Functional decline in the aphasic variant of Alzheimer’s disease

Abstract

INTRODUCTION:

Primary progressive aphasia (PPA) is a clinical dementia syndrome associated with frontotemporal lobar degeneration (FTLD) or Alzheimer’s disease (AD). Impairment in activities of daily living is essential for dementia diagnosis, yet less is known about the neuropathologic impact on functional decline in PPA, especially over time.

METHODS:

Activities of Daily Living Questionnaire (ADLQ) ratings were compared by suspected underlying pathology between 17 PPA^Aβ+ and 11 PPA^Aβ- participants at 6-month intervals for 2 years using a linear mixed-effects model. A general linear model examined associations between functional decline and cortical thickness at baseline.

RESULTS:

Groups did not differ in demographics or aphasia severity at baseline, yet overall and subdomain scores of the ADLQ were significantly worse for PPA^Aβ+ compared to PPA^Aβ- (p=0.015) at each interval across 18-months.

DISCUSSION:

Functional decline appears more pronounced and disrupts more aspects of life activities for individuals with non-semantic PPA with suspected AD versus non–AD neuropathology.

b). Background

Primary Progressive Aphasia (PPA) is a clinical dementia syndrome characterized by initially isolated deficits in understanding or expressing language, which will decline over time¹. It is most commonly associated with frontotemporal lobar degeneration (FTLD; 60%) or Alzheimer’s disease (AD; 40%) neuropathology. There is known variability in language symptoms and the emergence of additional cognitive and functional decline over time², which is incompletely understood and difficult to predict at the individual level. Such ambiguity has made it difficult to set realistic expectations for the pace of disease progression, which consequently impacts the ability for families to develop appropriate care plans to navigate these challenges.

Initial evidence suggests atrophy is more widespread and clinical and anatomical progression is accelerated in those with non-semantic PPA with suspected AD versus FTLD neuropathology³, suggesting that the type of neuropathology may provide one contributor to understanding pace of decline and prognosis. These findings indicate that the presence of AD in PPA may be associated with disproportional decline in functional abilities, but this has yet to be explored in the context of suspected neuropathology. Instead, the presence of differential functional decline in PPA has been characterized through the lens of phenotypic subtypes and has yielded inconclusive results on whether decline is more rapid in one variant over another^2,4–13.

The current study focused on filling this gap by using a prospective longitudinal design to characterize and assess functional decline in non-semantic PPA with and without AD biomarkers (PPA^Aβ+ and PPA^Aβ-) at clinically relevant 6-month intervals over the course of 24 months. Functional decline was measured using the Activities Daily Living Questionnaire (ADLQ), which was designed to assess daily function in patients living with dementia ^14,15. Decline in functional activity was examined using the overall composite score and by the six aspects of daily life activities that comprise the composite: self-care, household care, employment and recreation, shopping and money, travel, and communication.

Given the more expansive pattern of atrophy reported in PPA with suspected AD pathology³ we expected that participants with AD biomarkers (PPA^Aβ+) would show greater impairment at baseline and over time in activities of daily living than those without AD biomarkers (PPA^Aβ-). Based on previous work in a clinical sample without biomarkers¹⁶, we anticipated that both groups would experience decline in functional activities beyond communication, especially over time.

c). Methods

Participants

Participants were enrolled into Northwestern University’s PPA Research Program for a prospective study examining disease progression at 6 month-intervals over the course of two years. The study was approved by Northwestern University’s Institutional Review Board and informed consent was obtained from each participant and their informant.

All participants were right-handed and met established diagnostic criteria^17–19 for PPA. Clinical diagnoses and subtypes were confirmed by a behavioral neurologist and author (MMM). Participants presenting with spared single word comprehension and impaired repetition and naming were classified as logopenic PPA (PPA-L) (8 PPA^Aβ+, 2 PPA^Aβ-). Participants presenting with a primary impairment of grammar, effortful speech, and relatively spared single word comprehension were classified as agrammatic PPA (PPA-G) (8 PPA^Aβ+, 9 PPA^Aβ-). One participant in the PPA ^Aβ+ group was classified as anomic^1,20, indicating that word-finding deficits were present and object knowledge, grammar and repetition were intact. Individuals with mild-to-moderate PPA, as determined by a Western Aphasia Battery Aphasia Quotient²¹ (WAB AQ) ≥ 65 at baseline were included in this study.

