Objectives Frontotemporal dementia (FTD) is a heterogeneous clinical syndrome linked to diverse types of underlying neuropathology. Diagnosis is mainly based on clinical presentation and accurate prediction of underlying neuropathology remains difficult.
Methods We present a large cohort of patients with FTD spectrum diseases (n=84). All patients were thoroughly characterised by cerebrospinal fluid (CSF) Alzheimer’s disease (AD) biomarkers, neuroimaging, neuropsychological testing and standardised apraxia screening.
Results A potential AD pathology was found in 43% of patients with FTD. CSF AD biomarker levels positively correlated with AD-typical apraxia scores in patients with FTD. The discriminative power of apraxia test results indicative of AD pathology was high (sensitivity: 90%, specificity: 66%).
Conclusions Apraxia is common in neurodegenerative dementias but under-represented in clinical workup and diagnostic criteria. Standardised apraxia screening may serve as bedside test to objectify an AD-typical apraxia profile as an early and robust sign of AD pathology in patients with FTD.
- Alzheimer’s disease
- frontotemporal dementia
- primary progressive aphasia
- amyloid pathology
- default mode network
Statistics from Altmetric.com
- Alzheimer’s disease
- frontotemporal dementia
- primary progressive aphasia
- amyloid pathology
- default mode network
Frontotemporal dementia (FTD) is a heterogeneous clinical syndrome characterised by progressive deficits in behaviour, executive function and language.1 It represents the most common cause of young-onset neurodegenerative dementia following Alzheimer’s disease (AD).2 Its major clinical subtypes are the behavioural variant (bvFTD) and primary progressive aphasia (PPA). The latter is further subdivided into the non-fluent/agrammatic (nfPPA), semantic (svPPA) and logopenic (lpPPA) variants.3 4 Most cases of FTD are due to frontotemporal lobar degeneration (FTLD), while some cases may have underlying AD pathology.1 5 Clinicopathological correlation in patients with FTD is low: different pathologies cause similar clinical syndromes and vice versa, although rigorous syndromic classification on the basis of distinct clinical features and ancillary diagnostic information can predict the underlying pathology in some cases.6 7 The prediction of underlying pathology is important for appropriate patient counselling and therapeutic interventions. Nevertheless, the correlation of clinical syndromes with underlying pathology is further clouded by the presence of mixed pathologies.6 8–10 AD pathology has also been demonstrated in many patients with other types of neurodegenerative dementias. Although not representing the lead pathology in such cases, it has been shown to function as important modulator of their clinical course and presentation.11 12 Little is known about the contribution of concomitant AD pathology in patients with FTLD.8–10 13
Clinicopathological correlations in neurodegenerative dementias may be improved by the use of specific clinical features and molecular or imaging biomarkers.4 14 Recently, we have demonstrated that patients with typical AD and patients with bvFTD present with distinct forms of praxis impairment.15 Apraxia denotes the inability to perform specific predefined actions or to carry out learnt and purposeful movements due to acquired brain lesions.16 17 Importantly, apraxia occurs independently of sensorimotor or general cognitive deficits that could impair the comprehension of the task, the recognition of the stimulus or the implementation of the response.16 17 As such, apraxia screenings are among the few cognitive tests that can be applied to patients with diverse dementia syndromes with relative little confounding influence from comorbidities or impairment in other cognitive domains (eg, language or memory). In the present study, we performed comprehensive screening for praxis impairment in a large and well-characterised cohort of patients with FTD. By correlating apraxia scores with established cerebrospinal fluid (CSF) AD biomarkers, we demonstrate that specific types of praxis impairment are highly suggestive for underlying AD pathology, thus promoting standardised apraxia screenings as promising bedside tool for the prediction of AD pathology in patients with a clinical diagnosis of FTD.
