Dimensions of post-stroke depression and neuropsychological deficits in older adults

Abstract

Objective:

Post-stroke depression (PSD) has a heterogeneous presentation and is often accompanied by cognitive impairment. This study aimed to identify distinct dimensions of depressive symptoms in older adults with PSD and to evaluate their relationship to cognitive functioning.

Design:

Cross-sectional factor and correlational analyses of patients with post-stroke depression.

Setting:

Patients were recruited from the community and from acute inpatient stroke rehabilitation hospitals.

Participants:

Participants had suffered a stroke and met DSM-IV criteria for major depression (≥18 Montgomery Asberg Depression Scale; MADRS).

Intervention:

None

Measurements:

MADRS was used to quantify depression severity at study entry. Neuropsychological assessment at the time of study entry consisted of measures Global Cognition, Attention, Executive Function, Processing Speed, Immediate Memory, Delayed Memory, and Language.

Results:

The participants were 135 older adults (age ≥50) with PSD and varying degrees of cognitive impairment (MMSE Total ≥ 20). Factor analysis of the MADRS identified three factors, i.e. Sadness, Distress, and Apathy. Items comprising each factor were totaled and correlated with neuropsychological domain z-score averages. Symptoms of the apathy factor (lassitude, inability to feel) were significantly associated with greater impairment in executive dysfunction, memory, and global cognition. Symptoms of the sadness and distress factors had no relationship to cognitive impairment.

Conclusions:

PSD consists of three correlated dimensions of depressive symptoms. Apathy symptoms are associated with cognitive impairment across several neuropsychological domains. PSD patients with prominent apathy may benefit from careful attention to cognitive functions and by interventions that address both psychopathology and behavioral deficits resulting from cognitive impairment.

INTRODUCTION

Stroke is the leading cause of both serious and long-term disability and the fifth leading cause of death in the United States.(1, 2) Post-stroke depression (PSD) afflicts approximately one-fourth of stroke survivors and increases the likelihood of persistent disability.(3-6) Prompt identification and treatment of PSD can improve affective symptoms and stroke related disability.(7, 8) PSD often occurs in patients presenting emotional lability, anosognosia and cognitive impairment leading to a clinical presentation far more complex than that of idiopathic depression. Despite the apparent heterogeneity of PSD, few studies to date have attempted to identify distinct dimensions of depressive symptoms in PSD. One prior PSD study of adults with a wide age range(9) diagnosed with major depression due to medical reasons, used the post-stroke depression scale and reported three dimensions of depressive symptom patterns, i.e., depressive/anxious, lack of emotional control, and reduced motivation. Reduced motivation (i.e., a combination of symptoms of apathy and anhedonia) was significantly associated with older age in PSD patients.(8)

Little is known about the clinical heterogeneity of PSD in older adults. In the only study examining symptom patterns of PSD in older adults, a multidimensional profile emerged.(8)Factor analysis of items of the Montgomery Asberg Depression Rating Scale (MADRS) in 163 patients with PSD identified three factors, i.e. sadness, agitation and anhedonia.(8) The subjects of this study had mild depression and there have been no studies of more severe PSD to date.

The relationship of depressive symptoms to cognitive impairment is especially meaningful in PSD, as post-stroke cognitive deficits are common and widespread.(10, 11) Cognitive deficits in PSD include memory difficulties, executive dysfunction, poor attention, slowed information processing speed, and visuoperceptual/visuoconstructional impairments.(12-16) Executive dysfunction in particular is a persistent, chronic and prevalent cognitive deficit among stroke survivors(14, 17) and may worsen deficits in other cognitive domains (e.g., learning and memory).(18) Among executive functions, verbal fluency and conceptualization are impaired in 33%-66% of stroke patients.(18) Aging compounds stroke related cognitive deficits leading to worsening outcomes in older adults with PSD.(19)

While cognitive impairments improve or stabilize in many stroke survivors, cognitive function in patients with PSD often worsens.(12) Earlier reports indicate that patients with prominent anhedonia experience greater cognitive impairment, while those with prominent sadness exhibit greater sensorimotor and cranial nerve deficits.(8) Characterizing the relationship of distinct dimensions of depressive symptoms with cognitive impairments in older adults with PSD may help elucidate aspects of the neurobiology of distinct presentations of PSD, guide the selection of interventions and increase personalization of treatment.

