Introduction

Frontotemporal dementia (FTD) is a progressive brain condition defined by three canonical clinical presentations. These include a behavioural variant (bvFTD) [1], as well as two forms of primary progressive aphasia (PPA): semantic dementia (SD; or the semantic variant of PPA), and progressive nonfluent aphasia (or the non-fluent/agrammatic variant of PPA) [2]. While language features co-occur with behavioural symptoms [3], language changes in bvFTD have received limited attention to-date. This under-appreciation of language change may, in part, be due to the diagnostic criteria for bvFTD focusing on behavioural and functional impairments [1]. Some language change may be expected in this disease, however, given the location of predominant brain atrophy [4] and known associations between language performance and both temporal lobe atrophy [5–7] and frontotemporal activity [8].

Investigating language integrity in bvFTD is important for a number of reasons. Firstly, deficits may emerge early in the disease and affect daily functioning [9, 10]. Secondly, assessment of language in bvFTD may prove a helpful aid to clinical diagnosis. Reduced picture naming has been found in bvFTD patients with abnormal MRIs (i.e., “probable bvFTD”) but not in “MRI-normal” bvFTD patients (i.e., “possible bvFTD”) [11]. In cases where brain imaging is not available or possible, this clinical marker could help provide further certainty regarding the diagnostic category. In addition, bvFTD is known to arise from a number of pathologies that are difficult to distinguish in life [12, 13]. Some people with possible bvFTD may indeed have an alternative psychiatric condition, which becomes evident over time. Recent studies suggest that the examination of both expressive and receptive language may aid in the differentiation of bvFTD from other psychiatric conditions at presentation [14]. These findings suggest that examination of language integrity in bvFTD could help with diagnostic decisions.

To enable the development of language markers to guide diagnosis, separate subgroup analyses are needed when investigating language skills in bvFTD. To-date, few studies have clearly included both possible and probable bvFTD groups, with most either not mentioning the distinction or focusing solely on probable cases (see review by Geraudie et al. [15]). Studies that have analysed both possible and probable cases (or equivalents, e.g., “MRI-normal” vs “MRI-abnormal”; “bvFTD/PET+” vs “bvFTD/PET– “) have, to-date, reported conflicting findings regarding language deficits [11, 16, 17], signalling the need for further investigation.

Single-word processing in frontotemporal dementia

Assessment of language at the single-word level has been used effectively to characterise and distinguish impairments within language-led dementias (i.e., PPA) [18, 19]. In bvFTD, single-word-level language impairments have been identified in naming [15, 20] (found in 31 of 49 studies reviewed [15]), word comprehension (8 of 12 studies), and semantic knowledge (6 of 8 studies), but not word repetition (0 of 5 studies [15]). In two studies that compared bvFTD with other FTD subtypes, bvFTD patients showed less severe but similar deficits to individuals with SD [5, 6] (i.e., reductions in naming, word comprehension, and semantic association, but relative sparing of word repetition). Whether the inter-relationships between the naming and single-word comprehension deficits were similar between the groups, or underpinned by similar mechanisms, was not examined. Also, in many studies, small sample sizes (e.g., n < 30 participants) have impeded the ability to conduct more complex investigations within datasets.

In SD, deficits in naming and word comprehension are known to arise from degradation of the semantic memory system, associated with significant atrophy to the anterior and medial temporal lobes [2]. For bvFTD, explanations for language changes remain largely unconfirmed. Prior studies have shown, however, that the deficits in word retrieval (naming) and comprehension are not simply explained by general cognitive decline, given deficits remain after accounting for disease severity and non-verbal intellectual skills [5]. In addition, deficits in naming are not only observed late in the disease, but have been reported in pre-symptomatic and early phases [21, 22].

The relative contributions of executive function and linguistic factors to language change are currently undetermined. From an executive function perspective, poor performance could potentially arise from deficits in inhibition, working memory, or cognitive flexibility, all of which are known to occur in bvFTD [23–26]. Disinhibition could lead to more off-task responses, or impulsive errors. Poor working memory could impact lexical processing/language comprehension performance by reducing the ability to retain task requirements in mind [27]. Other forms of executive dysfunction could lead to reduced efficiency in searching the lexicon or increased perseverative responses. In support of an executive contribution, individuals with bvFTD who experience greater language deficits reportedly experience greater executive functioning impairment [5].

Alternatively, language changes in bvFTD may have a primarily linguistic basis and arise in response to the atrophy within key language processing areas of the brain (e.g., lateral temporal, insula, superior temporal, and mid-hippocampal regions) [6, 13]. Further studies that include both language and executive measures in bvFTD are needed to address this question.

Aims and hypotheses

The current study aimed to examine language abilities in large samples of individuals diagnosed with either possible and probable bvFTD, using a comprehensive assessment of single word language in combination with measures of executive function. In doing so, this study sought to compare language profiles of bvFTD and SD, and investigate the relative contributions of semantic functioning and executive measures to language performance. It was predicted that a greater range and severity of impairment would arise in probable bvFTD as compared to possible bvFTD, and that a broadly similar pattern of performance across single word processing skills would be observed between SD and bvFTD although not necessarily underpinned by the same cognitive abilities. Whereas performance in SD was expected to be primarily associated with semantic impairment, performance in bvFTD was expected to reflect contributions from both semantic and executive functioning. Accordingly, differences between bvFTD and SD were anticipated when examining specific relationships between tasks where different demands on semantic knowledge or executive control may be present.