Participants and study population
We screened all patients with early stage neurodegenerative diseases that presented to the Münster memory disorder unit between January 2012 and May 2017. Detailed inclusion and exclusion criteria are summarised in figure 1. We included all patients who received a clinical diagnosis of FTD (bvFTD, nfPPA, svPPA or lpPPA), according to current diagnostic criteria.3 4 Standard diagnostic workup for all patients included neurological examination, history taking, caregiver interviewing, extensive neuropsychological assessment and structural MRI. Sixty-one per cent of patients received additional fluorodeoxyglucose positron emission tomography (FDG-PET). Both MRI and/or PET imaging revealed no evidence for concurring aetiologies. Furthermore, only patients who had received both CSF biomarker analysis and standardised apraxia assessment (see below) were included. Exclusion criteria were advanced disease durations>5 years and relevant comorbidity (severe cerebral microangiopathy, stroke, psychiatric diseases, epilepsy, radiation therapy, inflammatory central nervous system diseases, motor disorders including parkinsonism, deficits in visual acuity). Additionally, we selected 68 patients with clinical presentation of early stage AD according to current diagnostic criteria and a CSF AD biomarker profile as reference cohort from the same patient population as detailed above.14 All patients with AD were characterised by the same standardised diagnostic workup as patients with FTD.
Apraxia assessment was performed during routine neuropsychological examination within 6 months after initial diagnosis by using the Cologne Apraxia Screening (CAS).18 The CAS consists of a total of 20 standardised items involving the praxis subdomains pantomime of object use (10 items), imitation of hand and finger postures (5 items) and imitation of face postures (5 items). Scoring was carried out by an experienced clinical neuropsychologist in accordance with the manual. Recently, we demonstrated that patients with typical amnestic AD and patients with bvFTD present with distinct apraxia profiles15: the former were shown to have pronounced deficits in imitation of hand and finger postures (CAS subscore 2.2), while imitation of face postures (CAS subscore 2.1) was spared. In contrast, patients with typical bvFTD had relatively preserved imitation of hand and finger postures but pronounced deficits in imitation of face postures.15 Therefore, we calculated a limb (CAS 2.2) minus face (CAS 2.1) functional score for differential diagnosis of AD versus FTD praxis impairment.
CSF biomarker assays
Amyloid beta (Aβ) and total Tau (T-tau) levels in the CSF were assessed by using the kits ‘Innotest β-amyloid(1-42)’ (Fujirebio) and ‘hTau total ELISA’ (Analytik Jena). The lab-internal cut-off value for both Aβ and T-tau was 500 pg/mL, with Aβ values below 500 pg/mL and T-tau values above 500 pg/mL being abnormal. A ‘CSF AD profile’ was defined as T-tau/Aβ ratio >0.52.19
Statistical analysis was performed using SPSS V.25. Figures were prepared using GraphPad Prism V.6. Demographics, disease severity variables, neuropsychological test results, CSF biomarkers and CAS scores were visually checked for outliers. Normality of distributions were tested by inspection of Q-Q plots and Kolmogorov-Smirnov test. Analysis of variance with post hoc Games-Howell tests was used in case of normally distributed variables. In case of non-normality, Kruskal-Wallis tests with post hoc Dunn’s tests were applied to test for between-group differences. For all post hoc comparisons, an alpha level corrected for the number of comparisons was applied. Praxis performance of patients with AD versus non-AD CSF profiles within each diagnostic category was contrasted using Mann-Whitney U tests. For correlation analyses of CSF biomarkers and apraxia scores, we computed non-parametric Spearman rank correlations. Optimal cut-off value, sensitivity and specificity for the CAS functional score to predict CSF biomarker profile was calculated using receiver operating characteristic (ROC) curves.
From a total of 673 patients who were screened according to the inclusion and exclusion criteria outlined in figure 1, we selected 84 patients with clinical diagnosis of possible or probable bvFTD or any of the three subtypes of PPAs (bvFTD n=43; nfPPA n=15; svPPA n=13; lpPPA n=13).3 4 Moreover, we included 68 patients with early stage AD as reference cohort, which fulfilled current diagnostic criteria for probable AD dementia with additional evidence of AD pathology based on CSF biomarkers.14 Demographics, clinical characteristics and neuropsychological test results of the selected patients are summarised in table 1. Age, years of education and disease duration did not differ between any of the patient groups. The total FTD cohort comprised 39 patients with young age of symptom onset (<65 years). There was a lower percentage of women among patients with bvFTD and svPPA, whereas the other clinical subgroups had an even sex distribution. As expected, patients of different clinical categories presented with significantly differing Mini-Mental State Examination (MMSE) scores (table 1) due to the well-known disproportionately high weighting of memory, verbal and visuoperceptual functions in the MMSE.