This study sought to identify dimensions of depressive symptoms in older adults with moderate to moderately severe PSD. Based on the limited, available literature, we hypothesized that: 1) the PSD syndrome of older adults has several distinct dimensions; and 2) the dimensions of the PSD syndrome have different relationships to cognitive deficits.

METHODS

Participants

The participants were recruited from local inpatient stroke rehabilitation units and from the community. The inclusion criteria were: 1) Age of 50 years and older; 2) history of ischemic, embolic or hemorrhagic stroke established by medical record review; 3) diagnosis of major depression by DSM-IV criteria; and 4) Montgomery Asberg Depression Scale (MADRS)(20) score of 18 or higher.

Exclusion criteria were: 1) Moderate or severe dementia (Mini Mental State Examination (MMSE(21)) score < 20); 2) greater than moderate aphasia (NIH Stroke Scale(22) Best Language > 1); 3) psychotic depression or bipolar disorder (by DSM-IV); 4) suicide intent or plan; 5) plans to enter a nursing home for treatment in the near future; and 6) inability to speak English. All participants signed informed consent approved by the Weill Cornell Institutional Review Board.

Assessment

Diagnostic evaluation was conducted with the Structured Clinical Interview- Revised (SCID-R).(23) Severity of depression was quantified with the MADRS.(20) The MADRS is a reliable and valid 10-item clinician rated scale, which places less emphasis on somatic symptoms and is frequently used as a depression severity rating instrument for older adults who have experienced a stroke.(24-27)

Global cognitive dysfunction was assessed with the MMSE,(21) a reliable and valid brief screening instrument, and with the Dementia Rating Scale(28) a more extensive instrument of cognitive dysfunction in older adults. Participants also received a neuropsychological battery consisting of scales assessing the following cognitive domains: 1) Executive Function (Controlled Oral Word Association Test (COWAT), Dementia Rating Scale (DRS) Initiation/Perseveration (IP) subscale, DRS Conceptualization subscale, Stroop Color Word Interference, Weschler Adult Intelligence Scale (WAIS –III) Digit Span Backward Span); 2) Processing speed (Stroop Color and Stroop Word subtests); 3) Attention (WAIS–III Digit Span- Digits Forward Total Span, DRS Attention subscale), 4) Immediate Memory (Hopkins Verbal Learning Test-Revised (HVLT-R) Immediate Recall), and 5) Delayed memory (HVLT-R Delayed Recall, DRS Memory subscale); and 6) Language (Semantic fluency/Animal Naming Test (ANT)). To account for effort as a confounder of performance, reliable digit span (RDS) was used and optimal effort was indicated by RDS>6.(29)

Statistical Analysis

Exploratory principal axis factoring(30) was performed on the 10 items of the MADRS scale using an oblique rotation procedure in SPSS 25.0 (Direct Oblimin). Three factors that explained 46.2% of the total variation were extracted by visual inspection of the scree plot and by requiring factor eigenvalues to be > 1. Factor loadings greater or equal to 0.2 were used to identify MADRS factors of depression. Factor (dimensions) totals were created by adding the score on each MADRS item within each factor grouping into a sum.

Multiple imputation(31) by Markov Chain Monte Carlo method (MCMC) was used to account for missing data within each neuropsychological measure (see table 2 for list of measures and missing data estimates by variable). Neuropsychological data were, then, normatively corrected according to respective normative tables (age and education corrected: FAS (32), ANT(32); age corrected: HVLT-R(33), WAIS-III Digit Span Backward and Forward(34), DRS (28), Stroop Color Word Interference Test(35). Individual normative scores were translated into z-scores. Domains of cognitive functioning were created by averaging z-scores of measures within each neuropsychological domain as defined above (Table 2). MADRS factor (dimensions) scores were, then, correlated with each neuropsychological domain. Descriptive and correlational analyses derived by pooled analysis of five imputation models. All tests were two-tailed, with results considered significant at p≤0.05.

Table 2.

Baseline Neuropsychological Characteristics by Cognitive Domain after Multiple Imputation N=135.

