ABSTRACT
Objectives
In a sample of community‐dwelling older adults, we examined the association of Alzheimer's Disease and Related Dementias (AD/ADRD) blood‐based biomarkers (BBMs) and neuropsychiatric symptoms (NPS) and whether informant type (i.e., spouse vs. child vs. other) modified that association.
Methods
This study included 430 participants with a cognitively unimpaired or mild cognitive impairment consensus diagnosis from the Wake Forest Alzheimer's Disease Research Center Clinical Core cohort. Informants reported NPS using the Neuropsychiatric Inventory Questionnaire. AD/ADRD BBMs included the Aβ42/40 ratio, p‐tau181, p‐tau217, NfL, and GFAP. Generalized linear models were used to examine the associations between AD/ADRD BBMs and NPS. Secondary models adjusted for age, sex, education, race, and cognitive status. Tertiary models adjusted for covariates in secondary models, as well as informant type. Interactions between informant type and AD/ADRD BBMs were examined.
Results
Higher p‐tau217 was associated with higher NPS in both unadjusted models and models adjusted for demographics and cognitive status. This association was attenuated and no longer statistically significant after additionally adjusting for informant type. Significant interactions of informant type and p‐tau181 or p‐tau217 on NPS were demonstrated, where p‐tau181 or p‐tau217 were more strongly associated with NPS reported by children compared to spouses.
Conclusions
Informant type modified the association between AD/ADRD BBMs and NPS, with stronger associations observed when symptoms were reported by child informants compared to spouse informants. These findings have important implications for earlier detection of individuals with AD/ADRD pathologies.
Keywords: Alzheimer's Disease and Related Dementias, blood‐based biomarkers, informant, neuropsychiatric symptoms
Key Points
P‐tau217 was significantly associated with NPS in unadjusted models and models adjusted for demographics and cognitive status. This association was attenuated and no longer statistically significant after additionally adjusting for informant type.
Informant type modified the associations of p‐tau181 or p‐tau217 on NPS, where the associations were stronger when NPS were reported by children compared to spouses.
These findings highlight the importance of considering informant type when investigating the association between AD/ADRD BBMs and NPS.
1. Introduction
By 2050, the number of people diagnosed with Alzheimer's Disease and Related Dementias (AD/ADRD) in the United States is predicted to nearly double [1]. It has been estimated that a 5‐year delay in AD/ADRD would result in a 50% reduced incidence [2], highlighting the importance of identifying individuals in the earliest stages of the disease for interventions to potentially delay or alter the disease course.
Neuropsychiatric symptoms (NPS) are non‐cognitive, behavioral symptoms that manifest in the early stages of AD/ADRD and become more severe with disease progression [3]. Recent evidence has supported the association of these symptoms with AD/ADRD pathologies even before the emergence of cognitive impairments [4]. Importantly, NPS have been shown to be associated with more deleterious outcomes for individuals with AD/ADRD and their caregivers. For example, increased NPS have been associated with increased cost of care for patients [5], more rapid cognitive declines [6], and increased mortality [7].
Given the significant impact and early emergence of NPS in AD/ADRD, assessments must be both consistent and accurate. As AD/ADRD symptomology typically presents as memory impairments that progressively increase with disease advancement [1], an individual's ability to accurately report symptoms may become compromised. One method commonly used in research settings to mitigate the influence of an individual's memory on NPS reporting is the reporting of symptoms being completed by a study partner or “informant”. However, prior research has suggested informant reporting of multiple constructs, including NPS and cognitive function, varies by informant type [8, 9, 10, 11]. Moreover, one study demonstrated that informant reports of cognition were more accurate when they were made by spouses compared to children [12]. Given that NPS are considered early clinical manifestations of AD/ADRD pathologies, it is crucial to investigate factors, such as informant type, that may affect reporting. At present, to our knowledge, there remains a dearth of research examining how the association of NPS and AD/ADRD pathologies varies by informant types.
The biological definition of AD characterizes the disease by its pathological hallmarks, including amyloid‐beta (Aβ) plaques, neurofibrillary tau tangles, and neurodegeneration [13]. Historically, the in vivo quantification of these pathologies was completed using magnetic resonance imaging (MRI) and positron emission tomography (PET) methodologies. However, the high cost and increased burden have limited the accessibility of these methodologies across both clinical and research settings. Recent advancements have enabled the in vivo assessment of AD/ADRD pathologies in larger, more diverse samples using blood‐based biomarkers (BBMs). AD/ADRD BBMs include Aβ42/40 ratio, phosphorylated tau 181 (p‐tau181), phosphorylated tau 217 (p‐tau217), neurofilament light (NfL), and glial fibrillary acidic protein (GFAP).
The current study was designed to examine the association of AD/ADRD BBMs and NPS in a sample of community‐dwelling older adults from the Wake Forest University (WFU) Alzheimer's Disease Research Center (ADRC) Clinical Core, and whether informant type (i.e., spouse vs. child vs. other) modified that association. We hypothesized that participants with more abnormal levels of AD/ADRD BBMs would demonstrate higher levels of reported NPS, and that this association would differ between informant types, specifically spouses compared to children or other.
