Background and purpose Right-lateralised semantic dementia (right SD) and behavioural-variant frontotemporal dementia (bvFTD) appear clinically similar, despite different patterns of underlying brain changes. This study aimed to elucidate distinguishing clinical and cognitive features in right SD versus bvFTD, emphasising emotion processing and its associated neural correlates.
Methods 12 patients with right SD and 19 patients with bvFTD were recruited. Clinical features were documented. All patients were assessed on standardised neuropsychological tests and a facial emotion processing battery. Performance was compared to 20 age-matched and education-matched controls. Grey matter intensity was related to emotion processing performance using whole-brain voxel-based morphometry analysis.
Results Patients with right SD exhibited disproportionate language dysfunction, prosopagnosia and a suggestion of increased obsessive personality/behavioural changes versus patients with bvFTD. In contrast, patients with bvFTD demonstrated pronounced deficits in attention/working memory, increased apathy and greater executive dysfunction, compared to patients with right SD. Decreased empathy, disinhibition and diet changes were common to both dementia subtypes. Emotion processing deficits were present in both FTD syndromes but were associated with divergent patterns of brain atrophy. In right SD, emotion processing dysfunction was associated with predominantly right medial and lateral temporal integrity, compared to mainly left temporal, inferior frontal and orbitofrontal and right frontal gyrus integrity in bvFTD.
Conclusions This study demonstrates comparable deficits in facial emotion processing in right SD and bvFTD, in keeping with their similar clinical profiles. These deficits are attributable to divergent neural substrates in each patient group, namely, right lateralised regions in right SD, versus predominantly left lateralised regions in bvFTD.
- FRONTAL LOBE
Statistics from Altmetric.com
Behavioural and emotional disturbances are common in frontotemporal dementia (FTD), particularly the semantic dementia (SD) and behavioural-variant FTD (bvFTD) subtypes.1–10 Characterised by severe left anterior temporal atrophy, SD, also referred to as semantic-variant primary progressive aphasia (PPA), typically presents with a breakdown in word and object knowledge.11 Approximately 25% of SD cases, however, present atypically,12 with predominant right temporal lobe atrophy (right SD). In contrast to typical left SD cases, behavioural, emotional and interpersonal dysfunctions are prominent early symptoms in the right SD variant.1–8 These features are similar to those seen in bvFTD, which is associated with increased apathy and changes in personality and behaviour, together with a marked reduction in empathy.10 Furthermore, these features outweigh classic language difficulties,1 ,2 ,8 making clinical distinction from bvFTD difficult.8
Emotion processing dysfunction is well documented in right SD2 ,7 and bvFTD,13 however, direct comparisons between these syndromes are scant. Of relevance here, recent studies suggest that aspects of socioemotional processing are differentially affected in bvFTD compared with the non-fluent variant of PPA,14 suggesting that different profiles may also exist between bvFTD and right SD. Whether the observed emotion processing deficits in bvFTD and right SD are underpinned by pathological involvement of similar brain structures, remains unknown. While right anterior temporal and prefrontal cortical atrophy has been associated with socioemotional deficits in FTD,15–17 no study has directly contrasted the neural substrates of emotion processing difficulties in right SD versus bvFTD.
Here, we aimed to explore the (1) clinical and cognitive features; (2) emotion processing profile; and (3) neural correlates underlying emotion processing in right SD compared to bvFTD. We hypothesised that significant overlap in clinical features would be evident between the patient groups. Similarly, for emotion processing, we expected to find significant deficits in both patient groups, although we predicted that emotion processing deficits may be attributable to divergent patterns of brain atrophy according to FTD subtype. In bvFTD, frontal regions are initially involved, however with disease evolution the temporal lobes become increasingly affected, whereas in right SD the converse pattern is observed. We therefore hypothesised that frontal regions would be predominantly implicated in emotion processing in bvFTD, whereas temporal regions would play a greater role in right SD.
