Behavioral variant frontotemporal dementia in patients with previous severe mental illness : a systematic and critical review

Leandro Boson Gambogi https://orcid.org/0000-0002-1268-8872; Henrique Cerqueira Guimarães https://orcid.org/0000-0002-6680-7099; Leonardo Cruz de Souza https://orcid.org/0000-0001-5027-9722; Paulo Caramelli https://orcid.org/0000-0002-4786-6990 Correspondence: Paulo Caramelli; Departamento de Clínica Médica Faculdade de Medicina da UFMG; Avenida Alfredo Balena, 190 / Sala 246; 30130-100 Belo Horizonte MG, Brasil; E-mail: caramelli@ufmg.br Conflict of interest: There is no conflict of interest to declare. Received 26 November 2018; Received in final form 05 May 2019; Accepted 11 May 2019. ABSTRACT Objectives: To explore the relationship between severe/serious mental illness (SMI) and the behavioral variant of frontotemporal dementia (bvFTD), as the patterns of symptoms and cognitive performance that characterize both disorders share similarities. Methods: We performed a systematic review investigating what has already been published regarding the relationship between bvFTD and SMI. Studies were selected from PubMed and LILACS databases, including those published up to February 12, 2018. The search strategy included the following terms: “frontotemporal dementia” plus “bipolar”, OR “frontotemporal dementia” plus “schizophrenia”, OR “frontotemporal dementia” plus “schizoaffective”. Publications without abstracts, case reports with absent genetic or histopathological confirmation, reviews and non-English language papers were excluded across the search process. Results: The search on PubMed retrieved 186 articles, of which 42 met eligibility criteria. On the LILACS database, none met the requirements. Generally, three major research aims were identified: 1) to look for frontotemporal lobar degeneration-associated genetic abnormalities in patients with prior SMI; 2) to compare the cognitive profile between patients affected by neurodegenerative disorders and schizophrenic patients; 3) to highlight the association between bvFTD and preceding psychiatric conditions and/or distinguish them both. The investigated mutations were found infrequently in the studied SMI samples. Cross-sectional studies comparing cognitive performance between bvFTD and psychiatric disorders mostly found no remarkable differences. There were only a few case reports identifying definite frontotemporal lobar degeneration in patients with previous psychiatric diagnoses. Conclusions: The available evidence demonstrates how fragile the current understanding is regarding the association between bvFTD and prior SMI.

