An Update On Semantic Dementia


Progressive and relatively circumscribed loss of semantic knowledge, referred to as semantic dementia (SD) which falls under the broader umbrella of frontotemporal dementia, was officially identified as a clinical syndrome less than 50 years ago. Here, we review recent neuroimaging, pathological, and genetic research in SD. From a neuroimaging perspective, SD is characterised by hallmark asymmetrical atrophy of the anterior temporal pole and anterior fusiform gyrus, which is usually left lateralised. Functional magnetic resonance imaging (fMRI) studies have revealed widespread changes in connectivity, implicating the anterior temporal regions in semantic deficits in SD. Task-related fMRI have also demonstrated the relative preservation of frontal and parietal regions alongside preserved memory performance. In addition, recent longitudinal studies have demonstrated that, with disease progression, atrophy encroaches into the contralateral temporal pole and medial prefrontal cortices, which reflects emerging changes in behaviour and social cognition. Notably, unlike other frontotemporal dementia subtypes, recent research has demonstrated strong clinicopathological concordance in SD, with TDP43 type C as the most common pathological subtype. Moreover, an underlying genetic cause appears to be relatively rare in SD, with the majority of cases having a sporadic form of the disease. The relatively clear diagnosis, clinical course, and pathological homogeneity of SD make this syndrome a promising target for novel disease-modifying interventions. The development of neuroimaging markers of disease progression at the individual level is an important area of research for future studies to address, in order to assist with this endeavour.


Semantic dementia (SD), a progressive neurodegenerative disorder affecting language, was empirically described only relatively recently. In the early 1970s, the conceptualisation of memory into two distinct systems, an episodic system and a semantic system by Tulving , coincided with the report of three individuals who presented with visual object agnosia, a profound inability to recognise or identify objects. In light of this new memory system and additional assessment, Warrington recognised that the constellation of symptoms of these patients could be conceptualised as an underlying loss of semantic memory. Since this seminal paper, the syndrome, which is characterised by circumscribed but profound loss of semantic knowledge, has been referred to as SD  and, more recently, as semantic-variant primary progressive aphasia (PPA) . Less than 50 years later, our understanding of this striking clinical syndrome has advanced. In this review, we will consider how recent studies in imaging, genetics, and pathology over the last decade have informed our knowledge of SD.

Contemporary consensus criteria for SD require individuals to first meet criteria for PPA; i.e. the most prominent clinical symptom to be in the domain of language, and evidence of subsequent impaired activities of daily living. Then, sub-classification as semantic-variant is based on impaired confrontation naming and single-word comprehension, with supportive features including impaired object knowledge, surface dyslexia or dysgraphia, spared repetition, and spared speech production. In a series of 100 cases all of whom underwent longitudinal follow-up, the mean age at presentation was 64.2 years but with a range of 40–79 years. There was a 50% survival of 12.8 years indicating a slower progression than in other forms of frontotemporal dementia . Studies of the prevalence and incidence of SD have been relatively limited; however, a recent epidemiology study estimated the prevalence of frontotemporal dementia at 10.8/100,000, with SD accounting for approximately one-third of these cases in line with previous estimates . Whether this prevalence is similar across countries, however, remains to be examined, as most existing epidemiological data hail from European studies.

Clinical Presentation And Cognitive Profile

Clinically, patients with SD show a speech profile that is relatively fluent but empty of content, producing a pattern of so-called logorrhoea. Importantly, loss of semantic knowledge is observed irrespective of testing modality. Impaired word comprehension is a mandatory feature and patients demonstrate word alienation in that they are able to repeat words such as “violin” or “caterpillar” but have no idea of their meaning. This deficit gradually progresses from low frequency and less familiar words, such as those mentioned, to more common words. Adlam et al. demonstrated that SD patients are also impaired on non-verbal semantic matching tasks, tests of colour knowledge, sound knowledge, and object-use knowledge, which do not require naming or verbal comprehension even from an early stage of the disease. Such findings have provided evidence that, in SD, symptomatology reflects a profound and progressive loss of conceptual knowledge which is not limited to performance on verbal tasks. There is also accompanying surface dyslexia: patients are unable to correctly pronounce irregular words such as pint which they read to rhyme with hint or flint.

