Abstract
Background:
Alzheimer’s disease (AD) and behavioral-variant of frontotemporal dementia (bvFTD) can present with an overlapping neuropsychological profile, which often hinders their clinical differentiation.
Objective:
To compare changes over time in memory, general cognition tasks, and functional scales between bvFTD and AD.
Methods:
Consecutive cases diagnosed with probable bvFTD (n = 22) and typical AD (n = 31) with at least two clinical visits were selected. Of these, 13 (9 AD, 4 bvFTD) underwent Pittsburgh compound B PET scan, which supported the clinical diagnosis in all cases. Mixed-model regressions were used to estimate the differential rate of decline on selected tasks between cohorts.
Results:
Analyses demonstrated that, despite equivalent baseline performance, bvFTD patients experienced a more rapid functional deterioration and a steeper decline in global cognition than AD patients. At baseline, both groups were impaired on executive function and memory tasks compared to controls, but these deficits were more marked in the bvFTD group. In addition, performance on these domains continued to decline more rapidly in this group.
Conclusions:
Neither the initial neuropsychological assessment nor projected performances can reliably distinguish the totality of bvFTD and AD individuals. Nevertheless, annual rates of progression on cognitive tasks provide valuable information and will potentially help establish the impact of future therapeutic treatments in these dementia syndromes.
Keywords
INTRODUCTION
Current consensus diagnostic criteria indicate that executive dysfunction with relatively sparing of episodic memory defines the neuropsychological profile of behavioral-variant of frontotemporal dementia (bvFTD) [1]. Nevertheless, executive dysfunction is often observed in Alzheimer’s disease (AD) [2 –5] and episodic memory deficits are reported in early bvFTD [6 –9], suggesting that those deficits are insufficiently specific to distinguish both groups. This contention, however, is based mostly on cross-sectional studies, and little is known about the stability and progression of these disease profiles over time. Existing evidence suggests that a relentless cognitive and functional decline occurs more rapidly in bvFTD than in AD [10 –14]. The reported rapid decline in bvFTD raises the possibility that distinctive features of this syndrome not evident at baseline may become more prominent as the diseaseprogresses. Another possibility is that the rapid progression and widespread cognitive deterioration in bvFTD overshadows any distinctive feature, making the differentiation between AD and bvFTD increasingly difficult. These findings, however, predate current diagnostic criteria and have relied on small study samples. How these syndromes evolve clinically over time needs to be re-visited in the light of recent advances in the understanding of the clinical and pathological processes of these disorders. Indeed, this information is crucial for improving differential diagnosis at different stages of disease severity, effectively evaluating the impact of therapeutic drugs, and providing accurate prognosis [12].
We investigated the progression in bvFTD and AD across several cognitive, behavioral, and functional measures over time, focusing on the longitudinal trajectory of memory. We hypothesized that each group would follow a distinctive pattern of progression reflecting divergent underlying pathological process.
MATERIALS AND METHODS
Participants
This study compared AD and bvFTD longitudinal cohorts selected from consecutive cases recruited from the Frontier, Frontotemporal Dementia Group at Neuroscience Research Australia between 2008 and 2013. As a part of the research activities, participants regularly attend annual visits. Patients with at least two annual visits were selected, whereas those patients with significant mental illness, history of drug or alcohol abuse, cerebrovascular disease, movement disorders, or limited English proficiency were excluded from the study. As a result, from an initial sample of 143 patients (AD = 76, bvFTD = 67), a cohort of 53 patients (AD = 31, bvFTD = 22) was suitable for the study. Clinical diagnosis was based on the current International Consensus Criteria for each condition [1, 15], and all patients underwent a comprehensive neurological assessment at baseline and follow-up visits, which involved history taking and neurological examination, plus neuropsychological testing and an MRI. Four bvFTD and 9 AD patients underwent a Pittsburgh compound B PET (PiB-PET). None of the bvFTD cases that underwent a PiB-PET scan showed amyloid retention, whereas positive retention was found in all AD patients scanned. Healthy controls who scored ≥88 on the Addenbrooke’s Cognitive Examination-Revised [16] (ACE-R) were selected from the Frontier pool ofcontrols and matched for age and education with the patient sample. All study participants underwent a full neuropsychological evaluation and a brain MRI.
