Abstract
Background
Non-Alzheimer's disease dementias, including frontotemporal dementia (FTD) can be difficult to characterize due to the predominance of distinct behavioral and neuropsychiatric symptoms. Widely used measurement tools lack structure and objectivity.
Objective
The purpose of this study was to use systematic direct observation of neuropsychiatric and behavioral symptoms, via the Neurobehavioral Rating Scale (NBRS), to characterize clusters of behavioral and neuropsychiatric symptoms in FTD and examine how selected symptom clusters correlate with structural neuroimaging.
Methods
We performed a factor analysis on the NBRS data from 172 patients with FTD and examined the neural correlates of the selected symptom clusters in a subsample of 67 patients.
Results
Six factors accounted for 56% of total variance across NBRS item scores: Apathy/Blunting, Agitation/Disinhibition, Cognitive/Language, Planning/Insight, Anxiety/Lability, and Psychosis. Symptom clusters showed significant associations with specific regions of cortical thinning: Agitation/Disinhibition with bilateral frontal regions, and Cognition/Language with the left bank of the superior temporal sulcus and supramarginal regions.
Conclusions
The selected symptom clusters associated with known regions of atrophy in FTD. The NBRS is an effective observational measure that may extend characterization and understanding of FTD.
Keywords
Introduction
Frontotemporal dementia (FTD) is a heterogenous neurodegenerative disorder characterized by progressive atrophy of the frontal and temporal lobes. Despite being one of the more common early-onset dementias, classification of FTD remains a challenge in today's healthcare system given the lack of comprehensive biomarkers and the variability of clinical phenotypes.1,2 Currently, diagnosis involves the integration of multiple assessments, including clinician evaluation, neuropsychological testing, neuroimaging, and biomarker and genetic testing. The behavioral and neuropsychiatric symptoms, such as apathy and disinhibition, which predominate the clinical presentation, can significantly reduce quality of life in patients and their families.3,4 However, there is an ongoing challenge in the field related to assessment and identification. Anosognosia, which is a pathological lack of insight into symptomology and a core characteristic of FTD, 5 limits the utility of patient reports, and thus the field currently relies on subjective reports from family members, assessment tools that do not cover key specific FTD symptoms (e.g., Neuropsychiatric Inventory [NPI]), 6 or largely unstructured clinician assessments. There is a key need for simple, predictive, and structured observational measures to be validated in FTD, to aid in the characterization and study of the phenomenologic heterogeneity of this disorder. 7
The Neurobehavioral Rating Scale (NBRS) 8 is one promising measure for assessing the neuropsychiatric symptoms of FTD. Relying on systematic direct observation and assessment of an extensive list of behavioral and neuropsychiatric symptoms by a trained third-party assessor, this tool offers a complement to caregiver-rated measures and physician-based interviews. The NBRS was initially developed for use in patients with traumatic brain injury, but has since been validated in various other neurological populations including in dementia cohorts, 9 and encompasses a broad range of behavioral, neuropsychiatric, and cognitive symptoms that are common in neurological illness.
Previous studies of the NBRS have examined its factor structure, or how the measured symptoms group or cluster together. Symptom clustering is an important approach to understanding and distinguishing many medical conditions, including neurodegenerative illnesses.10–14 While the factor structure of the NBRS in FTD patients has not yet been described, the NBRS was recently used in combination with other rating scales (NPI and Frontal Systems Behavior Scale)6,15 to examine symptom clustering in FTD. 14 The authors extracted three overarching factors, including Apathy, “Disinhibition vs. Depression”, and Psychosis. 14 Other studies have also sought to identify neuropsychiatric/behavioral symptom clusters in FTD, although this research has been based on participant and carer reports; for example, Schonecker et al. 13 identified five neurobehavioral factors in a cohort of genetic FTD (diverse behavioral, affective, psychotic, euphoric/hypersexual, and tactile hallucinations), and Borroni et al. 16 reported four phenotypic factors in behavioral variant FTD (disinhibited, apathetic, aggressive, language). Yet, to date, no study has established the unique factor structure of neuropsychiatric/behavioral symptoms in FTD using the NBRS, which covers a broader range of symptoms than more commonly used measures such as the NPI, and circumvents the subjectivity issues with family reports.