Although mortality is higher and life expectancy is far shorter in people with severe/serious mental illness (SMI), recent epidemiological studies have suggested that this group of individuals is also affected, albeit to a minor extent, by the same demographic changes that have driven population aging worldwide 1 .According to epidemiological research in other populations, it is reasonable to conceive that in these individuals with SMI, who are usually exposed to an unhealthy lifestyle, cognitive decline and overt dementia in the long run are probable outcomes if they reach old age.The relationship between severe psychiatric disorders and dementia has been recognized remotely, such as Kraepelin's description of "dementia praecox".Additionally, nowadays it is undisputed that psychiatric symptoms, such as mood changes, delusions, hallucinations, anxiety and others, constitute an essential hallmark of dementia.
The diagnosis of dementia in a person with a lifetime severe psychiatric disorder, such as severe relapsing unipolar depression, bipolar affective disorder (BD), and schizophrenia (SCZ) spectrum disorders is a challenging task.Most major neurocognitive disorders diagnostic proposals state that the patients' deficits "should not be better explained" by a psychiatric condition.In fact, there is a general tendency to attribute to a basal disorder, either neurological or psychiatric, novel symptoms or signs allegedly arising from brain malfunction.
Unfortunately, however, there are few reports that have systematically evaluated the issue of dementia superimposing lifetime psychiatric disorders.It is unclear whether the dementia symptoms observed in these patients represent a new condition, such as Alzheimer's disease (AD), but with phenotypic clinical features modified by the premorbid disorder; or whether in some individuals the observed cognitive decline should be regarded as a natural stage of the psychiatric disorder itself.
To accomplish this purpose, large cohorts of psychiatric patients should be followed for decades, with their cognitive and functional performance under objective evaluation, preferably with detailed longitudinal structural neuroimaging analyses and, critically, under a diagnostic framework that should recognize the presence of significant cognitive and functional decline as a dementia equivalent.To the best of our knowledge, a research project like this has not been conducted so far.
Meanwhile, the identification of genetic mutationsmicrotubule-associated protein tau gene (MAPT), progranulin gene (GRN), and chromosome 9 open reading frame 72 (C9orf72) gene, responsible for up to 15% of the frontotemporal lobar degeneration (FTLD) group of disorders 2 -provide a framework to investigate cross-sectionally the prevalence of a neurodegenerative disease in this select group of patients.The advent of biomarkers related to AD in the cerebrospinal fluid (CSF) and positron emission tomography (PET) scans with beta-amyloid or tau tracers also allow a better investigation of neurodegenerative pathological substrates in this group of individuals with SMI.
Severe/serious mental illness is a clinical concept generally established by its duration and by the functional decline that it produces, encompassing disorders that cause psychotic symptoms, such as SCZ and schizoaffective disorder (SZA), and severe/psychotic presentations of affective disorders, such as severe major depression and BD 3 .This concept is also replicated by the National Institute of Mental Health, which defines SMI as a mental, behavioral, or emotional disorder (excluding developmental and substance use disorders) of sufficient duration to meet diagnostic criteria specified within the Diagnostic and Statistical Manual of Mental Disorders (DSM), and resulting in serious functional impairment that substantially interferes with or limits one or more major life activities.
Bipolar affective disorder, SZA and SCZ, especially, may be grouped together as SMIs, not only by their severity, but also because they might be indistinguishable from a biological perspective 4 .Clementz et al. 4 , based on multivariate analyses from cognitive, electroencephalographic, and oculomotor paradigms, identified three neurobiologically distinct psychosis biotypes previously classified within the psychosis dimension, and documented that each biotype included all DSM psychosis categories.The same three biotype constructs were later the objects of a study that examined whole brain gray matter density measures in probands, their relatives, and healthy individuals, organized by their biotype, and then by DSM diagnoses.The research pointed to brain anatomy characteristics, measured by reduced gray matter density, consistent with their cognitive and sensorimotor profile first described 5 .The biotype 1 included individuals with a mean age of 35.3 years and mean age of illness onset of 20.5 years.The participants presented with severely impaired cognition and sensorimotor function, and had extensive and diffuse gray matter loss, more apparent in the frontal, anterior/middle cingulate cortex, and temporal regions.Pooroutcome individuals made up this group.The biotype 2 was typified by moderately impaired cognition and accentuated sensorimotor reactivity, with intermediate and more localized gray matter atrophy, more obvious in insula and frontotemporal regions.The mean age of the group was 35.4 years and mean age of onset of illness was 20.8 years; biotype 3, which had a mean age of 35.3 years and mean onset of illness at 20.3 years, presented with near-normal cognition and close to normal sensorimotor function, with small reductions in anterior limbic regions.Therefore, distinguishing an ongoing dementia process would be especially challenging for the most severely impaired groups, which already have brain anatomy abnormalities 5 .
The diagnostic challenge of recognizing dementia in a patient with lifetime severe psychiatric disorder is far more puzzling when considering the possibility of behavioral variant frontotemporal dementia (bvFTD), in which many patients do not routinely show gross cognitive deficits, particularly in their early stages.Several reports have documented that a large number of patients with bvFTD are, indeed, initially labeled with classical psychiatric conditions 6,7 .
To consider the relationship between SMI and bvFTD, we performed a systematic review investigating the available data regarding the interface between these conditions.First, we provide a very brief overview on the recent history of bvFTD and its neurobiological underpinnings, introducing the overlapping features between this type of dementia and primary psychiatric symptoms.Then we present the systematic review itself and discuss the available findings.Lastly, we present a critical analysis and propose a script to understand current evidence, which may be useful for future studies in the field.

