Late-Life Depression

LLD, defined as a major depressive episode occurring in an older adult (60 years or older), is a heterogeneous mood disorder that frequently presents with cognitive impairment. LLD encompasses both late-onset cases (i.e., first lifetime episode after age 60), as well as early-onset cases that recur or continue into later life. Epidemiological data suggest LLD prevalence of 1–4% among community-dwelling older adults, with higher prevalence (10–12%) among those hospitalized for medical or surgical reasons (Blazer, Reference Blazer2003). When sub-threshold (but still clinically significant) depressive symptoms are considered, these rates rise considerably, with prevalence estimates of 8–16% among community-dwelling elders, and greater than 30% among elders residing in hospital or long-term care facilities (Blazer, Reference Blazer2003). From a diagnostic perspective, LLD is often under-reported, under-recognized, misdiagnosed, and under-treated (Reynolds, Alexopoulos, Katz, & Lebowitz, Reference Reynolds, Alexopoulos, Katz and Lebowitz2001), due in part to characteristics that set it apart from mid-life depression. Compared to their younger counterparts, individuals with LLD are less likely to present with prominent sadness (Gallo, Rabins, Lyketsos, Tien, & Anthony, Reference Gallo, Rabins, Lyketsos, Tien and Anthony1997) and are more likely to exhibit agitation or somatic symptoms such as gastrointestinal distress, insomnia, or fatigue (Hegeman, Kok, van der Mast, & Giltay, Reference Hegeman, Kok, van der Mast and Giltay2012). In addition, while mid-life depression is more prevalent in women, such gender differences are greatly reduced in LLD (Steffens, Fisher, Langa, Potter, & Plassman, Reference Steffens, Fisher, Langa, Potter and Plassman2009). LLD is also often more chronic and difficult to treat to remission than mid-life depression, an observation that has been attributed in part to medical co-morbidity and cognitive dysfunction (Blazer, Reference Blazer2003). For these reasons and others, LLD is associated with significant mortality (including due to suicide), health care usage, physical disability, and functional decline (Beekman et al., Reference Beekman, Penninx, Deeg, de Beurs, Geerling and van Tilburg2002; Bruce et al., Reference Bruce, Ten Have, Reynolds, Katz, Schulberg, Mulsant and Alexopoulos2004; Steffens et al., Reference Steffens, Otey, Alexopoulos, Butters, Cuthbert, Ganguli and Yesavage2006).

Cognitive Functioning in Late-Life Depression

Cognitive deficits that often co-occur with LLD include impairments of episodic memory, speed of information processing, executive functioning, and visuospatial ability (Baudic, Tzortzis, Barba, & Traykov, Reference Baudic, Tzortzis, Barba and Traykov2004; Butters et al., Reference Butters, Whyte, Nebes, Begley, Dew, Mulsant and Becker2004; Elderkin-Thompson et al., Reference Elderkin-Thompson, Kumar, Bilker, Dunkin, Mintz, Moberg and Gur2003; Lockwood, Alexopoulos, & van Gorp, Reference Lockwood, Alexopoulos and van Gorp2002; Nebes et al., Reference Nebes, Butters, Mulsant, Pollock, Zmuda, Houck and Reynolds2000; Rapp et al., Reference Rapp, Dahlman, Sano, Grossman, Haroutunian and Gorman2005; Sheline et al., Reference Sheline, Barch, Garcia, Gersing, Pieper, Welsh-Bohmer and Doraiswamy2006). Of these domains, information processing speed and executive functioning appear particularly vulnerable, and several studies have reported that cognitive impairment associated with LLD appears to be predominantly mediated by slowed information processing and/or working memory deficits. Nebes and colleagues (Reference Nebes, Butters, Mulsant, Pollock, Zmuda, Houck and Reynolds2000) found that depressed older adults performed significantly worse on measures of both processing speed and working memory, and while performance on these measures improved in patients whose LLD remitted, the extent of improvement was no greater than that seen in comparison subjects who underwent repeat assessment. In a similar vein, Butters and colleagues (Reference Butters, Whyte, Nebes, Begley, Dew, Mulsant and Becker2004) reported that relative to non-depressed comparison subjects, LLD patients performed poorer in all cognitive domains, with information processing speed and visuospatial and executive abilities being the most broadly and frequently impaired. More recently, Sheline and colleagues (Reference Sheline, Barch, Garcia, Gersing, Pieper, Welsh-Bohmer and Doraiswamy2006) found that age, depression severity, education, race, and vascular risk factors all made significant and independent contributions to cognitive impairment in LLD, and that changes in information processing speed mediated the influence of predictor variables on other cognitive domains. In addition, patients with LLD who exhibit executive dysfunction may be at higher risk for poor outcomes, as deficits in executive functioning have been associated with poor or delayed antidepressant treatment response (Alexopoulos et al., Reference Alexopoulos, Kiosses, Heo, Murphy, Shanmugham and Gunning-Dixon2005) as well as higher relapse and recurrence rates in some studies (Alexopoulos et al., Reference Alexopoulos, Meyers, Young, Kalayam, Kakuma, Gabrielle and Hull2000), although not in others (Butters et al., Reference Butters, Whyte, Nebes, Begley, Dew, Mulsant and Becker2004).