Assessment of the CSF AD biomarkers, Aβ and T-tau, revealed significantly reduced Aβ levels and significantly increased T-tau levels in AD compared with the FTD group (figure 2A). An increased ratio of T-tau to Aβ with >0.52 as cut-off value was previously shown to serve as robust biomarker signature indicating the presence of AD pathology, being more reliable than any single biomarker.19 As expected, the T-tau/Aβ ratio was significantly higher in our AD cohort compared with the FTD spectrum cohort (figure 2A). In the total FTD cohort, 36 patients (43%) exhibited a CSF AD biomarker profile, that is, a T-tau/Aβ ratio >0.52 (figure 2B). Subgroup analysis demonstrated a CSF AD biomarker profile in 14 (33%), 6 (40%), 6 (46%) and 10 (77%) patients with bvFTD, nfPPA, svPPA and lpPPA, respectively (figure 2B). Patients with FTD with late onset (≥65 years) had significantly lower CSF Aβ levels (difference: 372 pg/mL, p<0.0001), higher T-tau levels (difference: 235 pg/mL, p=0.003) and a higher T-tau/Aβ ratio (difference: 1.07 pg/mL, p=0.001) compared with those with young onset, suggesting a higher prevalence of comorbid AD pathology.
In line with previous work, the total CAS score revealed no differences between any of the dementia subgroups analysed in this study (figure 3A). In contrast, analysis of apraxia subdomain scores demonstrated that patients with FTD scored higher in imitation of hand and finger postures, and lower in imitation of face postures compared with patients with AD (figure 3B, C). Likewise, the CAS functional score (difference of limb minus face imitation) reached higher values in all four clinical FTD subgroups compared with the AD reference group (figure 3D). Multiple comparison analysis revealed that this difference was highly significant only when comparing AD and bvFTD (figure 3D).
Correlation of apraxia and CSF AD biomarker levels
Having assessed both the CSF biomarkers and the apraxia profiles of our dementia cohorts, we next asked whether the CAS functional score can predict the presence of a CSF biomarker profile suggestive of AD pathology in diverse groups of clinically diagnosed FTD cases. To this aim, we dichotomised each of the four FTD groups according to their CSF AD biomarker profiles (see figure 2B). Interestingly, we found that the CAS functional scores were lower (and negative) in the CSF AD pathology subgroups compared with the CSF non-AD pathology subgroups in all four FTD syndromes, thus indicating more severe deficits in limb compared with face apraxia in patients with a CSF AD profile (figure 4A). Mann-Whitney U tests comparing the CSF AD group with the CSF non-AD group in the total FTD cohort demonstrated high statistical significance (n=84, p<0.0001). The statistical difference remained highly significant when omitting the lpPPA subgroup from analysis (n=71, p<0.0001), and when analysing both the bvFTD group and the total PPA group separately (n=43 and n=41, respectively; both p<0.01; figure 4A). Mann-Whitney U tests comparing each of the three PPA subgroups separately did not reach statistical significance, most likely due to small sample sizes (figure 4A). In line with these results, correlation analysis indicated a highly significant linear relationship between the CSF T-tau/Aβ ratio and the CAS functional score (figure 4B; p<0.0001).
Importantly, the correlation of apraxia scores and CSF biomarkers was not affected by age. When analysing patients with young-onset and late-onset dementias separately, results were similar to those in the total group of patients with FTD: in both groups, patients with a CSF AD profile had significantly lower CAS functional scores than patients with a CSF non-AD profile (young-onset FTD n=39, CSF AD profile n=10, p=0.002; late-onset FTD n=45, CSF AD profile n=26, p=0.001). Additionally, the correlation between the CAS functional score and the T-tau/Aβ ratio remained highly significant when statistically controlling for age (partial Spearman’s r=−0.473, p<0.0001).