2. Materials and Methods
2.1. Participants
The WFU ADRC Clinical Core recruited participants aged 55–85 years from the surrounding community; inclusion/exclusion criteria have been published previously [14]. Participants completed clinical evaluations using the Uniform Data Set (UDS) Version 3, neuropsychological testing, neuroimaging, and blood draws. We examined baseline visit data from Clinical Core participants diagnosed as cognitively unimpaired (CU) or mild cognitive impairment (MCI), and who had available BBMs and informant‐reported NPS data. Of the 747 Clinical Core participants enrolled in the cohort from 2016 to 2022, 317 participants were excluded due to incomplete data, dementia/other diagnoses, or self‐reported NPS (Figure S1), leaving 430 participants for the current analyses. Compared to participants excluded from the current analyses, those included were more likely to be White (80.7% vs. 68.1%, p < 0.001) or CU (62.1% vs. 29.0%, p < 0.001), and less likely to be Black/African American (18.1% vs. 29.97%, p < 0.001). There were no significant differences in age, gender, or years of education. The WFU Institutional Review Board approved study protocols. All participants provided written informed consent.
2.2. Adjudication
The cognitive status of Clinical Core participants was adjudicated during a weekly consensus review by an expert panel comprised of geriatricians, neurologists, neuropsychologists, and neuroimagers. Clinical diagnoses of CU, MCI, dementia, or other were adjudicated after review of all clinical and cognitive data in accordance with 2011 NIA‐AA diagnostic guidelines [15, 16]. Etiological causes underlying the clinical diagnoses were then discerned using all available biomarker data. As previously described, Alzheimer's disease is the primary etiological cause underlying the clinical diagnoses of Wake Forest ADRC Clinical Core participants [17].
2.3. Measures
2.3.1. NPI‐Q
The Neuropsychiatric Inventory‐Questionnaire (NPI‐Q) was obtained by informant‐report and provided a comprehensive measure of 12 domains: delusions, hallucinations, agitation/aggression, depression/dysphoria, anxiety, elation/euphoria, apathy/indifference, disinhibition, irritability/lability, motor disturbance, nighttime behaviors, and appetite/eating [18]. Informants first responded to a screening question for each of the 12 domains that reflected whether domain‐specific behaviors were exhibited by the participant in the past month. Positive responses were then further characterized with severity scores that ranged from “mild” to “severe” on a three‐point scale. Total severity scores were a summation of all items (0 = symptom absent, 1 = mild, 2 = moderate, 3 = severe) and had a possible range of 0–36 (hereafter referred to as NPS‐severity). Prior to entry in models, a constant of 1 was added to the values.
2.3.2. Informant Demographics
Data characterizing the informants of participants were collected using the NACC UDS Co‐participant Demographics Form A2 [19]. Information collected included the informant's age, gender, ethnicity, race, education, and relationship to the participant. The reported informant‐participant relationship was used to stratify informants into three groups: spouse or romantic partner (hereafter referred to as spouse), child, and other.
2.3.3. BBMs
As previously described [17], blood samples were collected from participants in the fasting state. Plasma Aβ42, Aβ40, p‐tau181, NfL, and GFAP were analyzed utilizing the Quanterix Neurology 4‐Plex E and p‐tau181 version 2 Advantage Kits on a Quanterix Simoa HD‐X. Concentrations of p‐tau217 were quantified using the ALZpath p‐tau217 assay on the Simoa HD‐X platform [20].
2.4. Statistical Analyses
2.4.1. Descriptive Statistics
Descriptive statistics for participants and informants are reported for the total sample and stratified by informant type in Tables 1 and 2, respectively. Categorical variables are reported as frequency (percentage). Continuous variables are reported as mean (standard deviation (SD)). Kruskal‐Wallis tests were used to investigate differences in demographics by informant type for continuous variables using the base R stats package [21]. Dunn's test was used to further characterize significant Kruskal‐Wallis results through the examination of pairwise differences using the dunn.test package [22]. Fisher's exact tests examined differences in categorical demographic variables across the different informant types using the base R stats package [21]. Significant results were further characterized through the examination of row‐wise/pairwise differences using the rstatix package [23].
TABLE 1.
Participant demographic statistics by informant type.
| Measures | Total (n = 430) | Spouse (n = 273) | Child (n = 56) | Other (n = 101) | p |
|---|---|---|---|---|---|
| Age (years) a , c , d | 70.5 (7.9) | 69.9 (7.2) | 73.6 (8.7) | 70.2 (8.8) | 0.018 |
| Women (n (%)) b , c , e | 281 (65.3%) | 146 (53.5%) | 47 (83.9%) | 88 (87.1%) | < 0.001 |
| Education (years) a , e | 15.9 (2.5) | 16.2 (2.4) | 15.3 (3.0) | 15.4 (2.4) | 0.011 |
| Race (n (%)) b | < 0.001 | ||||
| White d , e | 347 (80.7%) | 239 (87.5%) | 45 (80.4%) | 63 (62.4%) | |
| Black/African American d , e | 78 (18.1%) | 31 (11.4%) | 10 (17.9%) | 37 (36.6%) | |
| Other | 5 (1.2%) | 3 (1.1%) | 1 (1.8%) | 1 (1%) | |
| Cognitive status (n (%)) | 0.722 | ||||
| Cognitively unimpaired | 267 (62.1%) | 170 (62.3%) | 37 (66.1%) | 60 (59.4%) | |
| MCI | 163 (37.9%) | 103 (37.7%) | 19 (33.9%) | 41 (40.6%) | |
| NPS‐severity a , c , d , e | 1.4 (2.8) | 1.4 (2.7) | 2.6 (4.3) | 0.8 (2.0) | < 0.001 |
| Plasma biomarkers | |||||
| Aβ42/40 ratio | 0.05 (0.01) | 0.05 (0.01) | 0.05 (0.01) | 0.05 (0.01) | 0.697 |
| p‐tau181 (pg/mL) a , c , d | 3.3 (1.9) | 3.3 (2.1) | 3.6 (1.9) | 3.0 (1.3) | 0.037 |
| p‐tau217 (pg/mL) a , d , e | 0.4 (0.3) | 0.4 (0.3) | 0.5 (0.4) | 0.3 (0.2) | 0.008 |
| NfL (pg/mL) a , c , d | 16.0 (11.1) | 15.8 (11.9) | 19.6 (9.5) | 14.5 (9.3) | < 0.001 |
| GFAP (pg/mL) a , c , d | 126.3 (68.7) | 119.7 (60.1) | 166.7 (106.1) | 121.5 (56.8) | 0.001 |
Note: Statistics are presented as mean (SD).