All participants were accompanied by an informant who was knowledgeable of the person living with PPA’s abilities and day-to-day activities. At each visit, informants completed the ADLQ¹⁶, a survey designed to assess the functional abilities of people living with dementia. The ADLQ is measured in percentages from 0–100 where higher percentages indicate greater impairment (0–33%= none to mild, 34–66%=moderate, >66%=severe). Due to its high test-retest and concurrent validity, its ability to parse focal decline¹⁶, and ease of use, it is the ideal measurement for daily function in PPA. 86% of informants were a spouse or partner, 11% an adult child, and 3% a close friend. Informants were consistent across all visits except in the case of two PPA^Aβ- participants who split visits between two friends and two adult children, respectively.

Neuropsychological Measures at Baseline

Participants completed neuropsychological testing and were rated by the Global Clinical Dementia Rating (CDR) and Language CDR.

The WAB AQ²¹ measured aphasia severity based on spontaneous speech, auditory comprehension, repetition and naming. The Boston Naming Test²² was used to assess naming ability. Grammatical production was measured using the 10 canonical and non-canonical questions from the Northwestern Anagram Test²³.

Non-language domains were evaluated at the initial visit by clinical assessment, informant report, and/or formal test scores including; the Three Words Three Shapes Test²⁴ and the Rivermead Behavioural Memory Test (RBMT)²⁵ for retentive memory; a modified version of the Visual Verbal Test²⁶ for mental flexibility²⁷; Mesulam’s Symbol Cancellation Test²⁸ for visuospatial function; and the Benton Facial Recognition Test²⁹ for visual object perception. In keeping with the diagnostic criteria for PPA¹⁹, each of these domains was found to be relatively spared.

Apraxia of speech (AoS) may emerge within the context of PPA. Speech samples from each participant were reviewed by a speech and language pathologist (LR) or neuropsychologist (SW) for auditory characteristics consistent with AoS. AoS was present in 5 PPA^Aβ- and 4 PPA^Aβ+.

MR and PET Imaging

Structural MR scans at the initial visit were obtained for all participants and used to identify peak atrophy patterns and anatomic associations with functional decline at baseline. Tl-weighted 3D MP-RAGE sequences (TR = 2300 ms, TE = 2.91 ms, TI = 900 ms, flip angle = 9°, FoV = 256 mm), recording 176 slices at a slice thickness of 1.0 mm, were acquired on a 3T Siemens TIM Trio using a 12-channel birdcage head coil.

MR images were processed using pipelines from FreeSurfer (v6.0.0)³⁰. Topological surface errors were manually corrected according to established guidelines³¹ in an iterative fashion until resolved. Cortical thickness estimates were calculated by measuring surface distances between representations of the white-gray and pial–cerebrospinal fluid boundaries.

¹⁸F-florbetapir amyloid PET imaging was acquired on all participants and used to categorize participants into suspected AD (PPA^Aβ+, n=17, 8 female) versus non-AD (PPA^Aβ-, n=11, 6 female), respectively. Amyloid PET imaging for the PPA group was performed on a Siemens Biograph 40 TruePoint/TrueV PET-CT system. A computed tomography scan was acquired for attenuation correction followed by a 20-min dynamic PET acquisition 50 minutes after administration of 370 MBq 18F-florbetapir.

A visual read of each ¹⁸F-florbetapir PET scan was completed by a radiologist to assess the uptake of the ligand as either elevated (Aβ+) or not (Aβ-)³². Briefly, individuals were considered Aβ+ if there was increased retention of the tracer in the cortical gray matter, defined as either loss of gray-white contrast in at least two cortical regions or, as intense, focal uptake in at least one cortical region. The PET scan was considered Aβ- when there was preserved gray/white contrast throughout the cortex or loss of gray/contrast was isolated to one region only. This method showed strong test-retest reliability³³, and in an autopsy study, the visual reads were slightly more reliable than the quantitative reads³⁴. Visual reads are also important because previous reports^35,36 suggest PPA^Aβ+ individuals can show focal, asymmetric uptake (left > right), which may not be detected by the standard volumes of interest developed and tested in dementia of the Alzheimer’s type due to AD³³. Of the entire sample, 13 were autopsy confirmed with concordant amyloid PET status (8 PPA^Aβ+, 5 PPA^Aβ-).

