Despite a wealth of studies on FTD, much remains unknown about the disease, including the cause of the sporadic form. This is partly because of the heterogeneity of clinical presentation, age at disease onset and speed of progression. In addition, there is a wide diversity of underlying neuropathology in patients with similar clinical presentations, and lack of clinicopathological correlation in the majority of patients. The overlap with other neurological syndromes makes the disease even more complex. In this review we aim to clarify the terminology of FTD, outline the various clinical features and diagnostic criteria of sporadic and familial FTD syndromes and discuss the current major challenges in FTD research and clinical practice. We also outline potential areas for future research.

Frontotemporal dementia (FTD) is a clinically and pathologically diverse group of progressive neurodegenerative disorders leading to changes in behaviour, social conduct, language or speech because of atrophy of the frontal or anterior temporal lobes of the brain (or both). Although it occurs less frequently than Alzheimer's disease (AD), FTD is a common cause of young onset dementia, often affecting individuals below the age of 65 years. However, it also affects older individuals, and may be under‐diagnosed because of individuals being misdiagnosed with AD or other types of dementia (Onyike and Diehl‐Schmid 2013 ). The majority of cases have no known cause (‘sporadic’ FTD), but approximately a third is familial, secondary to autosomal dominant mutations in one of several FTD‐associated genes. There are two main clinical subtypes found in patients presenting with FTD: behavioural variant FTD (bvFTD), which primarily affects behaviour and social interaction, and primary progressive aphasia (PPA), which causes progressive impairment of speech and language. Both sporadic and familial FTD patients can also develop concurrent motor neuron disease (MND) (Devenney et al . 2015 ) or an atypical parkinsonian disorder such as corticobasal syndrome (CBS) or a progressive supranuclear palsy syndrome (PSPS) (Espay and Litvan 2011 ; Kertesz et al . 2011 ; Park and Chung 2013 ).

Mutations in three genes account for the majority of familial FTD, predominantly through autosomal dominant inheritance: progranulin ( GRN ) (Baker et al . 2006 ; Cruts et al . 2006 ), microtubule‐associated protein tau ( MAPT ) (Wilhelmsen et al . 1994 ; Hutton et al . 1998 ) and the chromosome 9 open reading frame 72 ( C9ORF72 ) (DeJesus‐Hernandez et al . 2011 ; Renton et al . 2011 ). More recently, mutations in TRAF tumour necrosis factor receptor‐associated factor family member‐associated NF‐kappa‐B activator (TANK)‐binding kinase 1 ( TBK1 ) have been identified in association with familial FTD (Freischmidt et al . 2015 ; Gijselinck et al . 2015 ; Le Ber et al . 2015 ; Pottier et al . 2015 ). Small numbers of patients possess mutations in other rare, FTD‐associated genes. Mutations in valosin containing protein ( VCP ) ‐1 are usually associated with a multisystem proteinopathy manifesting as inclusion body myopathy and Paget's disease of the bone (Watts et al . 2004 ), and mutations in charged multi‐vesicular body protein 2B ( CHMP2B ) (Skibinski et al . 2005 ) are found mainly in a Danish cohort. Rare genetic causes of FTD include transactive response DNA‐binding protein‐43 ( TARDBP ) (Synofzik et al . 2014 ), ubiquilin 2 ( UBQLN2 ) (Gellera et al . 2013 ), p62/sequestome1 ( SQSTM1 ) (Rubino et al . 2012 ; Le Ber et al . 2013 ; Miller et al . 2015 ), fused in sarcoma (FUS) (Kwiatkowski et al . 2009 ; Vance et al . 2009 ), dynactin‐1, associated with Perry syndrome (Munch et al . 2005 ) and coiled‐coil–helix‐coiled–coil‐helix domain containing 10 ( CHCHD10 ) (Bannwarth et al . 2014 ). Mutations in presenilin‐1 or amyloid precursor protein, both associated with familial AD, and prion protein, associated with familial prion disease, have also been associated with a clinical FTD syndrome (Rohrer and Warren 2011 ). A small proportion of patients (1.2% in one study of 334 patients) (van Blitterswijk et al . 2013 ) have dual mutations, for example, the C9ORF72 expansion as well as another mutation in one of the other FTD genes, for example, GRN (Lashley et al . 2014 ).

Currently, the only confirmed risk factors for FTD are mutations in certain genes. Between 30% and 50% of patients report a positive family history of FTD in at least one family member and a clearly autosomal dominant inheritance pattern is seen in 10–15% of patients (Goldman et al . 2005 ; Rohrer et al . 2009a ). BvFTD is significantly more heritable than PPA, with nfvPPA being much more likely to be familial than svPPA (Rohrer et al . 2009a ). CBS can be familial in some cases, and PSP is only very rarely familial. Estimates of heritability of FTD‐MND vary widely between studies and it remains unclear how many cases are genetic.