A complex relationship exists between the treatment of LLD and improvement (or resolution) of cognitive symptoms, complicated further by the widely varying temporal association between cognitive and depressive symptoms. Studies suggest that, following antidepressant treatment, a subset of individuals with LLD and cognitive dysfunction may exhibit improved cognition (Doraiswamy et al., Reference Doraiswamy, Krishnan, Oxman, Jenkyn, Coffey, Burt and Clary2003), although it is clear that a significant proportion will continue to experience cognitive impairment (Bhalla et al., Reference Bhalla, Butters, Mulsant, Begley, Zmuda, Schoderbek and Becker2006, 2009; Butters et al., Reference Butters, Becker, Nebes, Zmuda, Mulsant, Pollock and Reynolds2000; Lee, Potter, Wagner, Welsh-Bohmer, & Steffens, Reference Lee, Potter, Wagner, Welsh-Bohmer and Steffens2007; Murphy & Alexopoulos, Reference Murphy and Alexopoulos2004; Nebes et al., Reference Nebes, Pollock, Houck, Butters, Mulsant, Zmuda and Reynolds2003), even in individuals who achieve depression remission (Basso, Miller, Estevis, & Combs, Reference Basso, Miller, Estevis and Combs2013). In one intervention study, Bhalla and colleagues (Reference Bhalla, Butters, Mulsant, Begley, Zmuda, Schoderbek and Becker2006) found that despite reaching remission, 94% of patients who were impaired at baseline when depressed remained impaired one year later. Of patients who were cognitively normal while depressed, 23% developed impairment one year later. In a subsequent cross-sectional study comparing individuals with remitted LLD and never-depressed comparison subjects, nearly twice as many remitted depressed subjects were diagnosed with MCI or dementia (48% vs. 28%), suggesting that despite adequate depression treatment, cognitive impairment persisted (Bhalla et al., 2009). Others have similarly noted persistent cognitive impairment following treatment and/or remission of depressive symptoms (Bruce et al., Reference Bruce, Ten Have, Reynolds, Katz, Schulberg, Mulsant and Alexopoulos2004; Butters et al., Reference Butters, Becker, Nebes, Zmuda, Mulsant, Pollock and Reynolds2000; Doraiswamy et al., Reference Doraiswamy, Krishnan, Oxman, Jenkyn, Coffey, Burt and Clary2003; Lee et al., Reference Lee, Potter, Wagner, Welsh-Bohmer and Steffens2007; Reynolds et al., Reference Reynolds, Butters, Lopez, Pollock, Dew, Mulsant and DeKosky2011), with certain risk factors, including lower baseline cognitive function, older age, later age of depression onset, and greater vascular burden associated with less improvement in cognitive function (Barch et al., Reference Barch, D’Angelo, Pieper, Wilkins, Welsh-Bohmer, Taylor and Sheline2012). With regard to age, Arve and colleagues reported a doubling of co-morbid depressive symptoms and cognitive impairment for every 5-year increase in age (after 70), an observation that extrapolated to a prevalence of roughly 25% in individuals 85 and older presenting with both low mood and impaired cognition (Arve, Tilvis, Lehtonen, Valvanne, & Sairanen, Reference Arve, Tilvis, Lehtonen, Valvanne and Sairanen1999).