Thirty-one patients with dementia (19 bvFTD, 12 right SD) from a consecutive series of referrals, and 20 age-matched and education-matched controls were recruited through FRONTIER, the FTD research clinic located at Neuroscience Research Australia, Sydney. Ethical approval was obtained from the South Eastern Sydney Local Health District and the University of New South Wales ethics committees. Informed consent was acquired in accordance with the Declaration of Helsinki.
All participants had a structural 3T MRI at baseline with coronal images. Patients with bvFTD met International Consensus Criteria.10 In brief, individuals with bvFTD presented with behavioural disinhibition, reduced empathy, changes in appetite, perseverative behaviour and/or frontoexecutive dysfunction. Similar to previous methodology,8 patients with right SD were selected on the basis of a pattern of MRI changes typical of SD but with markedly greater right>left anterior temporal lobe atrophy on coronal sections on a 3T MRI as judged by the clinical team. The main presenting symptoms varied. Of the 12 right SD cases, 10 reported naming or word finding difficulties and fulfilled clinical diagnostic criteria for SD.11 Six presented with prosopagnosia as the principal symptom, although three cases had prosopagnosia in addition to other symptoms. No right SD cases presented with purely behavioural disturbance but 11 had behavioural changes and nine showed clinical features suggestive of possible bvFTD.10 For all participants, exclusion criteria included prior history of mental illness, significant head injury, movement disorders, cerebrovascular disease, alcohol/drug abuse, limited English proficiency and severe language disturbance that impeded sufficient understanding of tasks. The Frontotemporal Dementia Functional Rating Scale (FTDFRS18) evaluated functional decline. The Addenbrooke's Cognitive Examination, Revised (ACE-R19), assessed general cognitive functioning. Controls scored >88/100 on the ACE-R.
All patients and carers underwent a structured interview by the senior neurologist (JRH). This interview is designed to determine the presence/absence of clinical features including: memory impairment, language impairment, visuospatial difficulty, motor symptoms, as well as behavioural features such as disinhibition, apathy, loss of empathy, perseverative behaviour and neuropsychiatric features such as hallucinations and delusions. A neuropsychologist (JK) reviewed the clinical file of each patient and abstracted these clinical features from the structured interview as well as other available clinical records (eg, referral letters, neuropsychological reports). Clinical features were recorded as either present or absent concentrating particularly on features defined in recent International Consensus Criteria for bvFTD and for variants of PPA.10 ,11
Participants completed a neuropsychological battery of tests to assess: visual episodic memory recognition (Doors A, Doors and People test20) and recall (Rey Complex Figure Test (RCFT) 3 min delay21), basic attention/working memory (Digit Span Total, Wechsler Adult Intelligence Scale, Third Edition22), psychomotor speed and mental flexibility (Trail Making Test23), letter fluency (F,A,S24), language (object Naming and word Comprehension; Sydney Language Battery25) and visuoconstruction (RCFT Copy21). Face processing was assessed via two subtests from the Facial Affect and Identity Discrimination Task.26 In the Face-Perception task, participants view pairs of faces and must determine if the two images are identical or different (maximum score 40). In the Face-Matching task, participants view pairs of faces expressing different emotions and must determine if they belong to the same individual or not (maximum score 42). The Famous Faces task examines knowledge of well-known faces, whereby, over 20 trials, the participant points to the famous face (Recognition) and provides the corresponding name (Naming) of the famous face among three distractors.27
All participants were assessed on a comprehensive emotion processing battery that included:
Face-Emotion Matching Task: Participants were presented with pairs of faces of different individuals and determined whether the two faces expressed the same or different emotions (happy, sad, anger, fear, disgust, surprise and neutral) across 42 trials, following four initial practice trials.26
Emotion Selection Task: Participants were presented with seven different emotional faces (anger, disgust, fear, sadness, surprise, happiness, neutral) and pointed to the face that matched the emotion label that was provided orally, across 42 trials. Prior to the task, the participants completed a practice trial to ensure they understood the task.26
Ekman 60 task: Recognition of facial expressions across six basic emotions (anger, disgust, fear, happiness, sadness, surprise) was assessed using stimuli from the Pictures of Facial Affect series.28 Across 60 trials, stimuli were presented one at a time, pseudorandomly for 5 s, and participants selected the label that best described the emotional expression. Participants completed six practice trials before the task started.