In patients with overlap syndromes of FTD with MND, PSPS or CBS, behavioural and cognitive symptoms can develop before, after or simultaneously with motor symptoms (Kertesz et al . 2011 ; Park and Chung 2013 ; Siuda et al . 2014 ; Devenney et al . 2015 ). In clinical practice, there is often controversy or indecision about what diagnosis to give, or whether to revise the diagnosis when new symptoms appear, to capture this development of a new mixed phenotype. For example, a patient presenting with behavioural changes consistent with bvFTD who later develops falls, supranuclear gaze palsy and axial rigidity, may have their diagnosis changed to PSPS. Similarly, a patient with initial language dysfunction characterized by effortful and agrammatic speech, who is first diagnosed with nfvPPA, but later develops asymmetric limb apraxia, rigidity and myoclonus, may be later re‐diagnosed with CBS. In our experience, this changing of the diagnosis can be confusing for patients and their families (‘was the initial diagnosis wrong?’), and from a research point of view can lead to loss of important phenotypic information (e.g. in pathological or genetic studies it may be important to distinguish between PSPS cases who develop PPA and those who develop bvFTD). We would advocate the use of overlap terms such as PPA‐CBS or FTD‐PSPS to help clarify such confusion, as has been done with MND: patients with FTD who later develop MND are usually diagnosed with FTD‐MND (or FTD‐ALS), and those with initial MND and symptoms that later fit criteria for bvFTD or PPA are labelled as MND‐FTD (or ALS‐FTD). However, overlap of these disorders can be variable and one unresolved dilemma is how to classify patients that do not quite fulfil criteria for a particular disorder but have mild features. Although around 10–15% of patients with FTD develop MND (Lomen‐Hoerth et al . 2002 ; Burrell et al . 2011 ), there is an even higher prevalence of ‘subclinical’ evidence of MND, with electromyogram evidence of MND or subtle MND‐like clinical signs, such as fasciculations, in 60% of FTD patients (Lomen‐Hoerth et al . 2002 ). Conversely, while 10–20% of MND patients meet diagnostic criteria for FTD, at least 50% of patients presenting with MND develop cognitive or behavioural impairment, termed MNDci (or ALSci) and MNDbi (or ALSbi) (Strong et al ., 2009 ). Of the various MND phenotypes seen in FTD patients, the majority usually develop the ALS variant, but lower motor neuron (primary muscular atrophy) or upper motor neuron (primary lateral sclerosis) phenotypes are also seen rarely (Devenney et al . 2015 ). As with MND, patients with FTD may develop parkinsonian features (bradykinesia, rigidity, tremor and/or postural instability) not fully consistent with a particular clinical syndrome such as PSPS or CBS, and are often diagnosed with FTD with parkinsonism. Parkinsonism is seen in around 20% of patients, while a larger proportion may eventually develop this in end‐stage disease (Park and Chung 2013 ). One large study of 364 FTD cases (35 with pathological confirmation), demonstrated the presence of parkinsonism as an early feature in 16% (18% of bvFTD, 14% of nfvPPA and 11% of svPPA) (Seelaar et al . 2008 ). Very rarely, patients develop FTD, MND and parkinsonism, including some patients with an underlying C9ORF72 expansion mutation (Coon et al . 2011 ; Boeve et al . 2012 ; Mahoney et al . 2012 ; O'Dowd et al . 2012 ; Snowden et al . 2012 ).

FTD is now used as an umbrella term to describe the overall group of clinical syndromes, while it has previously been used to just mean the progressive behavioural syndrome now called bvFTD (Neary et al . 2005 ). Other terms used previously for this include frontal lobe dementia (Lund and Manchester Groups 1994 ) and frontal variant FTD. Evolution of PPA terminology has been more tortuous. Despite Pick's original FTD case being a patient with language difficulties (Pick 1892 ) and reports of a variant of FTD with predominant language or speech decline published in the early 20th century, explicit description and widespread acceptance of a language‐led variant of FTD remained elusive until the 1970s. In 1975 , Warrington characterized the presentation of patients with selective deficits in semantic memory, leading to the later description of semantic dementia (Snowden et al . 1989 ; Hodges et al . 1992 ). Mesulam ( 1982 ) also described a slowly progressive selective aphasia, later labelling this syndrome as PPA (Mesulam 1987 ). In an early consensus document of diagnostic criteria for behavioural and language variants of FTD, PPA was initially split into a fluent subtype (semantic dementia) and a non‐fluent subtype (progressive non‐fluent aphasia) (Neary et al . 1998 ). However, another subtype of PPA was subsequently recognized, called logopenic aphasia (LPA) or the logopenic/phonological variant of PPA (Gorno‐Tempini et al . 2004 , 2008 ), and this was subsumed into the most recent consensus diagnostic criteria for PPA (Gorno‐Tempini et al . 2011 ) which recognizes three variants: nfvPPA, svPPA and lvPPA. The nosology of the FTD spectrum is not entirely resolved and several studies of PPA have identified a group of patients that do not fit criteria for any of the three described variants (Sajjadi et al . 2012 , 2014 ; Wicklund et al . 2014 ).

