Assessing the Role of Past Depression in Patients with Mild Cognitive Impairment,with and without Biomarkers for Alzheimer’s Disease

INTRODUCTION

A diagnosis of major depressive disorder (MDD) increases the risk of dementia by two-fold or greater [1]. Furthermore, depression has been shown to accelerate progression from mild cognitive impairment (MCI), referencing an intermediate stage of cognitive impairment between normal cognitive aging and dementia, to dementia [2]. This finding has been replicated consistently in late-onset depression (LOD), with mixed results in early-onset depression (EOD) [3–5]. It has been posited that EOD and LOD represent distinct phenotypes and pathological processes. Early adverse events and genetic predispositions are implicated in EOD, while accumulation of vascular burden and neuropathological substrates contribute to LOD [6–8].

Late-life depression (LLD), which includes EOD that reoccurs in late-life, is associated with deficits in verbal and visuo-spatial memory, executive function, and processing speed [9, 10]. Furthermore, these cognitive deficits commonly persist after remission of depressive symptoms and appear to increase the risk of dementia through poorly understood neuropathological processes [11–13]. Inflammation and increased glucocorticoid production, which can contribute to amyloid deposition, accelerated hippocampal atrophy, and cerebrovascular disease (CVD) are implicated in LLD [6, 14]. There is conflicting evidence to support a direct link between amyloid-β (Aβ) and tau, two of the pathological hallmarks of Alzheimer’s disease (AD) and LLD [15–24]. Furthermore, many of these studies involve actively depressed individuals, with a paucity of research addressing the association of remitted depression and AD pathology [25].

CVD, as represented by white matter hyperintensities (WMH) on T2-weighted structural magnetic resonance imaging (MRI), is a common co-pathology of neurodegenerative diseases, particularly AD [26, 27]. White matter hyperintensities and AD pathology appear to have independent and additive effects on dementia risk [28, 29]. Among patients with normal cognition [30–32], MCI [30], and AD dementia [28–36], WMH are associated with lower levels of cerebrospinal fluid (CSF) Aβ₄₂ [28]. WMHs are also associated with depression [7, 36]. The link between depression and CVD appears to be bidirectional: depression increases the risk of CVD [37] and vascular damage disrupts frontal-subcortical-limbic neural pathways and connectivity in networks central to mood that predispose the development of depressive symptoms [38].

The interaction between AD, CVD, and depression is poorly understood. The aim of our study was to investigate the relationship between past depression and WMH burden in those with positive (AD+) and negative (AD-) CSF biomarker profiles for AD. We also aimed to better understand how CSF AD pathology and WMH burden relate to the clinical phenotypes of MCI and/or MDD. We compared AD CSF biomarker status (positive versus negative) and WMH burden among three groups: MCI, MCI with remitted MDD (MDD+MCI), and remitted MDD alone (MDD).

MATERIALS AND METHODS

RESULTS

DISCUSSION

Our study examined the relationship between AD CSF biomarkers (AD+ versus AD-), diagnosis (MCI, MDD, or MDD+MCI), WMH burden and neuropsychological performance. Among the MCI participants in our sample, more than half (7/12) were AD+, while only 2/12 of those with MDD+MCI were AD+. As expected, all participants with AD+ CSF were cognitively impaired and performed worse than the AD- participants in verbal memory. In AD- participants, we did not observe a distinct cognitive profile. Finally, nearly all participants had moderate or severe WMH burden, irrespective of their diagnosis or AD biomarker status; however, AD- participants had more severe WMH burden.

Depression has been identified as one of twelve modifiable risk factors for dementia [1], and there is evidence to support a direct link to AD pathogenesis [15–23]. The neuropathological processes associated with MDD and increased dementia risk are highly variable [39, 40]. Age of onset of first episode and severity and duration of depressive episodes serve as important mediators for dementia and AD risk [42–44]. As discussed above, older patients with MCI or a history of recent or remote MDD which has recurred (i.e., LLD) are considered to be at high risk for cognitive decline. There is evidence that LLD is a prodromal state of neurodegenerative disease [45–47]. The PACt-MD sample attempts to capture those with remitted EOD and LOD. Among the 19 LLD participants in our sample, all had EOD; 12 had MCI, with CSF AD pathology evident in only two participants. These results suggest that the etiology of cognitive impairment among participants with a history of MDD may be distinct from participants with MCI alone [6]. This may also differ between patients with remitted versus non-remitted LLD; since our study excluded patients with current MDE, we cannot comment on this group.

The role of WMH in cognitive decline and the etiology of dementia is well established [48, 49]. Nearly all participants in our sample (26/27) exhibited moderate or severe WMH, however a sizeable number (n = 7) of participants were cognitively intact despite significant CVD burden. Participants with AD+ CSF performed worse in verbal memory than AD- participants, consistent with the neurocognitive profile of AD [50, 51]. All participants with MCI exhibited slowed processing speed, which is typical of vascular cognitive impairment in the context of moderate to high WMH burden [52]. Variations in the spatial distribution of WMH are known to contribute to a range of clinical phenotypes [53]. Although our study did not assess WMH localization, periventricular WMH are more closely related to cognitive impairment than deep WMH [28–30]; frontal [31–34] and temporal [35] WMH are associated with depression. We hypothesize that vascular lesions in clinically meaningful loci may account for those exhibiting MCI, given that only 38% of those with MCI had positive CSF AD biomarkers (n = 9). The sizeable WMH burden in our sample suggest that protective processes such as cognitive reserve [62] may be contributing to intact cognition in the MDD group. Participants in our sample were highly educated, with a mean of 15.3 years of education, lending support to this hypothesis.

The ɛ4 allele of the apolipoprotein E (APOE4) is a well-established risk factor for AD [63]. In our sample, the frequency of APOE4 was as expected significantly higher among AD+ participants. The relationship between APOE and AD is poorly understood. One theory suggests that APOE4 may impair the clearance of Aβ, contributing to its accumulation [63]. There is mixed evidence that APOE4 carrier status may increase the risk of depression [64] and influence the association between WMH and cognitive decline in AD [65]. In our sample, the MDD+MCI group has a greater proportion of APOE4 carriers compared to the MDD group (41.7% versus 14.3%, respectively), however, our small sample size precludes any analysis of the relationship between APOE4 status, depression, WMH, and AD.

Study limitations include the limited generalizability of our results arising from the small sample size, and a predominantly White and highly educated demographic. We did not capture potentially salient details of prior MDEs, such as episode frequency, which could have potential correlations to WMH burden and an evolving vascular neurocognitive disorder. Anti-depressant use also appears to mediate improved cognitive performance among individuals with depression [66]. Anti-depressant use appeared to be more common in the MDD group compared to the MDD+MCI group (71.3% versus 58.3%, respectively); however, we were unable to perform statistical analyses to control for anti-depressant use due to our small sample size. We may also have underestimated WMH burden using non-fluid attenuated T2 images rather than FLAIR images, which is the gold standard for WMH quantification. Finally, despite the use of well-validated diagnostic criteria of the DSM-5, diagnosing a remote MDE remains unreliable [39, 67].

Further research with larger samples sizes are needed to examine the potential contribution of WMH in patients with a remitted MDD and MCI in conjunction with pathological biomarkers for AD and CVD. A causal link between WMH burden and depression with associated persistent cognitive impairment would provide a rationale for primary prevention of depression with control of vascular risk factors.