Conceptual framework for the definition of preclinical and prodromal frontotemporal dementia

The presymptomatic stages of frontotemporal dementia (FTD) are still poorly defined and encompass a long accrual of progressive biological (preclinical) and then clinical (prodromal) changes, antedating the onset of dementia. The heterogeneity of clinical presentations and the different neuropathological phenotypes have prevented a prior clear description of either preclinical or prodromal FTD. Recent advances in therapeutic approaches, at least in monogenic disease, demand a proper definition of these predementia stages. It has become clear that a consensus lexicon is needed to comprehensively describe the stages that anticipate dementia. The goal of the present work is to review existing literature on the preclinical and prodromal phases of FTD, providing recommendations to address the unmet questions, therefore laying out a strategy for operationalizing and better characterizing these presymptomatic disease stages.

therapeutic approaches, at least in monogenic disease, demand a proper definition of these predementia stages. It has become clear that a consensus lexicon is needed to comprehensively describe the stages that anticipate dementia. The goal of the present work is to review existing literature on the preclinical and prodromal phases of FTD, providing recommendations to address the unmet questions, therefore laying out a strategy for operationalizing and better characterizing these presymptomatic disease stages.

K E Y W O R D S
definition, frontotemporal dementia, frontotemporal lobar degeneration, mild cognitive and/or behavioral and/or motor impairment, mild cognitive impairment, preclinical, presymptomatic, prodromal which may form part of either a progressive supranuclear palsy (PSP) or corticobasal syndrome (CBS), 6 and there is considerable clinical overlap with motor neuron disease (MND). 7 The presymptomatic stages of FTD are still poorly defined and likely encompass a long accrual of progressive biological (preclinical) followed by clinical (prodromal) changes, antedating the onset of dementia. The heterogeneity of clinical presentations and the different neuropathological phenotypes have prevented a prior clear description of either preclinical or prodromal FTD. Recent advances in therapeutic approaches, at least in monogenic disease, make proper definition of these presymptomatic stages more urgent. As postulated for Alzheimer's disease (AD), the ability to intervene early may offer a chance to delay or even prevent neurodegeneration. In AD, the literature has suggested the conceptual framework of a preclinical biologically active process that precedes the onset of a prodromal or mild cognitive impairment (MCI) phase, which is then followed by dementia. 8,9 The heterogeneous presentation of FTD suggests that a wider set of clinical features might present in the prodromal phase compared to AD.
Nonetheless, a similar conceptual framework to MCI could be translated to the FTD field. In this view, we may define a preclinical FTD stage in those subjects with an ongoing neuropathological process but without clinical abnormalities, and a prodromal stage in those subjects with the onset and progression of subtle clinical symptoms.
A privileged point of view for studying the preclinical and prodromal phases of FTD is provided by its genetic forms. Indeed, familial aggregation has been reported in a significant proportion of people with FTD (up to 40% of cases), with mutations in the microtubule-associated protein tau (MAPT) and progranulin (GRN) gene, or a pathogenic expansion in the chromosome 9 open reading frame 72 (C9orf72) as the most common cause of monogenic disease. 10 Mutations in MAPT lead to abnormal tau accumulation, while GRN and C9orf72 pathogenic variations are associated with TDP-43 deposition. 11 The study of family members bearing a pathogenic mutation has allowed the naturalistic observation of the shift from preclinical and prodromal status to overt disease. There is a wide variation in the age at onset, both within mutation class and within families with the same mutation at least in GRN and C9orf72 mutations, 12 and possible disease modifiers have been recently reported, even though penetrance is high at age 75. 13 Moreover, several studies have faced the challenge of detecting a clinical, biological, or imaging signature preceding the onset of dementia. A major contribution in this field has been provided by the international consortia devoted to the extensive evaluation of presymptomatic subjects carrying pathogenic mutations.
The ongoing European-and Canadian-based Genetic Frontotemporal dementia Initiative (GENFI, www.genfi.org), the US-based Advanc-  [17][18][19][20] These studies collaborate together under the auspices of the FTD Prevention Initiative (FPI). 12 It is therefore important for observational studies and clinical trials to determine specific parameters and measures of preclinical and prodromal FTD, to share a common lexicon when identifying patients in the earliest phases of disease. However, several outstanding issues still need proper analysis and scrutiny. To this end, the goal of the

