Sage Journals: Discover world-class research

Abstract

Objective:

Depression has been shown to be a risk factor for Alzheimer’s disease (AD), and in older adults may provide a marker for the beginning of the prodromal phase of AD. The purpose of this systematic review is to examine the relationship between amyloid-β (Aβ), a key biomarker of AD, and depression in older adults.

Method:

The literature search was limited to studies conducted from 2006 to 2014 that were published in English in peer-reviewed journals. Studies were selected if they included a group of older adults who either met established criteria for Major Depressive Disorder or Dysthymia; or were assessed for depressive symptoms on a standardised measure. Studies were also required to include an outcome variable that was a direct measure of Aβ levels in either blood or cerebrospinal fluid (CSF) samples, or via neuroimaging techniques such as positron emission tomography (PET).

Results:

Nineteen studies were identified, 15 of which found significant differences in Aβ levels between depressed and non-depressed older adults. However, studies were limited by their cross-sectional design, reliance on blood-based measures of Aβ, and potential sample bias.

Conclusions:

Future investigations should consider prospective longitudinal design using neuroimaging and CSF measures of Aβ.

Keywords

Depression Amyloid-β Alzheimer’s disease

Converging clinical evidence suggests that in otherwise healthy older adults an increase in depressive symptoms can be an early or preclinical indicator of dementia (Byers and Yaffe, 2011). The most common type of dementia, Alzheimer’s disease (AD), accounts for 70% of cases worldwide (World Health Organisation, 2012). Evidence for a relationship between depression and AD comes from observations that a clinical diagnosis of depression in both early and late life increases the risk for AD diagnosis in healthy older adults (Da Silva et al., 2013; Vilalta-Franch et al., 2013) and in individuals who meet criteria for mild cognitive impairment (MCI) (Richard et al., 2013). Authoritative reviews also conclude that an increased number, longer duration and higher frequency of depressive episodes across the lifespan further increase the risk of developing AD (Byers and Yaffe, 2011; Piccinni et al., 2013). However, these reviews also emphasise that associations between depressive symptoms and clinically diagnosed AD are not evident in all studies (Luppa et al., 2013; Neubauer et al., 2013; Wilson et al., 2010). Thus, factors such as methodology, context, or the presence of subgroups of people with depression and additional risk factors for AD (e.g. apolipoprotein ε4 [APOE ε4] carriage) (Corder et al., 1993) may influence the extent to which depressive symptoms can predict clinical AD (Byers and Yaffe, 2011).

Several factors that complicate determination of the extent to which depressive symptoms are related to early AD have been identified. First there is overlap between the symptoms of depression and those that occur in early dementia. For example, 20–30% of patients with AD show lowered mood and decreased motivation (Tsuno and Homma, 2009). Second, subtle cognitive impairment, particularly in memory, is characteristic of both early AD (Albert et al., 2011; Sperling et al., 2011) and depression in older adults (Byers and Yaffe, 2011; Tsuno and Homma, 2009; Wells, 1979). Third, the insidious nature of AD means that symptoms develop very slowly (Dubois and Albert, 2004; Petersen et al., 2001). Therefore, prospective studies of older adults with depressive symptoms must be conducted over long intervals (e.g. decades) and include multiple clinical outcomes against which to determine any predictive validity of depression for AD. Finally, increased depressive symptoms in healthy older adults may not be a direct reflection of AD pathology but rather a consequence of individuals realising their cognitive function is failing (Byers and Yaffe, 2011). Considered together these data suggest that while depressive symptoms do increase the risk for AD, the nature and direction of the relationship between the two conditions remains unclear, as is the extent to which depressive symptoms predict incipient AD.

With a biological marker of AD, studies can now examine the extent to which depressive symptoms are associated with AD pathology over much shorter periods and with greater diagnostic specificity for dementia than previously achievable from studies dependent on clinical diagnosis of dementia. Advances in biomarker and neuroimaging techniques now make it possible to detect accumulation of amyloid-β (Aβ) in living non-demented older adults (Sperling et al., 2011). Prospective studies of older adults with increased cerebral Aβ show decline in episodic memory up to 20 years before clinical diagnosis of AD (Doraiswamy et al., 2012; Lim et al., 2013b, 2013c, 2014a, 2014b; Small et al., 2012; Villemagne et al., 2011). Study of the extent to which changes in mood occur in individuals with increased cerebral Aβ could inform models of the relationship between depressive symptomatology and AD.

To date several studies have measured levels of Aβ in older adults with clinically classified depression, and have observed changes in Aβ levels both in vivo (Cirrito et al., 2011; Sun et al., 2008, 2011) and post mortem (Rapp et al., 2006, 2011). However, these studies have quantified Aβ levels differently, using blood, cerebrospinal fluid (CSF) and Aβ neuroimaging biomarkers. Furthermore, depressive symptoms have been defined using categorical (e.g. DSM-IV criteria for major depression), continuous (e.g. score on a standardised measure such as the Geriatric Depression Scale [GDS]) or qualitative methods (e.g. clinical interview) (Byers and Yaffe, 2011). Finally, most studies measured relationships between depressive symptoms and Aβ on a single occasion (i.e. cross-sectional design). Consequently, development of a model of the extent to which Aβ biomarkers are associated with depressive symptoms, or clinical depression, has not been possible from considering the result of each study in isolation. Given the variation in methodology, a systematic review of outcomes across studies is warranted to improve understanding of relationships between depressive symptoms and Aβ and provide a foundation for studying such relationships.

The aim of this review was therefore to systematically examine relationships between depressive symptoms, or clinical depression, and Aβ in non-demented older adults.

