Approximately 30% of patients with amyotrophic lateral sclerosis (ALS) have cognitive impairment and 8%–14% fulfil the criteria for behavioural variant frontotemporal dementia (bv-FTD). The cognitive profiles of ALS and bv-FTD have been reported to be comparable, but this has never been systematically investigated. We aimed to determine the cognitive profile of bv-FTD and examine its similarities with that of ALS, to provide evidence for the existence of a cognitive disease continuum encompassing bv-FTD and ALS. We therefore systematically reviewed neuropsychological studies on bv-FTD patients and healthy volunteers. Neuropsychological tests were divided in 10 cognitive domains and effect sizes were calculated for all domains and compared with the cognitive profile of ALS by means of a visual comparison and a Pearson’s r correlation coefficient. We included 120 studies, totalling 2425 bv-FTD patients and 2798 healthy controls. All cognitive domains showed substantial effect sizes, indicating cognitive impairment in bv-FTD patients compared to healthy controls. The cognitive domains with the largest effect sizes were social cognition, verbal memory and fluency (1.77–1.53). The cognitive profiles of bv-FTD and ALS (10 cognitive domains, 1287 patients) showed similarities on visual comparison and a moderate correlation 0.58 (p=0.13). When social cognition, verbal memory, fluency, executive functions, language and visuoperception were considered, i.e. the cognitive profile of ALS, Pearson’s r was 0.73 (p=0.09), which raised to 0.92 (p=0.03), when language was excluded in this systematic analysis of patients with a non-language subtype of FTD. The cognitive profile of bv-FTD consists of deficits in social cognition, verbal memory, fluency and executive functions and shows similarities with the cognitive profile of ALS. These findings support a cognitive continuum encompassing ALS and bv-FTD.
- frontotemporal dementia
- motor neuron disease
Statistics from Altmetric.com
Eight to 14 per cent of patients with amyotrophic lateral sclerosis (ALS) have concurrent behavioural variant frontotemporal dementia (bv-FTD) and up to 30 per cent of patients with ALS are reported to have mild to moderate cognitive impairment.1–3 The mild cognitive deficits of ALS are attributed to frontotemporal dysfunction, because they include executive, memory and language deficits and impairment of social cognition.1 4–6 These findings, in combination with brain imaging abnormalities and pathological changes in the frontotemporal lobes of patients with ALS and pathogenic mutations associated with both diseases, have contributed to the hypothesis that ALS and (bv-)FTD are part of a disease continuum.6 7
The existence of this disease continuum would be strengthened by observations that mild cognitive deficits progress to bv-FTD in a proportion of patients with ALS. Although progressive paralysis complicates the performance of longitudinal studies in ALS, existing neuropsychological data, although sparse and including a limited number of patients with ALS, have failed to show progression of cognitive deficits.8 9 Another observation that questions the presence of a disease continuum between ALS and bv-FTD on the basis of cognition, is a discrepancy between the cognitive criterion of the international consensus criteria of bv-FTD, and the cognitive profile of ALS. The bv-FTD diagnostic criteria include relative sparing of memory and visuospatial functions, while these two cognitive domains, in particular memory, are not spared in patients with ALS, according to recent studies.10–15
Many neuropsychological studies have been performed in patients with bv-FTD, suggesting a cognitive profile including executive and social cognition deficits, with relative sparing of memory and visuospatial skills.10 A meta-analysis of these neuropsychological studies in bv-FTD could reveal new insights in the cognitive profile of bv-FTD, similar to our finding that the cognitive deficits of ALS are more extensive than previously thought.13 Second, such a meta-analysis in bv-FTD enables a comparison with the cognitive profile of ALS, which may increase our understanding of the nature and extent of brain involvement in these disorders.
Therefore, the aim of the current study was to define the cognitive profile of bv-FTD and to examine similarities with ALS. We performed a systematic review and meta-analysis of neuropsychological studies in patients with bv-FTD and healthy volunteers, and compared the results to our previous meta-analysis of the cognitive profile of patients with ALS.13
Medline (1966–2017), Embase (1970–2017) and PsycInfo (1970–2017) were searched on November 29th 2017 for neuropsychological studies of patients with bv-FTD, written in English, German, French or Dutch. Two authors screened all abstracts and evaluated full-text articles (EB and MKT for the original search in 2014, EB and RG for the updated search). References of articles were also screened and considered for inclusion. Keywords included frontotemporal dementia (FTD) and its synonyms, cognition and a selection of neuropsychological tests (online supplementary table e-1). ALS and its synonyms were also included as keywords, due to its association with FTD.