Characterizing differences between older adults with recurrent early-onset depression (EOD) versus those with late-onset depression (LOD) has been of great interest in recent years (Herrmann, Goodwin, & Ebmeier, Reference Herrmann, Goodwin and Ebmeier2007; Janssen et al., Reference Janssen, Hulshoff Pol, de Leeuw, Schnack, Lampe, Kok and Heeren2007; Rapp et al., Reference Rapp, Dahlman, Sano, Grossman, Haroutunian and Gorman2005), partly due to hypothesized differences in their neurobiological etiologies. Janssen and colleagues, for example, suggested that in EOD, stress-related neurotoxicity associated with repeated depressive episodes may result in hippocampal atrophy, while in LOD, large subcortical white matter lesions may serve as an intermediary between cerebrovascular disease and depression. Schneider and colleagues, however, have demonstrated that among community-dwelling older adults, individuals with dementia most often have mixed brain pathologies (Schneider, Arvanitakis, Bang, & Bennett, Reference Schneider, Arvanitakis, Bang and Bennett2007), suggesting that subtypes related to age of depression onset (EOD vs. LOD) may not correlate well with observed neuropathological etiologies in a real-world clinical population. Regardless of the underlying neurobiology—of which a detailed discussion is beyond the scope of this short review—LOD is generally defined as depressive illness that develops for the first time after age 60, while EOD indicates depression that first presented before this age. From a cognitive perspective, the literature has been inconsistent as to whether EOD and LOD are associated with worse or different cognitive profiles. Although both may present with cognitive impairment, most studies suggest that LOD may be associated with more executive and attentional deficits (Rapp et al., Reference Rapp, Dahlman, Sano, Grossman, Haroutunian and Gorman2005), as opposed to more primary episodic memory deficits in EOD (Salloway et al., Reference Salloway, Malloy, Kohn, Gillard, Duffy, Rogg and Westlake1996). A recent meta-analysis found that patients with LOD exhibit slower processing speed and poorer executive functioning than patients with EOD and non-depressed comparison subjects, although both depressed groups exhibit reduced functioning in all cognitive domains as compared to comparison subjects (Herrmann et al., Reference Herrmann, Goodwin and Ebmeier2007). While this suggests that pronounced executive deficits may be typical of LOD patients, the more robust finding is that older depressed patients demonstrate reduced cognitive functioning compared with non-depressed controls. Taken together, it is clear that LLD is a complicated, heterogeneous disorder, and although studying EOD and LOD phenotypes is appealing, the literature to date remains unclear as to whether consistent distinguishing characteristics exist.

Late-Life Depression as a Risk Factor for Cognitive Decline and Dementia

Although the temporal association between cognitive and depressive symptoms in LLD varies widely, increasing evidence suggests that depression contributes to the development of persistent cognitive dysfunction in a subset of individuals. Clinical, case-control, and epidemiologic studies have reported an association between LLD and progressive cognitive deficits, and it is clear that cognitive deficits accompanying LLD often persist following treatment of depressive symptoms (Bhalla et al., Reference Bhalla, Butters, Mulsant, Begley, Zmuda, Schoderbek and Becker2006; Butters et al., Reference Butters, Becker, Nebes, Zmuda, Mulsant, Pollock and Reynolds2000; Diniz, Butters, Albert, Dew, & Reynolds, Reference Diniz, Butters, Albert, Dew and Reynolds2013; Jorm, Reference Jorm2001; Murphy & Alexopoulos, Reference Murphy and Alexopoulos2004; Nebes et al., Reference Nebes, Butters, Mulsant, Pollock, Zmuda, Houck and Reynolds2000; Ownby, Crocco, Acevedo, John, & Loewenstein, Reference Ownby, Crocco, Acevedo, John and Loewenstein2006; Paradiso, Lamberty, Garvey, & Robinson, Reference Paradiso, Lamberty, Garvey and Robinson1997). Findings from a recent longitudinal observational study of women who were cognitively healthy at baseline, for example, showed that baseline scores on the Geriatric Depression Scale (Yesavage et al., Reference Yesavage, Brink, Rose, Lum, Huang, Adey and Leirer1982) were highly associated with subsequent impairment on all cognitive tests (immediate recall, delayed recall, psychomotor speed, executive functioning), suggesting that depression may be a risk factor for cognitive decline, and thus a potential target for diagnostic and therapeutic intervention (Rosenberg, Mielke, Xue, & Carlson, Reference Rosenberg, Mielke, Xue and Carlson2010). Other studies have found LLD to be associated with an approximately 50% increased likelihood of developing all-cause dementia, including Alzheimer’s disease and, in particular, vascular dementia (Diniz et al., Reference Diniz, Butters, Albert, Dew and Reynolds2013; Jorm 2001; Ownby et al., Reference Ownby, Crocco, Acevedo, John and Loewenstein2006). Some reports, however, have failed to find this association (Becker et al., Reference Becker, Chang, Lopez, Dew, Sweet, Barnes and Reynolds2009; Ganguli, Du, Dodge, Ratcliff, & Chang, Reference Ganguli, Du, Dodge, Ratcliff and Chang2006; Lindsay et al., Reference Lindsay, Laurin, Verreault, Hebert, Helliwell, Hill and McDowell2002). Furthermore, it is likely that the relationship between vascular disease and LLD (and related cognitive impairment) is, to a certain extent, bi-directional (Thomas, Kalaria, & O’Brien, Reference Thomas, Kalaria and O’Brien2004). However, the details of this complex issue are beyond the scope of the current review, and are discussed elsewhere (e.g., Naismith, Norrie, Mowszowski, & Hickie, Reference Naismith, Norrie, Mowszowski and Hickie2012; Taylor, Aizenstein, & Alexopoulos, Reference Taylor, Aizenstein and Alexopoulos2013).