Method

Study selection criteria

Type of study

This review focused on research investigating the relationship between Aβ and depression in older adults. Only peer-reviewed studies published in English since 2006 were considered. Included studies were of either cross-sectional or longitudinal design.

Participants

Studies were selected if the sample included a group of older adults aged 60 years or more and: (a) a group who met established criteria (Diagnostic and Statistical Manual of Mental Disorders [DSM] or International Classification of Diseases 10^th Revision [ICD-10]) for Major Depressive Disorder or dysthymia; or (b) a standardised measure of depressive symptoms was used to assess participants’ level of depression. Studies primarily focused on participants with Bipolar Disorders or disorders other than unipolar depression were excluded. Studies that did not directly assess participants’ depressive symptoms or diagnosis were also excluded.

Outcome measures

Studies were included if Aβ was measured using blood, CSF samples or positron emission tomography (PET) neuroimaging techniques. Studies of depression that measured indirect markers of Aβ, such as autoantibodies, or did not measure Aβ levels were excluded. In addition, studies that focused primarily on the risk of developing AD without assessing Aβ levels were excluded.

Search methods for identification of studies

Electronic database searches

Three electronic databases were subjected to systematic searches on 15 May 2014 using the search terms described below:

Configured for EBSCOhost (PsycINFO, PsycExtra, PsycArticles, MEDLINE):

Depress* (SU) OR Affect* (SU) OR Mood (SU) 2. Amyloid* (SU) OR A-beta (SU) OR Aβ* (SU) OR Beta-amyloid (SU) OR PiB (SU) OR ¹⁸F (SU) OR CSF (SU) OR Cerebrospinal (SU) 3. Alzheimer* (SU) OR MCI (SU) 4. Cognit* (TX) OR Memory (TX) OR “Executive Function” (TX) OR Attention (TX) OR “Processing Speed” (TX) OR Visuospatial (TX) OR Language (TX)

Configured for Scopus:

Depress* (KEY) OR Affect* (KEY) OR Mood (KEY) 2. Amyloid* (KEY) OR A-beta (KEY) OR Aβ* (KEY) OR Beta-amyloid (KEY) OR PiB (KEY) OR ¹⁸F (KEY) OR CSF (KEY) OR Cerebrospinal (KEY) 3. Alzheimer* (KEY) OR MCI (KEY) 4. Cognit* (ALL) OR Memory (ALL) OR “Executive Function” (ALL) OR Attention (ALL) OR “Processing Speed” (ALL) OR Visuospatial (ALL) OR Language (ALL)

Configured for Proquest:

Depress* (SU) OR Affect* (SU) OR Mood (SU) 2. Amyloid* (SU) OR A-beta (SU) OR Aβ* (SU) OR Beta-amyloid (SU) OR PiB (SU) OR ¹⁸F (SU) OR CSF (SU) OR Cerebrospinal (SU) 3. Alzheimer* (SU) OR MCI (SU) 4. Cognit* (ALL) OR Memory (ALL) OR “Executive Function” (ALL) OR Attention (ALL) OR “Processing Speed” (ALL) OR Visuospatial (ALL) OR Language (ALL)

Search strategy

The following process was used to search all three databases and to screen articles for relevance against the search criteria:

Subject terms for Depression, Aβ amyloid, AD, and Cognition were entered simultaneously, connected by the AND Boolean phrase

Non-English language and non-peer-reviewed articles (e.g. review, comment or opinion articles) were manually removed

Duplicates were removed

Titles and abstracts were screened for meeting inclusion and exclusion criteria

Remaining articles were then screened in full text

At this stage 15 articles were identified as meeting full inclusion criteria (Figure 1).

Figure 1.

Flowchart of search procedure and number of studies included and excluded from the analysis at each stage.

Additional searches

Studies meeting full inclusion criteria were subjected to reference list screening for key study citations. Studies not found by the current search strategy with a relevant title were manually retrieved and subjected to abstract and then full-text screening for relevance. Ten potentially relevant articles were screened, with four meeting inclusion criteria. The inclusion and exclusion process is illustrated in Figure 1.

Calculation of effect sizes

Only one of the 19 retained articles reported a measure of effect size (Cohen’s d) for relevant outcomes. To enable better comparison across studies, the magnitude of difference between sample means for relevant outcomes from each study were expressed using Cohen’s d (Cohen, 1988), a measure of effect that expresses the difference between means as a function of the pooled standard deviation of both groups. Six studies reported means and standard deviations and a further eight studies reported p-values and sample sizes and these were used to calculate Cohen’s d. Three studies did not provide sufficient information for calculation of Cohen’s d. For all studies the direction of differences were standardised to express outcome measures in the depressed group relative to controls.

Results

Nineteen studies satisfied the review inclusion/exclusion criteria. Design and sample characteristics for each study are summarised in Table 1. The outcome measures and relevant study results are summarised in Table 2.

Table 1.

Summary of study design and sample characteristics of included articles.