Statistical Analysis

All data analyses were generated using either SAS^TM37 software Version 9.4 or R Version 3.5³⁸. Welch t-tests were used to compare group characteristics (age at baseline, estimated age at symptom onset, sex, estimated symptom duration, education, and aphasia severity as determined by WAB AQ) at baseline. An ordered logistic regression model was fitted to compare Clinical Dementia Rating³⁹ scores between groups. A linear mixed-effect model was used to compare the two groups’ ADLQ ratings across all visits over time utilizing the terms 1) group, 2) time, and 3) group*time interaction. Secondary analysis included symptom duration and disease onset as covariates since these measures differed between the groups. No statistically significant differences were detected in demographics or language severity between the groups, limiting the need for other covariates. The PPA^Aβ- group completed all visits with the exception of two participants who missed the 18-month visit. 12 PPA^Aβ+ participants completed visits at all time points, two completed visits through 18 months, and three ceased participation after 12 months. Participant attrition was accounted for by the linear mixed models design.

Vertex-wise cortical thickness analyses using a general linear model were completed to identify peak atrophy in each PPA group relative to controls at the baseline visit. Relationships between whole-brain vertex-wise thickness measures and ADLQ scores at baseline were analyzed using a general linear model in FreeSurfer for the PPA^Aβ+ and PPA^Aβ- groups, controlling for participants’ aphasic severity.

d). Results

Group Demographics and Language Characterization

Table 1 summarizes the demographic and neuropsychological scores for the PPA^Aβ+ and PPA^Aβ- groups. All participants were non-Hispanic Caucasian and had completed at least 12 years of education. The PPA^Aβ- group reported a significantly older estimated age of symptom onset and had significantly fewer years of estimated symptom duration than the PPA^Aβ+ group. There was no significant difference among the 2 groups with respect to age at initial visit, years of education, sex, or WAB AQ at baseline. The groups did not differ significantly at baseline on language measures including the WAB-AQ, BNT and NAT (Table 2). The global CDR for both groups indicated minimal impairment, while the language CDR⁴⁰ showed mild functional deficits in that domain (1.0=mild impairment) at baseline.

Table 1.

Participant demographics, clinical characterization, and neuropsychological test performance. Groups did not differ in sex, age and education and differed significantly in age at onset and estimated symptom duration.

	PPAAβ-	PPAAβ+	P value
n (female)	11 (6)	17 (8)	-
Age at initial visit, years	70.0 ± 6.8 [61–82]	65.9 ± 5.8 [58–80]	0.123
Age at onset, years	66.6 ± 6.3 [58–79]	60.5 ± 5.5 [53–74]	0.012
Est. symptom duration, years	3.5 ± 1.5 [1.2–6.0]	5.4 ± 2.8 [1.4–10.9]	0.031
Education, years	14.8 ± 2.2 [12–19]	16.4 ± 2.4 [12–20]	0.098
Subtype  Logopenic  Agrammatic  Anomic	2 9 0	8 8 1

NOTE: Numbers are provided as means ± standard deviation and [Ranges]. Subtypes are displayed as counts.

Abbreviations: Est. symptom duration, years references the estimated symptom duration provided by participants at baseline.

Table 2.

Neuropsychological Language Test Scores and CDR at the Initial Visit.

	PPAAβ-	PPAAβ+	P value
WAB AQ (100)	84.5 ± 6.5	82.5 ± 9.0	0.513
BNT (60)	47.0 ± 11.9	40.5 ± 13.0b	0.192
NAT 10Q (10)	7.5 ± 2.5	7.1 ± 2.5	0.730
WabRep66 (66)	50.2 ± 9.7	41.7 ± 10.2	0.037
PPVT (36)	33.2 ± 2.6	33.8 ± 2.1	0.537
CDR Global (# 0/0.5/1)	7/4/0	4/11/2	0.015
CDR Language (# 0/0.5/1/2)	0/1/8/2	0/4/7/6	0.428

Abbreviations: WAB-AQ (Western Aphasia Battery-Aphasia Quotient), BNT (Boston Naming Test), NAT 10Q (10 canonical and non-canonical items on the Northwestern Anagram Test), WabRep66 (items 10–15 on the Western Aphasia Battery Repetition subtest), PPVT (a subset of moderately difficult items; 157–192, from the fourth edition of the Peabody Picture Vocabulary Test)

NOTE: Numbers are provided as means ± standard deviation except for CDR scores which are listed as counts of total participants by score.

^bOne PPA^Aβ+ participant discontinued the BNT at baseline due to frustration.

Functional Decline: Activities of Daily Living Questionnaire

Linear mixed models were used to estimate change in ADLQ severity over time in each group. On average, ADLQ ratings were significantly worse for PPA^Aβ+ compared to PPA^Aβ- at baseline and at each 6-month interval through the 18-month visit (Figure 1). The average rate of decline for the PPA^Aβ+ group, 4.96% every 6 months, was significantly more precipitous than the PPA^Aβ- (4.12%). While PPA^Aβ+ progressed to moderate functional impairment by the 18-month visit, PPA^Aβ- as a whole never reached moderate functional impairment throughout the 2-year follow-up period.

