Abstract
Background:
Cardiovascular risk factors (CVRFs) have been linked to both depression and cognitive decline but their role in neuropsychiatric symptoms (NPS) has yet to be clarified.
Objective:
Understanding the role of CVRFs in the etiology of NPS for prospective treatments and preventive strategies to minimize these symptoms.
Methods:
We examined the distribution of NPS using the Neuropsychiatric Inventory (NPI) scores in three cohorts from the Prevention of Alzheimer’s Dementia with Cognitive Remediation Plus Transcranial Direct Current Stimulation in Mild Cognitive Impairment and Depression (PACt-MD) study: older patients with a lifetime history of major depressive disorder (MDD) in remission, patients with mild cognitive impairment (MCI), and patients with combined MCI and MDD. We also examined the link between individual NPS and CVRFs, Framingham risk score, and Hachinski ischemic score in a combined sample.
Results:
Analyses were based on a sample of 140 subjects, 70 with MCI, 38 with MCI plus MDD, and 32 with MDD. There was no effect of age, gender, education, cognition, or CVRFs on the presence (NPI >1) or absence (NPI = 0) of NPS. Depression was the most prevalent affective NPS domain followed by night-time behaviors and appetite changes across all three diagnostic groups. Agitation and aggression correlated negatively while anxiety, disinhibition, night-time behaviors, and irritability correlated positively with CVRFs (all p-values <0.05). Other NPS domains showed no significant association with CVRFs.
Conclusion:
CVRFs are significantly associated with individual NPI sub-scores but not with total NPI scores, suggesting that different pathologies may contribute to different NPS domains.
Keywords
INTRODUCTION
Neuropsychiatric symptoms (NPS) are common in Alzheimer’s disease (AD), estimated to occur in approximately 80% of patients at some point over the course of the disease [1]. Their frequency in mild cognitive impairment (MCI) has been estimated to be lower although the range may be anywhere from 35% to 75% [2, 3], though studies of NPS in patients with MCI to date have been limited. These symptoms typically consist of a broad spectrum of behavioral changes and include anxiety, depression, agitation, aggression, motor behaviors, delusions, and hallucinations. In recent years, there has been a greater focus on NPS as early indicators of emergent dementia in at risk patients [2, 4]. Mild behavioral impairment, a relatively new construct, suggests that NPS in some cases may be the only symptom of an emerging dementia syndrome and may predate the onset of cognitive decline by many years [5]. In addition, studies suggest that depression may be a harbinger of subsequent cognitive decline and eventual development of AD [6, 7], though the role of remitted lifetime history of depression is less clear. Furthermore, there has been limited characterization of NPS burden in patients with remitted MDD. Therefore, to the extent that remitted MDD may represent a risk factor of an evolving dementia, it is possible that patients with MDD in remission may exhibit additional NPS, potentially with a similar profile to that observed in patients with MCI.
The Prevention of Alzheimer’s Dementia with Cognitive Remediation plus Transcranial Direct Current Stimulation (PACt-MD) study is a large Toronto based intervention study focused on prevention of cognitive decline in patients with MCI, and MDD in remission with and without MCI. Our study examined the association of total Neuropsychiatric Inventory (NPI) [8] scores with demographic factors, cognition, diagnosis, and CVRF. As prior studies have demonstrated an association between NPS and later progression to dementia [9], we hypothesized that NPI scores would be associated with correlates of future dementia including decreased cognitive test performance. In addition, we predicted that depression and MCI, both possible correlates of future dementia, would combined be associated with higher NPI scores compared to each condition individually.
Our study also attempted to delineate the relationship of cardiovascular disease burden with NPS burden. Cardiovascular disease and CVRF have been identified as a risk factor for both AD [10, 11] and MDD [12]. Additionally, fluctuations in blood pressure have recently been demonstrated to be associated with cognitive decline in patients with AD. Possible mechanisms of this may include a compromised blood-brain barrier and endothelial blood vessel wall damage [13–15]. However, there has been limited study of the relationship of CVRF and NPS. A potential mechanism by which cardiovascular disease could impact NPS may be through increased small vessel disease burden, which may disrupt brain network connectivity, potentially leading to NPS [16]. There is very little evidence to date, though Moon et al. found a correlation between hypertension and severity of apathy and depression in AD [17]. Similarly, Fischer et al. found a correlation between psychosis and vascular risk factors in AD patients with psychosis [18]. We thus predicted that the overall NPS burden would be associated with greater cardiovascular risk. Finally, given that specific NPS may arise from different neural mechanisms, we hypothesized CVRF would associate differentially with specific NPS domains.