Study(author & year)	Participants					Criteria for depression	Type of depression
	Groups (n)	Mean Age	Gender (%Male)	Mean Education	APOE ε4 (%carriers)
Blasko et al., 2010	331 non-demented, never depressed	75.8	43.5%	10.6 years	20.2%	DSM-IV Criteria	Major, minor and subclinical
Butters et al., 2008	8 unmatched controls8 depressed	71.8	50%	14.4 years	Not reported	DSM-IV Criteria	Major
Kita et al., 2009	30 matched controls30 depressed	55.9	42.5%	12.6 years	Not reported	DSM-IV Criteria	Major
Kramberger et al., 2012	51 not depressed SCI41 depressed SCI60 not depressed AD31 depressed AD	67.6	33.3%	11.7 years	Not reported	CSDD score >6	Major, minor and subclinical
Madsen et al., 2012	18 matched controls28 depressed	61.1	34.8%	12.2 years	Not reported	ICD-10 criteria with previous psychiatric hospital admission	Moderate and Major
Namekawa et al., 2013	81 matched controls89 depressed	67.3	29.4%	11.8 years	17.6%	DSM-IV Criteria	Major
Reis et al., 2012	8 healthy older adults20 depressed12 AD	71.3	30%	5.6 years	Not reported	DSM-IV Criteria	Major, first episode
Sun et al., 2008	647 non-depressed348 depressed	74.9	23.8%	63.8% completed high school or above	23.4%	CES-D score >16	Major, minor and subclinical
Baba et al., 2011	64 depressed 160 healthy controls	56.3	42%	12.9 years	20%	DSM-IV Criteria	Major
Kumar et al., 2011	19 matched controls20 depressed	67.0	43.6%	16.3 years	Not reported	HAM-D score >15DSM-IV criteria	Major
Moon et al., 2011	65 non-depressed58 depressed	76.0	74%	2.7 years	Not reported	SGDS-K score >8	Major
Pomara et al., 2012	19 matched controls28 depressed	67.3	53%	16.6 years	34%	DSM-IV Criteria	Major
Lavretsky et al., 2009	20 cognitively normal23 MCI	65.9	48.8%	17.2 years	Not reported	Not categorised, used GDS score continuously	N/A
Sun et al., 2009	285 depressed APOE ε4-520 non-depressed APOE ε4-80 depressed APOE ε4+175 non-depressed APOE ε4+	74.5	22.2%	60.5% completed high school or above	24%	CES-D score >16	Major, minor and subclinical
Wu et al., 2014	25 depressed11 non-depressed	69.0	20%	7.6 years	19%	DSM-IV Criteria	Major
Gudmundsson et al., 2007	70 healthy older adults 11 depressed	72.6	0%	Not reported	Not reported	DSM-III Criteria	Major
Pomara et al., 2006	47 depressed35 non-depressed	74.6	30.5%	Not reported	Not reported	DSM-IV Criteria	Major
Qiu et al., 2007	101 depressed354 non-depressed	75.7	23%	7.5 years	23.3%	CES-D score >16	Major, minor and subclinical
Sun et al., 2007	101 depressed223 non-depressed	74.6	24.6%	Not Reported	24.9%	CES-D score >16	Major, minor and subclinical

DSM: Diagnostic and Statistical Manual of Mental Disorders; CSDD: Cornell Scale for Depression in Dementia; ICD-10: International Classification of Disease 10^th revision; CES-D: Center for Epidemiologic Studies Depression Scale.

Table 2.

Summary of outcome measures, covariates, analytic methods and relevant findings.

Study(author & year)	Measure of depression	Measure of Aβ amyloid	Relevant findings and effect sizes (calculated manually except where indicated)
			Aβ₄₀	Aβ₄₂	Aβ_40:42	PET
Blasko et al., 2010	SGDS	Plasma Aβ₄₂	–	Newly depressed at 5 year follow up, d = 0.41*	–	–
Butters et al., 2008	SCID-1	PiB PET	–	–	–	Frontal cortex, d = 0.38Precuneus, d = 0.41
Kita et al., 2009	HAM-D	Serum Aβ₄₀Serum Aβ₄₂Serum Aβ_40:42	d = -0.04	d = -0.48	d = 1.01*	–
Kramberger et al., 2012	CSDD	CSF Aβ₄₂	–	SCI, d = 0.03AD, d = -0.30	–	–
Madsen et al., 2012	HAM-DGDS	PiB PET	–	–	–	Global, d = -0.18
Namekawa et al., 2013	HAM-D	Serum Aβ₄₀Serum Aβ₄₂Serum Aβ_40:42	EO, d = -0.20LO, d = 0.05	EO, d = -0.63LO, d = -0.36	EO, d = 0.49LO, d = 0.35	–
Reis et al., 2012	HAM-D – Brazilian version	CSF Aβ₄₂	–	No difference, unable to calculate d	–	–
Sun et al., 2008	CES-D	Plasma Aβ₄₀Plasma Aβ₄₂Plasma Aβ_40:42	d = 0.19	d = -0.28*	d = 0.34*	–
Baba et al., 2011	HAM-D	Serum Aβ₄₀Serum Aβ₄₂Serum Aβ_40:42	d = -0.13	d = -0.35*	d = 0.37*	–
Kumar et al., 2011	HAM-D SCID-1	[¹⁸F] FDDNP PET	–	–	–	d = 0.82* (reported in article)
Moon et al., 2011	SGDS-K	Plasma Aβ₄₂	–	d = 0.37SGDS-K correlation,r = 0.18	–	–
Pomara et al., 2012	HAM-D SCID-1	CSF Aβ₄₀CSF Aβ₄₂CSF Aβ_40:42	d =-0.57	d = -0.73HAM-D correlation,r = -0.42	–	–
Lavretsky et al., 2009	GDS	[¹⁸F] FDDNP PET	–	–	–	MCI LTR, GDS correlation,r = 0.44CN MTR, GDS correlation, r = 0.55
Sun et al., 2009	CES-D	Plasma Aβ₄₀Plasma Aβ₄₂Plasma Aβ_40:42	No difference, data not reported	APOE ε4non-carriers,d = -0.19*	APOE ε4non-carriers,d = 0.15*	–
Wu et al., 2014	HAM-D SCID-1	[¹⁸F] Florbetapir PET	–	–	–	Global, d = 0.46Parietal, d = 0.91Precuneus, d = 0.80
Gudmundsson et al., 2007	MADRS	CSF Aβ₄₂	–	d = -0.78*	–	–
Pomara et al., 2006	HAM-D	Plasma Aβ₄₀Plasma Aβ₄₂Plasma Aβ_40:42	No difference, data not reported	d = 0.57*	d = 0.79*	–
Qiu et al., 2007	CES-D	Plasma Aβ₄₀Plasma Aβ₄₂Plasma Aβ_40:42	No difference, unable to calculate d	Depressed group lower*, unable to calculate d	–	–
Sun et al., 2007	CES-D	Plasma Aβ₄₀Plasma Aβ₄₂Plasma Aβ_40:42	No difference, unable to calculate d	Depressed group lower*, unable to calculate d	–	–

indicates statistical significance.