FRONTOTEMPORAL DEMENTIA OVERVIEW
The predominant degeneration of the frontal and temporal cortex, first described in 1892 by Arnold Pick, usually induces a clinical presentation characterized by progressive behavioral and personality changes, and/or language impairment.Despite this long-standing description, formal diagnostic criteria and classification of clinical subtypes were first proposed only about 20 years ago and have been a subject of intense modification since then.Initially developed by a consensus of researchers in 1994, the criteria were subsequently refined by an international consensus in 1998 to include presentations with predominant language involvement.At that time, the term FTLD was proposed to define a group of a few pathological conditions associated with a broad spectrum of syndromes with overlapping clinical presentations, ranging from those with predominant behavioral changes to progressive language impairment.More recently, the clinical criteria for bvFTD diagnosis have been revised 8 , substantially raising its sensitivity, albeit with a compromised specificity when used to tell it apart from FTLD and psychiatric disorders 9 .Concurrently, the criteria for clinical characterization of primary progressive aphasia subtypes were detailed 10 .This brief overview of the history of the frontotemporal dementia (FTD) description provides a sense of how recently clinical presentations of FTLD have been characterized, which in part accounts for the diagnostic difficulties encountered in clinical practice.
In parallel with advances in clinical characterization, mutations associated with FTLD have been identified since 1998, when mutations in the MAPT gene on chromosome 17 were first recognized in families with FTD and parkinsonism.Since the identification of MAPT mutations, over 10 other genes have been associated with FTD spectrum disorders and, surprisingly, a single genetic mutation was found to account for completely different clinical pictures as well as for variable neuroimaging phenotypes 11 .
Despite this myriad of presentation possibilities, there is a clear tendency for a pattern of atrophy of the frontal and temporal brain regions.The main affected areas correspond to the medial frontal lobe and fronto-insular regions 12 .Temporal lobes are also frequently affected and are even the most common site of atrophy in MAPT mutation individuals.Contemporary neuroscience is still dedicated to understanding the reasons for this convergent topographic involvement, despite a heterogeneous pathological background.Even more incipient is the understanding of the neurobiological role played by these brain regions, which are affected very early in the degenerative process that characterizes the FTLD spectrum 13 , and can determine dramatically disturbing clinical presentations.
The frontal lobes and their multiple connections regulate essential aspects of human behavior, especially those resulting from the decision-making process.The frontal lobes also contribute to the processing of an additional repertoire of highly adaptive behaviors through afferents coming from the temporal lobes, especially from the polar region and the amygdala.The ability to regulate affective responses, adjusting them to the current social context and promoting adaptive prosocial attitudes, is also attributed to this interaction.
Such sophisticated behavioral adjustments are fundamental to a successful life, both from an individual and a societal point of view.This repertoire of behaviors is much vaster and ecologically far more relevant than the classical executive function performance, usually probed in artificially-set testing environments, through tasks relying on dorsolateral prefrontal cortex, which are heavily dependent on working memory and attentional control processes, and do not tap into the frontotemporal regions herein emphasized and critically important to FTD phenomenology.It is not surprising, therefore, that degenerative processes affecting these circuits, or psychiatric disorders that lead to their dysfunction, can be deeply disturbing to the patient's daily functioning, despite mildto-no impairment in classical neuropsychological tests.More recently, neuroimaging studies have demonstrated that there may be a common anatomo-functional underpinning between FTD and a few psychiatric conditions, at least in a small subset of patients 14 .

Methods
Studies were selected from PubMed and LILACS databases, including those published up to February 12, 2018.The search strategy included the following terms: "frontotemporal dementia" plus "bipolar", OR "frontotemporal dementia" plus "schizophrenia", OR "frontotemporal dementia" plus "schizoaffective".Neither a publication date restriction nor a search field filter was used.This search strategy was augmented with manual searching through the reference lists from the included studies and was independently performed by two investigators (HCG and LBG).
Titles and abstracts of the papers -including publications ahead-of-print -retrieved in the initial search were screened and subsequently classified according to their own design characteristics: 1) case-control, 2) case reports, 3) cohort and 4) cross-sectional design studies.Publications without abstracts, case reports with absent genetic or histopathological confirmation, review articles, and papers written in non-English languages were not selected across the systematic search process.