Supplementary file 1
All studies had to meet the following criteria:
Inclusion of a control group of healthy subjects.
Patients had to be diagnosed with bv-FTD according to validated clinical criteria (Neary or Rascovsky criteria).10 16 Patients with other forms of frontotemporal lobar degeneration (FTLD) were excluded (ie, patients with semantic dementia or progressive non-fluent aphasia). When these patients were also studied, data of patients with bv-FTD had to be reported separately, in order to allow extraction of their data. When patients with ALS or ALS-FTD were included, their data had to be reported separately.
Age-matched and education-matched controls had to be included or, when controls were not matched, norm scores corrected for age and education had to be used (ie, T-score, z-score).
Data of at least one validated neuropsychological test had to be reported. When studies had only performed the mini mental state examination (MMSE), dementia screening tests and/or IQ tests, they were not included. In order to allow conversion to effect sizes, the mean and SD of raw test scores, T-scores or z-scores for patients and controls had to be reported.
Studies had to report unique cohorts. If papers had reported data of the same cohort, the study with the largest sample was included. With the exception of patients with other forms of FTLD, there were no other exclusion criteria.
We extracted the following demographic and clinical variables from the articles: age (years), disease duration (months), educational level (years of formal education), disease severity (measured with validated dementia rating scales or, when these were not available, with results of dementia screening tests), use of psychoactive medication and depressive/anxiety symptoms.
We categorised all neuropsychological tests in 10 cognitive domains: social cognition, verbal memory, fluency, executive functions, visual memory, language, attention, visuoperception, psychomotor speed and visuoconstruction. A description of each cognitive domain and the included neuropsychological tests can be found in onlinesupplementary table e-2. We extracted all the neuropsychological test scores, that is, the mean, SD and number of participants of both the patient and control groups. To reduce extraction errors, data extraction was performed by two authors and differences were resolved by consensus (EB, MKT).17
Subgroup analyses were performed to find explanations for moderate or substantial heterogeneity in the cognitive domains. The following demographic and clinical variables were assessed in the subgroup analyses, based on a median split: age (years), educational level (years of formal education), disease duration (years) and disease severity (Dementia Rating Scale; that is, Clinical Dementia Rating Scale (CDR and CDR sum of boxes), Addenbrooke’s Cognitive Evaluation—revised (ACE-R) and the Mattis Dementia Rating Scale (MDRS)).
Demographic and clinical variables of the bv-FTD meta-analysis and ALS meta-analysis were summarised using simple descriptive statistics. When a study had presented longitudinal data, only data from the first visit was used for analysis. Effect sizes were expressed as Hedges’ g, that is, mean difference between patients with FTD and healthy controls, divided by the pooled SD and calculated in Review Manager.18 First, we divided all cognitive tests per study in the before-mentioned cognitive domains. Then, we calculated the effect sizes for all cognitive domains per study. When studies had reported data of more than one test within a specific cognitive domain, an averaged effect size was calculated, resulting in only one effect size per study in each cognitive domain. A random effects model was used to pool all effect sizes in each cognitive domain. A positive effect size indicated impairment of patients with FTD, compared with healthy controls. Effect sizes were considered small when <0.5, moderate when between 0.5 and 0.8, and large when >0.8.19 Statistical heterogeneity among studies was assessed with the Cochran’s Q test (χ2) and the I2 statistic. The χ2 test measures if differences in the results are compatible with chance alone. The I2 statistic is an estimate of the percentage of the variation across studies that is due to heterogeneity rather than chance. I2 cut-off points for low, moderate and substantial heterogeneity were 25%, 50% and 75%, respectively.20
To assess the impairment of patients with bv-FTD on specific neuropsychological tests and to explore heterogeneity, effect sizes of individual tests were calculated. This was performed for all tests that were administered in three or more studies. Statistical uncertainty was expressed in a 95% CI. We used IBM Statistics SPSS (V.20) and Review Manager (V.5.3) for the statistical analyses.18
Comparison of the cognitive profiles of bv-FTD and ALS
The effect sizes of the bv-FTD meta-analysis were compared with the effect sizes of our updated meta-analysis of the cognitive profile of ALS, encompassing 1287 patients and 10 cognitive domains.13 We expected to find some differences between the effect sizes of the bv-FTD and ALS meta-analyses. Due to exclusion of patients with dementia in the ALS meta-analysis, we expected the effect sizes of the bv-FTD meta-analysis to be systematically larger. However, we assumed that the effect size of language would be relatively larger in the ALS meta-analysis, because in the current meta-analysis on bv-FTD, patients with language variants of FTLD were excluded.