Note. The direction of differences were standardised to indicate the measure in the depressed group relative to controls.

SGDS: Short Form Geriatric Depression Scale; SCID-1: Structured Clinical Interview for DSM-IV; HAM-D: Hamilton Rating Scale for Depression; CSDD: Cornell Scale for Depression in Dementia; GDS: Geriatric Depression Scale; MADRS: Montgomery Asberg Depression Rating Scale; EO: Early Onset; LO: Late Onset; LTR: Lateral Temporal Region; MTR: Medial Temporal Region; PiB PET: ¹¹C-labeled Pittsburgh Compound-B Positron Emission Tomography.

Study characteristics

Sample sizes studied ranged from n = 16 (Butters et al., 2008) to n = 1060 (Sun et al., 2008), with around half of the studies having a sample size less than n = 50 (Butters et al., 2008; Kumar et al., 2011; Lavretsky et al., 2009; Madsen et al., 2012; Pomara et al., 2012; Reis et al., 2012; Wu et al., 2014). Five studies used healthy control groups matched on key demographic characteristics to the depressed group, while three studies used unmatched healthy older controls. In all control groups, individuals with clinically classified depression were excluded, as were those with history of psychiatric disease. Nine studies created a comparison group from their original sample using individuals who did not meet their defined criteria for depression. Two studies did not include comparison or control groups; one because they did not classify depression and the second because it used a prospective design.

Included studies also varied widely in their consideration of key risk factors for AD and the potential for these factors to bias study results. Half of the studies included samples that contained 40–60% more females than males. Average years of educational attainment ranged from 2.7 to 17.2 years. More than half (n = 10) of the studies did not assess or report the proportion of APOE ε4 carriers in their sample. Variation across study samples on these factors may confound outcomes causing inconsistency across study results.

Definition of depression

The majority of studies used established diagnostic criteria from either the DSM-III (n = 1), the DSM-IV (n = 10) or the ICD-10 (n = 1) to define depression. Four of these 12 studies used current inpatient admission for Major Depressive Disorder in conjunction with diagnostic criteria to define their depressed group. Eighteen studies utilised a standardised measure of depressive symptoms (e.g. the GDS or the Hamilton Depression Scale (HAM-D)) to assess severity of depressive symptoms, with six studies using an established cut score to define their depressive group rather than diagnostic criteria. Only one study (Lavretsky et al., 2009) used a standardised measure of depressive symptoms as a continuous variable without any categorisation.

Measurement of Aβ amyloid

Five studies measured Aβ amyloid using PET neuroimaging; two using PiB, two using the [¹⁸F] FDDNP compound, and one using the [¹⁸F] Florbetapir compound. CSF Aβ₄₂ was measured by four studies with one study also measuring CSF Aβ₄₀ or the ratio of CSF Aβ_40:42. All other studies used blood-based measures of Aβ, seven using plasma and three using serum. Only two of these 10 studies did not include measures of Aβ₄₀ or the ratio of Aβ_40:42.

Measurement of cognition

The majority of studies used the Mini Mental State Exam (MMSE) to either screen for cognitive impairments or to compare global cognitive function between groups. Only one study found a statistically significant difference between group mean scores of moderate magnitude (Cohen’s d = -0.31) and that was only for individuals who did not carry an APOE ε4 allele (Sun et al., 2009). Six studies included neuropsychological assessments other than the MMSE to assess cognitive functioning; however, the specific tests used and cognitive domains assessed varied widely between studies. Generally, measures included verbal or visual memory, and language, visuospatial and executive functioning. Comparisons of cognitive function between groups yielded inconsistent results. Two studies found depressed individuals performed significantly worse than non-depressed individuals on measures of memory, executive and visuospatial function. In contrast, two other studies found no significant differences between depressed and non-depressed individuals across a range of cognitive domains. In addition, one study found depressed individuals performed worse than non-depressed individuals on verbal memory, but only within a group who did not carry the APOE ε4allele. This difference was not apparent in a group of APOE ε4 carriers. One study reported a significant association between cognitive performance and the interaction of depressive symptoms and plasma Aβ_40:42 ratio; however, this association has not been examined in any of the other studies.

Findings

Nine of the 19 studies used a measure of Aβ₄₀ as a primary outcome measure, all of which failed to find a significant difference in Aβ₄₀ levels between depressed and non-depressed individuals despite finding significant effects with other Aβ amyloid measures. In addition, Aβ₄₀ was not correlated with scores on the HAM-D or with age of onset of depression in two of these studies (Kita et al., 2009; Namekawa et al., 2013).