Figure 1.

Daily function is significantly more impaired for PPA^Aβ+. NOTE: Group averages of the overall ADLQ are featured with standard errors. Open circles indicate significance and closed circles indicate that analysis yielded no significant differences between groups at that time point. The PPA^Aβ+ group has a higher group mean at baseline and this difference remains throughout all visits.

Figure 2 summarizes the group means for the ADLQ subscales of Self-Care, Household, Employment and Recreation, Shopping and Money, Travel, and Communication across all time points. Mixed-effect modeling was used to compare the overall slopes on the ADLQ subscale scores between PPA^Aβ+ and PPA^Aβ- groups. The overall slopes were statistically different between PPA^Aβ+ and PPA^Aβ- for Household (p=0.014), Employment and Recreation (p=0.013), Shopping and Money (p<.001), and Travel (p<.001). To compare the slopes between PPA^Aβ+ and PPA^Aβ-, we fit another model with an interaction term between PPA^Aβ+/PPA^Aβ- and Visit (time). Only Self-Care (p=0.011), Employment and Recreation (p=0.031), and Shopping and Money (p=0.018) were statistically different, indicating that these abilities were significantly worse for PPA^Aβ+ compared to PPA^Aβ- across all visits. To address the potential effects of the significant differences of age of onset and symptom duration between groups (Table 1), 2 additional mixed-effect models were fit by including these variables as covariates. The conclusions of the longitudinal data analysis remain the same as the models without these covariates. Therefore, given the subjective nature of family- or self-report of symptom onset (and therefore also symptom duration) in neurodegenerative dementias, we chose to present the primary results based on statistical modeling without these covariates.

Figure 2.

PPA^AB+ shows unique patterns of decline across functional subdomains.

ADLQ subscales by group over time. Darker shades represent the PPA^Aβ- group and lighter shades represent the PPA^Aβ+ group. The most notable difference between the two groups is in the area of Shopping and Money (p<.0001) followed by Travel (p<.001), Employment and Recreation (p=0.013) and Household Care (p=0.014).

It should be noted that statistical modeling with both random intercepts and random slopes was attempted, but due to the limited sample size (n=28), the model stabilization was best in the mixed-effect model with random intercept alone, which is reported herein. Furthermore, utilizing a mixed-effect model allowed for a more robust review of data across all timepoints than other repeated measures models such as Generalized Estimating Equations^41,42 even though some datapoints were missing.

Cortical Thickness Analysis

Quantitative MRI morphology was available for all participants (17 PPA^Aβ+ and 11 PPA^Aβ-). Cortical thickness of each PPA group was statistically compared against 25 (11 female) previously described³ right-handed, similarly-aged and -educated healthy individuals [Average age: 67.8 years (range: 60–80), average education: 16.1 years (range: 11–20)]. Differences in cortical thickness were calculated by conducting a general linear model on every vertex along the cortical surface and adjusted for false discovery rate⁴³ at a threshold of 0.05. The cortical thickness maps for PPA^Aβ+ and PPA^Aβ- at baseline are displayed in Figure 3. Similar to our previous study³, the pattern of atrophy at baseline was more widespread in individuals with suspected AD (PPA^AB+), including notable right hemisphere atrophy, compared to those with non-AD neuropathology (PPA^AB-). While controlling for aphasia severity at baseline, separate vertex-wise correlations of cortical thickness and ADLQ scores showed no significant regional associations in the PPA^Aβ+ or PPA^Aβ- groups, whether groups were analyzed separately or combined.

Figure 3.

Baseline atrophy is more widespread in PPA^Aβ+.

Red/yellow indicates vertices with significant cortical thinning at baseline for the PPA^Aβ- and PPA^Aβ+ groups compared to healthy individuals.

NOTE: LH=Left hemisphere. RH=Right hemisphere.

e). Discussion

This study examined the pace and quality of functional decline in 28 participants with non-semantic PPA with either autopsy or biomarker evidence of Alzheimer’s disease pathology as evidenced by increased amyloidosis through PET biomarkers or autopsy confirmation. Although both groups experienced declines in their daily functioning over the course of 2 years, the PPA^Aβ+ group’s ADLQ ratings were significantly worse than PPA^Aβ- across all visits through the 18-month visit, despite groups being comparable in age, sex distribution, education and aphasia severity at baseline. These data suggest the type of proteinopathy may be an important factor for prognosis and setting patient expectations about the pace and severity of functional decline.