Neuropsychological Domains (missing data prior to multiple imputation)	Pooled Descriptive Statistics
	Mean	SD	z- score	z-score SD
Global Cognition			−0.5	1.3
MMSE* Total (2 missing)	27.2	2.7	−0.5	1.9
Dementia Rating Scale (DRS) Total (24 missing)	129.2	13.1	−0.6	1.2
Attention			0.3	0.7
WA1S-III+ Forward Digit Span (25 missing)	6.3	1.5	0.1	1.1
DRS Attention Total (14 missing)	35.5	2.5	0.4	0.8
Executive Function			−0.4	0.6
DRS Initiation/Perseveration (19 missing)	31.5	6.1	−0.8	1.1
DRS Conceptualization (13 missing)	34.6	4.1	−0.2	0.9
Stroop Color/Word Interference++(30 missing)	2.7	6.6	0.3	0.7
WAIS-III Backward Digit Span (25 missing)	4.0	1.2	−0.3	1.0
Controlled Oral Word Association Test (FAS) Total (14 missing)	28.6	13.3	−1.0	1.1
Processing Speed			−1.5	0.7
Stroop Word Total++(26 missing)	84.0	15.5	−1.2	0.7
Stroop Color Total++(30 missing)	53.5	12.6	−1.7	0.7
Visuoconstruction			−0.6	0.8
DRS Construction (20 missing)	5.2	1.4	−0.6	0.8
Immediate Memory (Learning)			−1.5	1.3
Hopkins Verbal Learning Test- Revised (HVLT-R) Immediate Recall (11 missing)	19.3	6.2	−1.5	1.3
Delayed Memory (Retention)			−0.9	1.1
HVLT-R Delayed Recall (14 missing)	5.8	3.3	−1.6	1.5
DRSMemory (13 missing)	22.4	3.1	−0.2	1.2
Language			−0.8	1.2
Animal Naming Test Total (12 missing)	15.4	5.5	−0.5	1.3

^*MMSE=MiniMental State Examination

⁺Wechsler Adult Intelligence Scales – Third Edition

⁺⁺Age Corrected raw score

RESULTS

The participants were 135 consecutively recruited older adults with major depression and history of stroke (Table 1). The severity of their depression (MADRS) was in the moderate to moderately severe range (MADRS Total Range 18-35). Their global cognitive impairment (MMSE) scores ranged from 21 to 30 with 21 participants having an MMSE score below 24. Effort on neuropsychological measures was assessed by the reliable digit span index and deemed to be adequate (RDS score >6)(29) in 86% of the sample. Participant neuropsychological performance fell within the impaired to the average ranges (Table 2).

Table 1.

Clinical and Demographic Characteristics of the study sample (N = 135)

	Mean (SD)
Age, years	70.1 (11.2)
Education, years	15.7 (3.6)
Female, % (n)	59%
Living alone % (n)	36%
On Antidepressant Medication %	56%
MADRS*	23.2 (4.0)
Duration of Depressive Episode (Months)	17.5 (23.6)
Number of Previous Depressive Episodes	3.4 (10.8)
Adequate Effort % (RDS**)	86%

^*MADRS=Montgomery Aspberg Depression Rating Scale

^**RDS=Reliable Digit Span; adequate effort defined as RDS>6

Exploratory factor analysis of MADRS items resulted in a three-factor solution explaining 46.2% of the variance (Table 3). We labeled each factor based on item content and prior literature(8, 9) as: 1) Sadness (apparent sadness, reported sadness, pessimistic thoughts, suicidal thoughts;17.5% variance explained) 2) Distress (inner tension concentration difficulties, and reduced appetite; 15.9% variance explained) and 3) Apathy (lassitude, inability to feel;12.8% variance explained). Correlations between factors were as follows: Sadness and Distress r= −0.04, Sadness and Apathy r=−0.05, Distress and Apathy r= −0.02), One MADRS item (reduced sleep) did not load well on any one factor and was removed from the analysis.

Table 3.

