depressive episodes, residual depressive symptoms (Deng et al., 2018; Judd et al., 1998; Nierenberg et al., 2010; Nierenberg et al., 2003), sleep and anxiety symptoms (Andreescu et al., 2007; Deng et al., 2018; Reynolds III et al., 2006), which are all positively related to recurrence.
Further, the vascular depression hypothesis suggests that cerebrovascular disease can cause hypoperfusion and the formation WMHs in regions that are vital for cognition and emotional regulation, resulting in the dysfunction of these areas and the onset of LLD (Alexopoulos et al., 1997; Taylor et al., 2013). Neuroinflammation and microglial activation may be other key factors in the etiology of WMH (Solé-Guardia et al., 2023; Tomimoto et al., 1996) and should be considered in the study of LLD. Neuroinflammation involves the activation of microglia, leading to the release of inflammatory cytokines (i.e. IL-1, IL-6 and TNF-α) that increase the permeability of the blood-brain barrier, allowing for the infiltration of peripheral immune cells which activate a local inflammatory response resulting in oxidative stress and endothelial dysfunction that impairs cerebral blood flow (Serna-Rodríguez et al., 2022). Dysfunctional and/or chronic activation of neuroinflammatory mechanisms may result in neurovascular damage that could increase the risk of WMH (Farkas et al., 2004; Jalal et al., 2012). Hippocampal atrophy is a common biomarker of neurodegenerative diseases, such as Alzheimer’s Disease, and may play a role in the pathogenesis and recurrence of LLD (Buddeke et al., 2017; Taylor et al., 2014), however the potential role of hippocampal atrophy in the interplay of WMH and LLD recurrence is not clear.
While our study provides an important insight into the relationship between WMH volume and LLD recurrence, we acknowledge the limitations to our work. Previous studies have suggested that the change in WMH volume over time may be a significant predictor of relapse (Khalaf et al., 2015; Taylor et al., 2003), as white matter lesions are generally progressive and increase in volume. Because our work uses a cross-sectional measure for WMH volume at baseline, we are unable to investigate whether individuals who experienced LLD relapse have greater rates of WMH accumulation over the study course. Our results will be better contextualized, especially in the consideration of early relapse, by performing further statistical analyses using longitudinal WMH data on the completion of our study. Moreover, the current results on early relapse are based on a relatively small sample size with relatively high variance in WMH volume, limiting the generalizability of our results to broader populations. When considering those who relapsed at any point in the study, the relapsed sample size doubled and the variance decreased, suggesting this group may be more representative of the larger patient population. Future research including other possible confounding variables (e.g., medical comorbidity, social determinants of health) and larger sample sizes across groups will be important for validating and expanding on the results of this study.
In summary, our work investigated the complex relationship between WMHs and recurrent LLD. While those with remitted LLD were found to have larger WMH burden than controls, WMH volumes were significantly associated with relapse up risk up to roughly 8-months post-remission, but not later than that. These results underscore the multifactorial nature of depression and the intricate dynamic of the biological markers underlying LLD treatment response and relapse. Further investigation is needed of the longitudinal interplay of neurobiological markers and psychosocial factors in relation to recurrent LLD, with the goal of improving outcomes for those affected by late-life depression.
Supplementary Material
Acknowledgements
We would like to thank participants for their participation in the study as well as study staff from the ARGO neuroscience of aging research group and study staff from WDT and OA groups.
Funding
This work was supported by National Institutes of Health grants R01 MH121619, R01 MH121620, and R01 MH121384; the National Institute of Mental Health grant T32 MH019986; and the National Center for Advancing Translational Sciences grants UL1 TR000445 and UL1 TR002243.
Appendix A. Supplementary data
Supplementary data to this article can be found online at https://doi.org/10.1016/j.jad.2025.04.116.
Footnotes
CRediT authorship contribution statement
Leigh B. Pearcy: Writing – review & editing, Writing – original draft, Formal analysis, Conceptualization. Helmet T. Karim: Writing – review & editing, Writing – original draft, Formal analysis, Conceptualization. Meryl A. Butters: Writing – review & editing. Robert Krafty: Writing – review & editing. Brian D. Boyd: Writing – review & editing. Layla Banihashemi: Writing – review & editing. Sarah M. Szymkowicz: Writing – review & editing. Bennett A. Landman: Writing – review & editing. Olusola Ajilore: Writing – review & editing, Funding acquisition, Conceptualization. Warren D. Taylor: Writing – review & editing, Funding acquisition, Conceptualization. Carmen Andreescu: Writing – review & editing, Writing – original draft, Funding acquisition, Conceptualization.