MATERIALS AND METHODS
Study design and participants
Participant demographic and clinical characteristics for this study were selected from the Prevention of Alzheimer’s Dementia with Cognitive Remediation plus Transcranial Direct Current Stimulation in Mild Cognitive Impairment and Depression (PACt-MD) database, an intervention study targeting patients with MCI and remitted MDD using cognitive remediation combined with transcranial direct current stimulation (tDCS). Participants for this study were recruited from 5 academic centers in the Toronto area: Centre for Addiction and Mental Health, University Health Network, St. Michael’s Hospital, Sunnybrook Hospital and Baycrest Hospital. As this is an on-going intervention trial, data collected up until October 2017 were analyzed. Inclusion criteria for this study were: patients must be diagnosed with MDD in remission (aged 65 or older) or MCI (aged 60 or older) or both. For the purposes of our study, we included in our analysis only the 140 patients with available NPI scores (70 excluded).
Diagnoses of MCI were arrived at clinically and confirmed by using standardized interviews and neuropsychological assessment. A diagnosis of MCI or mild neurocognitive disorder was made in accordance with DSM-5 [19] criteria. Major depressive disorder was diagnosed according to DSM 5 criteria using a standardized interview and severity measured with the Montgomery Asberg Depression Scale (MADRS) [20]. Baseline demographics were collected, including age, gender, race/ethnicity and education.
Measures
We measured NPS burden using the NPI-Q which is a scale administered to caregivers to measure the nature, severity and distress associated with NPS in patients with AD and related neurodegenerative disorders. Scores are given for 12 different behavioral domains: delusions, hallucinations, agitation/aggression, depression/dysphoria, anxiety, apathy/indifference, elation/euphoria, disinhibition, irritability/lability, motor disturbance, night-time behavior, and appetite/eating problems. Patients are rated 0 if the behavior is absent, 1 if present. The caregiver is asked to quantify the severity of the behavior on a scale of 1 to 3 and caregiver distress on a scale of 0 to 5. Patients are then given a total score (the number of domains rated as present), a total severity score (the sum of all severity scores for all positive behavioral domains) and a total caregiver distress score (the sum of all caregiver distress scores for all domains). The NPI-Q is validated and has been used widely in AD and MCI clinical populations [21]. We computed the Framingham Risk Scores (FRS) according to the Canadian Cardiovascular Society’s standards [22] on a subset of patients where the data was available, using combined data on age, sex, HDL cholesterol, total cholesterol, systolic blood pressure, diabetes and smoking status, and use of antihypertensives. We also collected data on individual CVRF including hypertension, cholesterol, smoking, hyperlipidemia, and diabetes in addition to the Hachinski ischemic score [23]. We collected and analyzed global measures of cognitive function that included Standardized Mini-Mental State Exam (S-MMSE) [24] and Montreal Cognitive Assessment (MoCA) [25].
Statistical analysis
We computed the baseline demographic and clinical characteristics of our sample. We then categorized our sample according to positive total NPI score (1 or greater) and negative NPI scores (0) to compare demographic factors, cardiovascular burden and cognition between the two groups. We then grouped the sample according to diagnosis and compared NPI scores and sub-scores among three groups: MCI, MDD in remission, and the two combined (MCI + MDD in remission). Subsequent sub-analyses were conducted on NPI sub-scores based on these findings. We examined the bivariate relationship of the total NPI score to measures of cardiovascular burden.