Prior to each task, the researcher checked the participant could sufficiently understand the language demands of the task. Where any concerns about comprehension ability were raised, the participant was excluded from the study.
Behavioural statistical analyses
The suitability of variables for parametric analyses was determined using Kolmogorov-Smirnov tests and were found to be normally distributed. Between-group differences in categorical variables, including the presence or absence of clinical features, were assessed with χ2 tests. For continuous variables, analyses of variance with Sidak post hoc comparisons were used. To compare between patient groups, independent samples t tests were conducted. Unless otherwise reported, significance level was set at p<0.01 to correct for multiple comparisons.
Structural MRI scans were acquired using a 3T Phillips MRI scanner with standard quadrature head coil (eight channels). The 3D T1-weighted images were acquired using the following sequences: coronal orientation, matrix 256×256, 200 slices, 1×1 mm in-plane resolution, slice thickness 1 mm, echo time/repetition time=2.6/5.8 ms, flip angle α=19°. Scans were available for nine patients with right SD, 13 patients with bvFTD and 14 Controls who had completed the emotion processing battery.
MRI data was analysed using FSLVBM, a voxel-based morphometry (VBM) analysis using the FSL software package (http://www.fmrib.ox.ac.uk/fsl/fslvbm/index.html), V.22.214.171.124–31 First, tissue segmentation was carried out using FMRIB's Automatic Segmentation Tool from brain-extracted images. The resulting grey matter partial volumes were aligned to the Montreal Neurological Institute standard space (MNI152) using non-linear registration approach via FNIRT, which uses a b-spline representation of the registration warp field. A study-specific template was created and the native images were re-registered non-linearly.32 Modulation of the registered partial volume maps was carried out by dividing them by the Jacobian of the warp field. The modulated, segmented images were smoothed with an isotropic Gaussian kernel with a sigma of 3 mm.
A voxel-wise general linear model was applied and permutation-based non-parametric analyses were conducted with 5000 permutations per contrast. First, differences in grey matter intensity between bvFTD and right SD and controls were examined using t tests. Next, to investigate correlations between regions of grey matter atrophy and performance on emotion processing tasks, patient groups were combined with Controls. This procedure serves to increase the study's statistical power to detect brain–behaviour relationships across the entire brain by achieving greater variance in behavioural scores.33 ,34 One bvFTD and two right SD patients were excluded from covariate analyses as emotion processing scores were unavailable. A covariate only statistical model with a  t-contrast was used. For all analyses, a voxel-wise approach was applied with results reported at p<0.001, uncorrected for multiple comparisons and with age included as a nuisance variable. For the covariate, emotion processing analyses, a cluster-extent threshold of 50 contiguous voxels was applied.
Groups were matched for age, education, and sex (table 1). Patients demonstrated significantly poorer overall cognitive functioning than Controls (both p values<0.001), with patients with right SD more impaired than patients with bvFTD (p<0.001). Patients with bvFTD exhibited greater functional decline than patients with right SD (p=0.029), with no difference between patient groups for disease duration (p=0.925).
Overlap in clinical features between right SD and bvFTD was evident. Both patient groups displayed comparable levels of reduced empathy, disinhibition, diet changes and decline in self-care (all p values>0.3; figure 1).
Several clinical features, however, differentiated the two patient groups. Patients with right SD presented with greater language disturbance (comprehension/word-finding difficulties; p<0.001), prosopagnosia (p<0.001), and a trend toward greater frequency of obsessive, rigid and irritable behaviour and personality changes when compared to patients with bvFTD (p=0.044). In contrast, apathy/social withdrawal was more frequent in bvFTD than in right SD (p=0.005), with the suggestion of greater memory disturbance (p=0.024) and executive dysfunction (poor judgment/planning; p=0.052) in the bvFTD group.