Principal Factor Analysis Pattern Matrix MADRS Item Factor (factor items in bold)

Items	Factor 1	Factor 2	Factor 3
	Sadness	Distress	Apathy
Apparent Sadness	0.57	−0.03	0.20
Reported Sadness	0.72	0.07	0.03
Pessimistic Thoughts	0.30	0.30	−0.03
Suicidal Thoughts	0.22	−0.01	−0.07
Reduced appetite	0.12	−0.72	0.00
Concentration difficulties	−0.26	0.47	0.17
Inner Tension	0.08	0.39	−0.07
Lassitude	−0.02	0.00	0.31
Inability to feel	0.14	−0.04	0.70

There was no significant correlation between effort during neuropsychological assessment (RDS performance) and Sadness (Spearman r=−0.03, N=135 p=0.8), Distress (Spearman r=0.06, N=135, p=0. 53) or Apathy (Spearman r=−0.11, N=135, p=0.24). Apathy was associated with greater executive dysfunction, worse immediate and delayed memory performance, worse language functioning and worse general cognitive functioning (Table 4). Neither Sadness nor Distress dimensions were significantly correlated with performance on any cognitive measures.

Table 4.

Pooled Spearman correlations between MADRS Subgroups and Neuropsychological Domains (N = 135)

Neuropsychological Domains	Sadness	Distress	Apathy
Global Cognition	−0.00	0.04	−0.34**
Attention	−0.04	0.00	−0.11
Executive Function	−0.00	0.16	−0.20*
Processing Speed	0.13	0.12	−0.00
Visuoconstruction	−0.10	0.02	−0.08
Immediate Memory	−0.07	0.08	−0.30**
Delayed Memory	−0.05	0.15	−0.23**
Language	0.11	0.07	−0.22*

^*Correlation is significant at the p<0.05 level (two-tailed)

^**Correlation is significant at the p<0.01 level

DISCUSSION

The principal finding of this study is that PSD has a heterogeneous presentation with symptoms clustered along three dimensions, i.e. sadness, distress, and apathy. Apathy was the only symptom cluster of PSD with a significant relationship to impairment in executive function, memory, and global cognitive function. To our knowledge this is the first study to characterize the dimensions of PSD in older adults with moderate to severe depression. Its findings are consistent with reports of previous studies, which factor analyzed the presentation of PSD in wide age range samples with mild severity of depression.(8, 9) Further, our observations are in line with earlier findings suggesting that post-stroke apathy is associated with cognitive impairment.(8, 36)

Apathy afflicts 20% and 25% of stroke patients.(36) As in post-stroke depressed patients of this study, post-stroke apathy has been associated with cognitive impairment.(36) Apathy is a disturbance of motivation(37) clinically expressed as paucity of goal-directed, behavior.(38) Goal-directed behavior requires focusing of attention on meaningful information, holding this information in working memory, suppressing irrelevant or conflicting information, and selecting appropriate responses.(39) Brain lesion studies in humans suggest that apathy results, in part, from failed encoding of salience (meaning) of an ongoing or forthcoming behavior(40) leading to disruption of the processing that attributes motivational value of a behavior and that orients decision-making. These processes are served by interactions of the salience network with the cognitive control network. The salience network includes an anterior insular and anterior cingulate cortical (ACC) circuit that monitors the environment for motivationally salient stimuli and transforms salient signals into an orienting response that selectively increases arousal and engages networks that govern attention, and working-memory.(41) Subsequent behavioral responses are mediated by activation of the cognitive control network, which unlike the salience network, is equipped to operate on identified salience. Thus the salience network, with its connections to limbic and subcortical structures, serves as a bottom-up processor of salient cognitive, homeostatic, or emotional information (39) that triggers subsequent activation of the cognitive control network.(42, 43)

Based on this understanding, the association of apathy with executive dysfunction in PSD may be understood as a consequence of lesions that disrupt the salience-cognitive control network communication along with pathways serving other cognitive functions. Consistent with this view are imaging studies suggesting that apathy mostly occurs after frontal or basal ganglia strokes and is associated with disruption of neural networks connecting the anterior cingulate gyrus, the dorsomedial frontal cortex, and the frontal pole with the ventral aspects of the caudate nucleus, the anterior and ventral globus pallidus, and the dorsomedial and intralaminar thalamic nuclei.(36)’ (44) Executive dysfunction may impair learning and memory by disrupting organization during information storage and retrieval and account in part for the association of apathy with memory impairment.