In contrast, recent studies have confirmed that episodic memory is relatively preserved in SD, particularly when tasks with minimal conceptual loading are employed . The intact performance on traditional non-conceptually loaded episodic memory tasks converges with the performance of SD patients on autobiographical memory tasks. Patients typically show relatively preserved recollection of recent autobiographical memory in the context of poorer remote autobiographical memory (known as the reverse temporal gradient or step-function), reflecting increased semanticisation of past events. This is in stark contrast to the compromised ability of SD patients to project forwards in time to imagine possible future events. These deficits in future-oriented thought are attributable to semantic processing impairments, and have led to the advancement of the semantic scaffolding hypothesis which proposes that semantic knowledge is required to impart structure and meaning during the process of future simulation.

Changes in behaviour and social cognition are increasingly recognised in SD. Clinically, SD patients often show mental rigidity and inflexible behaviour. For example, patients may become obsessive in tasks they engage in (e.g. we have noticed patients spending hours completing jigsaw puzzles), food preferences (usually restricted to specific foods), or daily routines (e.g. clockwatching). In addition, SD patients may have increased apathy and changes in eating behaviour, as well as loss of empathy, impaired emotion perception and emotional memories, and reduced theory of mind capacity . Over time, many patients become essentially mute with only a limited repertoire of stereotypic phrases and a complete loss of word comprehension. Changes in emotional capacity as well as increased rigid behaviours are associated with higher carer burden, and progressive behavioural changes and/or increasing disability leads to residential care in most cases.


As this review reveals, despite a relatively short history much knowledge has been gained about the SD syndrome, particularly over the last decade. Indeed, SD appears to be one of the more straightforward frontotemporal dementia subtypes. It has a clear clinical course, which begins with language features and, with progression, affects behaviour and social cognition; this reflects early and relatively circumscribed neurodegeneration of the anterior temporal pole, which encroaches into medial prefrontal and posterior temporal regions as well as into the contralateral hemisphere with disease progression. Pathologically, it is most commonly associated with TDP43 type C and genetic causes are rare.

In spite of this progress, a number of outstanding questions remain which we hope research over the next decade will address. Clinical phenotype is clearly influenced by the laterality of pathology and associated atrophy in this syndrome, with left lateralised atrophy initially manifesting as loss of semantic knowledge (i.e. anomia) and right lateralised atrophy initially manifesting as loss of person knowledge (i.e. prosopagnosia, knowledge of social norms), giving important insights into the representation of conceptual knowledge across hemispheres. Currently, it remains unclear why only a subset of cases (~30%) present with right lateralised atrophy, with the majority of patients presenting with left lateralised neurodegeneration. Future studies that consider pre-clinical variables (e.g. handedness, occupational history, learning disabilities) with the potential to influence vulnerability of brain hemispheres to disease, may shed light on this issue . Indeed, it has been suggested that patients with PPA have a higher rate of pre-existing language disorders than would be expected in the general population, but this has not been investigated specifically in the SD clinical phenotype. In addition, from a management perspective, changes in behaviour, capacity to engage in social situations, and reduced empathy lead to increased burden and stress in carers, which is often greater than in carers of patients with the behavioural variant of frontotemporal dementia. This may reflect inadequate psychoeducation of carers regarding the manifestation of behavioural change in SD patients which is important for clinicians to consider when interacting with family members and carers of SD patients.

Clinically, one of the key issues on the horizon is the development of drugs that target the deposition or clearance of pathology. As these are likely to be specific to pathological subtypes, SD appears to be a promising syndrome for drug developers to target, given the striking clinicopathological concordance. Development of such agents, however, has been hindered by the lack of suitable animal models showing pathological changes that mirror those seen in SD. Knowledge is also lacking about the basic biological processes that underlie the deposition of TDP43 type C in SD. Recent advances in CSF studies may represent a promising avenue to identify FTLD pathological subtypes in vivo. The development of biomarkers such as these (e.g. CSF NfL, neuroimaging, blood biomarkers) are essential for discrimination, prognosis, and prediction of disease progression in SD. Improved understanding of the pathophysiological mechanisms which give rise to SD will also be essential for the development of novel drug interventions. From an imaging perspective, new techniques to measure change in brain integrity and function with disease progression have already started to make some headway. Applications of these techniques at the individual level are likely to be key to track disease progression and potentially measure the efficacy of interventions as these become available.