Significant overlap in cognitive deficits was also evident between both FTD subtypes. Compared to controls, patient groups were impaired on Face-Matching, episodic memory recall (RCFT 3 min), and letter fluency (all p values<0.002), with patients with right SD and bvFTD displaying comparable disturbances across all these measures (all p values>0.3; table 1). Basic Face-Perception was preserved for both groups (both p values>0.05).
Specific cognitive features also differentiated between the two FTD subtypes. Patients with right SD demonstrated disproportionate difficulties with language (Naming), naming of familiar faces and visual recognition memory, relative to bvFTD patients (all p values<0.001). Further, familiar face recognition and word Comprehension was significantly impaired in right SD relative to bvFTD (both p values<0.001), with patients with bvFTD not differing significantly from Controls on these measures (both p values>0.07). In contrast, patients with bvFTD showed pronounced difficulties with attention/working memory and executive function (both p values<0.002), compared to patients with right SD who scored in line with Controls (p>0.7). Table 2 summarises the discriminating and overlapping clinical and cognitive features in right SD and bvFTD.
When determining if two faces expressed the same emotion (Face Emotion Matching), patients with bvFTD exhibited significant deficits relative to Controls (p=0.003), while patients with right SD performed similarly to Controls (p=0.512), with no difference between patient groups (p=0.496; figure 2). In contrast, when required to select a specific emotion from six distractor emotions (either by pointing to the correct face expression that matched the oral emotion label (Emotion Selection), or by choosing the correct written emotion label that matched the facial expression (Ekman 60)), both right SD and bvFTD groups were significantly impaired compared to Controls (all p values<0.001), with no significant differences between the patient groups (both p values>0.4).
Patterns of atrophy
Compared to Controls, patients with right SD displayed widespread atrophy involving medial and lateral temporal lobes, with the right hemisphere more heavily affected than the left. Patients with bvFTD exhibited striking, largely symmetrical bilateral frontal and temporal atrophy relative to Controls, encompassing the bilateral orbitofrontal cortex and bilateral frontal pole, extending to the medial and lateral temporal lobes (see online supplementary file 1).
Neural correlates of emotion processing
Correlations between behavioural performance on the Emotion Selection Task and the Ekman 60 emotion processing tasks and grey matter intensity were conducted, as both groups performed significantly worse than controls on these tasks (figure 3, see online supplementary file 2). Impairments on the Emotion Selection Task in the right SD group were associated with grey matter intensity decrease primarily in right temporal and frontal regions including the right lateral and medial temporal lobes and right orbitofrontal cortex, with minimal left anterior temporal lobe involvement. In contrast, in the bvFTD group, reduced performance was associated with decreased grey matter intensity in predominantly left lateralised structures including the left inferior frontal and orbitofrontal cortices and the left temporal lobe.
On the Ekman 60 task, poorer performance in the right SD group was associated with decreased grey matter intensity in the right medial and lateral temporal lobes and right orbitofrontal cortex, with some left lateral temporal involvement. In contrast, in the bvFTD group reduced Ekman 60 performance was associated with grey matter intensity decrease primarily in the left temporal lobe and frontal pole, as well as the superior frontal gyri bilaterally.
For patients with right SD, on a non-emotion processing control task, the ability to copy a complex geometric figure (RCFT Copy) correlated with grey matter intensity decrease in the right precuneus cortex and frontal pole (see online supplementary file 3). Similarly in bvFTD, performance on the RCFT copy was associated with a distributed set of regions including the right superior frontal gyrus, left lateral occipital cortex and the left frontal pole. Crucially, this contrast demonstrates that right temporal regions do not, by default, correlate with all cognitive tasks administered to the right SD group.