It is now well recognized that there are two main initial clinical presentations seen in patients with FTD: bvFTD describes those who develop progressive behavioural change, inappropriate social conduct and executive dysfunction, and PPA describes those who have progressive language decline and speech difficulties. There are three variants of PPA: semantic variant PPA (svPPA) leading to fluent speech with anomia, impaired single word comprehension and surface dyslexia due to loss of semantic memory; non‐fluent (or agrammatic) variant PPA (nfvPPA), leading to effortful speech production with agrammatism, apraxia of speech and impaired sentence comprehension; and logopenic variant PPA (lvPPA) leading to word‐finding pauses and impaired sentence repetition (Gorno‐Tempini et al . 2011 ). LvPPA is mostly associated with AD pathology and is therefore not always included within the FTD spectrum, while in a minority of cases it can be associated with FTLD pathology (Lashley et al . 2015 ).

The terminology of FTD can be confusing, and has evolved significantly since the first description of a patient with progressive language disturbance and left superior temporal gyrus atrophy by Pick ( 1892 ). Histopathological presence of argyrophilic globular neuronal cytoplasmic inclusions (later termed Pick bodies) was actually described not by Pick but by Alzheimer ( 1911 ) and the concept of FTD as ‘Pick's disease’ by a Dutch group (Gans 1925 ) and by a German group (Onari and Spatz 1926 ). By 1956, it had become evident that true Pick's pathology was underlying only around 20% of clinical FTD cases (Escourolle 1958 ) and subsequent studies confirmed that there were multiple other pathologies associated with atrophy of the frontal and/or temporal lobes in patients with the clinical syndrome of FTD (Brun 1987 ; Mann et al . 1993 ). The historical term for bvFTD, Pick's disease, is now reserved for cases of FTD with Pick type pathology. The term ‘frontotemporal lobar degeneration’ (FTLD) was therefore designated to describe a heterogeneous group of neurodegenerative diseases characterized by selective frontal and/or temporal lobe atrophy (Neary et al . 1998 ), and who have non‐Alzheimer's disease neuropathology (Lashley et al . 2015 ).

Clinical syndromes of FTD

In this section we summarize the clinical features of bvFTD and PPA variants, with reference to most recent diagnostic consensus criteria (Gorno‐Tempini et al. 2011; Rascovsky et al. 2011). Specific features of the various familial FTD phenotypes will be discussed in the later section entitled ‘Clinical syndromes of familial FTD’.

Behavioural variant FTD BvFTD presents with progressive decline in social skills, difficulties with planning and higher level thinking due to executive dysfunction and distinct changes in behaviour with relative preservation of other cognitive areas such as episodic memory and visuospatial function in the early stages. Patients with a PPA subtype or PSPS/CBS overlap disorder can also display similar behavioural features, as discussed below, but by definition they are not predominant at initial presentation. Patients with bvFTD often lack insight into their problems, and may seem indifferent or annoyed when brought to medical attention as they feel that there is nothing the matter with them. It is usually the patients’ relatives or close friends who notice that something is wrong, usually because of a breakdown in their relationship with the patient or complaints from friends or work colleagues about odd behaviour or increasingly poor performance at work. Relatives’ reports of the patient having a ‘poor memory’ usually refer more to their perception of a change in the patient's level of personal and social functioning rather than true memory problems, and unfortunately can lead to a misdiagnosis of AD or repeated misdiagnoses such as ‘stress’, anxiety or depression by the non‐specialist. Careful questioning in clinic, and particularly of the accompanying relative when they are alone, will reveal the true nature of cognitive changes and a history of progression of symptoms over time, both essential for aiding correct diagnosis. The most recent diagnostic criteria for bvFTD (Rascovsky et al. 2011) were developed by the Frontotemporal Dementia Consortium to summarize more succinctly the key features of behavioural change seen in this subtype, while recognizing that other cognitive features such as episodic memory can be affected, albeit less commonly. Criteria for a diagnosis of possible bvFTD are displayed in Table 1; patients must attain any three out of the six key clinical features: five behavioural (disinhibition, apathy or inertia, loss of sympathy or empathy, stereotyped or compulsive behaviours or hyperorality) and one cognitive (predominant executive dysfunction on neuropsychological assessment). The sensitivity and specificity of these diagnostic criteria for correct diagnosis of bvFTD have been reviewed in a number of studies with confirmation of FTLD or non‐FTLD pathology, which have established that the criteria have 85–95% sensitivity and 82% specificity for a diagnosis of possible bvFTD and 75–85% sensitivity and 95% specificity for probable bvFTD (Rascovsky et al. 2011; Harris et al. 2013b). Table 1. Summary of behavioural and cognitive symptoms within the current diagnostic criteria for behavioural variant frontotemporal dementia (bvFTD) and other commonly seen features Behavioural/cognitive symptoms – diagnosis of possible bvFTD requires at least three of the following symptoms to be fulfilled: Examples of specific symptoms Early behavioural disinhibition ≥ 1 of Socially inappropriate behaviour Staring, inappropriate physical contact with strangers, inappropriate sexual behaviour, verbal or physical aggression Loss of manners or decorum Lack of social etiquette, insensitive or rude comments, preference for crass jokes and slapstick humour, inappropriate choices of clothing or gifts Impulsive, rash or careless actions New gambling behaviour, driving or investing recklessly, overspending, gullibility to phishing/Internet scams Early apathy or inertia ≥ 1 of Apathy Reduced drive, stops previous hobbies, stops going out, reduced bathing or personal care Inertia Lack of persistence or completion of an activity, does not initiate activities or conversations Early loss of sympathy or empathy ≥ 1 of Diminished response to other people's needs and feelings Selfish or hurtful comments or actions, inability to perceive when someone is upset, embarrassed, or in pain, reduced appreciation of sarcasm or sophisticated humour Diminished social interest, interrelatedness, or personal warmth Emotionally cold or detached, lack of rapport in conversation, loss of interest or affection in relationships with friends or family members, reduced interest in sex Early perseverative, stereotyped or compulsive or ritualistic behaviour ≥ 1 of Simple repetitive movements Repetitive rocking, tapping, clapping, or rubbing Complex compulsive or ritualistic behaviours Hoarding, strict grooming or walking routines, timekeeping and counting, checking or sorting items, cleaning or tidying, new obsessions or interests (usually spiritual, religious, artistic, or musical) Stereotypy of speech Habitual repetition of particular words, sentences or topics Hyperorality and dietary changes ≥ 1 of Altered food preferences Sweet tooth (sweets, biscuits, ice cream), carbohydrates, or obsessive food fads Binge eating, increased consumption of alcohol or cigarettes Cramming food into mouth, overeating or messy eating, new addictions to alcohol or smoking Oral exploration or consumption of inedible objects Pica Neuropsychological profile – all three of Deficits in executive tasks Vary as per neuropsychological assessment used Relative sparing of episodic memory Relative sparing of visuospatial skills Other features of bvFTD (not in diagnostic criteria) Examples of specific symptoms Loss of insight Lack of awareness of own condition or symptoms Impaired social cognition Poor response to social or emotional cues; impaired performance on tests of theory of mind or emotion recognition Altered sensitivity to pain Heightened perception of a non‐painful stimulus or reduced response to painful stimulus; hypochondriasis or overly focusing on mild physical complaints Altered tolerance of temperature Inappropriate clothing for the ambient temperature, such as wearing multiple coats or blankets Psychotic features Delusions (usually somatic or paranoid) and hallucinations (usually visual or tactile) There are a number of clinical features, however, which are not part of the Rascovsky criteria and yet are relatively common in bvFTD. In particular, virtually all patients with bvFTD have impaired social cognition, with reduced ability to use a ‘theory of mind’ to see another person's point of view or imagine their feelings (Kumfor and Piguet 2012). Several studies have shown that patients with bvFTD also display significant impairments in emotion recognition, even when tested across multiple modalities, and have more difficulty recognizing negative emotions (Lavenu et al. 1999; Rosen et al. 2004; Fernandez‐Duque and Black 2005; Lough et al. 2006). They also have difficulty in expressing meaningful emotions, resulting in ‘emotional blunting’ (Neary et al. 1998; Sturm et al. 2011). Several groups have observed altered perception of surrounding environmental and internal somatosensory stimuli, including changes in tolerance of pain or temperature. Patients with bvFTD tend to have blunted perception of pain (Bathgate et al. 2001; Snowden et al. 2001; Carlino et al. 2010; Fletcher et al. 2015) and temperature (Ahmed et al. 2015; Fletcher et al. 2015). In bvFTD patients this can anecdotally manifest as wearing inappropriately heavy clothing or blankets in a warm clinic. Others have developed altered perception of sound or music (Seeley et al. 2008; Warren and Rohrer 2009; Barquero et al. 2010; Mahoney et al. 2011; Fletcher et al. 2013), with some patients merely developing heightened sensitivity to noise (Fletcher et al. 2013) and others frank musicophilia (Fletcher et al. 2013) or amusia (Barquero et al. 2010). These phenomena suggest that a variety of networks involved in sensory input processing and integration may be affected in FTD. Neuropsychiatric manifestations such as delusions or hallucinations are found in sporadic bvFTD and may be the sole presentation in patients with familial bvFTD (particularly those with C9ORF72 or GRN mutations, as discussed later). In a review of 751 cases of FTD published in 199 publications from 1950 to 2007, 46 (6%) of patients presented with schizophrenia, schizoaffective disorder, bipolar affective disorder, psychotic depression or another psychotic disorder; with 98% of these patients presenting aged < 60 (Velakoulis et al. 2009). In another large study of patients with a variety of neurodegenerative disorders, including bvFTD, nfvPPA, svPPA, AD, PSPS, CBS and ALS, 28.5% of patients had received a previous psychiatric diagnosis (usually depression), and this was much more common in patients who turned out to have bvFTD (50.7%) than nfvPPA (11.8%), svPPA (24.4%) or AD (23.1%) (Woolley et al. 2011). The typical