Significant differences from Kruskal‐Wallis tests are reported between participants with different informant types.
Significant differences from Fisher's exact tests are reported between participants with different informant types.
Significant pairwise differences are reported between participants with spouse informants vs. child informants.
Significant pairwise differences are reported between participants with child informants vs. other informants.
Significant pairwise differences are reported between participants with spouse informants vs. other informants.
Abbreviations: Aβ, amyloid‐beta; GFAP, Glial fibrillary acidic protein; MCI, Mild cognitive impairment; NfL, Neurofilament light; NPS‐Severity, neuropsychiatric symptoms total severity scores; p‐tau181, phosphorylated tau 181; p‐tau217, phosphorylated tau 217.
TABLE 2.
Informant demographic statistics by informant type.
| Measures | Total (n = 430) | Spouse (n = 273) | Child (n = 56) | Other (n = 101) | p |
|---|---|---|---|---|---|
| Age (years) a , c , d | 65.5 (12.8) | 68.9 (8.3) | 44.8 (11.8) | 67.6 (12.8) | < 0.001 |
| N (missing) | 1 | 0 | 0 | 1 | |
| Women (n (%)) b , e | 250 (58.1%) | 135 (49.5%) | 37 (66.1%) | 78 (77.2%) | < 0.001 |
| Education (years) | 15.8 (2.4) | 15.9 (2.4) | 16.0 (2.4) | 15.5 (2.3) | 0.254 |
| N (missing) | 2 | 0 | 1 | 1 | |
| Race (n (%)) b | < 0.001 | ||||
| White e | 341 (79.3%) | 233 (85.3%) | 44 (78.6%) | 64 (63.4%) | |
| Black/African American e | 78 (18.1%) | 34 (12.5%) | 10 (17.9%) | 34 (33.7%) | |
| Other | 11 (2.6%) | 6 (2.2%) | 2 (3.6%) | 3 (3%) | |
| Lives w/Participant (n (%)) b , c , e | 287 (66.7%) | 267 (97.8%) | 11 (19.6%) | 9 (8.9%) | < 0.001 |
| Length of relationship (years) | 40.4 (17.4) | 40.0 (14.8) | 43.9 (12.2) | 39.5 (24.8) | 0.296 |
| N (missing) | 2 | 1 | 0 | 1 |
Note: Statistics are presented as mean (SD). Of the 101 other informants, 44 were family members and 57 were friends or other social acquaintances.
Significant differences from Kruskal‐Wallis tests are reported between spouse, child, and other informants.
Significant differences from Fisher's exact tests are reported between spouse, child, and other informants.
Significant pairwise differences are reported between spouse informants versus child informants.
Significant pairwise differences are reported between child informants versus other informants.
Significant pairwise differences are reported between spouse informants versus other informants.
2.4.2. Model Selection
Generalized linear models (GLMs) were used to examine the relationship between each AD/ADRD BBM and NPS‐severity. GLMs are advantageous modeling techniques that can be used when dependent variables violate the assumption of normality and the use of ordinary least squares regressions may result in biased or inefficient results [24]. We examined diagnostic plots and model quality indices to determine the appropriate distribution family using the performance package [25]. These results informed the models described in Section 2.4.3.
2.4.3. Analyses
GLMs with gamma distributions and identity links were used in initial analyses. Log links are commonly used in GLMs with dependent variables that are gamma distributed; however, we chose to first utilize identity links for the interpretability of the results [26, 27]. For analyses that did not converge with identity links, secondary models with log links were examined. Utilizing a model‐based approach, we conducted the analyses in four steps. We first examined the relationship of BBMs and NPS‐severity in unadjusted models. Secondary models adjusted for age, gender, education, race, and cognitive status. Tertiary models adjusted for covariates in secondary models and informant type. Men, White participants, CU participants, and participants with spouse informants were the referent groups for the gender, race, cognitive status, and informant type variables, respectively. Next, to determine the effect of informant type on these relationships, we included an interaction of informant type with each BBM. Significant interactions were further investigated in models stratified by informant type. As the stratification of the sample by informant type yielded different distributions of the dependent variable (NPS‐severity), we re‐examined the appropriate distribution families for the two subsamples of interest (i.e., participants with spouse informants and participants with child informants) using diagnostic plots and model quality indices. These results indicated that gamma distributions were most appropriate for both samples (i.e., participants with spouse informants and participants with child informants). The initial models with identity links in the subsample of participants with child informants did not converge; thus, results from secondary models with log links are reported. The threshold for statistical significance was p < 0.05. All analyses were conducted in R 4.3.1 [21].
3. Results
3.1. Demographics
3.1.1. Participant Demographics
Participant demographic statistics are described in Table 1 for the total sample and stratified by informant type. The mean (SD) age of participants was 70.5 (7.9) years. Women comprised 65.3% of the total sample (n = 281). The mean (SD) years of education was 15.9 (2.5). The sample was comprised of 80.7% White participants (n = 347), 18.1% Black/African American participants (n = 78), and 1.2% participants of other races (n = 5). The mean (SD) of NPS‐severity was 1.4 (2.8) and the median (range) was 0 (0–25).