We compared the bv-FTD and ALS effect sizes in two different manners: a visual comparison and a comparison using correlation coefficients.
First, we depicted the effect sizes for each cognitive domain of both disorders into a graph and visually compared the sizes and directions of the effect sizes between bv-FTD and ALS. If the cognitive profiles would be similar, a comparable pattern would emerge: for example a higher effect size for the executive domain compared with the verbal memory domain in both disorders.
Second, we used correlation coefficients to assess the association between the effect sizes of bv-FTD and ALS. We calculated three Pearson’s rs; the first included all cognitive domains except the domains with bias due to motor impairment (ie, psychomotor speed and visuoconstruction); the second included all, so-called ALS-specific cognitive domains and the third included all ALS-specific cognitive domains except language. The latter was done because we expected a difference between the bv-FTD and ALS effect size of this domain, given our exclusion criterion (patients with other forms of FTLD, that is, semantic dementia and progressive non-fluent aphasia were excluded). The ALS-specific domains were chosen based on recent literature of neuropsychological deficits in patients with ALS.13 21 Accordingly, the domains social cognition, verbal memory, fluency, executive functions, language and visuoperception were considered ALS-specific. The latter domain is included in the ALS-specific domains because it is typically spared in ALS.13
Risk of bias
We assessed the quality of included articles with the Newcastle-Ottawa Quality Assessment Scale (maximum score 8 points, higher scores indicating better quality).13 22
We investigated the probability of a relevant publication bias in all cognitive domains with the fail-safe N, which was calculated with Rosenberg’s fail-safe number calculator.23 24 This represents the number of studies without an effect that would need to be added to the meta-analysis to obtain a non-significant pooled effect size, that is, P>0.05. We also calculated the tolerance level, as a benchmark to assess the observed fail-safe N (5 × number of studies + 10). If the fail-safe N is large in comparison to the tolerance level, the observed result is considered to be a reliable estimate of the real effect.
The literature search resulted in 9862 hits, of which 848 articles underwent full-text evaluation (figure 1). One hundred and twenty articles were included in the meta-analysis and their characteristics are shown in online supplementary table e-3. The reasons for exclusion of the remaining 728 articles can be found in the supplementary material (online supplementary table e-4).
A total of 2425 patients with bv-FTD (64.6% male) and 2798 healthy volunteers were included in the meta-analysis. The mean (SD) age and educational level of the patients was 64.1 (3.4) years and 12.8 (2.7) years, respectively. Disease duration was reported in a limited number of studies (n=62, 51.2%), with a median (range) of 4.2 (1.0–9.2) years. Disease severity, measured with the CDR and CDR sum of boxes, ACE-R and MDRS, was reported in 65 studies (54.2%) and was moderate in most patients (online supplementary table e-3).25–27 The patients included in the bv-FTD and ALS meta-analyses had approximately the same mean age, educational level and sex distribution (table 1). The median MMSE was lower in the bv-FTD meta-analysis (25.3, range 19.6–28.3) compared with the ALS meta-analysis (28.0, range 25.7–29.0).
Effect sizes of cognitive domains
All cognitive domains showed substantial pooled effect sizes, that is, impairment of patients with bv-FTD compared with healthy volunteers. The three largest effect sizes were found for the domains social cognition (1.77), verbal memory (1.67) and fluency (1.53)(figure 2 and table 2).
For all cognitive domains pooled effect sizes of individual tests within the domains were calculated (table 2). The three largest effect sizes were found for the recognition memory test—words (2.31; verbal memory), Hayling test (2.19; executive functions), the awareness of social inference test (1.96; social cognition) and Faux pas test (1.96; social cognition).