previous psychiatric diagnosis in bvFTD patients in this study was schizophrenia or bipolar disorder. Young onset apparently sporadic bvFTD cases with FUS pathology have a particularly high (up to 50%) rate of psychiatric symptoms. The lack of other cognitive or neurological features early on commonly leads to young and older patients with sporadic and familial bvFTD being referred to and assessed within a psychiatric or psychogeriatric setting rather than in a specialist cognitive neurology or memory clinic (Lanata and Miller 2015). The obvious overlap between early bvFTD symptoms (lack of insight, prominent apathy, obsessive or compulsive behaviours, inappropriate sexual behaviour, binge eating, gambling and substance misuse, emotional lability or blunting, delusions and hallucinations) and psychiatric presentations (depression, obsessive–compulsive disorder, bipolar affective disorder and schizophrenia and other psychotic disorders) can initially lead to misdiagnosis of a neurodegenerative disease as a psychiatric disorder. Younger patients with ‘later than usual’ onset of neuropsychiatric disease, atypical or prominent behavioural features and any suggestion of multiple family members with significant psychiatric disease (e.g. needing long‐term or permanent admission to a mental health facility), ‘early onset dementia’, AD, FTD or MND, should be carefully assessed with a neurological examination, detailed family history and, wherever possible, formal neuropsychology and detailed magnetic resonance imaging (MRI). Patients with bvFTD classically have preserved episodic memory, at least early on in disease, helping differentiation from AD, but this is not always the case. Patients often have deficits in verbal and visual memory on neuropsychological assessment, even if they do not report memory problems; their performance is often worsened by poor strategy during assessments because of concurrent executive dysfunction and distractibility. However, a significant proportion of pathologically confirmed bvFTD cases have presented with a predominant amnestic syndrome (Hodges et al. 2004; Graham et al. 2005; Piguet et al. 2009; Irish et al. 2013), perhaps because of a higher occurrence of hippocampal sclerosis in the older age group (Baborie et al. 2011; Balasa et al. 2015) or mixed pathology (Balasa et al. 2015). This emphasizes the difficulties clinicians face in making a correct diagnosis, particularly in later onset cases, despite using currently available consensus criteria. The requirements for functional decline and neuroimaging abnormalities for diagnosing probable bvFTD, compared with previous diagnostic criteria (Neary et al. 1998), are particularly useful for excluding patients with a so‐called phenocopy syndrome, who may have bvFTD‐like symptoms but do not have bvFTD. Relatives of patients with phenocopies often seem certain that there is progression over time, but most neuropsychological assessments tend to dispute this, often showing normal values or mild but stable impairment (Hornberger et al. 2009). These patients have no or minimal atrophy on MRI and normal nuclear medicine imaging with PET (positron emission tomography) or SPECT (Single‐photon emission computed tomography) scans, and either remain stable or improve over time, without significant disruption of function (Davies et al. 2006). Previous diagnostic criteria for bvFTD (Neary et al. 1998) potentially allowed phenocopy cases to be falsely diagnosed as having bvFTD, as they presented almost identically on core diagnostic criteria (Hornberger et al. 2009), and these cases are notoriously difficult to tell apart, particularly at initial assessment. Repeated assessments of functional abilities over a 12‐month period (Mioshi and Hodges 2009), neuropsychological assessments of executive function (Hornberger et al. 2008) and social cognition (Kipps et al. 2009b), and neuroimaging using combined MRI and FDG‐PET (Kipps et al. 2009a) appeared most helpful in differentiating phenocopy syndromes from true bvFTD, hence their incorporation into the revised criteria (Rascovsky et al. 2011). One proviso to this is that some patients can occasionally actually have an atypical, very slowly progressive form of FTD, with very slow deterioration on repeated neuropsychological assessments over at least 15 years, and neuropathological confirmation of typical FTLD pathology (Brodtmann et al. 2013), which further confuses the clinical picture.

Primary progressive aphasia The key clinical characteristic of PPA is progressive and insidious language decline affecting at least one of speech production, object naming, syntax, or word comprehension (Gorno‐Tempini et al. 2011). Other cognitive or behavioural deficits can develop either early on or in late disease, but must not be the initial and predominant complaint and language must also remain the most impaired domain throughout the disease course (Mesulam 1982, 2003). The most recent diagnostic criteria for PPA have specified that three inclusion criteria, based on criteria developed by Mesulam (1982, 2001, 2003), must first be fulfilled for its diagnosis (Gorno‐Tempini et al. 2011): (i) the most prominent clinical feature is difficulty with language, (ii) these deficits are the principal cause of impaired daily living activities and (iii) aphasia should be the most prominent deficit at symptom onset and for the initial phases of the disease. In addition, none of the following criteria should be met: (i) the pattern of deficits is better accounted for by other non‐degenerative nervous system or medical disorders (e.g. tumour or stroke); (ii) the