Primary progressive aphasia: a clinical approach

The primary progressive aphasias (PPA) are a diverse group of disorders that collectively present with relatively focal degeneration of the brain systems that govern language. Despite much recent attention in the scientific literature [1, 2], these ‘language-led dementias’ remain daunting for even experienced clinicians to diagnose and manage. This is not surprising: PPA is uncommon (estimated prevalence is conservatively around three cases per 100,000 [3, 4]), the underlying pathology is heterogeneous and generally inaccessible and the functions principally targeted are uniquely complex. Although patients with PPA have been described for well over a century [5], the true significance of these disorders was only appreciated quite recently [6, 7] and the paradigm of selective brain network degeneration caused by pathogenic protein spread has transformed our understanding of neurodegenerative disease [8]. While challenging, accurate clinical diagnosis of PPA is worth striving for: these patients are often affected in late middle life, with devastating implications for family life, work and social functioning.

In this review, we outline an approach to the diagnosis and management of PPA in the clinic and at the bedside, distilled from our accumulated experience of meeting and caring for these patients. We firstly present a clinical framework for assessing language functions, tailored in particular to the major syndromic presentations of PPA (Tables 1 and 2, Figs. 1 and 2). We then consider these presentations in detail. Three major forms of PPA—nonfluent–agrammatic variant (nfvPPA), semantic variant (svPPA) and logopenic variant (lvPPA)—comprise the canonical syndromes currently recognised in consensus diagnostic criteria [9] (see Table S1, Supplementary Material on-line). These syndromes are distinguished by the language deficits with which they present and associated cognitive, neurological and neuroanatomical profiles and tend to have distinct neuropathological substrates. Key features of PPA syndromes are summarised in Tables 1, 2 and 3 and Fig. 1; additional ‘clinical pearls’ that we have found useful in diagnosis of each syndrome (but which are not widely discussed in the literature of these conditions) are presented in Table 4. Following the taxonomy of classical (stroke) aphasiology, nfvPPA might be anticipated to align with Broca’s aphasia, svPPA with transcortical sensory aphasia and lvPPA with Wernicke’s or conduction aphasia. However, such clinical correspondences are loose, at best. This probably reflects the very different nature of the underlying disease processes, and most pertinently, the distributed neural network basis of PPA [10]. One important corollary is that PPA syndromes extend (cognitively and neuroanatomically) beyond the province of language, to involve other complex behavioural functions. The clinical challenges posed by PPA foreshadow significant unresolved issues in the nosology and neurobiology of these conditions [1, 2, 10‐16]. Here we highlight potential diagnostic pitfalls including atypical variant presentations of PPA not well captured by standard criteria (Table 1) and propose a diagnostic ‘roadmap’ (Fig. 3). After outlining principles of management of PPA, we conclude with a prospect for future developments.

When confronted by an aphasic patient, it is important firstly to establish the context of the language disturbance. This usually requires the help of an informant who knows the patient well and can supply reliable background information. A diagnosis of PPA requires that speech or language dysfunction was the initial and most salient clinical complaint (see Table S1). However, the patient’s previous verbal skills (including formal education, occupation, bilingualism or any specific developmental difficulties such as stammering or dyslexia) are relevant to interpreting current deficits. It is also necessary to determine the extent of any uncorrected peripheral hearing or visual impairments as these can impact significantly on everyday communication and performance on language tests. In defining the history of the language problem, it is essential to establish the circumstances of onset and very first symptoms (often noticed by the patient’s family), overall duration and tempo. The length of the history bears strongly on the interpretation of deficits, since PPA syndromes tend to converge over time [17]. In PPA, a history of gradual, but unrelenting decline over a number of months or several years is typical, but some apparent fluctuation is not uncommon, particularly under conditions that stress the language system, such as public speaking or conversations by telephone or in a non-native tongue. There may have been a sentinel event such as a family celebration or minor head injury that first drew attention to the patient’s difficulties; informants may interpret this as an acute onset but a searching history usually reveals a more insidious prodrome.

The profile of the patient’s language dysfunction then allows the clinico-anatomical syndrome to be characterised (see Table 1). Fundamentally, language supports communication—the understanding, creation and delivery of messages. In assessing a patient’s speech, it is useful to analyse the various stages at which the idea for a message is first generated, the content (or vocabulary) of the message, its structure (assembly) and delivery. Similarly, in assessing understanding of language, it is useful to analyse the separable stages at which messages are perceived and then invested with meaning. These operations are differentially vulnerable to particular PPA syndromes and can be explored using targeted questions on history and a small set of core language tests (Table 2). The patient’s use of written language typically echoes the speech disorder as the illness evolves. Examples of patients’ transcribed spoken and written productions are presented in Table 3 (corresponding speech sound files are provided in Supplementary Material on-line).