There was moderate to substantial heterogeneity in the domains visuoconstruction, psychomotor speed, visual memory, executive functions, attention, verbal memory and social cognition (I2=51%–83%). Low to moderate heterogeneity was found for the domains language and fluency (I2=40%–47%). Relatively low heterogeneity was found in the domain visuoperception (I2=10%).
Subgroup analyses were performed for age, disease duration, educational level and disease severity (according to two validated dementia ratings scales, or, when these were not available, with results of dementia screening tests) and are shown in online supplementary tables e-5–e-8. There was little influence of age, disease duration and education on any of the cognitive domains. Four analyses showed significant differences; more language deficits in younger patients, more attention deficits and less visuospatial deficits in patients with a lower educational level and more visuoconstruction deficits in patients with a shorter disease duration. As expected, the subgroup analysis for disease severity showed that patients with a more advanced disease had more cognitive deficits, although the only statistically significant difference was found in the verbal memory domain.
Comparison of the cognitive profiles of bv-FTD and ALS
First, we visually compared the effect sizes of the bv-FTD and ALS meta-analyses, as shown in figure 2. As expected, the effect sizes of most cognitive domains and relevant neuropsychological tests were smaller in patients with ALS than in patients with bv-FTD. In both disorders social cognition and fluency showed high effect sizes and visuoperception showed the lowest effect size. The CI of the effect sizes of visuoconstruction in bv-FTD overlapped with that of ALS, indicating that the patients with ALS were as impaired on these domains as the patients with bv-FTD. The CIs of psychomotor speed in bv-FTD and ALS almost overlapped, indicating that both patient groups are similarly impaired in this domain. When the differences between the remaining effect sizes for each domain are considered (upper eight effect sizes in figure 2), the difference between the effect sizes of language between bv-FTD and ALS is smaller, compared with that of other domains.
Second, Pearson’s r correlation coefficient calculated for all cognitive domains without bias due to motor impairment was 0.58 (P=0.13). We also calculated Pearson’s r correlation coefficient for ALS-specific cognitive domains (social cognition, verbal memory, fluency, executive functions, language and visuoperception), which was 0.73 (P=0.09). Pearson’s r of the ALS-specific domains without language was 0.92 (P=0.03).
Risk of bias
The quality assessment with the Newcastle-Ottawa Scale showed scores between 4 and 7 points (maximum score 9). In all but one study, patients and healthy volunteers were matched for age and education. The study without matching of patients and healthy volunteers used standardised scores. In all cognitive domains, the risk of a publication bias was low, that is, the fail-safe N was at least two times higher than the tolerance level, which indicates that the presence of a publication bias is unlikely to have influenced the results.
The current meta-analysis was performed to define the cognitive profile of bv-FTD and to examine similarities with that of ALS. We chose to exclude language variants of FTLD to reduce heterogeneity and because they occur less often in combination with ALS.6 The cognitive profile of bv-FTD consists of deficits in social cognition, fluency and other executive functions and also includes verbal memory impairment, together with a relatively spared visuoperception. A comparison of the cognitive profiles of bv-FTD and ALS shows similarities, notwithstanding cognitive deficits are more severe in patients with bv-FTD in most domains. In both disorders, considerable impairment in social cognition, fluency and verbal memory was found, whereas impairment of visual memory and attention was less prominent with visuoperception showing the lowest level of impairment. The domains psychomotor speed and visuoconstructive functions behave in a different manner in bv-FTD and ALS, which can be partly explained by physical impairment resulting in overestimation of psychomotor speed and visuoconstructive function deficits in patients with ALS.
