Transdiagnostic approach to language and cognition in the logopenic variant of primary progressive aphasia and Alzheimer’s disease

Abstract

Language is commonly affected in many dementias as a dominant or adjunctive symptom. Speech and/or language deficits are the primary symptom in primary progressive aphasia (PPA), associated with frontotemporal lobar degeneration and Alzheimer’s disease (AD). Language deficits are also frequently found in other dementias like typical, amnestic AD. Progressive language deterioration negatively affects individuals’ quality of life, carer burden, and outcome. Enabling early and accurate diagnosis is a priority for inclusion in clinical trials and targeted speech-language interventions. In my thesis, I discuss barriers to the assessment of progressive aphasias and present exemplars of transdiagnostic approaches to improve clinical characterisation of language-based dementias. The clinical motivation behind Chapter 2 was the wide use of picture description tasks across disciplines but the lack of a validated tool to base the interpretation on, further deeming this otherwise clinically valuable task impractical for non-specialist settings. Using data-driven methods, I developed a novel aphasia checklist designed to be simple and sensitive in detecting a primary or associated language impairment with a one-minute speech sample. In addition to finding excellent within-sample four-fold cross-validation, I mitigated the potential limitation of a small sample size (n = 74) by conducting predictive validity testing on an unseen dataset (n = 33). This level of validation is not typically done for patients with rare diseases like PPA, PSP and CBS, and my current models have the potential to further generate specific word checklists for other picture description tasks, different languages, and diverse patient groups. In Chapter 3 to 5, I posed three key unanswered questions that challenge our current assumptions and understanding about the logopenic variant of PPA (lvPPA), also known as the “language-variant AD”: (1) If people with lvPPA have semantic deficits, what is the nature of the deficit (semantic presentation versus semantic control; verbal versus non-verbal) and when does the impairment occur?; (2) Beyond short-term memory impairment, do people with lvPPA have an impairment within phonology itself?; and (3) Is lvPPA a distinct nosological entity or part of a graded multidimensional AD clinical spectrum? To this end, I recruited nine patients who had typical, amnestic presentation of AD (tAD) and 18 with lvPPA. The lvPPA sample was further grouped into those meeting strict consensus criteria for PPA and lvPPA (n = 10) and others who may have previously satisfied definitions of lvPPA but had progressed with multi-domain cognitive impairments (herein referred to as “lvPPA+”; n = 8). The inclusion of tAD patients facilitated a direct comparison of cognitive and neuroanatomical profiles across typical and atypical presentations of AD and the distinction between lvPPA and lvPPA+ patients offered a window into cognitive decline across lvPPA disease severity. Results from Chapter 3 showed that like post-stroke semantic aphasia and Wernicke’s aphasic patients, (i) lvPPA patients’ performance declines in line with increasing semantic control demands, and (ii) this pattern of decline is true for both verbal and non-verbal semantic tasks. Findings from Chapter 4 highlighted that instead of having a core phonological impairment, lvPPA patients have a working memory/buffering impairment that adversely affects their performance on length-dependent working memory tasks. In my deep phenotype study of lvPPA in Chapter 5, I concluded that: (i) the multidimensional cognitive pattern of impairment is present in some mild lvPPA cases; (ii) memory deficits are not only present in lvPPA, but also equally impaired relative to tAD (iii) and this is true for verbal and non-verbal memory tests, as well as for both clinic- and research-level assessments; and (iv) the patterns of cognitive decline in lvPPA remain largely graded (dimensional) over time. Together these Chapters contribute to the ongoing debate about how to consider the atypical (e.g., aphasic, visual) and typical (amnestic) AD phenotypes and support the hypothesis that AD subtypes represent positions within a graded multidimensional space. While AD phenotypes may present with a predominant domain of symptoms and signs (e.g., language in lvPPA and memory in tAD), there are gradations, overlap and only partial distinctions between these phenotypes. Crucially, given the transdiagnostic symptoms shared across AD phenotypes, (1) AD clinical trials should be inclusive of all AD subtypes including lvPPA, and (2) lvPPA-focussed clinical trials, extending beyond the non-pharmacological speech-language intervention, should also increase in number. Furthermore, the current diagnostic criteria for lvPPA could be aided by adopting additional features, such as semantic representation versus control-impaired subtypes, and differentiating between phonological representations (which is relatively intact) versus short-term memory/information buffering, which may be the primary cognitive mechanism underlying the core deficit of phrase and sentence repetition in lvPPA. This thesis provides improved clinical characterisation of lvPPA across disease severity which has the potential to help patients, families, and clinicians to set expectations, manage symptoms, and initiate support.