Evidence from several studies suggests that risk for MCI or dementia is proportional to an individual’s cumulative depression burden, as measured by lifetime duration of depression or number of depressive episodes. A large Danish case-registry study, for example, found that the rate of a subsequent diagnosis of dementia was significantly correlated with the number of prior depressive episodes. On average, the rate of dementia increased by 13% with every depressive episode that led to an inpatient admission (Kessing & Andersen, Reference Kessing and Andersen2004). Several studies (Geerlings, den Heijer, Koudstaal, Hofman, & Breteler, Reference Geerlings, den Heijer, Koudstaal, Hofman and Breteler2008; Palsson, Aevarsson, & Skoog, Reference Palsson, Aevarsson and Skoog1999; Speck et al., Reference Speck, Kukull, Brenner, Bowen, McCormick, Teri and Larson1995) have found that a longer interval between onset of first depressive episode and diagnosis of cognitive dysfunction increases the risk for developing dementia—supporting the aforementioned hypothesis regarding cumulative depressive burden and risk for cognitive dysfunction—while others have suggested that a shorter interval between the two leads to greater dementia risk (Green et al., Reference Green, Cupples, Kurz, Auerbach, Go, Sadovnick and Farrer2003; Steffens et al., Reference Steffens, Plassman, Helms, Welsh-Bohmer, Saunders and Breitner1997; Wetherell, Gatz, Johansson, & Pedersen, Reference Wetherell, Gatz, Johansson and Pedersen1999). Others, still, have found no relationship between the two (Ganguli et al., Reference Ganguli, Du, Dodge, Ratcliff and Chang2006; Godin et al., Reference Godin, Dufouil, Ritchie, Dartigues, Tzourio, Peres and Alperovitch2007; Lindsay et al., Reference Lindsay, Laurin, Verreault, Hebert, Helliwell, Hill and McDowell2002; Mendez et al., Reference Mendez, Underwood, Zander, Mastri, Sung and Frey1992). Overall, these mixed findings emphasize the heterogeneity of LLD and its relationship to cognitive comorbidity and sequelae, and further raise the question of whether prior depression is a true etiologic risk factor for dementia, or instead a prodromal clinical manifestation of underlying dementia neuropathology. Complexities of depression as a syndrome, and variability in the measures, methodologies, and treatments used in these studies, likewise contribute to inconsistent findings.

To shed light on this topic, a recent review examined sixteen clinical studies that investigated the association between dementia and depression, and concluded that earlier-life depression was a consistent risk factor (and unlikely prodrome) for dementia (Byers & Yaffe, Reference Byers and Yaffe2011). In addition, while current research findings suggest an association between LLD and risk of dementia, several inconsistencies across individual studies exist, perhaps due to methodological, cultural, or sub-population differences across the studies (Byers & Yaffe, Reference Byers and Yaffe2011). Furthermore, the designs of many of these studies make it difficult (if not impossible) to determine whether depressive episodes were a symptom of prodromal dementia, or instead the psychological consequence of an already present cognitive deficit.