HOW DO WE DEFINE THE ONSET OF PRECLINICAL DISEASE?
The onset of a preclinical disease stage may be theoretically defined by the occurrence of first signs of protein misfolding, presumably initially without either neuronal dysfunction or neurodegeneration, and with no clinical FTD-related symptoms. One of the key questions in the current literature is therefore how we define this switch from a "no disease" stage to a "preclinical stage" with available markers (see Figure 1).
Conceptually, while the disease process may be initiated through misfolded proteins forming neurotoxic oligomers, the first identifiable hallmark of a preclinical disease stage is the abnormal accumulation of pathogenic protein aggregates within cells, including (1) hyperphosphorylated tau, (2) TDP-43 immunoreactive inclusions, (3) FET family proteins (consisting of FUS, Ewing's sarcoma protein [EWS], and TATAbinding protein associated factor 2N [TAF15]), (4) dipeptide repeat proteins (DPR), or (5) still-to-be-defined proteins in those with frontotemporal lobar degeneration-ubiquitin proteasome system (FTLD-UPS) pathology. 1,21 Reliable in vivo biomarkers able to predict the two main proteinopathies, namely tau or TDP-43, are not yet available. No TDP-43 positron emission tomography (PET) tracer has been investigated as of yet, and tau PET imaging studies have led to variable results, with the main limitation in the primary tauopathies being the non-specific/offtarget binding and variable affinity for different tau species. 22,23 Sim-ilarly, fluid biomarkers of tau and TDP-43 in cerebrospinal fluid (CSF) or blood have not shown specificity for FTLD pathology. While blood phosphorylated tau (p-tau 181 and p-tau 217 ) assays have recently been shown to be useful to identify AD, they do not identify primary tauopathies including FTLD. [24][25][26][27] Markers of blood and CSF TDP-43   measurements have been developed but are not specific for TDP-43 pathology. 28,29 Phosphorylated TDP-43 markers and CSF TDP-43 realtime quaking-induced conversion reaction (RT-QuIC) may improve specificity, 30,31 but these results await confirmation. TDP-43 aggregates may be found even in a subset of AD patients, or in other neurodegenerative disorders or in some aged people, thus TDP-43 biomarkers may be not completely specific. 32,33 Markers for the FET proteins have also not yet been developed.
Recent work has identified the presence of a CSF measure that is specific to C9orf72 expansion carriers. One of the key pathophysiological mechanisms in C9orf72-related disease is the accumulation of sense and antisense transcripts of the expanded repeats. These RNA transcripts serve as templates for the synthesis of DPRs through repeat associated non-ATG (RAN) translation. So far, only one of these, the glycine-proline-repeating protein or poly(GP), has been shown to be measurable in CSF, [34][35][36] being increased in C9orf72 expansion carriers in both the presymptomatic and symptomatic phase, and normal in controls. This suggests it could be useful as a preclinical biomarker in genetic FTD. [37][38][39] Importantly, reports of autopsy studies in C9orf72 expansion carriers have also described widespread DPR protein pathology prior to the formation of TDP-43 inclusions and neuronal loss, [40][41][42] suggesting that at least for C9orf72 expansion carriers, the onset of the preclinical stage is defined by the presence of DPR proteins rather than TDP-43 pathology.
There is also a need for more studies examining the extent of neuropathological findings consistent with FTLD in healthy older people.

Recommendation:
The preclinical phase of FTD should theoretically extend from the earliest signs of protein misfolding to the onset of the first clinical symptom of FTD. Based on current knowledge, the onset of a preclinical stage cannot be reliably identified with available biomarkers at this time except potentially for those with C9orf72 expansions. We recommend that ongoing research aims to identify both PET tracers and fluid biomarkers that can sensitively and specifically show the presence of tau, TDP-43, and FET pathology.