Previous presentations
Organization for Human Brain Mapping 2024 Annual Meeting, Seoul, South Korea, June 24-27, 2024.
Consent statement
All participants gave written informed consent prior to starting study procedures. This study was approved by IRBs of all three institutions (VUMC, PITT, UIC).
Declaration of competing interest
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
Data availability
Data is available upon request from study site principal investigators.
References
- Alexopoulos GS, Meyers BS, Young RC, Campbell S, Silbersweig D, Charlson M, 1997. ’Vascular depression’ hypothesis. Arch. Gen. Psychiatry 54 (10), 915–922. [DOI] [PubMed] [Google Scholar]
- Alexopoulos GS, Kiosses DN, Heo M, Murphy CF, Shanmugham B, Gunning-Dixon F, 2005. Executive dysfunction and the course of geriatric depression. Biol. Psychiatry 58 (3), 204–210. [DOI] [PubMed] [Google Scholar]
- Andreescu C, Lenze EJ, Dew MA, Begley AE, Mulsant BH, Dombrovski AY, Pollock BG, Stack J, Miller MD, Reynolds CF, 2007. Effect of comorbid anxiety on treatment response and relapse risk in late-life depression: controlled study. Br. J. Psychiatry 190 (4), 344–349. [DOI] [PubMed] [Google Scholar]
- Andreescu C, Aizenstein HJ, Butters MA, Krafty R, 2019. Disruption of neural homeostasis as a model of relapse and recurrence in late-life depression. Am. J. Geriatr. Psychiatry 17 (12), 1316–1330. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Bockting CL, ten Doesschate MC, Spijker J, Spinhoven P, Koeter MW, Schene AH, 2007. Continuation and maintenance use of antidepressants in recurrent depression. Psychother. Psychosom 77 (1), 17–26. [DOI] [PubMed] [Google Scholar]
- Buddeke J, Kooistra M, Zuithoff N, Gerritsen L, Biessels G, Van der Graaf Y, Geerlings M, Group SS, van Petersen R, Dinther B, et al. , 2017. Hippocampal volume and the course of depressive symptoms over eight years of follow-up. Acta Psychiatr. Scand 135 (1), 78–86. [DOI] [PubMed] [Google Scholar]
- Butters MA, Whyte EM, Nebes RD, Begley AE, Dew MA, Mulsant BH, Zmuda MD, Bhalla R, Meltzer CC, Pollock BG, et al. , 2004. The nature and determinants of neuropsychological functioning in late-lifedepression. Arch. Gen. Psychiatry 61 (6), 587–595. [DOI] [PubMed] [Google Scholar]
- Casey BJ, Cannonier T, Conley MI, Cohen AO, Barch DM, Heitzeg MM, Soules ME, Teslovich T, Dellarco DV, Garavan H, et al. , 2018. The adolescent brain cognitive development (ABCD) study: imaging acquisition across 21 sites. Dev. Cogn. Neurosci 32, 43–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Chen PS, McQuoid DR, Payne ME, Steffens DC, 2006. White matter and subcortical gray matter lesion volume changes and late-life depression outcome: a 4-year magnetic resonance imaging study. Int. Psychogeriatr 18 (3), 445–456. [DOI] [PubMed] [Google Scholar]
- DeCarli C, Fletcher E, Ramey V, Harvey D, Jagust WJ, 2005. Anatomical mapping of white matter hyperintensities (WMH): exploring the relationships between periventricular WMH, deep WMH, and total WMH burden. Stroke 36 (1), 50–55. 10.1161/01.STR.0000150668.58689.f2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Deng Y, McQuoid DR, Potter GG, Steffens DC, Albert K, Riddle M, Beyer JL, Taylor WD, 2018. Predictors of recurrence in remitted late-life depression. Depress. Anxiety 35 (7), 658–667. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Farkas E, Donka G, de Vos RA, Mihály A, Bari F, Luiten PG, 2004. Experimental cerebral hypoperfusion induces white matter injury and microglial activation in the rat brain. Acta Neuropathol. 108, 57–64. [DOI] [PubMed] [Google Scholar]
- Firbank MJ, Lloyd AJ, Ferrier N, O’Brien JT, 2004. A volumetric study of MRI signal hyperintensities in late-life depression. Am. J. Geriatr. Psychiatr 12 (6), 606–612. [DOI] [PubMed] [Google Scholar]
- Greenwald BS, Kramer-Ginsberg E, Krishnan KRR, Ashtari M, Auerbach C, Patel M, 1998. Neuroanatomic localization of magnetic resonance imaging signal hyperintensities in geriatric depression. Stroke 29 (3), 613–617. [DOI] [PubMed] [Google Scholar]