Social cognition in bv-FTD
The domain social cognition showed a large effect size in bv-FTD and is thus an important component of the cognitive profile of bv-FTD. Social cognition is a complex entity including theory of mind, which is the ability to infer mental states of others, such as their thoughts, beliefs and intentions.28 Social cognition has a complex anatomical basis, including the frontostriatal network, temporo-occipital tracts, ventromedial prefrontal cortex and the amygdala, and therefore tests within this domain are highly mentally demanding.29 The faux pas test, for example, consists of stories of approximately six to eight sentences and requires adequate attention and working memory (closely related to executive functioning).30 In ALS, the presence of an association between deficits in social cognition and executive dysfunction is unclear: a negative study (which specifically addressed this issue) is refuted by a recent meta-analysis of social cognition in ALS that showed an association between deficits in social cognition and executive functions.5 31
In conclusion, social cognition involves high-level cognitive functioning and can be easily disrupted in bv-FTD and ALS. One could argue that social cognition deficits are therefore likely to occur in early stages of bv-FTD and ALS and may show progression over time.32 This should be investigated in a longitudinal study including only patients with early stage bv-FTD and ALS. If confirmed, this could attribute to the early detection of cognitive impairment in patients with ALS.29
Memory in bv-FTD
According to the consensus criteria for bv-FTD, episodic memory is relatively spared in bv-FTD.10 Our findings of a large effect size for the domain verbal memory suggests that memory, in particular verbal memory, is less spared in bv-FTD than previously thought. Previous findings of a similar degree of episodic memory deficits in patients with bv-FTD and patients with Alzheimer’s disease, and of severe memory deficits in amnestic variants of FTD, in combination with the findings of the current review, corroborate the thought that memory deficits are part of the cognitive profile of bv-FTD.33–36 Our data do not allow for a discussion of brain correlates of memory deficits in bv-FTD and ALS—that is, whether memory deficits are predominantly associated with executive deficits related to frontal lobe dysfunction, or based on encoding difficulties related to hippocampal dysfunction. Nevertheless, the assumption that memory changes in bv-FTD (and ALS) are primarily based on executive deficits is challenged by MRI studies in patients with early stage bv-FTD and ALS showing atrophy of the hippocampus, and pathological findings of inclusions in the dentate gyrus of the hippocampus in patients with ALS and bv-FTD.11 37–40
There was considerable statistical heterogeneity in most cognitive domains. This can be partly explained by differences in study populations as a result of clinical variability—reflected, for example, by the results of the subgroup analysis ‘disease severity.’
The clinical diagnosis bv-FTD, according to consensus criteria, is based on the presence of abnormalities in three out of six domains of behavioural/cognitive symptoms. The presence of cognitive symptoms constitutes one of these six domains. Consequently, cognitive symptoms do not have be present in order to diagnose bv-FTD, which has probably contributed to statistical heterogeneity in our study.10 This is further exemplified by wide CIs of some cognitive domains of patients with bv-FTD (eg, social cognition, verbal memory and visual memory).
Besides clinical variability, other explanations for statistical heterogeneity in our study include the grouping of various neuropsychological tests into one domain. The classification of neuropsychological tests in different domains can be argued, as tests are never process-pure. We aimed for a functional classification, by using a priori theory and convention to group the neuropsychological tests.41 Furthermore, we grouped neuropsychological tests in agreement with other meta-analyses on cognitive impairment, and our own previous ALS meta-analysis, in order to enable a comparison between the bv-FTD and ALS meta-analyses.42–44
Nevertheless, different sensitivities to pathology across tests can lead to diverging effect sizes of tests within a domain, and accordingly, a higher level of statistical heterogeneity. See, for example, different effect sizes in the social cognition domain of the Ekman test (Hedges’ g=1.89; CI 1.46 to 2.51) and theory of mind (Hedges’ g=0.85; CI 0.41 to 1.29).
Comparison of the cognitive profiles of bv-FTD and ALS
Our study shows similarities between the cognitive profiles of bv-FTD and ALS. The domains social cognition, verbal memory and fluency are most impaired in both diseases, with relative sparing of visuoperception. We also showed less severe language deficits in patients with bv-FTD, compared with, for example, impairments in social cognition and verbal memory. In the ALS meta-analysis, we have shown that language impairment occurred frequently with one of the largest effect sizes. This difference between bv-FTD and ALS can be explained by exclusion of language variants of FTD in the current meta-analysis. Although it is well known that a minority of patients with ALS develops semantic dementia or progressive non-fluent aphasia without behavioural changes, in this study we have chosen to include patients with the behavioural subtype of FTLD only.45–47 We felt that inclusion of the language subtypes of FTLD would have led to a high level of heterogeneity (in particular regarding the language domain) and difficulties in interpreting the results. Furthermore the incidence of bv-FTD in combination with ALS is higher than that of the language subtypes of FTLD.48
The presence of a cognitive disease continuum between bv-FTD and ALS could be corroborated when trajectories of cognitive deterioration are similar between the disorders. A relatively low number of longitudinal studies, however, prevented a meta-analysis of the course of cognitive decline in ALS. A limited number of longitudinal neuropsychological studies in ALS so far, failed to show a consistent pattern of progression of cognitive deficits—longitudinal studies on behavioural changes in ALS are even more scarce. These kind of studies in patients with ALS are hampered by loss to follow-up due to progression of motor symptoms, including nocturnal hypoventilation or daytime hypercapnia due to respiratory insufficiency.49–51 The latter complicates the interpretation of the nature of cognitive and behavioural changes in end-stage ALS. A future study in patients with early onset ALS (ie, disease duration <1 year) with longitudinal (eg, home-based) assessment of cognitive, behavioural and respiratory functions in late-stage ALS, may add relevant new information to the ongoing debate on the progression of cognitive deficits in ALS and the cognitive disease continuum between ALS and bv-FTD.