Compared to participants with spouse informants, participants with child informants were significantly older, were more likely to be women, had higher NPS‐severity, and higher levels of p‐tau181, NfL, and GFAP (Table 1). Compared to participants with other informants, participants with child informants were significantly older, had a higher proportion of White participants, had higher NPS‐severity, and higher levels of p‐tau181, p‐tau217, NfL, and GFAP (Table 1). Participants with spouse informants, compared to those with other informants, were less likely to be women, had more years of education, had a higher proportion of White participants, higher NPS‐severity, and higher levels of p‐tau217 (Table 1).
3.1.2. Informant Demographics
Informant demographic statistics for the total sample and by informant type are described in Table 2. Informants included 273 spouses, 56 children, and 101 other. The mean (SD) age of informants was 65.5 (12.8) years and education years was 15.8 (2.4). Women comprised 58.1% of informants (n = 250). The sample of informants was comprised of 79.3% White informants (n = 341), 18.1% Black/African American informants (n = 78), and 2.6% informants of other races (n = 11). Moreover, 66.7% of informants lived with their participants (n = 287).
Compared to spouse informants, child informants were significantly younger and less likely to cohabitate with participants (Table 2). Compared to other informants, child informants were significantly younger (Table 2). Spouse informants, compared to other informants, were less likely to be women, had a higher proportion of White informants, and were more likely to cohabitate with participants (Table 2).
3.2. AD/ADRD BBMs and NPS‐Severity Analyses
3.2.1. Unadjusted and Adjusted Analyses
In unadjusted analyses, higher levels of p‐tau217 or NfL, but not other BBMs, were associated with higher NPS‐severity (Table 3). The association between p‐tau217 and NPS‐severity remained after adjusting for age, gender, education, race, and cognitive status (Table 3). However, there were no associations between any BBMs and NPS‐severity after additionally adjusting for informant type (Table 3).
TABLE 3.
Total sample: Relationship between BBMs and NPS‐severity.
| Model 1 | Model 2 | Model 3 | ||||
|---|---|---|---|---|---|---|
| Beta | p | Beta | p | Beta | p | |
| Aβ42/40 ratio | −11.110 | 0.404 | 3.355 | 0.748 | 4.903 | 0.617 |
| p‐tau181 | 0.160 | 0.075 | 0.004 | 0.951 | −0.034 | 0.550 |
| p‐tau217 | 1.794 | 0.003* | 1.071 | 0.041* | 0.650 | 0.180 |
| NfL | 0.056 | 0.002* | 0.011 | 0.469 | 0.006 | 0.659 |
| GFAP | 0.004 | 0.064 | 0.001 | 0.555 | 0.001 | 0.737 |
Note: Model 1 was unadjusted. Model 2 was adjusted for age, gender, education, race, and cognitive status. Model 3 additionally adjusted for informant type.
Abbreviations: Aβ, amyloid‐beta; BBMs, blood‐based biomarkers; GFAP, glial fibrillary acidic protein; NfL, neurofilament light; NPS‐severity, neuropsychiatric symptoms total severity scores; p‐tau181, phosphorylated tau 181; p‐tau217, phosphorylated tau 217.
indicates p < 0.05
A significant main effect of informant type was demonstrated across all five models, such that child informants reported higher NPS‐severity compared to participants with spouse informants.
3.2.2. Adjusted Analyses With Interaction Term and Stratified Analyses
We next assessed whether informant type modified the association between BBMs and NPS‐severity (Table 4; Figure 1). The main effect of informant type was attenuated and no longer associated with NPS‐severity after inclusion of the informant type by BBM interaction term for each of the five BBM models (Table 4). Additionally, the main effects of each of the BBMs remained nonsignificant in adjusted analyses with the interaction term (Table 4). However, we observed a significant interaction of informant type by p‐tau181 and p‐tau217 on NPS‐severity, indicating that the relationship between p‐tau181 or p‐tau217 and NPS‐severity differed between participants with spouse informants and participants with child informants (Table 4; Figure 1). Specifically, higher levels of p‐tau181 or p‐tau217 were more strongly associated with NPS‐severity reported by child informants compared to NPS‐severity reported by spouse informants. There were no significant interactions between informant type and Aβ42/40 ratio, NfL, or GFAP in relation to NPS‐severity (Table 4; Figure 1).
TABLE 4.
Total sample: Interaction of informant type and BBMs on NPS‐severity.
| Aβ42/40 ratio | Beta | p |
|---|---|---|
| Aβ42/40 ratio | 5.559 | 0.645 |
| Child informant | 2.240 | 0.312 |
| Other informant | −0.329 | 0.774 |
| Aβ42/40 ratio × child informant | −24.065 | 0.542 |
| Aβ42/40 ratio × other informant | 0.506 | 0.981 |
| p‐tau181 | Beta | p |
|---|---|---|
| p‐tau181 | −0.062 | 0.235 |
| Child informant | −1.244 | 0.136 |
| Other informant | −0.317 | 0.479 |
| p‐tau181 × child informant | 0.650 | 0.019* |
| p‐tau181 × other informant | −0.016 | 0.907 |
| p‐tau217 | Beta | p |
|---|---|---|
| p‐tau217 | 0.163 | 0.759 |
| Child informant | −0.937 | 0.125 |
| Other informant | −0.194 | 0.572 |
| p‐tau217 × child informant | 4.594 | 0.017* |
| p‐tau217 × other informant | −0.580 | 0.500 |
| NfL | Beta | p |
|---|---|---|
| NfL | −0.008 | 0.392 |
| Child informant | −0.322 | 0.724 |
| Other informant | −0.566 | 0.117 |
| NfL × child informant | 0.074 | 0.153 |
| NfL × other informant | 0.018 | 0.407 |
| GFAP | Beta | p |
|---|---|---|
| GFAP | 0.002 | 0.409 |
| Child informant | 0.057 | 0.944 |
| Other informant | 0.312 | 0.514 |
| GFAP × child informant | 0.005 | 0.339 |
| GFAP × other informant | −0.006 | 0.112 |
Note: Models were adjusted for age, gender, education, race, and cognitive status.