ROC curve analysis was performed to determine the diagnostic properties of the CAS functional score to detect the presence of AD pathology in each of the four FTD syndromes (figure 4C). The calculated areas under the curve alongside sensitivity and specificity values based on the CAS limb-face imitation difference using an optimal cut-off value of ≤0 for the presence of AD pathology (Youden index: 0.6) are reported in table 2. For all clinical subgroups the mean sensitivity of the CAS functional score to detect the presence of AD pathology was 89% with an average specificity of still 67%. Sensitivity was particularly high in the svPPA group (100%), while the specificity was lower in the nfPPA group (44%).
In summary, we present large cohorts of patients with early stage AD and distinct clinical variants of FTD that match current diagnostic criteria and are well characterised by CSF biomarker analysis, neuroimaging, detailed neuropsychological testing and apraxia screening. The findings of this study can be summarised as follows: (1) patients with typical AD have a distinct profile of praxis impairment compared with patients suffering from different clinical variants of FTD, with limb imitation being relatively more affected than facial praxis; (2) a relatively large fraction of patients with early stage FTD exhibit a CSF biomarker profile that is compatible with underlying AD pathology; (3) there is a significant positive correlation between AD-typical praxis impairment and an underlying AD pathology indicated by CSF AD biomarkers; and (4) apraxia screenings may thus be used as simple bedside test to predict a presence of AD pathology in patients with distinct clinical variants of FTD. Our results are of high clinical relevance as lumbar puncture is often not feasible, but accurate prediction of underlying pathology in dementia syndromes is crucial for optimal patient counselling and management.
The clinical and pathological heterogeneity of neurodegenerative dementias
Different types of neurodegenerative dementias can present with similar clinical syndromes and differential diagnosis remains difficult, despite the availability and refinement of clinical diagnostic criteria.3 4 14 In particular, the differential diagnosis of AD and FTD is challenging, especially in early disease stages.20 Even in elderly patients with pathologically confirmed FTLD, prominent memory deficits have been reported and seem to be rather common and easily misinterpreted as AD.21–25 The distinction between these two neurodegenerative spectrum disorders has been further obscured by CSF biomarker studies and neuropathological data, demonstrating significant amounts of AD pathology in up to one quarter of patients with clinical diagnosis of FTD.8–10 13 Moreover, inferring from other types of neurodegenerative proteinopathies, there is mounting evidence that distinct types of pathology (eg, Aβ, tau and α-synuclein pathology) act as significant contributors to dementia, exerting additive effects on cognitive dysfunction over the effect of single pathologies.11 12
In our cohort, the proportion of patients with an AD CSF biomarker signature is relatively high despite clearly fulfilling current clinical criteria for FTD variants and absence of MRI or FDG-PET evidence for concurring aetiologies. As expected, the proportion is highest in patients with a clinical diagnosis of lpPPA (77%), thus supporting the scarce neuropathological evidence that AD pathology is the most common pathological substrate of this clinical syndrome.26 27 Interestingly, the proportion is still as high as 37% in the three remaining FTD subtypes (bvFTD, nfPPA, svPPA), which may be related to the relatively large fraction of patients with late symptom onset in our FTD cohort. These three FTD syndromes are typically caused by either of the two distinct proteinopathies FTLD-tau or FTLD-TAR DNA binding protein (TDP). Although neuropathological studies have ascribed approximately 10% of bvFTD and nfPPA cases to atypical focal AD pathology, this proportion seems to be significantly higher in our cohort, which is based on CSF biomarker results alone. However, due to the lack of established positive molecular biomarkers of FTLD pathologies, we cannot distinguish between FTD cases with underlying FTLD and concomitant AD pathology, and cases in which AD pathology represents the main underlying aetiology. In fact, patients with a clinical diagnosis of FTD and CSF AD biomarker profile are likely a heterogeneous group of patients, including (A) patients with FTD due to AD pathology (eg, most patients with lpPPA); (B) patients with FTD due to FTLD but with significant amounts of additional AD pathology; (C) patients with FTD and preclinical AD; and (D) patients with atypical AD, in particular the relative poorly defined frontal variant, which may present similarly to FTD syndromes.22
Clinical implications and practical considerations