Data were examined for normality and homogeneity of variance using the Shapiro-Wilk Test and Levene Test, respectively. Non-parametric Mann-Whitney U tests were conducted on baseline demographic and clinical variables between NPI positive and negative groups as they did not follow a normal distribution. This same test was also conducted as a sub-analysis on NPI sub-domain scores with cardiovascular and cognitive parameters. Subsequently, we compared cardiovascular risk measures across the three diagnosis groups using the non-parametric Kruskal-Wallis 1-way ANOVA and a pairwise comparison with adjusted significance levels (Bonferroni correction for multiple tests). A sub-analysis ensued, in which MCI and MCI + MDD were combined into one diagnosis group and compared to the MDD group with respect to cardiovascular burden measures using the Mann-Whitney U test. All comparisons across NPI and diagnosis groups were stratified according to age, gender, and education (contribution of these factors mentioned only when significant). To measure the strength of the hypothesized relationship between NPI scores (total, severity, distress) and measures of cardiovascular burden (Framingham Risk Score, Hachinski Ischemic Score, total and HDL-cholesterol), we conducted the Spearman rank order correlation coefficient.
For all tests, the statistical significance threshold was determined at the 0.05 level. The statistical software used was IBM SPSS Statistics version 25 (IBM Corp, Armonk, NY, USA).
RESULTS
Baseline sample characteristics of the sample are provided in Table 1. The mean age was 73.66±6.25 years, 87 subjects were female, and 94 subjects had completed high school or greater. Seventy subjects had MCI, 38 had MDD in remission, and 32 had combined MCI plus MDD in remission (see Table 1). The mean MOCA score was 24.25±3.04 and the mean MMSE score was 27.85±1.80. The mean total NPI score was 1.60±1.70, the mean severity score was 2.24±2.71, and the mean distress score was 2.24±2.71. The mean Framingham score was 18.84±8.94 and hypertension and hypercholesterolemia were the most common vascular risk factors. Ninety-three subjects scored positive for one NPS or greater and there was no significant relationship with age, education, cognition, Framingham scores, psychoactive medication use or diagnostic group on the presence of NPS (see Table 2). Night-time behaviors were noted to be more common in MCI+MDD in remission relative to MDD in remission alone (p = 0.039), although this was no longer significant following correction for multiple comparisons (p = 0.058). However, when MCI and MCI + MDD diagnosis groups were combined, the prevalence of night-time behaviors in the expanded MCI group became significant in comparison to that of the MDD group alone (p = 0.014). Depression was the most common symptom in all groups followed by night-time behaviors and appetite disturbance (see Table 3).
Baseline demographics and clinical characteristics of sample (N = 140)
Comparison of NPI Positive versus NPI Negative patients across entire sample (N = 140, Mann-Whitney U test, df = 1)
*When stratified according to gender, p = 0.030 for Males (n = 53).
Correlations between total NPI Scores and measures of cardiovascular burden (Spearman rank order correlation)
Prevalence of NPI sub-scores across the diagnostic groups (Kruskal-Wallis 1-way ANOVA, df = 2)
**When pairwise comparisons adjusted according to the Bonferroni correction, the p-value was no longer significant (p = 0.058).
We conducted a sub-analysis to further determine the effect of CVRF on different NPI sub-scores and found that disinhibition and irritability correlated positively with low HDL cholesterol (p = 0.028) and high Framingham risk scores (p = 0.039). Night-time behaviors were positively associated with hypertension (p = 0.03) and high Framingham scores (p = 0.017). Anxiety symptoms correlated positively with Hachinski ischemia scores (p = 0.031) while agitation and aggression were negatively associated with hyperlipidemia (p = 0.044). Night-time behaviors correlated with decreased global cognitive function including MMSE (p = 0.011) and MOCA (p = 0.001). Appetite disturbance correlated with low MOCA scores (p = 0.041) and delusions correlated with poor performance on MOCA (p = 0.016) and MMSE (p = 0.047). No other relationships were significant.