Finally, while symptom clusters alone are a powerful tool in illuminating disease presentation, integrating symptom burden with neuroimaging findings of cortical atrophy offers valuable complementary insight into disease presentation in patients with FTD.12,17,18 For instance, specific regions of frontotemporal atrophy are associated with clusters of neuropsychiatric/behavioral symptoms in FTD; specifically, apathy-related symptoms tend to be associated with degeneration of frontal regions, and disinhibition is generally correlated with ventral frontal, temporal and subcortical (including nucleus accumbens and striatum) atrophy.14,16,19 Investigating the neural correlates of symptom clusters is important for understanding broader patterns of cortical degeneration, disease burden, and disease progression. 12
The purpose of this study was to assess and characterize FTD by investigating: (1) how behavioral and neuropsychiatric items cluster together in FTD using the NBRS in this population, and (2) how the selected symptom clusters are associated with structural neuroimaging. Toward these aims, we performed a factor analysis using NBRS data obtained from 172 patients with FTD and assessed the neural correlates of the revealed symptom clusters in a subsample of FTD patients.
Methods
Participants
All participant data was collected through the Cognitive Neuroscience Section of the National Institute of Neurological Disorders and Stroke (NINDS) of the National Institute of Health (NIH) in Bethesda, Maryland, between 1998 and 2010. The study included a nine-day research visit during which patients and their family members, including a study partner with Durable Power of Attorney, participated voluntarily in clinical, neuropsychological, and neurological examinations in addition to completing a T1-weighted 1.5 T structural magnetic resonance imaging (MRI). The study was granted ethical approval by the NIH Institutional Review Board and all participants provided assent individually and through the consent of their Legal Representative.
Scale responses and factor structure were analyzed based on the observation of 172 participants for whom the NBRS scale was completed during this study. Participants were recruited based on a clinical diagnosis of FTD. Diagnoses were confirmed using the criteria current at the time 5 by a study-affiliated neurologist, psychiatrist, and neuropsychologist. The group was differentiated by FTD subtype: 138 were diagnosed with behavioral variant FTD (bvFTD), 28 with nonfluent variant primary progressive aphasia (nfPPA), and 6 with semantic dementia or semantic variant primary progressive aphasia (svPPA). Demographic and clinical data are detailed in Table 1 (FTD subtype, age, sex, ethnicity, years of education, age of onset, years since first symptoms, NBRS total score).
Patient demographics and clinical characteristic
bvFTD: behavioral variant frontotemporal dementia; svPPA: semantic variant primary progressive aphasia; nfPPA: non-fluent variant primary progressive aphasia; MDRS-2: Mattis Dementia Rating Scale-Second Edition; NBRS: Neurobehavioral Rating Scale.
NBRS
The NBRS was completed based on observations over a period of 5–9 days by a trained research assistant. It is a 27-item instrument developed to assess behavioral disturbance, psychiatric symptoms and cognitive deficits.8,9 The NBRS has strong interrater reliability and validity.8,20 Neurobehavioral changes are ranked from 0 to 6—‘Not Present’ to ‘Extremely Severe’—by trained research coordinators with extensive experience with patients with FTD and who had received additional training on the completion of the NBRS based on observation of patient characteristics and functioning during the study visit. NBRS total scores can range from 0 to 162, with higher scores representing a greater presence and severity of neurobehavioral symptoms.
Factor analysis
Factor analysis was conducted based on responses to the NBRS. A principal components analysis (PCA) was conducted in SPSS and R to define clusters of NBRS items. Orthogonal rotation was employed based on the presumed independence of scale items, and a scree plot was used to determine the number of factors for analysis and interpretation. Similarity between factor structures was calculated as the proportion of items that loaded onto equivalent factors.
Initial confirmatory tests were conducted: a KMO value of 0.695 and Bartlett's Sphericity significant at <0.1 suggested suitability of the data for a factor analysis. Through visualization of the resulting scree plot, the number of factors for the analysis was set as 6 based on an inflection point >1 after which the difference between eigenvalues flattened, differing by <0.1. Upon running the factor analysis, items were assigned to factors based on correlation coefficients >0.4. The frequency within this FTD sample for each symptom cluster was also calculated based on the summation of item-level scores.
Neuroanatomical correlates
A subgroup of 67 patients with a diagnosis of bvFTD, complete scores on the NBRS, and an MRI were included for the presented neuroanatomical analyses. Patients with low-quality MRI, based on visual inspection, were excluded. Aside from diagnosis, this subset was socio-demographically representative of the larger sample. There were no statistically significant differences in age, sex, education, ethnicity, or race between the subset of patients used for the MRI analysis and the rest of the participants in this study.