In addition to its strengths, that is, a systematic analysis of a large number of studies, data extraction by two authors and a well-defined patient group, the current meta-analysis has some limitations, of which considerable heterogeneity in most cognitive domains has been described above.
The data do not allow for meta-analysing subgroups based on biomarker, genetic or pathological findings: in 8 (out of 120) included studies, the authors included and/or excluded (a proportion of) participants based on CSF findings (Aβ42, τ and p-τ, n=7) or amyloid PET (n=3) results. Pathogenic mutations (MAPT, C9orf72 and PGRN) were reported in five articles. Therefore, it is not possible to draw conclusions about the cognitive profile of bv-FTD and ALS subtypes based on these criteria.
Furthermore, some of the included studies were performed by the same research group. In order to prevent including the same patient group twice, we carefully checked the method sections of different papers by the same research group, in particular when longitudinal data were presented. Also, demographic characteristics and results of neuropsychological tests batteries performed by the same research group were compared to exclude entirely similar cohorts. Although we cannot completely rule out the possibility that the same patient cohort was included twice, we are fairly confident that this is not the case. Also, because of the large number of participants and the robust results of the effect sizes and CIs, the inclusion of the same patient cohort would probably not significantly influence our results.
Finally, our findings do not exclude the possibility that the shared cognitive profile of ALS and bv-FTD is a reflection of a common cognitive profile of neurodegenerative diseases in general, because we did not include other neurodegenerative disorders. To further investigate this we recommend comparisons of cognitive profiles of bv-FTD and ALS with those of other neurodegenerative diseases, such as Parkinson’s disease and Huntington’s disease.
In conclusion, we have determined the cognitive profile of bv-FTD and have indicated that the cognitive profiles of bv-FTD and ALS show similarities. The findings support the existence of a cognitive continuum between bv-FTD and ALS and may direct cognitive testing of patients with ALS and bv-FTD. Based on our meta-analysis results of individual tests, a combination of neuropsychological tests could be proposed to be used in patient care and scientific studies of patients with bv-FTD and ALS. The following tests show relatively large effects sizes and low measures of heterogeneity within domains and, together, cover most of the cognitive changes in bv-FTD and ALS: Ekman test (social cognition), auditory verbal learning test (verbal memory), letter fluency (fluency), Hayling test (executive functions), Doors test (visual memory) and Boston Naming Test (language).
The authors thank René Spijker for his help with the literature search.
Contributors EB was involved in the study concept, literature search, analysis and interpretation of the data, statistical analysis, drafting/revising the manuscript for content and study coordination. JR was involved in the study concept, analysis and interpretation of the data and drafting/revising the manuscript for content. MKT was involved in the literature search, analysis and interpretation of the data and drafting/revising the manuscript for content. RG was involved in the literature search, analysis and interpretation of the data and drafting/revising the manuscript for content. YALP was involved in the analysis and interpretation of the data, statistical analysis and drafting/revising the manuscript for content. RJdH was involved in the analysis and interpretation of the data, statistical analysis and drafting/revising the manuscript for content. MdV was involved in the analysis and interpretation of the data, statistical analysis and drafting/revising the manuscript for content. BAS was involved in the analysis and interpretation of the data, drafting/revising the manuscript for content and study coordination.
Competing interests None declared.
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.