Abbreviations: Aβ, amyloid‐beta; BBMs, blood‐based biomarkers; GFAP, glial fibrillary acidic protein; NfL, neurofilament light; NPS‐severity, neuropsychiatric symptoms total severity scores; p‐tau181, phosphorylated tau 181; p‐tau217, phosphorylated tau 217.
indicates p < 0.05
FIGURE 1.
Forest plots depicting the interaction coefficients of informant type and BBMs on NPS‐severity. Models were adjusted for age, gender, education, race, and cognitive status. Interaction coefficients and 95% confidence intervals are provided. Aβ, amyloid‐beta; BBMs, blood‐based biomarkers; GFAP, glial fibrillary acidic protein; NfL, neurofilament light; NPS‐severity, neuropsychiatric symptoms total severity scores; p‐tau181, phosphorylated tau 181; p‐tau217, phosphorylated tau 217.
The significant interactions were further examined in analyses stratified by informant type (i.e., spouse vs. child; Table 5; Figure 2). In the subsample of participants with spouse informants, there were no associations between p‐tau181 or p‐tau217 and NPS‐severity. Among participants with child informants, higher levels of p‐tau217, but not p‐tau181, were associated with higher NPS‐severity.
TABLE 5.
Relationship between BBMs and NPS‐severity stratified by informant type.
| Child informant | n | Beta | p |
|---|---|---|---|
| p‐tau181 | 56 | 0.097 | 0.170 |
| p‐tau217 | 56 | 0.785 | 0.016* |
| Spouse informant | n | Beta | p |
|---|---|---|---|
| p‐tau181 | 273 | −0.069 | 0.188 |
| p‐tau217 | 273 | −0.087 | 0.872 |
Note: Models were adjusted for age, gender, education, race, and cognitive status.
Abbreviations: BBMs, blood‐based biomarkers; NPS‐severity, neuropsychiatric symptoms total severity scores; p‐tau181, phosphorylated tau 181; p‐tau217, phosphorylated tau 217.
indicates p < 0.05
FIGURE 2.
Scatterplots of the relationship between BBMs and NPS‐severity stratified by informant type. The association between p‐tau181 or p‐tau217 and NPS‐severity stratified by informant type. For all variables, age, sex, education, race, and cognitive status have been regressed out. Trendlines include 95% confidence intervals. BBMs, blood‐based biomarkers; NPS‐severity, neuropsychiatric symptoms total severity scores; p‐tau181, phosphorylated tau 181; p‐tau217, phosphorylated tau 217.
4. Discussion
In the present study, we investigated the association between BBMs and NPS‐severity in a large sample of community‐dwelling older adults, and determined whether informant type (i.e., spouse, child, or other) modified the association. We hypothesized that higher levels of AD/ADRD BBMs would be associated with higher NPS‐severity, and that the association would be stronger among spouses. However, this hypothesis was not supported by our findings. We observed associations between p‐tau217 with NPS‐severity in unadjusted models and models adjusted for age, gender, education, race, and cognitive status; however, this association was attenuated and no longer significant after adjusting for informant type. We also observed significant interactions of informant type and p‐tau181 or p‐tau217 on NPS‐severity, where the association between p‐tau181 or p‐tau217 with NPS‐severity was stronger among child informants compared to spouse informants. In additional stratified analyses by informant type (i.e., spouse vs. child informants), higher levels of p‐tau217, but not p‐tau181, were associated with higher NPS‐severity among participants with child informants, but not those with spouse informants.
4.1. AD/ADRD BBMs and NPS
Our findings highlight the complexities of investigating the association between AD/ADRD biomarkers and NPS‐severity and emphasize the importance of considering informant type in these investigations. In this study, higher p‐tau217 was associated with higher NPS‐severity in unadjusted models and models adjusted for demographics and cognitive status. This finding is consistent with prior work that reported an association between p‐tau217 and NPS‐severity after adjusting for demographics [28]. However, in the current study, this association was attenuated after additional adjustment for informant type. Additionally, we demonstrated a main effect of informant type in all adjusted models (i.e., models of Aβ42/40, p‐tau181, p‐tau217, NfL, and GFAP), suggesting that reports of NPS are influenced by informant type even after controlling for participant demographic variables. This highlights the importance of considering informant type when assessing the association between AD/ADRD BBMs and NPS‐severity.
4.2. Study Implications and Future Research
Informant‐reports are widely used to assess NPS in clinical practice, clinical trials, and research settings [29]. However, there is limited evidence examining how NPS reporting may vary by informant type, and whether informant type affects the diagnostic or prognostic value of NPS. One recent study found that NPS endorsed by a spouse or child informant were associated with faster progression to dementia compared to NPS endorsed by other informants [8]. In contrast, the effect of informant demographics and relationship to participant on reporting has been more extensively examined regarding cognitive function, with inconsistent results [10, 30, 31]. Further research is needed to investigate the underlying reasons for these differences in reporting.