cognitive disturbance is better accounted for by a psychiatric diagnosis; (iii) there are prominent initial episodic memory, visual memory and visuoperceptual impairments and (iv) there is prominent, initial behavioural disturbance. Once these inclusion and exclusion criteria have been satisfied, one can go on to subdiagnose the syndrome as one of three PPA variants (svPPA, nfvPPA or lvPPA), as per diagnostic criteria for each variant (see Gorno‐Tempini et al. 2011, for a summary). Clinical diagnosis can be supplemented by information from neuroimaging analyses (leading to the more firm category, ‘imaging‐supported diagnosis’). If there is a clinical diagnosis of PPA (with or without neuroimaging support) and presence of either a known pathogenic gene mutation on DNA analysis or specific neurodegenerative pathology on histopathological analysis, this leads to a diagnosis of PPA ‘with definite pathology’. Different language features can be used to differentiate between the three variants and these are summarized in Table 2. However, not all patients clearly fit into a particular variant as they present with a number of features from across the spectrum of language dysfunction, and previously have been termed as having mixed disease (PPA‐M) (Mesulam et al. 2009). More recently patients have been diagnosed with ‘PPA‐not otherwise specified’ and over time the evolving syndrome may or may not become clearer (Harris et al. 2013b), while some are associated with a GRN mutation (Rohrer et al. 2010a). There is also a syndrome within the non‐fluent aphasia spectrum called progressive primary apraxia of speech, which progressively affects speech articulation and production because of impaired motor programming, but typically patients lack aphasia initially (Josephs et al. 2012). It can be associated with development of features of PSPS (Rohrer et al. 2010b; Josephs et al. 2014) or, less frequently, CBS (Josephs and Duffy 2008; Assal et al. 2012). Here, we discuss the clinical features of each PPA variant in turn, the language and behavioural features that overlap between them and with bvFTD, the presence of other common clinical features and the overlap with other conditions such as PSPS, CBS and MND. Table 2. Summary of clinical features across the syndromes of primary progressive aphasia Clinical features svPPA nfvPPA lvPPA Spontaneous speech (fluency, errors, grammar, prosody) Fluent, garrulous and circumlocutory, semantic errors, intact grammar and prosody Slow and hesitant, effortful ± apraxic, phonetic errors, may be agrammatic, aprosodic Hesitant, not effortful or apraxic, frequent word‐finding pauses and loss of train of sentence, intact grammar, intact prosody Naming Severe anomia with semantic paraphasias Moderate anomia with phonetic errors and phonemic paraphasias Mild to moderate anomia with occasional phonemic paraphasias Single word comprehension Poor Intact early on, but affected later on Intact early on, but affected later on Sentence comprehension Initially preserved, later on becomes impaired as word comprehension is impaired Impaired if grammatically complex Impaired, especially if long Single word repetition Relatively intact Mild to moderately impaired if polysyllabic, otherwise intact Relatively intact (compared with sentence repetition) Sentence repetition Relatively intact Can be effortful, impaired if grammatically complex Impaired, with length effect Reading Surface dyslexia Phonological dyslexia ± phonetic errors on reading aloud Phonological dyslexia Writing Surface dysgraphia Phonological dysgraphia Phonological dysgraphia

SvPPA SvPPA accounts for around 20% of cases of FTD (Johnson et al. 2005). It is a predominantly sporadic disorder (Rohrer et al. 2009a) and presents with a mean age of onset of 60, with a range of 40–79 years (Hodges et al. 2010), although is likely under‐diagnosed in older people, particularly as the semantic memory deficits can develop insidiously and are usually well‐masked by the perception of fluent speech, and use of commonly used empty speech terms such as ‘thing’ (Fletcher and Warren 2011; Hsieh et al. 2012). It is associated with not only bilateral, but markedly asymmetrical anterior temporal lobe atrophy at presentation, particularly affecting the inferior and middle temporal gyri, but also the anterior hippocampus and amygdala (Hodges et al. 1992; Mummery et al. 1999, 2000; Whitwell et al. 2005; Rohrer et al. 2009b). The majority of patients present with predominant left temporal lobe atrophy, which leads to the classical language disorder of svPPA, characterized by early loss of semantic memory and resultant language dysfunction (Snowden et al. 1989; Hodges et al. 1992). Less frequently patients present with predominant right temporal lobe atrophy at onset, often called right temporal lobe atrophy (RTLA) FTD or ‘right SD’ cases (Evans et al. 1995; Thompson et al. 2003; Chan et al. 2009). The language impairment in svPPA initially manifests as reduced semantic knowledge for words, objects and concepts, which affects spoken and written language through development of a reduced vocabulary and resultant anomia (Warrington 1975). As atrophy worsens and extends across to the right temporal lobe and to the inferior frontal lobe, insula, and more posterior left temporal lobe (Seeley et al. 2005; Brambati et al. 2009), it impairs semantic function across multiple modalities, leading to associative agnosia for visual, auditory (Bozeat et al. 2000; Goll et al. 2010), tactile (Coccia et al. 2004), olfactory (Rami et al. 2007) and gustatory (Piwnica‐Worms et al. 2010) stimuli. Patients