- Hawkins MA, Gathright EC, Gunstad J, Dolansky MA, Redle JD, Josephson R, Moore SM, Hughes JW, 2014. The MoCA and MMSE as screeners for cognitive impairment in a heart failure population: a study with comprehensive neuropsychological testing. Heart Lung 43 (5), 462–468. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Herrmann LL, Le Masurier M, Ebmeier KP, 2008. White matter hyperintensities in late life depression: a systematic review. J. Neurol. Neurosurg. Psychiatry 79 (6), 619–624. [DOI] [PubMed] [Google Scholar]
- Jalal FY, Yang Y, Thompson J, Lopez AC, Rosenberg GA, 2012. Myelin loss associated with neuroinflammation in hypertensive rats. Stroke 43 (4), 1115–1122. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Judd LL, Akiskal HS, Maser JD, Zeller PJ, Endicott J, Coryell W, Paulus MP, Kunovac JL, Leon AC, Mueller TI, et al. , 1998. Major depressive disorder: a prospective study of residual subthreshold depressive symptoms as predictor of rapid relapse. J. Affect. Disord 50 (2–3), 97–108. [DOI] [PubMed] [Google Scholar]
- Khalaf A, Edelman K, Tudorascu D, Andreescu C, Reynolds CF, Aizenstein H, 2015. White matter hyperintensity accumulation during treatment of late-life depression. Neuropsychopharmacology 40 (13), 3027–3035. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Krishnan KRR, Goli V, Ellinwood EH, France RD, Blazer DG, Nemeroff CB, 1988. Leukoencephalopathy in patients diagnosed as major depressive. Biol. Psychiatry 23 (5), 519–522. [DOI] [PubMed] [Google Scholar]
- Milani SA, Marsiskeb M, Cottler LB, Chen X, Striley CW, 2018. Optimal cutoffs for the Montreal cognitive assessment vary by race and ethnicity. Alzheimers Dement. Diagn. Assess. Dis. Monit 10, 773–781. 10.1016/j.dadm.2018.09.003. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Miller MD, Towers A, 1991. A Manual of Guidelines for Scoring the Cumulative Illness Rating Scale for Geriatrics (CIRS-G). University of Pittsburgh, May, Pittsburgh, PA, pp. 2–31. [Google Scholar]
- Miller MD, Paradis CF, Houck PR, Mazumdar S, Stack JA, Rifai AH, Mulsant B, Reynolds CF III, 1992. Rating chronic medical illness burden in geropsychiatric practice and research: application of the cumulative illness rating scale. Psychiatry Res. 41 (3), 237–248. [DOI] [PubMed] [Google Scholar]
- MoCA Cognition. (n.d.). MoCA cognition. https://mocacognition.com/.
- Montgomery SA, Åsberg M, 1979. A new depression scale designed to be sensitive to change. Br. J. Psychiatry 134 (4), 382–389. [DOI] [PubMed] [Google Scholar]
- Nierenberg AA, Petersen TJ, Alpert JE, 2003. Prevention of relapse and recurrence in depression: the role of long-term pharmacotherapy and psychotherapy. J. Clin. Psychiatry 64 (15), 13–17. [PubMed] [Google Scholar]
- Nierenberg A, Husain M, Trivedi M, Fava M, Warden D, Wisniewski S, Miyahara S, Rush A, 2010. Residual symptoms after remission of major depressive disorder with citalopram and risk of relapse: a STAR* D report. Psychol. Med 40 (1), 41–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Reynolds CF III, Dew MA, Pollock BG, Mulsant BH, Frank E, Miller MD, Houck PR, Mazumdar S, Butters MA, Stack JA, et al. , 2006. Maintenance treatment of major depression in old age. N. Engl. J. Med 354 (11), 1130–1138. [DOI] [PubMed] [Google Scholar]
- RR CPM, 1992. Revised NEO Personality Inventory and NEO Five Factor Inventory: Professional Manual. Psychological Assessment.[Google Scholar], Odessa, FL. [Google Scholar]
- Schmidt P, Gaser C, Arsic M, Buck D, Forschler A, Berthele A, Hoshi M, Ilg R, Schmid VJ, Zimmer C, et al. , 2012. An automated tool for detection of FLAIR-hyperintense white-matter lesions in multiple sclerosis. Neuroimage 59 (4), 3774–3783. [DOI] [PubMed] [Google Scholar]