Despite the known discrepancies, these studies are essential in understanding participants' symptomology, even early in the AD/ADRD disease continuum when cognitive impairments may not yet be present. For example, a cross‐sectional study of CU participants demonstrated that informant‐reported apathy, but not self‐reported apathy, was associated with AD/ADRD biomarkers [4]. Our study expands upon this research by using AD/ADRD BBMs to show that the associations between p‐tau181 or p‐tau217 with NPS were stronger when NPS were reported by children compared to spouses. However, only the association between p‐tau217 and NPS remained significant in the analyses stratified by informant type (i.e., spouse vs. child). This finding suggests that reports of NPS made by children may better reflect underlying changes in AD/ADRD biomarkers compared to those made by spouses.
The present findings highlight the need for additional investigation into the underlying reasons for differences between child and spouse reports of NPS and their association with AD/ADRD biomarkers. We speculate that several mechanisms may underlie the differences seen between NPS reports from children and spouses and their association to AD/ADRD biomarkers. First, it is possible that gradual changes in NPS are more likely to be perceived and reported by informants who have less frequent, or less consistent, interactions with the participant. That is, children, who often do not cohabitate with participants, may be more perceptive to subtle declines, while spouses, who interact daily, might become habituated to the changes. This hypothesis is supported by prior research demonstrating that informants with higher levels of perceived closeness to a participant reported significantly lower NPS [32]. However, the low levels of NPS reported by other informants in the present study suggests that the association between informant type and NPS may be more complex. Second, spouses may underreport symptoms as a result of denial or a desire to “cover” for the participant. One study reported that higher levels of caregiver burden was associated with higher levels of reported NPS [33]. However, it has also been found that spouse informants, despite reporting higher levels of caregiver burden, reported lower levels of NPS, cognitive impairment, and functional impairments compared to other informants, including children [11]. Furthermore, studies comparing NPS reported by spouses to clinicians have demonstrated that spouses often underreport the severity of NPS [34].
At present, it is unclear whether the findings of the current study reflect differences in the perception of changes in NPS or an obscuration of NPS by spouses during reporting. Future research, particularly in cohorts with reports of NPS from both the spouse and child of a participant, will provide further context in which the findings of this study can be interpreted and help elucidate the underlying mechanisms contributing to these discrepancies in NPS reporting.
4.3. Strengths and Limitations
There are multiple strengths to the study. First, the large, community‐based sample of older adults enhanced the generalizability of the findings. Second, the use of GLMs with gamma distributions addressed the skewed distribution of NPS without dichotomizing the variable, allowing us to utilize a measure of NPS severity instead of the presence/absence of NPS that is more commonly used. Third, the interaction between informant type and AD/ADRD BBMs in relation to NPS‐severity provided valuable insights regarding the impact of informant type on the association between AD/ADRD BBMs and NPS‐severity.
However, limitations also warrant consideration. First, NPS were reported by one informant instead of multiple informant types for each participant. Consequently, despite corrections for covariates, the findings may reflect actual differences in NPS levels among participants rather than variations in reporting. Second, this sample was moderately sized and exhibited lower levels of NPS, which may have limited the ability to find significant associations. Third, the cross‐sectional study design limited our ability to assess causality. It is of interest to examine these relationships longitudinally to compare the utility of reports of NPS as made by different informants with regard to predicting longitudinal changes in AD/ADRD BBMs and cognitive decline. Fourth, we did not examine this relationship using specific NPI‐Q items. It is unclear how these results would translate to item‐based analyses. Finally, we did not adjust our analyses for multiple comparisons, which may affect the robustness of our findings.
5. Conclusion
In summary, we observed an association of p‐tau217 with NPS‐severity in a sample of community‐dwelling older adults. Moreover, this association was stronger when reports of NPS were made by children of participants compared to spouses, suggesting that reports of NPS made by the former informant type may better reflect underlying AD/ADRD pathologies. These results should be replicated in other cohorts where reports of NPS have been provided by two or more informants for each participant and other AD/ADRD biomarkers are available, including PET or cerebrospinal fluid (CSF). These findings have important implications for earlier detection of individuals with AD/ADRD pathologies.
Ethics Statement
The study was approved by the Wake Forest University School of Medicine IRB.
Consent
All participants provided written informed consent.
Conflicts of Interest
Dr. Mielke has served on scientific advisory boards and/or has consulted for Acadia, Biogen, Eisai, LabCorp, Lilly, Merck, PeerView Institute, Novo Nordisk, Roche, Siemens Healthineers and Sunbird Bio. All other authors declare no conflicts of interest.
Supporting information
Figure S1: Study flow chart of included and excluded participants.
Acknowledgments
The authors thank all the Wake Forest Alzheimer's Disease Research Center Clinical Core participants for their participation in this study.
Bacci, Julia R. , Rudolph Marc D., Craft Suzanne, Bateman James R., Lockhart Samuel N., and Mielke Michelle M.. 2025. “The Relationship of Alzheimer's Disease and Related Dementias Blood‐Based Biomarkers and Informant‐Reported Neuropsychiatric Symptoms Differs by Informant Type in Older Adults Without Dementia.” International Journal of Geriatric Psychiatry: e70140. 10.1002/gps.70140.
Funding: Funding was provided by the National Institutes of Health/National Institute on Aging (P30 AG072947, U24 AG082930, RF1 AG077386, RF1 AG079397, RF1 AG69052, and T32 NS115704). No funding was received for the publication of this article.