Fourteen studies used a measure of Aβ₄₂ as a primary outcome. Eleven of these reported statistically significant differences in Aβ₄₂ levels between depressed and non-depressed individuals, and three did not. Ten of the 11 studies that observed a significant difference consistently observed lower Aβ₄₂ levels in depressed individuals relative to non-depressed individuals with the magnitudes of group differences ranging from moderate to large (d = -0.19 to -0.73). Interestingly, one study (Moon et al., 2011) observed moderately higher (d = 0.37) Aβ₄₂ levels in depressed individuals relative to controls. However, as the samples of all studies were not representative of the broader population, this inconsistency in the direction of the group difference between Aβ₄₂ levels may be due to specific biases in a particular sample. Importantly, only one study identified a significant difference between groups when considering individuals without cardiovascular disease and not taking antidepressant medication (Sun et al., 2008). Another study identified a significant difference between groups only after exclusion of individuals who did not carry an APOE ε4 allele (Sun et al., 2009).

Six studies also included a measure of Aβ_40:42 ratio as a primary outcome. All six studies indicated that depressed individuals had higher Aβ_40:42 ratios than non-depressed individuals (mean d = 0.53), with the magnitude of differences reported by studies ranging from small to large (d = 0.15 to 1.13). Of note, one study only observed this difference in individuals who did not carry an APOE ε4 allele and another study only when considering individuals without cardiovascular disease and not taking antidepressant medication (similar to the differences in Aβ₄₂ level discussed above). However, these effects were identified consistently in both late and early onset depression. Moreover, a significant moderate negative correlation (r = -0.20) was also identified between Aβ_40:42 ratio and age of onset of depression by one study (Namekawa et al., 2013).

Five studies used PET neuroimaging as a primary outcome measure, with three observing statistically significant relationships between neuroimaging results and depression status. One study found a large difference (d = 0.82) in FDDNP binding between depressed and non-depressed individuals, with depressed individuals having higher FDDNP binding indicating higher levels of global Aβ amyloid (Kumar et al., 2011). However, another study found moderate to large differences in amyloid binding between groups in the parietal and precuneus regions, but failed to find a significant difference globally (Wu et al., 2014). One study divided their depressed sample into MCI and cognitively normal groups and considered the correlation between FDDNP regional binding and scores on the GDS (Lavretsky et al., 2009). Within the MCI group they observed a moderate positive correlation between GDS scores and FDDNP binding in the lateral temporal region (r = 0.44), while in the cognitively normal group they observed a moderate positive correlation in the medial temporal region (r = 0.55). Interestingly, the two studies that failed to find any relationship between Aβ level and depression used the PiB compound, rather than and ¹⁸F compound, to measure amyloid. Therefore variations in results may be due to differences in the two compounds. Further study is required to establish whether differences in Aβ deposition in the brain between depressed and non-depressed individuals can reliably be detected using neuroimaging techniques.

Discussion

This review aimed to improve understanding of relationships between depressive symptoms and AD by reviewing the literature that has explored associations between depressive symptoms, or clinical depression, and levels of Aβ in non-demented older adults. After identifying 535 studies only 19 met inclusion criteria for the current review. Studies were mainly excluded due to non-peer review, non-inclusion of key variables, or animal model or treatment studies (see Figure 1). Results of studies included suggested that individuals with depression tended to have lower Aβ₄₂ levels and higher Aβ_40:42 ratios in their blood and CSF when compared with individuals without depression. Furthermore, Aβ₄₀ levels were generally equivalent between those with and without depression. While individuals with depression may also show higher amyloid binding on PET, reports of this were less consistent. This pattern of decreased Aβ₄₂ levels and increased Aβ_40:42 ratio in blood and CSF, as well as increased amyloid binding on PET, observed in individuals with depression concords with the pattern of Aβ changes typically observed in patients with AD (Alves et al., 2012; Prvulovic and Hampel, 2011). This suggests the differences observed in individuals with depression may be indicative of the beginning of the pathological cascade of AD and that depression may indeed be an early symptom of AD.

One important caveat is that these results have been observed most consistently using blood-based (plasma or serum) measures of Aβ even though the clinical utility of these measures has yet to be established (Humpel, 2011; Thambisetty and Lovestone, 2010). The ability to detect a difference in Aβ levels between individuals with and without depression using CSF or neuroimaging measures has been less consistent, although when detected, differences observed are consistent with those reported in studies using blood measures of Aβ. This is particularly relevant given CSF and neuroimaging markers have been shown to have good validity and utility in detecting changes in Aβ indicative of early AD (Prvulovic and Hampel, 2011; Visser et al., 2009). One factor that may be driving this inconsistency in detecting differences in Aβ between measurement methods is sample size. Neuroimaging studies have all used relatively small samples (n = 16–47), as have studies using CSF markers (n = 40–183), while studies using blood-based markers have had considerably larger samples (n = 120–1060). Studies using blood markers of Aβ therefore have greater statistical power, resulting in greater ability to detect subtle differences that may occur very early in the AD pathological process. Also of note, 10 of the 19 included studied used blood measures of Aβ, while only five used neuroimaging and four used CSF. Given the higher number of studies, there is greater opportunity to detect and establish a difference in blood-based measures across studies. Differences in sample sizes and the proportion of studies using blood measures of Aβ may be driven by the invasiveness and cost associated with neuroimaging and CSF analysis. Despite the limitations of blood-based measures of Aβ, detection of significant effects in a consistent direction for the majority of studies suggests study findings may be valid. Moreover, as the direction of difference was consistent with patterns observed in AD, this suggests this pattern may be potentially indicative of prodromal AD. Further investigation with neuroimaging and CSF in larger samples is still required to validate this pattern of change in Aβ levels for individuals with depression.