HOW DO WE DEFINE FURTHER STAGES OF PRECLINICAL DISEASE?
The preclinical disease stage may be characterized by when protein accumulation and misfolding is initiated, but later preclinical stages can also be defined. Accumulation of toxic proteins leads to neuronal dysfunction with multiple cellular mechanisms being affected, including the function of mitochondria and stress granules, autophagy, and transcription. The outcome of this is neuronal loss, that is, neurodegeneration. Both dysfunction and loss of neurons occur prior to the onset of clinical symptoms (see Figure 1). 18  Nonetheless, MRI represents one of the most powerful tools to study in vivo neurodegenerative disorders, with a wide range of possible approaches to explore incipient neurodegeneration. 52,53 The majority of imaging studies in preclinical FTD have used volumetric T1-weighted MRI to investigate changes in gray matter structure and to measure brain volume, the rate of brain atrophy, and the volumes of specific brain regions of interest. [54][55][56][57][58][59][60] In monogenic FTD, volumetric MRI analysis shows significant brain atrophy, first detectable in the insula, at least 10 years before expected symptom onset. 14 Diffusionweighted MRI detects white matter damage including axonal loss. In genetic FTD, changes to diffusivity have been found in white matter tracts many years before symptom onset. 61  In the presymptomatic period, analysis seems to suggest that levels change not long prior to symptom onset, increasing by 3-to 4-fold during conversion. [62][63][64] While longitudinal NfL measurements could be used to identify mutation carriers approaching symptom onset, 65 NfL needs to be further studied on a single subject basis, and in particular, studies showing whether it is sensitive enough to detect neurodegeneration prior to early symptoms (i.e., prior to a prodromal stage).

IS THERE A "NO DISEASE" PHASE IN GENETIC FTD PRECEDING THE ONSET OF PRECLINICAL DISEASE?
The conceptual timeline of FTD natural history typically includes a healthy stage, with "no disease," followed by preclinical and prodromal disease to overt dementia (see Figure 1). In monogenic FTD subtypes, some biomarkers appear to be altered from birth and many are abnormal even in young adulthood. This raises the question whether there is a neurodevelopmental dimension to FTD, and the existence of a stage that is without disease, or without neuropathological abnormalities. By analogy with another genetic dementia, Huntington's disease, there may even be fetal neurodevelopmental abnormalities. 66 Pathogenic loss-of-function mutations in GRN lead to haploinsufficiency, with blood and CSF levels of progranulin reduced to < 50% of normal levels. [67][68][69][70][71][72] Low serum, plasma, or CSF progranulin levels have high accuracy in detecting pathogenic GRN mutations, 72-75 with low levels observed from the earliest time period in GRN mutation carriers, likely antedating TDP-43 neuropathology. At present, studies have not been performed in children (< 18 years) to understand whether levels are low from birth, but the assumption is that they are, given the known pathophysiology. [73][74][75] As mentioned above, C9orf72 expansion carriers have widespread DPR protein pathology early in life. [40][41][42] While similarly to GRN mutation carriers studies of fluid biomarkers show abnormal levels (here of raised poly[GP] concentrations) from at least the fourth decade of life, 37,39,76,77 and no studies have been performed in children, there is a less clear assumption of abnormal levels from birth and studies in a pediatric cohort would be highly informative.
Recommendation: Based on current knowledge it is not clear if a "no disease" stage exists after normal childhood development, for some forms of genetic FTD. For people with GRN mutations, there may well be a phase during which a biological disruption is ensuing, but which is not accompanied by an abnormal accumulation of specific pathologic proteins. For people with C9orf72 expansions, the accumulation of DPRs appears to occur at least in young adulthood, but how early is unknown. Considering also the higher rate of developmental disorders in offspring of patients with FTD, [78][79][80][81] this has suggested the hypothesis of some forms of genetic FTD being neurodevelopmental disorders, in which the boundary with "no disease" is even more indistinct. Studies in pediatric at-risk genetic FTD cohorts, while ethically more complex, will be required to answer these questions more fully.