lose their grasp for increasingly imprecise or broad semantic terms and concepts, with responses to stimuli becoming more general (e.g. ‘poodle’‐‐>‘dog’‐‐> ‘animal’‐‐> ‘don't know’) over time. Patients with svPPA generally report word‐finding difficulties, which may start off as being only for specialist, low‐frequency (rarely used) words (such as names of flowers for a gardener, or facial anatomical terms for a dentist). This worsens to affect commonly used words. Patients may ask relatives to explain the meaning of a word someone has said or that they have read (Fletcher and Warren 2011; Warren et al. 2013), which at first is usually an unusual word, such as ‘orangutan’. Relatives may report the patient does not seem to understand what is being said to them, or ‘appears deaf’, asking for instructions to be repeated several times (Rohrer et al. 2008). Clinically, the language dysfunction in svPPA is characterized by fluent speech, which is often garrulous or difficult to interrupt but has frequent circumlocutions (e.g. ‘thing’ or ‘whatsit’), circumlocutory phrases (imprecise phrases that contain vague descriptions or explanations of the word aimed for, e.g. ‘the thing with the tail that you ride’ for ‘horse’) and semantic paraphasias (similar but incorrect words often from within the same category, e.g. ‘cat’ for ‘dog’), used by the patient to work around their lack of vocabulary (Hodges and Patterson 2007). There can be brief hesitations during word‐finding moments, but overall the speech is much more fluent than the effortful speech in nfvPPA or speech with significant pauses in lvPPA. On assessment, patients have anomia on confrontation naming tasks (which may appear subtle without detailed probing by a full neuropsychological assessment), and impaired comprehension of the meaning of single words, particularly on low‐frequency items such as ‘monocle’. Later on, there is anomia and impaired comprehension of pictures, sounds, smells and tastes. Patients often have difficulties with reading and writing, particularly with irregularly spelt words, leading to the phenomenon of a surface dyslexia or surface dysgraphia (Warrington 1975; Baxter and Warrington 1987). For example, patients will pronounce ‘sew’ as ‘soo’ or ‘yacht as ‘yatched’, as they have lost semantic knowledge of the word meaning (and hence the atypical rule for how it should be pronounced), relying on sounding out the word as written using superficial rules only (Rohrer et al. 2008). Other cognitive domains are usually unaffected, including episodic and topographical memory, visuoperceptual function, praxis, calculation and non‐verbal executive function (Warrington 1975; Cipolotti and Maguire 2003; Gordon et al. 2010). Other aspects of language such as speech articulation and prosody, and repetition of spoken words and phrases are also usually preserved. Grammar is intact, although as vocabulary declines, grammar can sound abnormal, because of the use of ‘paragrammatic’, circumlocutory phrases and broad classes of terms, which disrupt the normal flow of the sentence (Gorno‐Tempini et al. 2011). Patients with typical svPPA evolve over time (due to spread of disease) to develop behavioural changes, which can make it difficult to differentiate these patients from bvFTD patients clinically if presenting late in the disease. However, behavioural changes particularly found in typical svPPA include obsessionality, mental rigidity, narrowed interests (often affecting eating behaviour, daily routines and fixations on specific activities, e.g., jigsaw puzzles) (Snowden et al. 2001; Thompson et al. 2003), more compulsive and complex repetitive behaviours (Snowden et al. 2001), heightened perception of pain and sensory stimuli leading to hypochondriasis and increased sensitivity to temperature (Fletcher et al. 2015). Cases of svPPA with altered auditory perception such as hyperacusis and persistent tinnitus (of central rather than peripheral origin) have also been observed (Mahoney et al. 2011). Patients with the right temporal variant can be difficult to identify purely from a clinical assessment as they often have early behavioural changes and less prominent semantic difficulties initially (Chan et al. 2009). The key distinguishing feature of RTLA cases is early prosopagnosia (impaired recognition of familiar faces) (Tyrrell et al. 1990; Evans et al. 1995; Gainotti et al. 2003; Thompson et al. 2003; Joubert et al. 2006), but when compared with cases with predominant left temporal lobe atrophy, RTLA cases also report more difficulties with topographical memory (potentially due to right hippocampal atrophy) (Chan et al. 2009), and may have a more bizarre affect (Thompson et al. 2003). They also tend to have less insight into their disease (Thompson et al. 2003) and can develop other unusual features such as hyper‐religiosity (Edwards‐Lee et al. 1997; Chan et al. 2009). Not all RTLA patients will develop semantic impairment and initial studies have suggested that there are at least two RTLA variants: one that is the mirror analogue of svPPA with disease spread occurring inter‐hemispherically to the left temporal lobe, and another with behavioural symptoms where atrophy spreads intra‐hemispherically, predominantly affecting the right frontal and parietal lobe (Kamminga et al. 2015). These tend to have differing underlying pathologies as well: the right SD cases have FTLD‐TDP type C pathology, whereas patients with bvFTD rarely have this subtype (Rohrer et al. 2011a; Lashley et al. 2015), potentially affecting accurate targeting of future treatments towards the different disease groups.