RESULTS
The Figure shows the selected studies and the Table presents their findings.
It was possible to identify three major research aims among the selected studies: 1) to look for FTLDassociated genetic abnormalities, mostly C9orf72 expansions, in psychotic, schizophrenic or bipolar patients; 2) to compare the cognitive profile between patients affected by neurodegenerative disorders and schizophrenic patients; 3) to highlight the association between bvFTD and preceding psychiatric conditions and/or distinguish them both The C9orf72 expansion was detected in 0.67% of a SCZ/SZA sample (n = 298) and in 0.57% of a sample of psychotic individuals (n = 697) (15, 20).However, this expansion was not found in four other cohorts of BD (n = 206), SCZ (n = 192, n = 466) and in treatment-resistant SCZ (n = 386) 16,18,19,22 .Cross-sectional studies comparing the neurocognitive profile between bvFTD and psychiatric disorders mostly found no remarkable differences 44,45,47,53 , or pointed to less severe deficits in bvFTD 49,55 .Lastly, there were several case reports that established a definite FTLD diagnosis, through genetic or histopathological examination, in patients with previous diagnoses or clinical histories compatible with BD/SCZ/SZA or atypical psychiatric symptoms 28,29,30,31,32,33,34,35,36,37,38,39 .
As shown in the Table, investigated mutations were found infrequently in the studied samples.Additionally, almost every large cohort studied did not look for mutations in a control group.The attempt to compare psychiatric disorders and neurodegenerative conditions clinically usually fails, considering that a few of these psychiatric patients might have already become demented by the time of the research evaluation, especially samples of severely affected individuals.Conversely, many case reports suggest a pathophysiological association between psychiatric disorders and FTLD, but more evidence is warranted to support this assumption.

Figure.
Selection of studies.Flowchart depicting selection of items for systematic review on PubMed and LILACS databases using the terms "frontotemporal dementia" plus "bipolar", OR "frontotemporal dementia" plus "schizophrenia", OR "frontotemporal dementia" plus "schizoaffective".

DISCUSSION
Most of the evidence gathered from the systematic review process does not clarify the link between lifelong SMI and dementia, precluding reliable assumptions regarding this disputed issue.Nevertheless, a few studies suggested the possibility, under a structured diagnostic proposal, of formally recognizing an equivalent to dementia syndrome in selected psychiatric patients.
Here we present a critical analysis, combining the main findings of this review and other related publications in the field suggesting directions for future research.

Severe mental illness natural outcomes
Making a diagnosis of dementia in a patient with lifelong SMI is challenging.Cognitive and functional impairment constitute plausible outcomes from the premorbid mental illness itself and impairment in cognitive performance does not exclude primary psychiatric diagnoses in a dementia workup 42 .Indeed, long-term cognitive and functional outcomes from severe psychiatric illness-BD, SZA or SCZ-are still a source of an intense debate.Accordingly, the fifth edition of the Diagnostic and Statistical Manual of Mental Disorders (DSM-5), in its "Functional Consequences of Bipolar I and II Disorders" topics, recognizes functional impairment in approximately 30% of patients with BD type I and in at least 15-20% of those with BD type II.The DSM-5 also reserves a topic to discuss the "Functional Consequences of Schizophrenia" and states that significant social and functional impairment are associated with SCZ 55 .
The very concept of SCZ emerged in 1896, with Kraepelin's description of dementia praecox.He emphasized the progressive nature of the disease, which almost invariably proceeded to cognitive and behavioral deterioration.At that time, the unfortunate prognosis associated with this disorder seemed to suggest to him the idea of a possible neurodegenerative condition.Additionally, in the so-called manic-depressive insanity, Kraepelin noticed the relapsing nature of mood disorders; however, recognizing the ordinary possibility of full recovery between depressive or manic episodes.Nevertheless, within a scenario of decreasing intercrisis intervals and increasing duration of episodes along the disease course, Kraepelin considered a worse prognosis, admitting the possibility of a "psychic decline" in these patients.It is interesting to point out that Kraepelin's description of patients with chronic mania share many behavioral features that constitute core diagnostic symptoms of bvFTD 56 .
In SCZ, the remission rate from a first psychotic episode is estimated to be around 70% in the first year but remains almost unchanged over the course of the disease.Functional recovery is less frequently achieved (18-40%), namely living independently without psychotic or negative symptoms while working or studying, but this rate does not change over time, contradicting the idea of a progressive condition.Patients with psychotic disorders showed IQ deficits and developmental delays in specific cognitive domains during the first two decades of life, in a cohort study 57 .Hence, the cognitive impairment of these patients is suggested to be linked to a neurodevelopmental cause.Failed to find differences between groups on pattern of memory impairment.