DISCUSSION
Our study examined the distribution of NPI scores across three groups in the PACt-MD study—patients with MCI alone, patients with MDD in remission alone, and patients with combined MDD in remission and MCI. Contrary to our predictions, the distribution of NPI scores and sub-scores was similar across all groups, with the one exception of night-time behaviors being more prevalent in patients with MCI with and without MDD relative to MDD alone. The lack of variation of NPS across the diagnostic subgroups suggests that lifetime history of MDD in remission may not play a major role in NPS burden, at least when combined with MCI. Not surprisingly, depression was the most frequent NPS, followed by night-time behaviors and appetite disturbance. These results are consistent with previous studies demonstrating that mood, anxiety and depressive symptoms are the most common NPS in MCI [2].
We also examined the association between NPS and CVRF in a combined sample. Prior studies have identified cardiovascular disease burden as an important predictor of cognitive decline [10, 11], and CVRF to be associated with late life psychiatric disorders such as MDD [12]. Moreover, fluctuations in blood pressure have recently been identified as an important risk factor for cognitive decline in patients with AD. Proposed mechanisms include damage to the endothelial blood vessel walls as well as compromised blood-brain barrier. As our study focused only on baseline data we are unable to comment on the effect this may have had on our results [13–15]. Contrary to our predictions, the presence of CVRF, including the Framingham risk score and individual risk factors, was not associated with the presence of NPS as measured by the total NPI. In addition, age and level of cognitive decline did not have an effect on the presence or absence of NPS. Night-time behaviors were the only factor to associate positively with vascular risk factors of dementia including hypertension and Framingham score and with decreased memory, possible correlates of future dementia. Agitation was the only behavior to associate inversely with vascular risk factors while anxiety, disinhibition and irritability associated positively.
Collectively, there was no correlation between cognitive test scores and total NPI scores, though there was an association with specific sub-scores. Appetite disturbance and delusions correlated with worse cognitive performance but not increased CVRF, suggesting that alternate disease mechanisms possibly linked to neurodegenerative pathology may explain these symptoms. Delusions and other psychotic symptoms are associated with increased risk of dementia and accelerated rates of cognitive decline [26]. The one behavior that showed the greatest variation across the groups, the strongest association with cardiovascular disease burden and cognitive decline was night-time behaviors. Sleep disorders and abnormal night-time behaviors may play an important role in promoting dementia based on recent research [27]. Night-time behaviors may be both a manifestation of cardiovascular disease and a cause. Many symptoms had no association with cardiovascular disease burden (including elation, anxiety, delusions, hallucinations, and increased motor behavior) while frontal behaviors (disinhibition, anxiety and irritability) tended to correlate most strongly with CVRF including Framingham scores, Hachinski scores, and low HDL cholesterol levels. A potential mechanism for this observed frontal lobe dysfunction may be small vessel disease, which may be associated with abnormal cholesterol levels [28].
While the results of this study are of interest, there are some limitations that should be highlighted. This was a cross-sectional study that looked at baseline data and thus we are not able to speculate on the trajectory of cognitive decline over time. We were unable to obtain full data on all participants. We used the Framingham Risk Score which evaluates cardiovascular disease risk as opposed to the Framingham stroke score. In addition, the MCI sample was somewhat heterogeneous in nature and this may have confounded the results. We also combined patients with MCI and remitted MDD, two distinct diagnoses, which may have also affected our results, though the NPI score distribution did not vary among groups. As well, we looked at vascular risk factors but did not incorporate any imaging markers of cerebrovascular disease. Finally, we did not look at longitudinal variables, such as fluctuation in blood pressure measurements, which has recently been identified as a risk factor for AD progression, given we looked exclusively at baseline data.
In spite of the above-mentioned limitations, our study demonstrated a number of conclusions worthy of further investigation. We were able to demonstrate that both the prevalence and profile of NPS in MCI is similar to that of MDD in remission. We showed low HDL cholesterol, high Hachinski scores, and high Framingham scores may mediate NPI sub-scores such as anxiety, irritability, and disinhibition, suggesting a potential mechanism of increased small vessel disease. In addition, we showed a link among hypertension, overall Framingham score, and night-time behaviors which also correlated with cognitive decline, thus suggesting that night-time behaviors may be symptomatic of emergent dementia. Future longitudinal studies examining the impact of CVRF and blood pressure variation on individual NPS sub-scores and incorporating imaging and disease biomarkers are required.