HOW DO WE DEFINE ONSET OF PRODROMAL DISEASE?
Prodromal FTD may be defined as the presence of subtle cognitive and/or behavioral changes (see Figure 1). Based on studies from large genetic cohorts, the cognitive prodromal phase may start with gradual and progressive executive dysfunction, occurring in isolation or associated with other cognitive changes, such as impaired social cognition or language disturbances. These may be accompanied by behavioral symptoms, such as apathy, disinhibition, loss of empathy, compulsive behavior, and change in appetite or subtle motor deficits, 14,65,[82][83][84][85][86][87][88][89][90] which are observed by the patient, informant, or clinician, and represent a clear change from the person's usual behavior (see Box 2).
Unlike in AD, for which the concept of MCI was developed to define the prodromal stages, [91][92][93] no detailed characterization of prodromal FTD has been reported. The direct application of the term MCI to FTD is fraught with difficulties given the complex clinical presentation of FTD, which can be heralded by different phenotypes. Attempts to define MCI-like or prodromal stages in FTD have been undertaken with mixed results. Initial criteria for mild behavioral impairment (MBI) excluded serious memory complaints, ignoring cognitive functioning, despite its apparent importance for the early and accurate detection of FTD. 94,95 The term frontotemporal-MCI (FT-MCI) was later proposed, with criteria including also behavioral symptoms but not requir-B O X 2 Proposed recommendation for clinical features of prodromal FTD Gradual and progressive cognitive and/or behavioral and/or motor changes compared to prior functioning and reported by patient or informant, with preservation of independence in functional abilities of daily living, occurring along with one or more of the following features: • Objective evidence of a dysexecutive syndrome, occurring in isolation or associated with other cognitive changes, such as impaired social cognition, as measured by tests with established specificity for FTD • Language deficit, as measured by tests with established specificity for FTD • Behavioral changes: apathy, disinhibition, loss of empathy, compulsive behavior, and change in appetite • Signs and symptoms of parkinsonism or motor neuron disease ing the onset to be insidious and progressive, creating potential confusion with delirium, mania, and other conditions. 96 The phonological similarity in naming with Petersen MCI criteria could also generate confusion. 97 Finally, provisional MBI criteria have been recently proposed, excluding patients younger than 50 years and not including cognitive disturbances. 98 Thus, a unifying characterization of prodromal FTD is currently lacking.

Recommendation:
The onset of prodromal FTD is characterized by gradual and progressive cognitive and behavioral symptoms, which may be observed by the patient, informant, or clinician, as representing a clear change compared to prior functioning (see Box 2). Given that the onset of prodromal FTD can present with any of behavioral, cognitive, motor or language change, we suggest the label of mild cognitive and/or behavioral and/or motor impairment (MCBMI) to capture the complexity of the clinical phenotype under a single unifying characterization (see next section).

HOW MAY WE ASSESS MCBMI DUE TO FTD?
As with many other neurodegenerative conditions, behavioral and cognitive changes may be present in FTD years before the onset of manifest dementia. These changes clearly describe the switch from preclin-

HOW DO WE INCLUDE THE PRODROMAL NEUROPSYCHIATRIC FEATURES (PARTICULARLY OF C9orf72) WITHIN THIS FRAMEWORK?
A growing body of evidence describes neuropsychiatric symptoms as early markers of decline along the neurodegenerative spectrum. 121 This is of particular interest in prodromal FTD, in which behavioral symptoms represent the core feature of the disease. What is emerging is that, alongside behavioral symptoms already described in cur-rent clinical criteria for FTD, such as disinhibition, apathy, loss of empathy, perseverative or compulsive behavior, and hyperorality, other neuropsychiatric symptoms are frequently reported. These manifestations, which are still not defined as FTD core symptoms, should be sought during evaluation and should be considered possible presenting symptoms in the prodromal stages. 122 In particular, anxiety and depression as well as hallucinations and delusions may be present in people with FTD, the latter highly expressed in C9orf72 expansion carriers compared to the other FTD subtypes. 123  Neuropsychiatric evaluation tools will have to consider past psychiatric or personality profiles to reliably identify new emerging prodromal symptoms.

HOW DO WE INCLUDE MILD FEATURES OF PARKINSONISM OR MOTOR NEURON DISEASE WITHIN THIS SCHEME?
A significant percentage of patients with FTD have associated extrapyramidal symptoms, which can be nonspecific, not meeting criteria for a particular disorder, or may fit the criteria for either PSP (Richardson syndrome) or CBS. 5,6,12,[124][125][126][127][128][129][130][131] In both sporadic and genetic FTD, movement disorders can sometimes be the initial presentation. 132,133 There is also considerable clinical overlap with MND. 7 Considering that all these diseases are included under the frontotemporal lobar degeneration umbrella term and that most pathogenic mutations may lead to one of these clinical syndromes, Recommendation: Motor symptoms are a common feature in FTD, and it may be argued that the onset of isolated movement disorders in the absence of cognitive symptoms could also be defined as a prodromal phase of FTD. We propose a unified approach, potentially including motor features in the prodromal FTD construct, that is, MCBMI. Further studies assessing isolated initial motor symptoms at the onset of sporadic FTD are required.