Structural imaging data were acquired on a 1.5 T GE MR scanner (GE Medical Systems, Milwaukee, WI) with standard quadrature head coil. A T1-weighted spoiled gradient echo sequence was used to generate 124 contiguous 1.5-mm-thick axial slices (repetition time = 6.1 msec; flip angle = 20°; field of view = 240 mm; matrix size = 256 × 256 × 124).
Cortical thickness was assessed with FreeSurfer, which calculates cortical thickness of a brain region using structural MRI data.21,22 Using previously described methods, a 3-dimensional cortical surface model was created from Tl-weighted MRI with the automatic FreeSurfer “recon-all” pipeline.21,23 Surface models of the outer limit of the grey matter (pial surface) and white matter were mapped onto the MRI. 21 These surface reconstructions were visually inspected for accuracy and reconstructions and major errors that could not be corrected were excluded. Cortical thickness was defined as the distance between the pial and white matter surfaces. 21 To ensure that measurements of cortical thickness were comparable across the different scanners present in the data, the ComBat model 24 was applied to the thickness measurements obtained from FreeSurfer. ComBat is a data harmonization approach that has been shown to reduce scanner effects in structural and functional neuroimaging data while maintaining relevant biological and behavioral relationships.25–27 For region of interest analyses, mean cortical thickness was calculated for regions of interest within each hemisphere using the Desikan-Kiliany cortical atlas. 28
The primary analysis focused on frontotemporal regions of interest defined in Freesurfer including: medial orbitofrontal, lateral orbitofrontal, pars orbitalis, pars triangularis, pars opercularis, frontal pole, rostral middle frontal, caudal middle frontal, superior frontal, supramarginal, superior temporal, caudal anterior cingulate, precentral, and the banks of the superior temporal sulcus regions. Regions of interest were identified based on prior studies suggesting that atrophy in these regions is related to some of the key phenomenology seen in FTD (e.g.,29–31). Linear regression was used to quantify associations between regional cortical thickness and participants’ severity on each NBRS symptom factor. For each factor, and region, a regression model was run with each cortical thickness as a predictor and participants’ age at scanning visit as a covariate. Multiple comparisons across regions of interest were accounted for using false discovery rate (FDR) correction. Bonferroni adjustment was additionally applied to account for the analysis of six NBRS factors.
Results
Factor analysis
Six identified factors were determined to account for 56% of total variance across NBRS item scores through the PCA with orthogonal rotation. Six items did not load onto any factor and two items loaded onto more than one factor despite the orthogonal rotation (Table 2). Factors were named based on included items (Table 2). The six factors identified in the analysis were Apathy/Blunting, Agitation/Disinhibition, Cognitive/Language, Planning/Insight, Anxiety/Lability, and Psychosis in descending order of percent of variance captured by the factor in this FTD population (Table 2). Symptom frequency by factor within the present sample is described in Table 3.
NBRS selected factors & components in FTD.
*signifies a significant negative association with an item to the cluster. Italic text indicates an item that has loaded onto multiple clusters.
NBRS symptom frequency in FTD sample.
*Total = sum of item-level scores. Factors are ordered from most frequent to least frequent.
As an additional point of comparison, we replicated the factor analysis excluding all patients with svPPA and nfPPA (n = 34). This was done to determine whether patients with PPA were largely responsible for the language factor. The factor analysis could not be conducted within PPA due to the small sample size of 34. 32 The language factor remained when PPA variants were excluded, suggesting the importance of language dysfunction in patients with bvFTD and affirming the overlap in phenotypes of FTLD disorders that has been established in the literature. 33
Neuroanatomical correlates
Table 4 displays means, standard deviations, and ranges of NBRS factor scores. In a linear regression analysis controlling for age, Agitation/Disinhibition was associated with reduced cortical thickness in regions of the left and right hemispheres, most strongly with the right hemisphere superior frontal (β = −0.56, adjusted pFDR < 0.001), caudal middle frontal (β = −0.49, p = 0.002), and pars opercularis (β = −0.45, p = 0.007) regions (see Figure 1). Cognition/Language deficits were associated with less cortical thickness of left hemisphere in the supramarginal (β = −0.45, p = 0.023) and the banks of the superior temporal sulcus (β = −0.46, p = 0.023) regions. Additionally, Anxiety/Lability was associated with greater cortical thickness in right hemisphere regions of interest, most strongly with the superior frontal (β = 0.53, p = 0.001), pars opercularis (β = 0.51, p = 0.001), pars triangularis (β = 0.50, p = 0.001), and caudal middle frontal (β = 0.50, p = 0.001). See Table 5 for all regression results, and Figure 1 for a visual representation. Apathy/Blunting, Planning/Insight, and Psychosis were not significantly associated with cortical thickness in any of the regions of interest.