This study is the first to directly investigate the neural correlates of emotion processing in the syndrome of right SD compared with bvFTD. Our results indicate that the atypical (right-lateralised) presentation of SD is associated with marked alterations in complex emotion processing, at a similar level to that seen in bvFTD, which concords with the significant behavioural and interpersonal disturbances observed clinically in right SD and bvFTD.7 ,15 Notably, our results highlighted a subtle differentiation in emotion processing profiles; when determining if two faces expressed the same emotion, only patients with bvFTD were impaired, whereas when selecting a specific emotion from a series of distractor emotions, both groups showed comparable impairments.
The most striking finding to emerge from this study is that divergent neural changes underpin emotion processing deficits in these FTD syndromes. In right SD, difficulty in emotion processing was predominantly associated with degeneration of the right medial and lateral temporal lobes. Through the use of a control task, we demonstrated that right anterior temporal lobe atrophy correlates specifically with tasks that tap emotion processing abilities. Although mounting evidence indicates the importance of the right anterior temporal region in socioemotional processing,16 ,35 our study is the first to highlight the contribution of this region in mediating emotion processing disturbances in right SD. In contrast, emotion processing performance in bvFTD was associated with integrity of the left inferior frontal pole and left orbitofrontal cortex, with additional right middle/superior frontal lobe involvement. Emotion processing performance was also significantly associated with the integrity of the left temporal lobe in bvFTD, to a greater extent than that observed in right SD. Our findings corroborate previous studies pointing towards the importance of the bilateral frontal lobes15 and left temporal cortices36 for facial emotion processing in bvFTD. Furthermore, the potential role of interactions between frontoinsular and temporal lobe regions in supporting successful contextual social cognition has recently been emphasised.37 While these regions have previously been implicated in bvFTD, our results provide important evidence that a key node within the Social Context Network Model, namely the temporal lobe, is also implicated in face and emotion processing deficits in right SD. The finding of divergent neural correlates underpinning impaired performance suggests that individuals with bvFTD and right SD likely fail these emotion processing tasks for different underlying reasons. The ability to perceive and interpret facial information is known to involve a network of regions from early visual perception in the occipital lobe, to extended processing in limbic and temporal regions involved in the identification of emotion and identity.38 We suggest that the development of more precise tools to disentangle the contribution of face perception, emotion decoding, and higher-level executive functioning for emotion processing will prove particularly useful for the differential diagnosis of the bvFTD and right SD phenotypes.
In addition to common deficits in emotion processing, both FTD subtypes exhibited overlap in clinical features including disinhibition, reduced empathy, changes to diet, decline in self-care, as well as aspects of cognition, such as recall memory and letter fluency. Clinical reports and objective cognitive testing, however, revealed a number of potential discriminating features in each dementia subtype. Suggestion for an increased preponderance of obsessive, rigid and irritable behaviour and personality change was present in right SD. Patients with bvFTD, in contrast, exhibited marked apathy and attention/working memory difficulties, as well as the suggestion of increased reports of executive dysfunction and memory impairments, when compared to right SD. In addition, patients with right SD exhibited disproportionate language dysfunction and visual recognition memory difficulties versus patients with bvFTD. Although both patient groups showed some impairment in basic face perception (ie, determining whether two faces showing different expressions were the same person), only patients with right SD exhibited widespread evidence of prosopagnosia encompassing clinical reports of prosopagnosia (ie, impaired recognition of familiar faces), and objective difficulties on tasks assessing familiar face recognition/naming. The potential contribution of basic face perception to higher-level identity recognition and emotion recognition from facial stimuli has been surprisingly unexplored in this patient cohort.26 ,39 Nevertheless, our results suggest that examination of the presence of prosopagnosia may assist in differentiating right SD from bvFTD.