HOW DO WE DEFINE PHENOCONVERSION?
Applying the definition of "dementia," namely the presence of cognitive deficits that are significant enough to interfere with instru- to define conversion to FTD. In psychiatry, the presence of a mental disorder is defined as a condition that causes significant distress or impairment of personal functioning in social, occupational, or family activities, and must not be merely an expectable response to common stressors and losses. 137,138 It is worth noting that the National Institute on Aging-Alzheimer's Association criteria specifically state that a diagnosis of dementia is appropriate in the setting of interference with the ability to function at work or at usual activities, and that the change represents a decline from prior functioning, with changes in personality or behavior plus one other more classic cognitive domains. 139 As such, conversion to dementia could be defined by symptoms that lead to one or more of the following consequences: (1)

WHAT MODIFIES STAGE AND PROGRESSION OF DISEASE?
The risk of progression and natural history of preclinical and prodromal FTD may depend on modulating factors, for which the magnitude and interaction have yet to be determined. It has been postulated that certain lifetime experiences, including education, leisure activities, and occupational attainment, may be proxies of cognitive reserve and may modulate brain resistance and resilience. 140,141 In prodromal FTD, it has been shown that higher educational achievements are associated with greater gray matter volumes, suggesting that subjects with higher education are able to better counteract the detrimental effects of a pathogenetic mutation. 13 Bilingualism, another emerging aspect of cognitive reserve that has been shown to have an impact also in AD, [142][143][144] has been found to delay the onset of dementia in bvFTD but not in PPA. 145 Longitudinal studies have shown that increased education, but also active lifestyles, may also facilitate both brain reserve and brain maintenance in the prodromal stages of genetic FTD, 146,147 suggesting that cognitive reserve may confer clinical resilience, even in autosomal dominant FTD.
Along with modifiable modulators, even non-modifiable genetic factors have been identified and associated with age at disease onset in FTD. The most established genetic factor, at least in TDP-43 proteinopathies, is the transmembrane protein 106B (TMEM106B) gene. 148 It has been suggested that the TMEM106B rs1990622 polymorphism might modulate progranulin plasma levels, thus affecting age at symptom onset in GRN mutation carriers. 149,150 Accordingly, subjects with prodromal FTD due to GRN mutations and bearing the TMEM106B TT genotype showed greater functional brain damage than those with CT/CC TMEM106B genotypes. 13,151 In prodromal FTD-TDP-43 due to C9orf72 expansion, the relationship is less clear, and it has been suggested that TMEM106B might be able to affect disease pathology, but with an opposite association. 13,152,153 This effect may be an example of the general phenomenon of epistasis, in which a genetic variant is beneficial on some genetic backgrounds but deleterious in others. 152,154 In the same view, other genetic modifiers, such as apolipoprotein E genotype or MAPT haplotypes, should be considered. There are some issues that should be considered regarding MCBMI-FTD. FTD is a relatively rare disorder 156 and with a stronger genetic trait than AD. 12  which the pre-test probability that behavioral or cognitive symptoms will lead to FTD is much lower, needs to be further addressed. Initial findings suggest that clinical presentations (including cognitive, behavioral, and motor) are very similar between genetic and sporadic FTD. 158,159 Second, in MCBMI-FTD due to pathogenetic mutations we do not need diagnostic markers, but require prognostic markers, while in sporadic MCBMI-FTD we need both.
Most importantly, we should consider genetic MCBMI-FTD and sporadic MCBMI-FTD as distinct entities regarding treatment approaches. Finally, as with the symptomatic FTD stage, MCBMI-FTD also requires markers of phenotype prediction and markers of proximity to disease onset.
Several issues remain unanswered, including: how do we account for FTD phenocopies; what are the ethical issues in making an earlier diagnosis, informing subjects about biomarkers when it is still uncertain if it will progress to clinical FTD?
All the above considerations represent the roadmap of the recently established GENFI FTD Staging Working Group, whose main objectives will be to answer exhaustively the outstanding issues reported in the present proposal, to identify biomarkers in preclinical and prodromal FTD, and to plan larger collaborative international studies to test the utility and validity of this proposed new approach.
Our ability to carefully characterize the preclinical and prodromal stages of FTD will help in early disease detection, in enabling patient stratification, and in tailoring therapeutic selection for each patient.