Cortical thickness associated with NBRS factor scores.
NBRS factor score characteristics.
Relationship between NBRS factor scores and cortical thickness.
Linear regression results predicting NBRS factor scores with cortical thickness of each region of interest entered as a predictor. Participant age at scanning visit included as a covariate. Bolded values indicate statistically significant results at p < 0.05 after correction for multiple comparisons.
Discussion
This study found six factors underlying item-level responses on the NBRS: Apathy/Blunting, Agitation/Disinhibition, Cognitive/Language, Planning/Insight, Anxiety/Lability, and Psychosis. Additionally, within the sample, the presence and severity of identified symptom factors associated with differences in cortical thickness across brain regions. Collectively, these findings highlight the utility of the NBRS as an assessment tool for FTD, and meaningfully extend existing research on the characterization of FTD and related disease.
That the different identified factors were associated with decreased cortical thickness in specific frontotemporal brain regions extends literature on possible neural underpinnings of these important symptoms that may arise early in disease course. For instance, cognition/language deficits were associated with reduced cortical thickness of the left bank of the superior temporal sulcus and supramarginal regions, which are located in close proximity to Wernicke's area and have been shown to play a role in language processing. 34 Because previous studies have demonstrated worse survival outcomes for bvFTD patients based on their cognitive and language deficits, these findings may reinforce the importance of careful clinician assessment of patients’ cognitive and language function and atrophy of relevant regions on MRI.35,36
Further, in our study, cognitive/language deficits were more associated with left cortical atrophy, and behavioral and neuropsychiatric (disinhibition/agitation, anxiety/lability) symptoms with right cortical atrophy. This bears some similarity to previous research in bvFTD, for example, one study found that disinhibition was associated with pathology of the right orbitofrontal cortex while deficits on a language-based executive task were associated with left ventral-lateral temporal lobe pathology. 37 The association found here between anxiety and mood lability with preserved cortical thickness within right hemisphere brain regions also adds to mixed literature regarding anxiety and the frontotemporal regions in bvFTD 38 and in populations without dementia.39,40 More severe symptoms of agitation/disinhibition were associated with reduced cortical thickness of several frontal regions: the right superior frontal, pars opercularis, caudal middle frontal, pars triangularis, rostral middle frontal and the left superior frontal and caudal anterior cingulate regions, similar to previous research.14,41–43 Notably, although our results are somewhat consistent with previous research insofar that frontal regions appear to play an important role in disinhibition and agitation, we did not find the specific ventral frontal associations with disinhibition identified in previous FTD studies 14 (e.g.,42,44). This could be a result of modest differences in how disinhibition was defined across studies, and the fact our current disinhibition factor was combined with agitation. Standardization of relevant symptom measures will continue to be meaningful direction for the field, a goal toward which we hope this work contributes. Still, our findings add to this literature by suggesting that agitation and disinhibition in FTD may have a common substrate resulting from degeneration in a wide frontal network. Further work exploring functional connectivity changes in FTD will elucidate network-level disruptions (e.g.,29,45) and the prognostic implications of atrophy in these regions.