NfvPPA Approximately 25% of patients with FTD present with nfvPPA (Johnson et al. 2005). The classical neuroimaging feature is atrophy of the left posterior (and inferior) frontal lobe and insular cortex (Rohrer et al. 2009b). In contrast to the fluent aphasia observed in svPPA, patients with nfvPPA have non‐fluent speech, with the two core features being agrammatism and slow laboured speech production (‘effortful speech’) (Gorno‐Tempini et al. 2011). In some patients the former impediment is dominant, and in others, the latter, but in most cases the disease evolves to result in both features (Rohrer et al. 2010c). Patients with nfvPPA tend to present earlier than patients with svPPA, as speech is obviously disrupted and sounds abnormal early on (Hsieh et al. 2012). Importantly, single word comprehension and object knowledge are preserved, as semantic memory is intact, and this particularly helps to differentiate from the semantic variant in early disease (Gorno‐Tempini et al. 2011). Patients report word‐finding difficulties and do display anomia, but the anomia is less severe than svPPA cases. Speech agrammatism manifests as use of short, simple phrases which can sound muddled and ‘telegraphic’, because of omission of short connecting words and other function words, use of words in the wrong order and misuse of word endings, verb tenses, pronouns, prepositions and conjunctions (Mesulam 2003; Rohrer et al. 2008). There is also difficulty in comprehending grammar, leading to impaired sentence comprehension, particularly if sentences are long and syntactically complex (Grossman and Moore 2005). Patients can develop binary word reversals (typically yes/no, or pronouns, e.g., he/she) saying the opposite word to what they intended (Frattali et al. 2003) or utter sudden, unintended, stereotyped responses such as ‘don't know’ to different questions before giving the correct answer (Snowden and Neary 1993). Speech apraxia impairs the patient's ability to programme and plan the motor aspects of speech production properly, leading to effortful trial and error ‘groping’ of orofacial movements in the effort to produce the correct sounds (Duffy 2006; Josephs et al. 2006). Some patients perseverate on consonants or syllables, leading to a new ‘stuttering’ quality to speech as the initial presenting symptom (Kertesz et al. 2003). The variability in an apraxia of speech can lead to a misdiagnosis of a ‘functional stutter’ or functional speech disorder, i.e., of a non‐organic basis by non‐specialists, particularly as symptoms can fluctuate and become worse with anxiety or effort. The prosody of speech is also disrupted, thereby affecting its natural rhythm, rate (commonly leading to slowing), volume, or intonation (Josephs et al. 2006). There are typically distorted speech sounds (phonetic errors) because of errors in execution of programmed speech sounds, typified by syllable or consonant deletions, insertions, substitutions, distortions, repetitions and prolongations such as ‘capititain’ rather than ‘captain’ (Duffy 2006; Gorno‐Tempini et al. 2011), which can all make speech sound ‘jumbled up’ to the patient and their relatives. Writing can be intact or show grammatical errors later on in disease. With these core speech production features, the agrammatism affects language in a broader sense. Repetition of single words is relatively preserved (except for more complex multi‐syllabic words which becomes effortful), but repetition of longer sentences that are grammatically complex is affected. Over time speech deteriorates to a point where the patient has extreme difficulty making them understood and eventually mutism ensues, while this can be an early feature in some cases (Gorno‐Tempini et al. 2006). Many patients switch to non‐oral methods of communication such as writing on a notepad, or electronic language applications on handheld tablet computers. Orofacial/buccofacial apraxia, is also seen, which impairs the patient's ability to plan oral movements, leading to difficulty initiating swallowing, coughing and yawning (Tyrrell et al. 1991). On bedside testing, patients are unable to perform these actions to command, usually responding by repeating the word ‘cough’ or ‘yawn’ rather than the action itself (Tyrrell et al. 1991). Many patients display limb apraxia, particularly affecting the right side. Although subtle initially, this often worsens, progressively impairing hand function (Mesulam 2003).

LvPPA Around 30% of patients with PPA have the more recently described syndrome lvPPA (Kertesz et al. 2003; Gorno‐Tempini et al. 2004, 2008; Rosen et al. 2006). The hallmark imaging feature is left posterior temporoparietal atrophy encompassing the posterior superior temporal lobe, inferior parietal lobe, precuneus and mesial temporal lobe (Gorno‐Tempini et al. 2004; Rohrer et al. 2013a). The syndrome is thought by some to be an atypical and unihemispheric presentation of AD (Ahmed et al. 2012; Rohrer et al. 2012), although associated pathology is not universally ‘AD‐like’ (Harris et al. 2013a; Mesulam et al. 2014) and it is difficult to predict based on clinical features which patients have underlying AD versus other pathology (Chare et al. 2014). The key clinical features of lvPPA are frequent word‐finding pauses, anomia and impaired sentence (rather than single word) repetition (Gorno‐Tempini et al. 2004, 2008). There is also preserved single word comprehension and object knowledge as semantic memory is intact, but impaired comprehension of longer sentences, without agrammatism or apraxia of speech (Gorno‐Tempini et al. 2008; Rohrer et al. 2012). On speech assessment, there are frequent pauses (as the patient tries to retrieve the right word rather than apraxia of speech), and phonological errors (which are well articulated and not distorted, but definitely incorrect, such as ‘coptain’ rather than ‘captain’) due to difficulty with the phonology of the anticipated word. These phonological errors also appear in writing, and there may be a phonological dyslexia, affecting reading of new or non‐sense words. The short‐term, phonological memory deficit in lvPPA also characteristically impairs sentence repetition in a length‐dependent manner, but spares single word repetition (Gorno‐Tempini et al. 2008). Although comprehension of single words is intact, there can be difficulty in comprehending longer sentences, because of the deficit in phonological memory, but this is not affected by grammatical complexity like in nfvPPA. Patients with severe lvPPA can be difficult to differentiate from patients with non‐fluent PPA. However, the key differentiating features for lvPPA are lack of agrammatism, lack of apraxia speech, lack of orofacial apraxia, preserved prosody and impaired sentence repetition (Gorno‐Tempini et al. 2008; Chare et al. 2014). Limb apraxia is often present due to parietal involvement (Rohrer et al. 2012).