Distinguishing right SD from bvFTD is critical, given the likely different underlying neuropathology40 and advances in potential therapeutics. Such differentiation remains problematic, however, as no consensus criteria currently exist for this atypical presentation of SD. Furthermore, our data points toward marked heterogeneity within the right SD presentation. This withstanding, the majority of right SD cases demonstrated behavioural and/or language deficits, which, in the absence of brain imaging information, may lead the clinician to suspect either bvFTD or typical SD (PPA).8 A major presenting feature in half of cases was a difficulty recognising familiar or famous people, and this was a comorbid feature in a further quarter of cases. Our results suggest that comprehensive examination of classic frontally-mediated abilities are more likely to detect bvFTD cases, while assessment of language processing and prosopagnosia across multiple domains may assist in the early clinical diagnosis of right SD. Additionally, the assessment of emotion processing tasks may further elucidate the right temporal lobe presentation of SD.
Several caveats should be kept in mind when interpreting this work. Our sample sizes are relatively small, although this must be considered in the context of right SD being an uncommon clinical presentation. Additionally, although our patient groups were well-characterised, pathological confirmation of dementia aetiology was not available. Replication of our results in a larger patient cohort with neuropathological data is important for future studies. Further, it will be important to explore the potential differences in complex social cognitive processes such as interpersonal functioning and emotion reactivity, in which deficits have been previously documented in bvFTD15 and right SD7 separately. Consideration of additional socioemotional domains, such as theory of mind, may prove particularly insightful, given mounting evidence implicating the right anterior temporal lobe in this aspect of social cognition.16 ,35 Finally, some research suggests that emotion processing ability differs according to gender.41–43 Although our groups were matched for gender, it will be of interest for future studies to consider whether individuals with bvFTD and right SD show divergent changes in socioemotional functioning, according to gender.
In summary, this study is the first to objectively demonstrate comparable deficits in emotion processing in right SD and bvFTD. Although preliminary, our findings have important implications for the management of patients with right SD, with these deficits likely contributing to caregiver burden to a similar magnitude as that observed in bvFTD.44 These emotion-processing deficits, however, are attributable to divergent patterns of neural atrophy specific to each FTD syndrome: the right temporal lobe in right SD and the left temporal and frontal cortices in bvFTD. Considerable overlap in the clinical and cognitive presentation of patients with right SD and bvFTD underscores the inherent difficulty in discriminating between these FTD syndromes using existing clinical tools. We suggest that future studies incorporating targeted assessments of complex socioemotional processes such as theory of mind,16 or more basic face-processing, may assist in the differential diagnosis of these syndromes, as such tasks are thought to differentially recruit the right temporal lobe.
The authors would like to thank the participants and their families for supporting their research.
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
JK and FK contributed equally to this study.
Contributors JK contributed to design and conceptualisation of the study, analysis and interpretation of the data, drafting and revision of the manuscript. FK contributed to the data analysis and interpretation, and revision of the manuscript. JRB contributed to revision of the manuscript. OP and JRH contributed to design and conceptualisation of the study and revision of the manuscript. MI contributed to design and conceptualisation of the study, data analysis, interpretation of the data and revision of the manuscript.
Funding This work was supported by funding to Forefront, a collaborative research group dedicated to the study of frontotemporal dementia and motor neurone disease, from the National Health and Medical Research Council (NHMRC) of Australia programme grant (#1037746) and the Australian Research Council (ARC) Centre of Excellence in Cognition and its Disorders Memory Node (#CE110001021). Other grants include the NHMRC Project grant (#510106); NHMRC Early Career Fellowship (APP1072451; to JRB); Motor Neuron Disease Research Institute of Australia Grant (Grant in Aid; to JRB); Pfizer Neuroscience Research Grant; and Servier Staff “Barry Young” Fellowship (administered by Royal Australasian College of Physicians Foundation; to JRB); NHMRC Career Development Fellowship (APP1022684; to OP); and ARC Discovery Early Career Research Award (DE130100463; to MI).
Competing interests None.
Ethics approval South Eastern Sydney Local Health District and the University of New South Wales.
Provenance and peer review Not commissioned; externally peer reviewed.
Data sharing statement We will consider sharing our data on request.
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.