This research was approved by the South Eastern Sydney Local Area Health District and the University of New South Wales ethics committees. Written informed consent was provided by all participants or their caregiver.
Outcome variables: Cognitive and functional impairment measurement
The ACE-R was used as measures of global cognitive capacity (max score 100) where higher scores reflect better cognitive ability. ACE-R subdomain scores (attention/orientation, memory, fluency, language, and visuospatial) were used as outcome variables to examine the rate of cognitive decline over time across domains. Because the maximum possible scores vary across ACE-R subdomains, scores were standardized using healthy control performance as a reference to allow comparisons of rates of decline across subdomains.
Integrity of episodic memory function was assessed using the following tests: Doors section of the Doors and People Test (immediate correct recognition on Part A) [17]; Rey-Osterrieth Complex Figure Task (3-min recall and percent retention score; i.e., 3-min recall score corrected for copy score) [18]. In addition to this measure, ACE-R memory items, with and without semantic items, were used to ascertain episodic memory. For executive function, the following tests were administered: Hayling test (overall scaled score) [19]; Phonemic (letter) fluency (total number of correct words F, A and S) [20]; Trail Making Test A and B (time to complete) [21]; WAIS-III Digit Span Forward and Backward (max span) [22]. To establish the magnitude of change over time, all scores were normalized using performances of matched healthy controls as a reference.
Disease severity was assessed using the Functional Rating Scale [23] (FRS), a carer-based questionnaire, comprising 7 subscales (behavior, outing and shopping, household chores and telephone, finances, medications, meal preparation and eating, and self-care and mobility). FRS-Rasch scores are derived which can be categorized in to 6 stages of severity to facilitate clinical interpretation: very mild (5.39 to 4.12), mild (4.11 to 1.92), moderate (1.91 to –0.40), severe (–0.41 to –2.58), very severe (–2.59 to –4.99), and profound (–5.0 to –6.66). The FRS has demonstrated its usefulness in staging a wide spectrum of dementia phenotypes, including language and behavioral presentations of frontotemporal dementias and AD. It also appears to be more sensitive in detecting mild changes in bvFTD than other similar instruments, such as the Clinical Dementia Rating scale [23].
Statistical analysis
Baseline
Data were analyzed using IBM SPSS 21.0. Comparisons of continuous variables between groups were conducted using analysis of variance (ANOVA), followed by pair-wise contrasts where appropriate, Bonferroni corrected for multiple comparisons. Differences in gender distribution across groups were estimated using χ 2 tests. Total number of follow-up visits was expressed as a median; hence, a nonparametric approach was used for this variable. Time scores for Trail Making Test Part A and B were transformed into logarithmic scale to achieve Gaussian distribution.
Longitudinal
Imputation of missing data was conducted prior to carrying out longitudinal analyses, following a procedure described previously [24]. Briefly, missing observations were assumed to be not random, and regarded as non-ignorable, if they were not explained by the magnitude of the previous observation or length of follow-up. Only one non-ignorable missing observation per case was imputed in each variable as long as the missing observation occurs after at least two real observations. The imputation was conducted using simple linear regression calculated for each variable.
Linear mixed effect models were derived for each outcome variable over time. The fixed effects of this model included diagnostic group, follow-up time, and the interaction of both effects. The individual variability of the patients at baseline and follow-up were included as random effect in the model. Estimated parameters were all determined by the fixed and random effects in the model. A significant effect of follow-up time indicated a linear change in the measured outcome with time, and a significant interaction between follow-up time and diagnosis indicated that this linear change in the measured outcome over time differed between bvFTD and AD. Residual errors from the model and the random intercepts for patients’ baseline responses were assumed to be normally distributed. All patients were assumed to be independent. Because of the focus on episodic memory, longitudinal analyses on the ACE-R memory subscore were carried out with and without the 4 questions measuring semantic memory integrity, in order to account for a potential confounding effect of semantic knowledge on memory change over time.
RESULTS
Baseline assessment
Between-group differences in age and education were observed, with the bvFTD group being younger than healthy controls, and the bvFTD group having lower education levels than the other two groups. Total estimated length of disease duration, sex distribution, length of follow-up, and number of visits did not differ between bvFTD and AD groups (Table 1).