Neurobiological Pathways Linking Late-Life Depression to Cognitive Impairment and Dementia

Despite ongoing efforts to elucidate their underpinnings, the neurobiological mechanisms linking depression to increased risk for cognitive decline and dementia remain poorly understood. While a comprehensive discussion of this topic is beyond the scope of this review, some of these mechanisms are highlighted in Figure 1, and are described elsewhere in greater detail (e.g., Butters et al., Reference Butters, Young, Lopez, Aizenstein, Mulsant, Reynolds and Becker2008, Byers & Yaffe, Reference Byers and Yaffe2011). The current literature strongly suggests that from diagnostic (Diniz et al., Reference Diniz, Butters, Albert, Dew and Reynolds2013) and neuropathological (Sweet et al., Reference Sweet, Hamilton, Butters, Mulsant, Pollock, Lewis and Reynolds2004) perspectives, LLD is most commonly associated with increased risk for clinical Alzheimer’s disease (AD). However, there is also evidence that non-AD neuropathological changes (such as Lewy bodies, vascular disease, and hippocampal atrophy related to hypercortisolemia) occur in individuals with LLD at greater rates than in non-depressed individuals (Schneider et al., Reference Schneider, Arvanitakis, Bang and Bennett2007; Sweet et al., Reference Sweet, Hamilton, Butters, Mulsant, Pollock, Lewis and Reynolds2004). Figure 1 depicts mechanisms by which depression may contribute both directly to AD pathology and indirectly, by reducing brain reserve, to the earlier appearance of clinical dementia symptoms. The figure does not, however, present a comprehensive model of all neurobiological mechanisms involved (including AD pathogenesis and various clinical outcomes), and does not depict all interrelationships among constituent parts. For example, it does not depict the possibility that in some individuals, depression may interact directly with AD causal risk factors to cause clinical manifestations of AD; in this scenario, depression may only be a risk factor in the presence of another established risk factor (Geda et al., Reference Geda, Knopman, Mrazek, Jicha, Smith, Negash and Rocca2006).

Fig. 1

Potential pathways linking late-life depression to clinical onset of dementia. Caption: Adapted from Butters, M.A. et al. (2008). Pathways linking late-life depression to persistent cognitive impairment and dementia. Dialogues in Clinical Neuroscience, 10(3), 345–357. Reproduced with permission of the publisher (Les Laboratoires Servier: Suresnes, France).

The model in Figure 1 depicts three interacting pathways, based on well-established findings that LLD is associated with hypothalamic-pituitary-adrenal (HPA) -axis dysfunction leading to elevated adrenal glucocorticoid production, as well as chronic inflammation (Butters et al., Reference Butters, Young, Lopez, Aizenstein, Mulsant, Reynolds and Becker2008). There are two potentially independent pathways—each involving HPA-axis dysfunction—through which depression may lead to increased risk for, or incidence of, AD. In the first, HPA-axis dysfunction may directly contribute to AD pathology, based on evidence from rodent studies (Dong & Csernansky, Reference Dong and Csernansky2009; Rothman & Mattson, Reference Rothman and Mattson2010). In the second, HPA-axis dysfunction in the form of hypercortisolemia may, over repeated episodes, lead to apoptosis of hippocampal cortisol receptors, resulting in hippocampal atrophy. The third pathway in the model involves depression-related chronic inflammation leading to ischemia, especially in frontostriatal pathways. The combination of hippocampal atrophy, generalized ischemia, and frontostriatal dysfunction likely lowers brain reserve, which in the context of underlying AD neuropathology hastens the clinical presentation of dementia. In this model, it is the depression-related increase in AD pathology, together with the reduction in brain reserve caused by depression-associated neurotoxocity, which reduces time to expression of clinical dementia (thereby accounting for the frequently reported increased risk of cognitive decline and dementia among individuals with a history of depression).

A detailed discussion of the literature summarizing the above-mentioned pathways is beyond the scope of this brief review. For recent work on the relationship between stress, HPA-axis dysfunction, and AD pathology in rodents, the interested reader is referred to reviews by Dong and Csernansky (Reference Dong and Csernansky2009) and Rothman & Mattson (Reference Rothman and Mattson2010). For detailed discussions of the relationship between hippocampal volume and depression, the reader is referred to publications by Videbech & Ravnkilde (Reference Videbech and Ravnkilde2004), Janssen and colleagues (Reference Janssen, Hulshoff Pol, de Leeuw, Schnack, Lampe, Kok and Heeren2007), and Sheline, Gado, and Kraemer (Reference Sheline, Gado and Kraemer2003). Lastly, for a review of important work on the relationship between frontostriatal lesions and depression, the reader is referred to studies by Herrmann, Le Masurier, and Ebmeier (Reference Herrmann, Le Masurier and Ebmeier2008) and Krishnan and colleagues (Reference Krishnan, O’Brien, Firbank, Pantoni, Carlucci, Erkinjuntti and Inzitari2006).