**1. Introduction**

Dementia, one of the most common neurodegenerative disorders, is a devastating illness characterized by significant cognitive decline that induces interference in daily life and behavioral disturbances [1]. Alzheimer's disease (AD) is the most common dementia type, with worldwide patients expected to increase from 82 million in 2030 to 152 million in 2050 [2]. One in every 2–3 people over the age of 85 will develop AD-related dementia [3], and most AD patients experience mild cognitive impairment (MCI), which is the preclinical status of dementia with modest cognitive decline without dysfunction in daily life [4,5]. Several studies have established that the accumulation of amyloid β, hyperphosphorylation of tau proteins, and neuroinflammation affect the neurodegeneration seen in AD [6,7]. However, there is no effective drug for both delaying onset and restoring cognitive function. Therefore, delaying disease onset or progression could provide a significant reduction in the social and economic burden of these diseases [8]. For delaying or preventing AD, previous studies have found several modifiable risk factors, including diet, midlife hypertension, type 2 diabetes mellitus, smoking, cognitive/physical inactivity, traumatic brain injury, and depression [9–11].

Depression is the most prevalent coexisting noncognitive feature that occurs along with cognitive deficits and is associated with neurodegenerative disorders and cognitive decline [12–14]. Because a major depressive disorder (MDD) is a heterogeneous diagnostic

category that features differences in symptom profiles, comorbidities, and the course of disease [15,16], late-life depression (LLD) with an age of depression onset over 60 years has received a great deal of attention [17,18]. Moreover, the global number of individuals with LLD has increased by 27.1% from 2007 to 2017 [19]. Therefore, elucidating the link between the two disorders will help doctors and families understand and manage AD. Epidemiologic data have shown that LLD increases the risk of AD [20,21], and LLD is a risk factor that affects the progression of dementia from the normal cognition to MCI and from MCI to dementia [12,21–23]. Additionally, the risk of conversion from MCI to AD may vary due to the symptom severity of LLD or its successful treatment [24]. Individuals with LLD and high amyloid β levels exhibited a shortened conversion time than those without depression and with high amyloid β levels [25,26]. Altered levels and metabolism of amyloid β seen in AD were also reported in individuals with LLD [27]. Although these findings support previously suggested mechanisms that connect depression and dementia [28], a previous systematic review pointed out that these results are not consistent with other studies [29]. This discrepancy may be due to the study population differences or methodologic differences between the various studies [30]. Therefore, it is necessary to subdivide the study population and conduct research associated with more specific criteria.

Cognitive impairments in individuals with depression have been consistently reported in meta-analyses and reviews [31–34]. Based on these results, difficulties with concentration and making decisions have been described as part of major depressive disorder (MDD) [1]. Cognitive impairments in MDD were reported across most domains [35,36]. These cognitive impairments in MDD patients are usually normalized after remission of the MDD [35,36]. However, studies using a comprehensive neuropsychological battery have reported that cognitive impairment in remitted LLD patients persisted in executive function and episodic memory compared to healthy controls [37–39]. In addition, a longitudinal study has reported that LLD patients exhibit a significant decline in all domains, and three-month remitters also exhibited a significant decline in verbal fluency and executive function [17], suggesting that certain aspects of executive functioning are associated with the traits of LLD. Although other studies reported inconsistent results with no difference in LLD [40–42], these inconsistent results may be attributed to the differences in cognitive tests. Episodic memory is the other main impaired cognitive domain in individuals with MCI [43]. Impairment in these cognitive domains was usually exhibited to a greater extent in individuals with LLD+MCI (and those with AD), relative to individuals with LLD [44–47].

In recent years, using improved neuroimaging technology, we can investigate brain structure and function through neuroimaging tools, magnetic resonance imaging (MRI), computed tomography, and positron emission tomography (PET). Among them, functional MRI (fMRI) can provide information about the properties of functional connectivity (FC) that is, collections of brain regions that are coactivated to support shared functions—during a task or rest (i.e., in the absence of stimuli) through measuring the blood oxygenation level-dependent (BOLD) signal [48,49]. More specifically, previous studies have suggested resting-state (rs)-fMRI as a promising method for investigating the behavioral characteristics including psychological states: sustained attention [50], personality [51], temperament traits [52], creative ability [53], and cognitive ability, such as working memory and motor performance [54]. These newer methods provide reproducible results and reflect stable trait-like neurobiological signatures [55,56]. Recent work also presents that the patterns of resting-state FC are uniquely related both to specific symptoms and to respond to different forms of treatment [57,58]. Thus, reviewing rs-fMRI results seems to be suitable for understanding the links between AD/MCI and LLD.