Both patient groups were impaired on the ACE-R and all sub-scores compared to controls (Table 2). Similarly, on memory tasks, both AD and bvFTD groups demonstrated impaired performance on all measures when compared with healthy controls, and no distinctive group performance profiles were observed between the groups. To rule out confounding effects of verbal encoding on overall memory performance, we compared scores on the learning task of ACE-R (i.e., learning of a name and address over three trials; maximum score: 7 points). We found no significant difference between the AD (mean±SD = 5.8±1.5) and bvFTD (6.2±1.2), t = 1.1, p = 0.3) groups on this measure. On executive tasks, bvFTD patients performed more poorly than the AD group on Hayling and Letter Fluency tasks.
Given the group differences in age and education, analyses were repeated after co-varying for age and education. These investigations revealed a similar pattern of results across groups, except in AD who showed lower performance on visuospatial domain ACE-R after co-varying for age.
Given our a priori interest on memory, regression analyses were conducted to investigate possible contributions of executive function (TMT Part B, Digit Span Backward) to memory performance (ACE-R Memory and Doors A) with each memory score investigated separately. Both regression models failed to reach significance denoting that verbal and visuospatial memory performance were independent from executive function.
Longitudinal
Both patients groups demonstrated decline over time on almost all neuropsychological tests, with theexception of the Hayling and Rey Complex Figure recall and retention scores (Table 3). Despite equivalent ACE-R baseline performance in both dementia groups, the bvFTD group demonstrated an annual decline that was almost twice that of the AD group (Table 3 and Fig. 1). Changes over time on the ACE-R subscales did not differ between groups, with the exception of memory, where the bvFTD group showed an annual decline that was one some measures at least twice the rate of that observed in AD (e.g., ACE-R Memory subscores and Doors). On the executive tasks, the worse baseline performance on the Hayling and letter fluency tests in bvFTD than AD remained stable over time. In contrast, performance on Trail Making Test B and Digit span backwards declined more quickly in the bvFTD than in the AD group (Fig. 1). In addition, the annual decline on Digit Span Forward task was faster in the bvFTD than the AD group, despite a similar baseline performance. Finally, the early functional deficits in the bvFTD group identified on the FRS continued to evolve more rapidly in this group compared with AD (Fig. 1).
DISCUSSION
This study compared the baseline and longitudinal performances on memory, executive function, and attention tasks commonly used in the clinic between bvFTD and AD, the two most common younger-onset dementia syndromes. Baseline assessment demonstrated widespread cognitive impairments in both groups compared to healthy controls. Importantly, severity of these deficits was equivalent on a number of cognitive measures including memory. The tasks on which the groups differed; however, the bvFTD group performed significantly worse than the AD group, as was the case for executive function tasks (e.g., verbal generativity (verbal fluency), ability to inhibit a prepotent response (Hayling). In addition, although overall group performance on executive tasks was worse in bvFTD than in AD, at least a third of AD cases showed deficits of similar severity on these tasks. Functionally, the bvFTD group was also more impaired than the AD group despite a similar performance on a cognitive screening instrument (ACE-R).
Longitudinal analyses demonstrated that, with time, both groups experienced relentless cognitive and functional decline. Importantly, the bvFTD group showed a more rapid deterioration over time overall compared to AD, as well as on specific tasks of memory and executive function. While such deterioration has been previously reported [9 –14], these studies were antecedent to the current diagnostic criteria and, therefore, may have included possible cases with uncertain prognosis.
Our analyses revealed different patterns of cognitive change in bvFTD and AD with disease progression, depending on the tasks. On the majority of tests, both dementia groups were impaired to a similar degree compared with controls at baseline and either declined at the same rate (TMT part A, Digit Span Forward, ACE-R Attention, Language and Visuospatial components), or declined faster in bvFTD than in AD (Digit Span Backwards, Doors A, TMT Part B, ACE-R Total and Memory component). On a few tests (Letter Fluency, ACE-R Fluency component, Hayling, FRS), bvFTD performed worse at baseline than AD and declined faster. On no task, the AD group performed worse at baseline, or declined at a faster rate compared with the bvFTD group.