Continuation
In cognitive testing, patients with well-controlled SCZ usually display performances between 1.0 and 1.5 standard deviations below a reference control group 58 .This deficit corresponds to a mild neurocognitive disorder and is generally incompatible with dementia, where performance is typically 2.0 standard deviations or more below the mean, according to the DSM-5 55 .Conversely, approximately 25% of SCZ patients can be categorized as having a poor cognitive and functional prognosis.Nonetheless, studies that have addressed this issue have not ventured into characterizing a formal diagnosis of dementia in these individuals 59 .The group with a poor outcome would encompass "Kraepelinian" SCZ, with permanent cognitive impairment and performance similar to bvFTD patients 44 .In searching for the overlapping diagnosis of dementia and SCZ, Nicolas et al. found 14 individuals who fulfilled the criteria for dementia in a cohort of 96 SCZ patients followed for 20 months.Interestingly, four of them met criteria for bvFTD 21 .
Schizophrenic patients usually have social cognition disturbances as a manifestation of the illness.Therefore, Kosmidis and colleagues compared bvFTD and SCZ individuals using videotaped scenarios of social interactions illustrating sincere, sarcastic and paradoxical statements.The results indicated that the SCZ group had impaired performances on all theory of mind conditions, despite understanding sincere statements.Nonetheless, bvFTD patients had remarkable impairment in identifying sarcasm, and seemed to benefit from verbal tips that indicated the appropriate social context, suggesting that these patients were particularly impaired in detecting discrete nonverbal signs from the scene 54 .In another study, Bediou et al. explored facial emotional recognition performance in different clinical conditions, including SCZ and bvFTD.Their results showed impaired efficiency in both groups of patients when compared with their respective controls.The bvFTD group displayed poorer performance in comparison with the SCZ group 48 .
Patients with BD seem to present with cognitive impairment in all phases of the disease, including periods of remission, and it has been suggested that cognitive performance declines according to the number of accumulated mood disorder episodes 60 .Patients with BD share a similar cognitive profile with those with SCZ, including impairment in verbal memory, as well as in executive function and verbal fluency tests.Despite this similarity, there seems to be a severity spectrum across these disorders, in which BD patients display a better performance than those with SCZ, and SZA patients stand in between them 58 .