The higher ADLQ scores in the PPA^Aβ+ group (i.e., poorer function) holds implications for prognosis and care planning for individuals and their families. At baseline, despite maintaining similar levels of language competence, overall ADLQ scores and subdomain scores for Household Care, Employment and Recreation, Shopping and Money, and Travel were significantly worse at baseline for the PPA^Aβ+. These differences could be due in part to the more distributed pattern of atrophy seen in PPA^Aβ+3 potentially impacting other cognitive abilities and consequently functional ability. It makes sense that groups were not significantly different in the areas of communication (as they were matched for aphasia severity) and self-care (as people living with PPA have demonstrably lower levels of impairment in this area versus other dementia syndromes¹⁶).

The severity of impairment on the ADLQ did not show a linear relationship with cortical atrophy, perhaps because it is an aggregate measure of functional capacity which correlates with composite global measures of dementia severity (e.g., MMSE⁴⁴ and CDR³⁹)¹⁴. Longitudinal imaging studies examining temporal relationships with functional decline, perhaps by subdomain, may be helpful for disentangling specific anatomic associations.

In the current study, both the PPA^Aβ+ and PPA^Aβ- groups contained logopenic participants. In a pathological study of non-semantic PPA from our Center, of 58 autopsy cases, 45% had AD. Of those cases whose primary pathology was AD, 78% were PPA-L, indicating a strong, but not conclusive relationship between PPA-L and AD pathology⁴⁵. The current study showed a similar phenotypic distribution, reinforcing that phenotypic presentation is not sufficient to determine underlying pathology^3,46–48. This underscores the need for reliable and accessible in vivo neuropathologic biomarkers in neurodegenerative dementia syndromes.

The groups did not differ on their level of aphasia severity on the WAB-AQ, which is an accepted objective measurement, while age of onset, repetition, and global CDR differed between the groups. It should be noted that due to the insidious onset of PPA, symptom onset relies on subjective rather than objective report and is susceptible to bias and inconsistency. Should the participant choose the first moment they remember forgetting a word, or should they consider a date when language problems were more pervasive? When accounting for age of onset or symptom duration, conclusions did not differ, suggesting these factors were not primary drivers of the differences between the PPA groups. The differences in repetition between the groups is likely reflective of the larger percentage of PPA-L participants in the PPA^Aβ+ group, since repetition deficits are a common feature for this subtype¹⁹. Global CDR score takes into account activities of daily living (Personal Care, Home and Hobbies, and Community Affairs) and has been shown to correlate with ADLQ¹⁴; therefore, greater global CDR impairment in the PPA^AB+ group could be indicative of a greater level of functional impairment, which is consistent with our finding that the PPA^AB+ group was more significantly impacted in activities of daily living at baseline.

This study used ¹⁸F-florbetapir amyloid PET imaging to delineate between suspected AD and non-AD participants. Participants did not undergo a tau PET scan, which would account for presence of tau tangles, the other hallmark of AD. However, in persons scanned with florbetapir within 1 year of autopsy, sensitivity and specificity was high (≥96%)⁴⁹. There is a potential risk that despite having positive florbetapir scans, participants may have additional pathologies such as FTLD or TAR DNA-binding protein 43. However, the likelihood of a dual diagnosis is low since our previous report found only one such case out of 58 consecutive PPA autopsies⁴⁵. In this study, 13 of the 28 participants have come to autopsy and all neuropathological reports were concordant with their florbetapir status (8 PPA^Aβ+ and 5 PPA^Aβ-).

Determining the rate and quality of functional decline in PPA is relevant for understanding the neurobiology of disease, setting expectations for the family, adding appropriate allied health professionals to the care team (e.g., occupational therapists, speech and language pathologists, social workers), and creates new opportunities for the development of targeted interventions based on the areas strengths and challenges for the individual living with a diagnosis of PPA.

To our knowledge, this study is the first of its kind to examine deterioration in activities of daily living in non-semantic PPA with AD biomarkers. Functional decline differs in both severity and quality for individuals with PPA due to Alzheimer’s disease versus non–Alzheimer’s disease pathology and this difference remains for a period of time. These data suggest that despite shared clinical presentations, the type of proteinopathy is a relevant factor when considering future functional decline and care planning and stresses the importance of having reliable in vivo biomarkers.