Samuel N. Lockhart is a full‐time employee of Perceptive Inc. Burlington MA, USA.
Data Availability Statement
The data that support the findings of this study are available on request from WakeSHARE at https://wakeshare.org/.
References
- 1.“2024 Alzhimer’s Disease Facts and Figures,” Alzheimer's & Dementia 20, no. 5 (2024): 3708–3821. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Jorm A. F., Korten A. E., and Henderson A. S., “The Prevalence of Dementia: A Quantitative Integration of the Literature,” Acta Psychiatrica Scandinavica 76, no. 5 (1987): 465–479, 10.1111/j.1600-0447.1987.tb02906.x. [DOI] [PubMed] [Google Scholar]
- 3. Lyketsos C. G., Lopez O., Jones B., Fitzpatrick A. L., Breitner J., and DeKosky S., “Prevalence of Neuropsychiatric Symptoms in Dementia and Mild Cognitive Impairment: Results From the Cardiovascular Health Study,” JAMA 288, no. 12 (2002): 1475, 10.1001/jama.288.12.1475. [DOI] [PubMed] [Google Scholar]
- 4. Burling J. E., Katz Z., Yuan Z., et al., “Study Partner Report of Apathy in Older Adults Is Associated With AD Biomarkers: Findings From the Harvard Aging Brain Study,” American Journal of Geriatric Psychiatry 32, no. 8 (2024): 909–919, 10.1016/j.jagp.2024.01.020. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5. Murman D. L., Chen Q., Powell M. C., Kuo S. B., Bradley C. J., and Colenda C. C., “The Incremental Direct Costs Associated With Behavioral Symptoms in AD,” Neurology 59, no. 11 (2002): 1721–1729, 10.1212/01.wnl.0000036904.73393.e4. [DOI] [PubMed] [Google Scholar]
- 6. Krell‐Roesch J., Syrjanen J. A., Machulda M. M., et al., “Neuropsychiatric Symptoms and the Outcome of Cognitive Trajectories in Older Adults Free of Dementia: The Mayo Clinic Study of Aging,” International Journal of Geriatric Psychiatry 36, no. 9 (2021): 1362–1369, 10.1002/gps.5528. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Peters M. E., Schwartz S., Han D., et al., “Neuropsychiatric Symptoms as Predictors of Progression to Severe Alzheimer’s Dementia and Death: The Cache County Dementia Progression Study,” Australian Journal of Pharmacy 172, no. 5 (2015): 460–465, 10.1176/appi.ajp.2014.14040480. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Guan D. X., Mudalige D., Munro C. E., Nosheny R., Smith E. E., and Ismail Z., “The Effect of Study Partner Characteristics on the Reporting of Neuropsychiatric Symptoms Across the Neurocognitive Spectrum,” International Psychogeriatrics 36, no. 8 (2024): 1–14, 10.1017/s1041610224000590. [DOI] [PubMed] [Google Scholar]
- 9. Lin A., Brook J., Grill J. D., and Teng E., “Participant–Informant Relationships Affect Quality of Life Ratings in Incipient and Clinical Alzheimer Disease,” American Journal of Geriatric Psychiatry 25, no. 3 (2017): 297–307, 10.1016/j.jagp.2016.10.007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Shen J., Gao S., Unverzagt F. W., et al., “Validation Analysis of Informant’s Ratings of Cognitive Function in African Americans and Nigerians,” International Journal of Geriatric Psychiatry 21, no. 7 (2006): 618–625, 10.1002/gps.1536. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Persson K., Brækhus A., Selbæk G., Kirkevold Ø, and Engedal K., “Burden of Care and Patient’s Neuropsychiatric Symptoms Influence Carer’s Evaluation of Cognitive Impairment,” Dementia and Geriatric Cognit¡ve Disorders 40, no. 5–6 (2015): 256–267, 10.1159/000437298. [DOI] [PubMed] [Google Scholar]
- 12. Cacchione P. Z., Powlishta K. K., Grant E. A., Buckles V. D., and Morris J. C., “Accuracy of Collateral Source Reports in Very Mild to Mild Dementia of the Alzheimer Type,” Journal of the American Geriatrics Society 51, no. 6 (2003): 819–823, 10.1046/j.1365-2389.2003.51263.x. [DOI] [PubMed] [Google Scholar]
- 13. Jack C. R. Jr., Bennett D. A., Blennow K., et al., “NIA‐AA Research Framework: Toward a Biological Definition of Alzheimer’s Disease,” Alzheimer's & Dementia 14, no. 4 (2018): 535–562, 10.1016/j.jalz.2018.02.018. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Coffin C., Suerken C. K., Bateman J. R., et al., “Vascular and Microstructural Markers of Cognitive Pathology,” Alzheimer's & Dementia: Diagnosis, Assessment & Disease Monitoring 14, no. 1 (2022): e12332, 10.1002/dad2.12332. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Albert M. S., DeKosky S. T., Dickson D., et al., “The Diagnosis of Mild Cognitive Impairment due to Alzheimer’s Disease: Recommendations From the National Institute on Aging‐Alzheimer’s Association Workgroups on Diagnostic Guidelines for Alzheimer’s Disease,” Alzheimer's and Dementia 7, no. 3 (2011): 270–279, 10.1016/j.jalz.2011.03.008. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. McKhann G. M., Knopman D. S., Chertkow H., et al., “The Diagnosis of Dementia due to Alzheimer’s Disease: Recommendations From the National Institute on Aging‐Alzheimer’s Association Workgroups on Diagnostic Guidelines for Alzheimer’s Disease,” Alzheimer's & Dementia 7, no. 3 (2011): 263–269, 10.1016/j.jalz.2011.03.005. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Rudolph M. D., Sutphen C. L., Register T. C., et al., “Associations Among Plasma, MRI, and Amyloid PET Biomarkers of Alzheimer’s Disease and Related Dementias and the Impact of Health‐Related Comorbidities in a Community‐Dwelling Cohort,” Alzheimer's and Dementia 20, no. 6 (2024): 4159–4173, 10.1002/alz.13835. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18. Kaufer D. I., Cummings J. L., Ketchel P., et al., “Validation of the NPI‐Q, a Brief Clinical Form of the Neuropsychiatric Inventory,” Journal of Neuropsychiatry and Clinical Neurosciences 12, no. 2 (2000): 233–239, 10.1176/jnp.12.2.233. [DOI] [PubMed] [Google Scholar]
- 19. Besser L., Kukull W., Knopman D. S., et al., “Version 3 of the National Alzheimer’s Coordinating Center’s Uniform Data Set,” Alzheimer Disease and Associated Disorders 32, no. 4 (2018): 351–358, 10.1097/wad.0000000000000279. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20. Ashton N. J., Brum W. S., Di Molfetta G., et al., “Diagnostic Accuracy of a Plasma Phosphorylated Tau 217 Immunoassay for Alzheimer Disease Pathology,” JAMA Neurology 81, no. 3 (2024): 255, 10.1001/jamaneurol.2023.5319. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. R Core Team . “R: A Language and Environment for Statistical Computing,”2023, https://www.R‐project.org/.