While there were not significant associations between the identified Apathy/Blunting factor and neuroanatomical regions of interest in this present study, that it was revealed as part of the FTD structure through the NBRS is important. Apathy is a key symptom in FTD, 33 and previous studies have found that apathy is associated with pre-symptomatic genetic forms of FTD and is an indicator of poor prognosis.46–48
Overall, our findings indicate that the NBRS may be useful in the characterization of neuropsychiatric symptom patterns in FTD. However, the present study is limited in several important ways. The current study was unique given the nature of data collection over a week-long, observational study period, where research coordinators had ample time to observe behavior. The NBRS may be especially well-suited to such studies, including inpatient-based clinical trials where long-term relationships are formed between study personnel and participants, but may not be as appropriate in contexts where patient contact is brief and rudimentary. Another limitation of this study was that the sample was largely highly-educated (mean of 15.5 years of education), white, and had self-selected to participate in the intensive, multi-day study from which the findings were drawn. This limits the generalizability of findings to the broader populations seen in standard, community-based clinical care. The findings outlined here could be strengthened by replication in larger and more diverse FTD populations. Further, these findings are most representative of bvFTD patterns given that this was predominant diagnosis in the sample. It would be helpful to analyze differences and similarities in the NBRS factors for bvFTD versus PPA phenotypes with larger and more reliable sample sizes, as evidence suggests that neurobehavioral symptoms may vary among FTD subtypes (e.g., bvFTD versus PPA, and within PPA phenotypes).49–51
There are also relevant limitations to consider regarding the neuroimaging findings. Namely, the smaller subset of patients entered into the analysis may have been underpowered and therefore vulnerable to type II error. Also, although the range of symptoms, severity and atrophy in the population did still allow for meaningful neuroimaging comparisons to be made—the lack of a control group prevented cortical comparisons to a healthy population. Finally, subcortical regions of the brain were not analyzed in this study and may play a significant role in the neuropsychiatric symptoms of FTD; for example, the striatum has been linked with disinhibition in FTD. 14 Additionally, as genetic information was not available on these participants, we were not able to differentiate between those with FTD due to pathogenic variants vs. sporadic FTD. Future research should consider using the NBRS to examine heterogeneity among different pathogenic genetic mutation carriers with FTD. In addition, the only information on disease severity that was available was from the Mattis Dementia Rating Scale, as FTD-specific disease severity scales (e.g., CDR®+NACC FTLD) were not yet commonplace when the study was conducted. This information would have enriched interpretation of the present study. Ultimately, future research replicating and extending these findings could be impactful. Larger samples, direct in-study comparison of different dementia types and healthy controls, and integration of modern neuroimaging technology could very meaningfully expand upon these findings.
The findings of the current study suggest that measurements based on direct observation by trained raters, such as those captured by the NBRS, can complement caregiver-rated and physician assessments, as raters may spend more time with the patients in the research setting than research physicians, and may have more expertise in FTD than caregivers, uniquely positioning trained raters to capture important phenomenological information. The NBRS may also prove to be a useful tool to identify key symptoms in prodromal bvFTD, as the current factors align closely with the recently proposed research criteria for prodromal bvFTD. 46 Indeed, five of our six factors (all except psychosis) are represented in the Barker et al. 46 diagnostic criteria. These findings reinforce the utility of the NBRS as a simple yet effective tool for identifying key symptoms that may help identify and understand FTD.
In conclusion, this study identified six clusters of FTD symptoms based on factor analysis of the NBRS and explored how these symptom clusters correlated with distinct patterns of cortical atrophy. The NBRS is a simple and predictive observational measure, and since it is structured and based on observable behaviors, it circumvents problems with family reports and unstructured interviews. It may expand the characterization of phenomenologic heterogeneity in FTD, with substantial implications for patients, families, and providers.
Footnotes
Acknowledgments
We thank Karen Detucci and Alyson Cavanagh for patient testing. We thank Amelia Boehme, PhD, MSPH, for her support with database review. We thank the patients and caregivers who participated in this study, without whom this work would not be possible.
ORCID iDs
Author contributions
Corinne Sejourne (Conceptualization; Formal analysis; Methodology; Writing – original draft); Megan Barker (Conceptualization; Supervision; Writing – original draft); Madison Heath (Formal analysis; Writing – original draft); Yunglin Gazes (Formal analysis; Methodology; Writing – review & editing); Rachel Fremont (Conceptualization; Writing – review & editing); Yedili Genao Perez (Writing – review & editing); Luke Hearne (Conceptualization; Visualization; Writing – review & editing); Eric M Wassermann (Investigation; Resources; Writing – review & editing); Michael C Tierney (Investigation; Resources); Masood Manoochehri (Project administration; Writing – review & editing); Edward D Huey (Conceptualization; Funding acquisition; Methodology; Resources; Supervision; Writing – review & editing); Jordan Grafman (Conceptualization; Data curation; Funding acquisition; Methodology; Resources; Supervision; Writing – review & editing).
Funding
This work was supported by the Intramural Research Program of The National Institutes of Health/The National Institute of Neurological Disorders and Stroke, by a grant from the Division of Extramural Research of The National Institutes of Health / The National Institute of Neurological Disorders and Stroke to EDH [R00 NS060766].
Declaration of conflicting interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Data availability
Data is available upon request to Dr Jordan Grafman.