In neurology, the history generally suggests the diagnosis while the examination corroborates the historical suspicion. This precept is equally valid for language disorders, with the caveat that certain aspects of language are difficult to differentiate on the story alone. One key example (not often called upon in everyday communication) is the ability to repeat messages verbatim, which is central to the characterisation of PPA (see Table 1) and should be examined explicitly. Like the testing of pupillary and spinal reflexes in general neurology, certain language tests such as speech repetition or picture naming rapidly assay a number of connected neural operations: if such tests are performed normally, this demonstrates the overall integrity of the system but if a problem is found, it is necessary to establish where in the system it lies. The most important principle in examining speech is to obtain an adequate sample; for this purpose, it is convenient to carry a picture that will encourage the patient to talk and provide a prop for directed tests (one example is shown in Fig. 2).

Alongside the core clinical tests in Table 2 we list some more formal equivalents that might be administered by a neuropsychologist. However, neuropsychological assessment does not simply endorse the bedside impression. If available, it adds considerable value, particularly in quantifying language capacities in relation to standardised population norms and in the context of estimated premorbid ability, in tracking change in language functions over time and in measuring associated capacities that together with aphasia define the overall cognitive phenotype and may also affect the assessment of language.

Accurate and early identification of PPA syndromes is essential for clinical counselling and planning appropriate management. Beyond clinical characterisation, molecular stratification will become increasingly important with trials of candidate disease-modifying treatments on the horizon [35]. However, most patients presenting with a language complaint will not have PPA and further, a number of patients with PPA (as many as 40% in some series) do not conform closely to one of the canonical syndromic diagnoses [3, 13, 15, 16, 34]. Some of these less common, atypical variants are in Table 1.

In Fig. 3, we present a ‘roadmap’ for diagnosis of PPA syndromes that we have found useful in clinic. Key diagnostic decision ‘forks’ rest on the historical and examination findings outlined in Tables 1 and 2 and on neuropsychological assessment where available. Structural brain imaging (ideally, MRI) is essential in all cases of suspected PPA, both to rule out other causes of progressive language failure and more positively, to identify features of particular radiological phenotypes (Fig. 1). Where initial MRI features are borderline (notably in nfvPPA and lvPPA) it may be helpful to repeat the scan after an interval of a year or so, as directly comparing serial studies of the culprit brain region is often revealing. Functional MRI and FDG-PET in PPA are not in routine clinical use though there is considerable potential interest in applying such techniques to define aberrant and compensatory language network changes, with a view to future therapeutic trials [74, 75].

Having arrived at a syndromic diagnosis, it may be appropriate to investigate further to determine the underlying proteinopathy—including CSF examination or amyloid-PET scanning for Alzheimer neurodegeneration markers. These ancillary investigations may be relevant, for example, in deciding to trial a cholinesterase inhibitor in a patient with nfvPPA or more generally, in forecasting the overall outlook of the illness in earlier stage disease. We discuss the possibility of genetic risk in all younger patients with nfvPPA and genetic screening should be considered in any patient with PPA who has a relevant family history, particularly where there this raises suspicion of a frontotemporal dementia. Discussions in clinic around genetic testing broach a number of sensitive issues and should ideally involve other family members.

We next consider some common potential pitfalls on the path to diagnosis.

Patients with PPA generally live for a number of years following diagnosis, with evolving deficits and specific needs at each stage of the illness. Widely accepted clinical staging markers are presently lacking, however, the major PPA syndromes collectively raise similar management challenges and these broadly require the integration of non-pharmacological and pharmacological approaches.

Recognition of the major PPA syndromes has transformed our understanding of the language system and given us a new picture of selective neural vulnerability in degenerative brain disease. However, this dramatic progress should not obscure the many remaining difficulties surrounding these conditions. Ultimately, effective treatment of PPA will depend on both earlier and more accurate diagnosis (improved syndrome characterisation), more accurate disease and disability staging and identification of new biomarkers that can target tissue pathology and track therapeutic effects dynamically, in advance of irrecoverable brain atrophy (improved signalling of underlying proteinopathy [93]). A successful PPA biomarker will need to encompass substantial individual variation within syndromic categories (see, e.g., Table 3).

To realise these ambitions may require a reformulation of the ‘language-led dementias’ in more fundamental, pathophysiological terms. Emerging evidence suggests that generic disorders of nonverbal auditory information processing may underpin the canonical PPA syndromes [40, 94‐96] and that these syndromes may further be stratified by profiles of autonomic reactivity to emotional and other salient stimuli [61, 97, 98]. This evidence dovetails with the recent finding that implicit auditory sequence learning is retained in nfvPPA [99]. Such observations could motivate novel biomarkers and treatment strategies that do not depend on specific language capacities (a practical advantage in mounting large-scale, international trials in PPA). There is currently considerable interest in the application of molecular ligand neuroimaging techniques (including new tracers that can demonstrate tissue deposition of pathogenic proteins other than beta-amyloid) in PPA and other dementias [100]. Such techniques promise to delineate proteinopathies in vivo more reliably than is currently feasible; multivariate approaches combining ligand imaging with pathophysiological indices may constitute a powerful and dynamic signal of the underlying disease. Finally, more work is needed to establish rational retraining and functional rehabilitation interventions in PPA: in future, such interventions may amplify disease modifying therapies and bridge the gap from clinic to patients’ wider lives.