- 22. Dinno A.. dunn.test: Dunn’s Test of Multiple Comparisons Using Rank Sums. 2014;1.3.6.
- 23. Kassambara A., “Rstatix: Pipe‐Friendly Framework for Basic Statistical Tests,”2023, https://rpkgs.datanovia.com/rstatix/.
- 24. Mall S., Lund C., Vilagut G., Alonso J., Williams D. R., and Stein D. J., “Days out of Role due to Mental and Physical Illness in the South African Stress and Health Study,” Social Psychiatry and Psychiatric Epidemiology 50, no. 3 (2015): 461–468, 10.1007/s00127-014-0941-x. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Lüdecke D., Ben‐Shachar M., Patil I., Waggoner P., and Makowski D., “Performance: An R Package for Assessment, Comparison and Testing of Statistical Models,” JOSS 6, no. 60 (2021): 3139, 10.21105/joss.03139. [DOI] [Google Scholar]
- 26. Lee C. S. and Conway C., “The Role of Generalized Linear Models in Handling Cost and Count Data,” European Journal of Cardiovascular Nursing 21, no. 4 (2022): 392–398, 10.1093/eurjcn/zvac002. [DOI] [PubMed] [Google Scholar]
- 27. Polgreen L. A. and Brooks J. M., “Estimating Incremental Costs With Skew: A Cautionary Note,” Applied Health Economics and Health Policy 10, no. 5 (2012): 319–329, 10.1007/bf03261866. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Krell‐Roesch J., Zaniletti I., Syrjanen J. A., et al., “Plasma‐Derived Biomarkers of Alzheimer’s Disease and Neuropsychiatric Symptoms: A Community‐Based Study,” Alzheimer's & Dementia: Diagnosis, Assessment & Disease Monitoring 15, no. 3 (2023): e12461, 10.1002/dad2.12461. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Cummings J., “The Neuropsychiatric Inventory: Development and Applications,” Journal of Geriatric Psychiatry and Neurology 33, no. 2 (2020): 73–84, 10.1177/0891988719882102. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Vargas‐Gonzalez J.‐C., Chadha A. S., Castro‐Aldrete L., Ferretti M. T., and Tartaglia M. C., “Informant Characteristics Influence Clinical Dementia Rating Sum of Boxes Scores‐Based Staging of Alzheimer’s Disease,” Nature Aging 4, no. 11 (October 2024): 1538–1543: Epub ahead of print 25, 10.1038/s43587-024-00732-x. [DOI] [PubMed] [Google Scholar]
- 31. Morrell L., Camic P. M., and Genis M., “Factors Associated With informant‐Reported Cognitive Decline in Older Adults: A Systemised Literature Review,” Dementia 18, no. 7–8 (2019): 2760–2784, 10.1177/1471301218759836. [DOI] [PubMed] [Google Scholar]
- 32. Vernon E. K., Cooley B., Rozum W., et al., “Caregiver‐Care Recipient Relationship Closeness Is Associated With Neuropsychiatric Symptoms in Dementia,” American Journal of Geriatric Psychiatry 27, no. 4 (2019): 349–359, 10.1016/j.jagp.2018.11.010. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Majer R., Adeyi O., Bagoly Z., et al., “Neuropsychiatric Symptoms, Quality of Life and Caregivers’ Burden in Dementia,” Open Medicine 15, no. 1 (2020): 905–914, 10.1515/med-2020-0124. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Zaidi S., Kat M. G., and De Jonghe J. F. M., “Clinician and Caregiver Agreement on Neuropsychiatric Symptom Severity: A Study Using the Neuropsychiatric Inventory – Clinician Rating Scale (NPI‐C),” International Psychogeriatrics 26, no. 7 (2014): 1139–1145, 10.1017/s1041610214000295. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Figure S1: Study flow chart of included and excluded participants.
Data Availability Statement
The data that support the findings of this study are available on request from WakeSHARE at https://wakeshare.org/.