Psychiatric presentations in bvFTD
Personality changes and inappropriate social behaviors are core features of bvFTD, including decline in interpersonal and social conduct, emotional blunting and loss of insight.Almost routinely, the behavioral symptoms precede marked cognitive decline.As a result, patients with early behavioral changes of bvFTD are more likely to seek specialized psychiatric help and usually receive a cardinal psychiatric disorder diagnosis 6 .In this sense, the use of a behavioral inventory appears to be the best assessment method to diagnose bvFTD 17 .Some of the initial symptoms of bvFTD, such as apathy, disinterest and lack of initiative, are often mistakenly diagnosed as major depression.On the other hand, disinhibition, hypersexuality, compulsive behaviors and decreased need for sleep may be confused with the mania/hypomania symptoms of BD.Indeed, after major depressive disorder, BD was the most frequently-given diagnosis in a cohort of bvFTD patients, of whom almost 50% were diagnosed with a primary psychiatric disorder before bvFTD was finally recognized 7 .It is worth noting that Gossink and colleagues found different results (past psychiatric disorder in only 8.7% of individuals), albeit using a different set of inclusion criteria.They prospectively evaluated the prevalence of a past psychiatric disorder in a large cohort of bvFTD in a structured interview 61 .Nonetheless, this study did not consider the severe lack of insight in patients with bvFTD, and investigated the past psychiatric history of the patients by interviewing them directly, without the contribution of a trustworthy proxy.
Psychotic symptoms have been reported in about 10-20% of bvFTD patients 6 , in spite of this not being considered a core symptom in diagnostic criteria.Therefore, this presentation may easily be mistaken for schizophrenic symptoms, particularly in those with C9orf72 gene expansions 43 .The emotional blunting displayed by bvFTD patients is akin to the one classically described by Bleuler in the "Group of Schizophrenias", a cluster of symptoms later categorized as part of the negative presentation of the disease in 1974.In 1988, Weinberger noted that those symptoms were "phenomenologically similar to many of the characteristics of patients with disease of the frontal lobe", later confirmed by Ziauddeen and colleagues 49 .
In fact, primary psychiatric disorders and bvFTD may be almost indistinguishable under the concept of late-onset frontal lobe syndrome.Vijverberg and colleagues have investigated which demographic, clinical, neuropsychological, neuroimaging, and CSF biomarkers are important in distinguishing primary psychiatric disorders presenting as lateonset frontal lobe syndrome, from bvFTD 9 .These authors suggested that late-onset primary psychiatric disorders may show neuroimaging abnormalities, especially metabolic deficits akin to FTD 63 , in FDG-PET, which poses further difficulty in obtaining a correct differential diagnosis on a cross-sectional basis.Nevertheless, they found that specific CSF biomarkers-total tau, phosphorylated tau, amyloid-β 1-42 peptide and neurofilament light chain concentrations-may aid clinicians in the diagnostic workup (64).Also, in an attempt to contribute to the distinction between primary psychiatric disorders and bvFTD, Metin and colleagues used quantitative electroencephalography for differentiating bvFTD from late-life BD with 76% overall accuracy 41 .

Definite FTLD disguised as primary psychiatric disorders
Velakoulis et al. 6 hypothesized that very young-onset bvFTD could masquerade as a SCZ or BD presentation.In a clinicopathological study, they found that five out of 17 FTLD patients had received a previous diagnosis of psychotic illness.Genetic analysis in one of these patients showed a mutation in the GRN gene.Additionally, in their literature review, the authors also found that a third of FTD patients aged 30 years or less, and a quarter of those aged 40 years or less, had received a primary diagnosis of psychosis 6 .Indeed, early-onset bvFTD patients, whose presentation included psychotic symptoms, social withdrawal, emotional blunting, functional decline and mild executive dysfunction, were likely to have received a diagnosis of SCZ.
Velakoulis et al. 52 also investigated the presence of TDP-43 neuropathology, related to FTLD, in brain sections of 12 deceased SCZ and BD patients.They identified three individuals with TDP-43 neuropathological changes.Interestingly, all three individuals had late-onset psychotic features, notable impairment in decision-making and a positive family history for SMI 52 , similar to bvFTD.Following the same reasoning, Meisler and colleagues demonstrated that the C9orf72 repeat expansion may be associated with a classic clinical presentation of BD and a later progression to neurodegenerative disease 23 .The same mutation was considered a possible, albeit rare, cause of SCZ in European and North American cohorts 15,20 .Conversely, this mutation was not found in three other studies, two of them in Asian cohorts 16,18,22 .
There are frequent reports of lifelong psychiatric disorders, especially BD, describing patients in whom a definite bvFTD diagnosis was made through the identification of fully penetrant genetic abnormalities (C9orf72 and GRN), suggesting the possibility of a diagnostic superimposition of these two conditions, and the need for a debate on the relationship between them 28,32,35 .In the pursuit of the association between FTLD and primary psychiatric disorders, Schoder and colleagues investigated the risk of SCZ and SZA in first-degree relatives of patients with FTLD, and found a higher risk in this group of individuals in comparison with relatives of AD patients 50 .