- Serna-Rodríguez MF, Bernal-Vega S, Camacho-Morales A, Perez-Maya AA, et al. , 2022. The role of damage associated molecular pattern molecules (DAMPs) and permeability of the blood-brain barrier in depression and neuroinflammation. J. Neuroimmunol 371, 577951. [DOI] [PubMed] [Google Scholar]
- Sheehan DV, Lecrubier Y, Sheehan KH, Amorim P, Janavs J, Weiller E, Hergueta T, Baker R, Dunbar GC, et al. , 1998. The Mini-international neuropsychiatric interview (MINI): the development and validation of a structured diagnostic psychiatric interview for DSM-IV and ICD-10. J. Clin. Psychiatry 59 (20), 22–33. [PubMed] [Google Scholar]
- Sheline YI, Barch DM, Garcia K, Gersing K, Pieper C, Welsh-Bohmer K, Steffens DC, Doraiswamy PM, 2006. Cognitive function in late life depression: relationships to depression severity, cerebrovascular risk factors and processing speed. Biol. Psychiatry 60 (1), 58–65. [DOI] [PubMed] [Google Scholar]
- Sole-Guardia G, Custers E, de Lange A, Clijncke E, Geenen B, Gutierrez J, Küsters B, Claassen JA, De Leeuw F-E, Wiesmann M, et al. , 2023. Association between hypertension and neurovascular inflammation in both normal-appearing white matter and white matter hyperintensities. Acta Neuropathol. Commun 11 (1), 2. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Szymkowicz SM, Gerlach AR, Homiack D, Taylor WD, 2023. Biological factors influencing depression in later life: role of aging processes and treatment implications. Transl. Psychiatry 13 (1), 160. 10.1038/s41398-023-02464-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Taylor WD, 2014. Depression in the elderly. N. Engl. J. Med 371 (13), 1228–1236. [DOI] [PubMed] [Google Scholar]
- Taylor WD, Steffens DC, MacFall JR, McQuoid DR, Payne ME, Provenzale JM, Krishnan KRR, 2003. White matter hyperintensity progression and late-life depression outcomes. Arch. Gen. Psychiatry 60 (11), 1090–1096. [DOI] [PubMed] [Google Scholar]
- Taylor WD, MacFall JR, Payne ME, McQuoid DR, Steffens DC, Provenzale JM, Krishnan RR, 2005. Greater MRI lesion volumes in elderly depressed subjects than in control subjects. Psychiatry Res. Neuroimaging 139 (1), 1–7. [DOI] [PubMed] [Google Scholar]
- Taylor WD, Aizenstein HJ, Alexopoulos GS, 2013. The vascular depression hypothesis: mechanisms linking vascular disease with depression. Mol. Psychiatry 18 (9), 963–974. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Taylor WD, McQuoid DR, Payne ME, Zannas AS, MacFall JR, Steffens DC, 2014. Hippocampus atrophy and the longitudinal course of late-life depression. Am. J. Geriatr. Psychiatry 22 (12), 1504–1512. [DOI] [PMC free article] [PubMed] [Google Scholar]
- Taylor WD, Ajilore O, Karim HT, Butters MA, Krafty R, Boyd BD, Banihashemi L, Szymkowicz SM, Ryan C, Hassenstab J, Landman BA, Andreescu C, 2024. Assessing depression recurrence, cognitive burden, and neurobiological homeostasis in late life: design and rationale of the REMBRANDT study☆. J. Mood Anxiety Disord 5, 100038. 10.1016/j.xjmad.2023.100038. [DOI] [PMC free article] [PubMed] [Google Scholar]
- ten Doesschate MC, Bockting CL, Schene AH, 2009. Adherence to continuation and maintenance antidepressant use in recurrent depression. J. Affect. Disord 115 (1–2), 167–170. [DOI] [PubMed] [Google Scholar]
- Tomimoto H, Akiguchi I, Suenaga T, Nishimura M, Wakita H, Nakamura S, Kimura J, 1996. Alterations of the blood-brain barrier and glial cells in white-matter lesions in cerebrovascular and Alzheimer’s disease patients. Stroke 27 (11), 2069–2074. [DOI] [PubMed] [Google Scholar]
- Wang L, Leonards CO, Sterzer P, Ebinger M, 2014. White matter lesions and depression: a systematic review and meta-analysis. J. Psychiatr. Res 56, 56–64. [DOI] [PubMed] [Google Scholar]
- Wu M, Rosano C, Butters M, Whyte E, Nable M, Crooks R, Meltzer CC, Reynolds CF III, Aizenstein HJ, 2006. A fully automated method for quantifying and localizing white matter hyperintensities on MR images. Psychiatry Res. Neuroimaging 148 (2–3), 133–142. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Data is available upon request from study site principal investigators.