Apathy is common in PSD(7) but also in idiopathic late-life depression.(45) Studies of resting state functional connectivity (rsFC) in apathy of idiopathic late-life depression suggest a disruption of communication between and within the salience, reward, and cognitive control networks.(46) Relative to non-apathetic depressed elderly, depressed apathetic subjects had decreased rsFC of the right anterior insula (central node of the salience network) to dorsal ACC and to subcortical/limbic components of the salience network. Depressed elderly subjects with high apathy also had increased rsFC of the right anterior insula to right dorsolateral prefrontal cortex and right posterior cingulate cortex when compared to non-apathetic depressed elderly.(47) Further, depressed older apathetic patients had lower rsFC of the nucleus accumbens with the amydgala and with subcortical structures than non-apathetic patients and increased rsFC with the dorsomedial prefrontal cortex, the superior frontal cortex, and the insula.(48)

Both apathy and executive dysfunction are disabling. Post-stroke apathy has a chronic course characterized by progressive disability and it responds poorly to antidepressants.(36) Even in idiopathic late-life depression, apathy is correlated with disability.(45) During antidepressant treatment, change in apathy remains correlated with disability, but improvement of apathy is at best modest.(45, 49) Similarly, executive dysfunction is associated with disability (50) and predicts poor response of late-life depression to antidepressants.(51-53)

The findings of this study may guide the selection of interventions for PSD. The association of apathy with executive and other cognitive dysfunctions and its modest response to antidepressants points to the need for alternative treatment strategies. Given the relationship between PSD, apathy and executive dysfunction in our sample, augmenting traditional interventions for PSD may improve outcomes. Small clinical studies and anecdotal reports on dopaminergic agonists(54) and cholinesterase inhibitors(55) suggest that these treatments may improve response of apathy symptoms of PSD to antidepressants. However, high quality randomized controlled trials are lacking. Adherence to medications in individuals with PSD is challenging.(56) Cuing interventions using reminders, checklists, and electronic devices may help apathetic PSD patients to initiate action. Cognitive remediation is a promising approach found to reduce apathy in healthy older adults(57) and reduce depression in older adults with executive dysfunction.(58, 59) Problem solving therapy can reduce both depression (60) and disability (61) in patients with idiopathic major depression and executive dysfunction and needs to be studied in patients with apathetic PSD and executive dysfunction. A reasonable approach in PSD patients with prominent sadness may be a streamlined behavioral activation treatment, (62, 63) while PSD patients with prominent distress may benefit from relaxation training.(64, 65)

The findings of this study should be viewed in the context of its limitations. The study included participants with and without history of depression prior to stroke. This decision was based on findings that history of depressive episodes predisposes to depression after stroke.(66) However, some of this study’s subjects may have had depression unrelated to stroke. Another limitation is the exclusion of individuals with aphasia. It is possible that depressive profiles may differ in patients with aphasic syndromes secondary to stroke, limiting the generalizability of our findings to PSD without pronounced aphasia. Further, our use of the MADRS to identify dimensions of PSD may limit generalizability of our findings to this particular instrument; it is possible that use of another scale to quantify depression in PSD could have identified different symptom dimensions. It is conceivable that variability in symptoms of depression may be related to variables other than neuropsychological function (e.g., disability, stroke location); however, examining these relationship was beyond the scope of our analysis and should be explored further in future studies. Finally, our study is cross-sectional and it does not offer information on the stability of the relationship between depressive symptoms and cognitive impairments.

This study has sought to clarify the relationship of affective and cognitive symptoms in PSD. Our findings highlight the variability in symptom presentation of PSD, and underscore apathy as a meaningful dimension of PSD related to executive and other cognitive impairments. Findings from non-PSD populations implicate a disruption of communication between and within the salience, reward, cognitive control networks in apathy but studies of the neurobiology of apathetic in PSD are needed. Antidepressants have modest efficacy in both depression with apathy and executive dysfunction. However, behavioral interventions are available for apathetic syndromes. Preliminary findings suggest that some pharmacological agents may improve apathy and need to be investigated in apathy of PSD.

Highlights.

What is the primary question addressed by this study?

• Post- stroke depression (PSD) is prevalent, challenging to diagnose and often presents with widespread cognitive deficits. This study characterizes dimensions of post-stroke depression in an older adult sample with moderate to moderately severe depression.

What is the main finding of this study?

• Among the three dimensions identified, apathy was related to worse functioning on multiple cognitive domains, while sadness and distress were not related to any cognitive variables.

What is the meaning of this finding?

• Our findings indicate that PSD is a heterogeneous entity with clinically meaningful dimensions that if further explored can inform personalized therapeutic interventions.