Neuroprogression models in severe mental illness and the neuroinflammatory pathway
Increased blood levels of a wide range of cytokines in patients with primary psychiatric disorders, such as depression, BD and SCZ, have been documented, indicating a possible role of pro-inflammatory pathway neurotoxicity 65 .
In SCZ, neurotoxicity associated with a psychotic crisis was suggested by Wyatt 66 .Moreover, the duration of untreated symptoms has been correlated to a reduction of gray matter volumes in several reports 67 .Although neuroimaging studies showed structural developmental changes in the first psychotic episode, with subsequent reduction in brain tissue volume over the years, secondary causes, such as use of antipsychotics and other drugs, could alternatively explain the observed atrophy 68 .
In BD, there have been increasing suggestions that chronic inflammatory processes, both from periphery and within the brain, are involved in its pathophysiology.Post proposed the foundations of our current knowledge regarding neuroprogression 69 .This phenomenon was further elaborated on by Kapczinski et al. to explain long-term outcomes of BD in brain structure, patient cognition, functionality and response to treatments 70 .It is noteworthy that the neuroprogression and staging models of severe psychiatric disorders accepted stages compatible with a diagnosis of dementia late in the course of these diseases 70 .These are theoretical models that do not clarify the mechanism that underpins this impressive change in the disease course.They also fail to contemplate the possibility that another superimposed disease might be in progress.Additionally, these models do not elucidate why many individuals do not progress to dementia as an end-stage BD.
Dols and colleagues described four patients with lifetime BD 71 who evolved to a bvFTD phenotype, showing a behavioral profile very different from these individuals' previous levels of functioning.The authors admitted that the clinical picture could not be better explained by the patients' primary psychiatric disease course nor could it be explained by BD staging models, as the patients had few prior severe mood disorder episodes.The authors did not propose a neurodegenerative condition as a reliable explanation for their findings.After all, their patients had cognitive performances that were regarded as only mildly impaired, had no neuroimaging abnormalities, and did not show cognitive decline during follow-up.It is noteworthy that our group retrospectively studied patients with a similar clinical history to Dols 71 , but in the findings by Gambogi et al. 72 , including neuroimaging and cognitive testing, there were only subtle differences in a typical bvFTD patient.Gambogi et al. found that the frequency of antipsychotic drug use, primitive reflexes, apathy, stereotypic/compulsive/ritualistic behaviors, and family history of psychosis were statistically higher in bvFTD patients with a prior history of SMI than in patients with typical bvFTD.The former group of patients also showed a higher severity of neuropsychiatric symptoms and worse performance in the Frontal Assessment Battery 72 .
Unfortunately, studies of progressive neuroimaging changes in BD patients showed unreplicated and contradictory findings, such as increased gray matter volume in the prefrontal cortex, limbic and subcortical structures 73 .On the other hand, some reports pointed to a decrease in gray matter volume in orbital and medial prefrontal cortex, in addition to smaller volumes in medial temporal cortex and ventral striatum 74 .Moreover, some patients may have greater volumes in the third and fourth ventricles, and these structural changes tend to be worse in patients with recurrent episodes 75 .
Although the effective interplay between peripheral mediators and the inflammatory phenomena of brain tissue is not precisely known, the intense communication between circulating cytokines and leukocytes and the central nervous system is undisputed nowadays.This phenomenon is processed across the blood-brain barrier through intense communication among neurons, glial cells and endothelial cells, whose luminal surface is in permanent contact with the vascular content 76 .In fact, the role of inflammatory tissue activation in the vicinity of neuropathological markers of classic neurodegenerative diseases has been recognized for some time 77 .Furthermore, animal models of some of these diseases have shown that the peripheral inflammatory stimulus may aggravate previous cognitive deficits, and even accelerate the neuropathological process per se.In prospective studies with elderly human cohorts, documentation of episodic systemic infection, such as pneumonia, increased the subsequent incidence of dementia 78 .Furthermore, in observational studies the use of nonsteroidal anti-inflammatory drugs was found to be associated with lower risk for dementia in exposed individuals 78 .
These combined data suggest that systemic inflammatory factors, with SCZ or BD hypothetically included, may contribute to the phenomenon of progressive cognitive decline.