The more rapid decline of memory function in bvFTD than in AD cannot be attributed solely to a faster spread of brain pathology. Rather, the prominent decline in both executive function and memory decline in this group may be explained by the continuing degradation of orbitofrontal cortex integrity, accompanied by an involvement of the medial and dorsolateral aspects of the prefrontal cortices and their connections [25]. This pattern of progression further accords well with the similar sequential distribution of pathological changes in cases with bvFTD and positive TDP-43 inclusions [26]. This contrasts with the relative sparing of posterior brain regions, resulting in the comparatively slow decline in performance observed on visuospatial and attention/orientation tasks. This observation supports the notion that progression of neurodegeneration in the brain is not diffuse or stochastic, but it rather follows a predictable and specific pattern determined in large extent by the involvement of large-scale networks [27].
The presence of an early and continuing decline in memory function in bvFTD raises the important issue of diagnostic specificity. Emergence of episodic memory deficits is one of the cardinal and early manifestations of AD, and the most common complaint of individuals presenting to memory clinics. While presence of memory deficits no longer rules out a diagnosis of bvFTD, current diagnostic criteria for the disease still advocate a relative sparing of memory function compared to executive function [1]. Biologically, pathological involvement of the ventromedial prefrontal cortex may account in part for the memory deficits observed in bvFTD [28].
Importantly, the severity of the episodic memory deficits in the bvFTD group was commensurate with that found on tasks of executive function, and appeared to progress at a faster rate than in the AD group. These findings confirm that presence of memory deficits is not specific to AD and that performance on executive function tasks appear more helpful in distinguishing bvFTD from AD [29, 30]. Furthermore, as the regression analyses suggest, memory deterioration occurs independently from executive dysfunction, a finding which is in line with the pathological involvement of the mesial temporal lobe in bvFTD [30]. The progressively diverging pattern of pathological brain changes and rate of decline on cognitive tasks over time suggest that the neuropsychological profiles associated with these disorders would become more distinct as the disease progresses. Clinically, however, the distinction between these two clinical entities based solely on neuropsychological profiles remains challenging, even when longitudinal data points are available.
These results need to be interpreted with some caution. First, the relative small sample size and variable lengths of follow-ups may have limited the detection of small group differences. The use of mixed-model regressions, however, is a powerful tool that accounts for attrition and irregularities in follow-up visits. The robustness of our results is increased further by the inclusion of patients with at least two annual visits. Multiple neuropsychological testing data provide a robust estimate of the change over time on the tasks under consideration. Second, it may be argued that these results may have been biased by the inclusion of patients incorrectly diagnosed. Ascertainment of brain pathology in life in dementia remains difficult, even in highly-specialized dementia clinics [31]. Presence of individuals with brain pathologies inconsistent with the clinical diagnosis cannot therefore be entirely ruled out. In this study, however, the subset of bvFTD and AD patients who underwent PiB-PET scan all had results that were consistent with their clinical diagnosis. In other words, all AD patients revealed positive PiB retention, denoting abnormal amyloid-β protein deposition in the brain, one of the pathological hallmarks of AD pathology, while none of the bvFTD patients did. This finding provides us with increased confidence that the clinical diagnosis reflected the underlying pathological process. Finally, the similar progression of decline on memory tasks between the two groups may also in part be accounted for by a floor effect on some of the tasks. For example, score change over time on the Rey Figure recall was limited in the AD group because of the low baseline score. As such, memory tasks that have a large score range but with less sensitivity are needed to map progression of severity of memory deficits in this group.
Our findings highlight the importance of comprehensive cognitive and behavioral evaluations at multiple time points. Indeed, the combination of steeper deterioration of executive function andfunctional status over time, accompanied by the presence of changes in other domains, such as behavior or social cognition, are important features that will orient to a diagnosis of bvFTD [1], whereas the progressive impairment in memory seems not to contribute to distinguish bvFTD from AD. In addition, our results have important implications for clinical trials. They provide valuable information regarding the annual rate of change in these dementia groups, information that may help establish the impact of novel treatments in future drug trials.
Footnotes
ACKNOWLEDGMENTS
The authors are grateful to the patients and their families for their continued support of our research.
This work was supported by funding to ForeFront, a collaborative research group dedicated to the study of frontotemporal dementia and motor neuron disease, from the National Health and Medical Research Council (NHMRC) (APP1037746) and the Australian Research Council (ARC) Centre of Excellence in Cognition and its Disorders Memory Program (CE11000102), NHMRC project grants 630489 and 510106, and ARC Discovery Project grant DP1093279. CEL is supported by DVC Sydney University Postdoctoral Fellowship, (S0716 U2644). OP is supported by an NHMRC Senior Research Fellowship (APP1103258).