The relationship between depression and Aβ, and the contribution of this relationship to the development of AD, is likely to be complex and underpinned by multiple factors. In addition to increased Aβ deposition, several other biological mechanisms have been proposed to link depression to AD including increased glucocorticoids, vascular disease, alterations in nerve growth factors, hippocampal atrophy, and pro-inflammatory changes (Byers and Yaffe, 2011). Moreover, as such mechanisms may interact to increase an individual’s susceptibility to the development of AD (Byers and Yaffe, 2011), studies with larger samples are required to allow for sophisticated analytical techniques capable of considering variance that may be contributing to group differences other than that resulting from depression. Importantly, none of the 19 included studies were representative of the broader population and many failed to consider the potential influence of factors such as gender, education, premorbid intelligence, APOE ε4 carriage, or other potential mechanisms that might be involved such as glucocorticoid levels or vascular disease.

One important limitation to conclusions drawn from the majority of studies included in this review is that they were of cross-sectional design, whereas AD is a progressive disease. The current model that informs our understanding of AD describes a pathological cascade where biological changes occur over time from within normal ranges to maximally abnormal (Jack et al., 2010). Therefore measurement of relationships between depression and Aβ at a single time point may not effectively capture the role of depression in the development or presentation of AD. This is particularly the case in the early or preclinical stages of AD where changes are only beginning to occur, potentially differing from normal ranges only marginally. Restricting assessments to a single time point also prevents ascertainment of whether participants progress to clinical and cognitive symptoms of AD, or the role depression might play in the emergence of these signs. Furthermore, it is unclear whether Aβ levels in depressed individuals continue to change in line with the pathological trajectory of AD, remain stable over time, or even return to normal ranges.

Consideration of the trajectory of cognitive function over time would also be useful in elucidating the contribution of depression to the AD pathological process, and in particular its relationship to Aβ changes. The hallmark of AD is progressive cognitive decline (Albert, 2011), and current criteria for identification of prodromal AD is decline in performance on objective cognitive measures in addition to a biomarker of AD, such as Aβ levels (Albert et al., 2011; Sperling et al., 2011). As such, it is important to consider the role of cognition in the relationship between depression and Aβ, particularly in non-demented older adults, to confirm the syndrome observed is AD in its prodromal phases. While all studies included at least a cognitive screening measure, these measures are unlikely to be sensitive to the subtle changes that occur in older adults in the prodromal phases of AD. Consistent with investigations into cognitive impairment in prodromal AD (Harrington et al., 2013; Hedden et al., 2013; Mathis et al., 2013; Sperling et al., 2013), differences in cognitive function between groups were not consistently detected even with a more comprehensive neuropsychological assessment (Blasko et al., 2010; Kumar et al., 2011; Madsen et al., 2012; Pomara et al., 2012; Sun et al., 2008). Longitudinal assessment of cognitive function is necessary to detect early changes in function as a result of AD pathology (Albert et al., 2007; Doraiswamy et al., 2012; Lim et al., 2013a, 2013c; Small et al., 2012; Stomrud et al., 2010). In addition, it may be useful in differentiating cognitive impairment secondary to depressive symptoms from true AD. If a specific cognitive profile differentiates those with depression in the prodromal stages of AD, and have changes in Aβ, from individuals with depression who will recover, differential diagnosis would be greatly improved.

Future directions

Given the wide variability in sample sizes and characteristics, and the lack of representativeness to the broader population of all study samples included in this review, results of these studies must be interpreted with caution. Although consistent results were reported by many studies, indicating individuals with depression show patterns of Aβ levels consistent with those shown by AD patients, further investigation is required to characterise the nature and magnitude of the relationship between depression and Aβ, particularly longitudinally. Future investigations should control for the effects of sample characteristics on study outcomes. More importantly, future studies should consider using neuroimaging and CSF techniques to assess Aβ levels, as these have been established to have good utility and validity. In addition, future studies should report the statistics and characteristics, such as effect sizes, to allow for inclusion in meta-analysis.

Longitudinal assessment would allow differentiation of effects of the interaction between depression and Aβ from other factors and help in identifying individuals with depression who go on to develop AD and those who do not. The Australian Imaging Biomarkers and Lifestyle (AIBL) Flagship Study of Ageing is one of several large cohort studies currently running around the world, with the aim of improving early identification of AD and consists of participants assessed at 18-month intervals over 4.5 years (Ellis et al., 2009). Investigation of the relationship between Aβ and depression in such a cohort would allow for consideration of how depression and Aβ accumulation develop in parallel and interact over time, and importantly how this interaction then affects cognitive function and subsequent progression to AD dementia.

Footnotes

Funding

This research received no specific grant from any funding agency in the public, commercial or not-for-profit sectors.

Declaration of interest

The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.

References

Albert

(2011) Changes in cognition. Neurobiology of Aging 32 (Suppl 1): S58–S63.

Albert

Blacker

Moss

. (2007) Longitudinal change in cognitive performance among individuals with mild cognitive impairment. Neuropsychology 21: 158–169.

Albert

DeKosky

Dickson

. (2011) The diagnosis of mild cognitive impairment due to Alzheimer’s disease: Recommendations from the National Institute on Aging and Alzheimer’s Association workgroup. Alzheimers and Dementia 7: 270–279.

Alves

Correia

ASA

Miguel

. (2012) Alzheimer’s disease: A clinical-practice-oriented review. Frontiers in Neurology 3: 63.

Baba

Nakano

Maeshima

. (2012) Metabolism of amyloid-β protein may be affected in depression. Journal of Clinical Psychiatry 73: 115–120.

Blasko

Kemmler

Jungwirth

. (2010) Plasma amyloid beta-42 independently predicts both late-onset depression and Alzheimer disease. American Journal of Geriatric Psychiatry 18: 973–982.

Butters

Klunk

Mathis

. (2008) Imaging Alzheimer pathology in late-life depression with PET and Pittsburgh Compound-B. Alzheimer Disease and Associated Disorders 22: 261–268.