Our results propose screening for praxis impairment as suitable ancillary diagnostic test for the detection of AD pathology in patients with FTD. Apraxia screening tests, such as the CAS, are easy-to-apply, non-invasive, time-efficient, resource-efficient and cost-efficient tools that can be administered in less than 12 min, even in an outpatient setting. These practical considerations warrant apraxia screening as useful and attractive tool for the detection of AD pathology compared with the current gold standard, namely CSF analysis, which is not available in all clinical settings. Moreover, CSF biomarkers unfurl their full diagnostic strengths mainly in the setting of cognitive impairment due to AD versus non-degenerative aetiologies, while their discriminative power is limited when it comes to the differentiation of distinct forms of neurodegenerative dementias.28 Further studies and the use of improved apraxia tests tailored to the intricacies of neurodegenerative diseases instead of stroke will likely result in an improved discrimination of the diverse dementia syndromes.29 Indeed, due to the lack of items on emblematic gestures which require semantic action knowledge or items testing for apraxia of speech, the CAS may have a lower capacity to detect certain qualitative differences in praxis performance than other apraxia screenings. This may explain the diverging discriminative accuracy of the CAS functional score in patients with PPA (figures 3 and 4A).
Taken together, our results argue to include praxis assessment into the routine neuropsychological workup of early neurodegenerative diseases. Moreover, we propose the incorporation of standardised apraxia screening as an additional tool in clinical diagnostic criteria of different dementias. Interestingly, apraxia was referred to as a supportive feature for the diagnosis of AD in the original diagnostic criteria, commissioned by the National Institute of Neurological and Communicative Disorders and Stroke, and the AD and Related Disorders Association in 1984,30 but not in the revised version, charged by the National Institute on Aging and the Alzheimer’s Association in 2011.14 Thus, although apraxia is known to be an early and common symptom in a wide range of neurodegenerative dementias, standardised examination of apraxia is still neglected in routine diagnostic workup.31
The pathophysiology of apraxia in patients with AD pathology
Apraxia in patients with AD is typically characterised by pronounced deficits in limb imitation and to a lesser degree in object pantomime, whereas buccofacial praxis is usually not affected.15 29 32 33 In contrast to this characteristic apraxia profile, AD presents as heterogeneous clinical syndrome with diverse phenotypes that share the unifying molecular pathology of proteinaceous aggregates composed of Aβ or tau. According to the emerging network paradigm, dysfunction of specific, large-scale, cortical-subcortical networks underlies the pathogenesis of neurodegenerative diseases.34 35 It has been hypothesised that network disintegration arises from predictable interactions of pathogenic proteins with morphological and functional characteristics of specific neuronal networks.35 As such, AD pathology is preferentially distributed along network connections that are part of the cortical-subcortical, large-scale ‘default mode network’ (DMN), which is intrinsic to healthy brain function.35–37 The DMN is anatomically widely distributed.34 Depending on the actual site of DMN network disintegration, characteristic clinical features associated with distinct forms of AD may arise.37 Based on these results, we hypothesise that the AD-specific apraxia profile results from early and preferential dysfunction of the DMN. In line with this, we previously showed that patients with typical AD showed associations between limb apraxia and (medial) parietal grey matter atrophy.38 Medial parietal cortex integrity (especially of the precuneus) has been described as essential for higher visuospatial cognition, spatial representations of body parts and visual imagery, and is also a key hub within the DMN.39 Despite the striking clinical heterogeneity of variants of AD, involvement of this network is a consistent finding across the AD spectrum in pathological and neuroimaging studies which may explain similar apraxia profiles in patients with underlying AD pathology, despite otherwise diverse clinical presentations.35 37 In line with this, patients with FTLD pathology have been shown to present with alterations of different neural networks.35 40
Conclusion and future directions
In conclusion, our results demonstrate differences in apraxia scores between patients with AD and language and behavioural variants of FTD. A relatively larger impairment in limb imitation as compared with face imitation correlated with the presence of a CSF AD biomarker profile irrespective of the patient’s clinical presentation. Importantly, the diagnostic properties of the applied apraxia screening herald its clinical application as bedside tool for the prediction of the presence of AD pathology, irrespective of the clinical FTD subtype diagnosis and irrespective of whether AD pathology represents the main neuropathological substrate or acts as bystander pathology. Accurate prediction of the underlying pathologies in dementia syndromes is important for patient counselling and management. In particular, the decision of the therapeutic application of cholinesterase inhibitors in patients with dementia hinges partly on the presence or absence of AD pathology.