Atypical presentations of FTLD
Recently, it has become possible to identify individuals carrying genetic mutations or expansions.Hence, a range of possibilities for research in FTLD has been opened, allowing in vivo investigation of endophenotypes in carrier individuals who do not yet present with the clinical disease itself.Studies comparing supposed asymptomatic MAPT mutation carriers with controls have highlighted brain connectivity changes years before the expected disease onset, according to their affected relatives' average clinical course 79 .Interestingly, connectivity changes share a similar topography with brain regions classically targeted by atrophy in symptomatic individuals.Indeed, the idea of SCZ as a prodrome of a neurodegenerative disease had already been proposed by Waddington et al. 39 and by Khan et al. 34 The former study established all the neuropathological hallmarks of Pick's disease in a biopsy from a woman with a typical schizophrenic picture 39 .The latter study confirmed MAPT mutation in a 35-year-old patient with an initial schizophreniform presentation and typical pattern of brain atrophy 34 .
Symptomatic patients carrying GRN mutations show reduced plasma progranulin levels and this same finding was evident in asymptomatic carriers in their second or third decades of life.In addition, mutations in the GRN gene had already been linked to SCZ and BD, with equal reduction of plasma protein levels in carriers 36 .Moreover, both MAPT and GRN genes mutation carriers may exhibit structural and functional connectivity changes that precede the first symptoms of bvFTD.These changes are characterized by progressive reduction of fractional anisotropy through diffusion tensor imaging, and resemble abnormalities found in patients with BD, especially in the uncinate fasciculus 80 .These findings support the idea that psychiatric manifestations could be the prodromal phase of neurodegenerative diseases.
The definite confirmation of a pathological substrate associated with FTLD in patients with a psychiatric history, and the inversely proportional relationship between age at onset and the schizophreniform presentation (the younger, the more prevalent the presentations), suggest the possibility of a phenomenon already observed in other neurodegenerative diseases such as leukodystrophy and Niemann-Pick Type C disease.For reasons still not understood, these diseases sometimes manifest in adults or in the elderly, indicating a slow progression.It is reasonable to consider that psychosis might be a nonspecific psychiatric symptom, resulting from the reaction of the still-resilient young brain to the underlying neurodegenerative process.
Finally, the slowly-progressive degenerative phenomenon has been repeatedly reported in the literature, especially in individuals with bvFTD carrying the C9orf72 expansion.The relevance of this finding for the scope of the discussion presented here is particularly important, given that this etiologic subtype of FTD has consistently been associated with psychotic presentations 43 .

CONCLUSIONS
The available evidence discussed herein demonstrates how fragile the current understanding is regarding the association between FTD and prior SMI.The issue of misdiagnosis between these conditions is strongly emphasized in many reports.
Dementia superimposed on a severe psychiatric clinical picture, where cognitive impairment and cerebral atrophy arise early from neurodevelopment issues, is a plausible outcome, particularly when considering cognitive and brain reserve contemporary concepts.The underlying neuropathology remains unknown though.However, cases like those described by Dols et al. 71 and by Gambogi et al. 72 could not be explained from this perspective.This group of patients deserves a different evaluation workup.If dementia ensues, prognostic predictions and therapeutic choices might be revised.Special attention should be paid to late onset, atypical phenotypes, and to those patients with an unfavorable evolution.In these patients, we suggest the development of investigative algorithms comprising neuropsychological assessment, multimodal neuroimaging, genetic tests, and other biomarkers, including CSF analysis and, perhaps, peripheral inflammatory biomarkers.There is an urgent need to reframe the clinical point of view that seems permissive toward progressive or severe cognitive and functional decline in supposed primary psychiatric conditions.

Table .
Synthesis of articles included in the present review.