Byers

Yaffe

(2011) Depression and risk of developing dementia. Nature Reviews Neurology 7: 323–331.

Cirrito

Disabato

Restivo

. (2011) Serotonin signaling is associated with lower amyloid-β levels and plaques in transgenic mice and humans. Proceedings of National Academy of Sciences U S A 108: 14968–14973.

10.

Cohen

(1988) Statistical power analysis for the behavioural sciences. New York: Academic Press.

11.

Corder

Saunders

Strittmatter

. (1993) Gene dose of apolipoprotein E type 4 allele and the risk of Alzheimer’s disease in late onset families. Science 261: 921–923.

12.

Da Silva

Gonçalves-Pereira

Xavier

. (2013) Affective disorders and risk of developing dementia: Systematic review. British Journal of Psychiatry 202: 177–186.

13.

Doraiswamy

Sperling

Coleman

. (2012) Amyloid-β assessed by florbetapir F 18 PET and 18-month cognitive decline: A multicenter study. Neurology 79: 1636–1644.

14.

Dubois

Albert

(2004) Amnestic MCI or prodromal Alzheimer’s disease? Lancet Neurology 3: 246–248.

15.

Ellis

Bush

Darby

. (2009) The Australian imaging, biomarkers and lifestyle study of aging: Methodology and baseline characteristics of 1112 individuals recruited for a longitudinal study of Alzheimer’s disease. International Psychogeriatrics 21: 672–687.

16.

Gudmundsson

Skoog

Waern

. (2007) The relationship betwen cerebrospinal fluid biomarkers and depression in elderly women. The American Journal of Geriatric Psychiatry 15(10).

17.

Harrington

Lim

Ellis

. (2013) The association of Aβ amyloid and composite cognitive measures in healthy older adults and MCI. International Psychogeriatrics 25: 1667–1677.

18.

Hedden

Younger

. (2013) Meta-analysis of amyloid-cognition relations in cognitively normal older adults. Neurology 80: 1341–1348.

19.

Humpel

(2011) Identifying and validating biomarkers for Alzheimer’s disease. Trends in Biotechnology 29: 26–32.

20.

Jack

Knopman

Jagust

. (2010) Hypothetical model of dynamic biomarkers of the Alzheimer’s pathological cascade. Lancet Neurology 9: 119–128.

21.

Kita

Baba

Maeshima

. (2009) Serum amyloid protein in young and elderly depression: A pilot study. Psychogeriatrics 9: 180–185.

22.

Kramberger

MGV

Jelic

Kåreholt

. (2012) Cerebrospinal Fluid Alzheimer Markers in Depressed Elderly Subjects with and without Alzheimer’s Disease. Dementia and Geriatric Cognitive Disorders Extra 2: 48–56.

23.

Kumar

Kepe

Barrio

. (2011) Protein binding in patients with late-life depression. Archives of General Psychiatry 68: 1143–1150.

24.

Lavretsky

Siddarth

Kepe

. (2009) Depression and anxiety symptoms are associated with cerebral FDDNP-PET binding in middle-aged and older nondemented adults. American Journal of Geriatric Psychiatry 17: 493–502.

25.

Lim

Ellis

Harrington

. (2013a) Cognitive consequences of high Aβ amyloid in mild cognitive impairment and healthy older adults: Implications for early detection of Alzheimer’s disease. Neuropsychology 27: 322–332.

26.

Lim

Ellis

Harrington

. (2013b) Cognitive decline in adults with amnestic mild cognitive impairment and high amyloid-β: Prodromal Alzheimer’s disease? Journal of Alzheimers Disease 33: 1167–1176.

27.

Lim

Maruff

Pietrzak

. (2014a) Effect of amyloid on memory and non-memory decline from preclinical to clinical Alzheimer’s disease. Brain 137: 221–231.

28.

Lim

Pietrzak

Ellis

. (2014b) Beta-amyloid and cognitive change: Examining the preclinical and prodromal stages of Alzheimer’s disease. Alzheimers and Dementia. Epub ahead of print, 28 February 2014.

29.

Lim

Pietrzak

Ellis

. (2013c) Rapid decline in episodic memory in healthy older adults with high amyloid-β. Journal of Alzheimers Disease 33: 675–679.

30.

Luppa

Luck

Ritschel

. (2013) Depression and incident dementia. An 8-year population-based prospective study. PLoS ONE 8: e59246.

31.

Madsen

Hasselbalch

Frederiksen

. (2012) Lack of association between prior depressive episodes and cerebral [¹¹C]PiB binding. Neurobiology of Aging 33: 2334–2342.

32.

Mathis

Kuller

Klunk

. (2013) In vivo assessment of amyloid-β deposition in nondemented very elderly subjects. Annals of Neurology 73: 751–761.

33.

Moon

Kang

. (2011) The correlation of plasma Aβ42 levels, depressive symptoms, and cognitive function in the Korean elderly. Progress in Neuro-Psychopharmacology and Biological Psychiatry 35: 1603–1606.

34.

Namekawa

Baba

Maeshima

. (2013) Heterogeneity of elderly depression: Increased risk of Alzheimer’s disease and Aβ protein metabolism. Progress in Neuro-Psychopharmacology and Biological Psychiatry 43: 203–208.

35.

Neubauer

Wahl

H-W

Bickel

(2013) Depressive symptoms as predictor of dementia versus continuous cognitive decline: A 3-year prospective study. European Journal of Ageing 10: 37–48.

36.

Petersen

Doody

Kurz

. (2001) Current concepts in mild cognitive impairment. Archives of Neurology 58: 1985–1992.

37.

Piccinni

Origlia

Veltri

. (2013) Neurodegeneration, β-amyloid and mood disorders: State of the art and future perspectives. International Journal of Geriatric Psychiatry 28: 661–671.