Additional studies to corroborate our findings are warranted. These should include (1) correlation with typical disease-related alterations in structural MRI, including voxel-based morphometry and microstructural diffusion tensor imaging; (2) inclusion of additional molecular biomarkers of AD and FTLD pathology; (3) correlation with resting-state functional MRI to test our hypothesis that an AD-typical profile of praxis impairment reflects disintegration of the DMN; (4) genetic test results for GRN, C9orf72 and MAPT mutations; and (5) postmortem neuropathological diagnosis. Finally, more sensitive apraxia tests would be desirable to provide higher discriminative properties and smaller overlap compared with the CAS screening, which was originally designed for the evaluation of praxis impairment in patients with left-hemispheric stroke. To this aim, we are planning to longitudinally assess a new cohort of patients with AD and FTD with our recently developed dementia apraxia test, which is custom tailored to apraxia assessments in neurodegenerative dementias.15 29
MP, TD and AJ contributed equally.
Contributors MP performed statistical analyses, created figures and tables, wrote and revised the manuscript. VJ screened clinical records and created first drafts of figures and tables. HW and SGM helped design the study and reviewed the manuscript for medical content. TD helped design and supervised the study, recruited and treated patients. AJ designed the study, conducted neuropsychological assessments, performed statistical analyses, supervised draft and revision of the manuscript.
Funding The authors have not declared a specific grant for this research from any funding agency in the public, commercial or not-for-profit sectors.
Competing interests The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest. MP, VJ and AJ have nothing to disclose. HW receives honoraria for acting as a member of Scientific Advisory Boards and as consultant for Biogen, Evgen, MedDay Pharmaceuticals, Merck Serono, Novartis, Roche Pharma and Sanofi-Genzyme, as well as speaker honoraria and travel support from Alexion, Biogen, Cognomed, F Hoffmann-La Roche, Gemeinnützige Hertie-Stiftung, Merck Serono, Novartis, Roche Pharma, Sanofi- Genzyme, TEVA and WebMD Global. HW is acting as a paid consultant for AbbVie, Actelion, Biogen, IGES, Novartis, Roche, Sanofi-Genzyme and the Swiss Multiple Sclerosis Society. His research is funded by the German Ministry for Education and Research (BMBF), Deutsche Forschungsgesellschaft (DFG), Else Kröner Fresenius Foundation, Fresenius Foundation, Hertie Foundation, NRW Ministry of Education and Research, Interdisciplinary Center for Clinical Studies (IZKF) Muenster and RE Children’s Foundation, Biogen, GlaxoSmithKline, Roche Pharma and Sanofi-Genzyme. SGM has received honoraria for lecturing, travel expenses for attending meetings and financial research support from Almirall, Bayer Health Care, Biogen, Diamed, Fresenius Medical Care, Genzyme, Merck Serono, Novartis, Novo Nordisk, ONO Pharma, Roche, Sanofi-Aventis and Teva. TD has received speaker honoraria, consultancy fees and travel expenses from Genzyme, Shire, Sanofi Aventis, Novartis, Actelion Pharmaceuticals and Amicus, research support from Genzyme, Shire, Amicus and Actelion Pharmaceuticals, and educational grants from Novartis, Roche and Biogen.
Patient consent Obtained.
Ethics approval Local ethics committee.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.