38.

Pomara

Bruno

Sarreal

. (2012) Lower CSF amyloid beta peptides and higher F2-isoprostanes in cognitively intact elderly individuals with major depressive disorder. American Journal of Psychiatry 169: 523–530.

39.

Pomara

NPM

Doraiswamy

Willoughby

. (2006) Elevation in plasma Abeta42 in geriatric depression: A pilot study. Neurochemical Research 31: 341–349.

40.

Prvulovic

Hampel

(2011) Amyloid β (Aβ) and phospho-tau (p-tau) as diagnostic biomarkers in Alzheimer’s disease. Clinical Chemistry and Laboratory Medicine 49: 367–374.

41.

Qiu

WQX

Sun

Selkoe

. (2007) Depression is associated with low plasma Aβ42 independently of cardiovascular disease in the homebound elderly. International Journal of Geriatric Psychiatry 22: 536–542.

42.

Rapp

Hellweg

Heinz

(2011) [The importance of depressive syndromes for incipient Alzheimer-type dementia in advanced age]. Der Nervenarzt 82: 1140–1144 (article in German).

43.

Rapp

Schnaider-Beeri

Grossman

. (2006) Increased hippocampal plaques and tangles in patients with Alzheimer disease with a lifetime history of major depression. Archives of General Psychiatry 63: 161–167.

44.

Reis

Brandão

Freire Coutinho

. (2012) Cerebrospinal fluid biomarkers in Alzheimer’s disease and geriatric depression: Preliminary findings from Brazil. CNS Neuroscience and Therapeutics 18: 524–529.

45.

Richard

Reitz

Honig

. (2013) Late-life depression, mild cognitive impairment, and dementia. JAMA Neurology 70: 383–389.

46.

Small

Siddarth

Kepe

. (2012) Prediction of cognitive decline by positron emission tomography of brain amyloid and tau. Archives of Neurology 69: 215–222.

47.

Sperling

Aisen

Beckett

. (2011) Toward defining the preclinical stages of Alzheimer’s disease: Recommendations from the National Institute on Aging and the Alzheimer’s Association workgroup. Alzheimers and Dementia 7: 280–292.

48.

Sperling

Johnson

Doraiswamy

. (2013) Amyloid deposition detected with florbetapir F 18 (18F-AV-45) is related to lower episodic memory performance in clinically normal older individuals. Neurobiology of Aging 34: 822–831.

49.

Stomrud

Hansson

Zettrberg

. (2010) Correlation of longitudinal cerebrospinal fluid biomarkers with cognitive decline in healthy older adults. Archives of Neurology 67: 217–223.

50.

Sun

Bhadelia

Liebson

. (2011) The relationship between plasma amyloid-β peptides and the medial temporal lobe in the homebound elderly. International Journal of Geriatric Psychiatry 26: 593–601.

51.

Sun

Chiu

Liebson

. (2009) Depression and plasma amyloid β peptides in the elderly with and without the apolipoprotein E4 allele. Alzheimer Disease and Associated Disorders 23: 238–244.

52.

Sun

Steffens

. (2008) Amyloid-associated depression: A prodromal depression of Alzheimer disease? Archives of General Psychiatry 65: 542–550.

53.

Sun

XDM

Mwamburi

Bungay

. (2010) Depression, Antidepressants, and Plasma Amyloid β (Beta) Peptides in Those Elderly Who Do Not Have Cardiovascular Disease. Biological Psychiatry 62: 1413–1417.

54.

Thambisetty

Lovestone

(2010) Blood-based biomarkers of Alzheimer’s disease: Challenging but feasible. Biomarkers in Medicine 4: 65–79.

55.

Tsuno

Homma

(2009) What is the association between depression and Alzheimer’s disease? Expert Review of Neurotherapeutics 9: 1667–1676.

56.

Vilalta-Franch

López-Pousa

Llinàs-Reglà

. (2013) Depression subtypes and 5-year risk of dementia and Alzheimer disease in patients aged 70 years. International Journal of Geriatric Psychiatry 28: 341–350.

57.

Villemagne

Pike

Chételat

. (2011) Longitudinal assessment of Aβ and cognition in aging and Alzheimer disease. Annals of Neurology 69: 181–192.

58.

Visser

Verhey

Knol

. (2009) Prevalence and prognostic value of CSF markers of Alzheimer’s disease pathology in patients with subjective cognitive impairment or mild cognitive impairment in the DESCRIPA study: A prospective cohort study. Lancet Neurology 8: 619–627.

59.

Wells

(1979) Pseudodementia. American Journal of Psychiatry 136: 895–900.

60.

Wilson

Leurgans

Boyle

(2010) Neurodegenerative basis of age-related cognitive decline. Neurology 75: 1070–1078.

61.

World Health Organisation. (2012) Dementia. Available at: www.who.int/mediacentre/factsheets/fs362/en/.

62.

Hsiao

Chen

. (2014) Increased brain amyloid deposition in patients with a lifetime history of major depression: Evidenced on 18F-florbetapir (AV-45/Amyvid) positron emission tomography. European Journal of Nuclear and Molecular Imagingg 41: 714–722.

Amyloid-beta and depression in healthy older adults: A systematic review

Abstract

Objective:

Method:

Results:

Conclusions:

Keywords

Method

Study selection criteria

Type of study

Participants

Outcome measures

Search methods for identification of studies

Electronic database searches

Search strategy

Additional searches

Calculation of effect sizes

Results

Study characteristics

Definition of depression

Measurement of Aβ amyloid

Measurement of cognition

Findings

Discussion

Future directions

Footnotes

Funding

Declaration of interest

References