Long-Term Severe Mental Disorders Preceding Behavioral Variant Frontotemporal Dementia: Frequency and Clinical Correlates in an Outpatient Sample

INTRODUCTION

The diagnosis of dementia in a subject with a lifetime severe psychiatric disorder, such as severe relapsing unipolar depression, bipolar affective disorder (BAD), and schizophrenia (SCZ) spectrum disorders, is still a challenging task. Most major neurocognitive disorders diagnostic proposals state that the patients’ deficits “should not be better explained” by a psychiatric condition [1, 2]. In fact, there is a general tendency to attribute to an existing disorder, either neurological or psychiatric, novel symptoms or signs allegedly arising from brain malfunction.

Considering worldwide population aging, which has also reached the individuals with severe mental disorders [3, 4], it is of utmost importance to establish the boundaries between lifetime psychiatric disorders symptoms and those attributed to another superimposing condition, such as a neurodegenerative dementias.

Tertiary outpatient neurological units usually receive patients referred from other specialized settings to verify a dementia diagnosis and to identify the underlying cause, including subjects with lifetime severe psychiatric disorders. Many of these latter patients display late-onset frontal symptoms (e.g., impulsivity, disinhibition, compulsive/stereotypical behavior) resembling behavioral variant frontotemporal dementia (bvFTD) phenotype [5, 6]. The recognition of dementia in these patients is especially challenging, since many classical bvFTD patients may not disclose gross cognitive deficits, particularly in its early stages [7]; otherwise, patients with severe psychiatric disorders usually display objective cognitive deficits, especially in executive functions [8]. On the other hand, previous reports have documented that a large number of patients with bvFTD are initially labelled with classical psychiatric conditions [9, 10]. Apathy and emotional blunting, which are core bvFTD symptoms, are typically observed in severe psychotic disorders [11], such as SCZ. Additionally, a considerable proportion of bvFTD patients carrying C9ORF72 mutations display overt psychotic symptoms [12]. Moreover, disinhibition, loss of insight, mood swings, and sleep disorders are frequent in bvFTD and resemble those observed in BAD [13].

Several reports, usually single case studies [14] and small case series [9], have highlighted the association between bvFTD and preceding psychiatric conditions, suggesting a possible stem shared by these disorders. However, to the best of our knowledge, only one study has yet prospectively evaluated the prevalence of past psychiatric disorder under a structured interview with a fairly large cohort of bvFTD [15]. Notwithstanding, this study neglected bvFTD severe lack of insight [16] and surveyed participants’ past psychiatric history directly with patients, without input from a reliable proxy. Even so, considering this limitation, Gossink and colleagues were still not able to find any individual with a past medical history of severe psychiatric disorder in their cohort of bvFTD participants. Their only finding of 8.7% frequency of unipolar depression in those patients reflect the typical lifetime prevalence of this disorder in population-based epidemiological surveys [17].

A detailed clinical profile of patients considered to have dementia following a lifetime psychiatric disorder, especially SCZ spectrum disorders and BAD, has been scarcely studied so far [18]. Moreover, modern proposals that admit a dementia stage to those severe psychiatric disorders have not clarified its clinical profile [19, 20]. Therefore, we set two primaries aims for the present study: 1) to investigate in a Cognitive and Behavioral Neurology outpatient unit, the frequency of a past medical history satisfying diagnostic criteria for BAD, SCZ, and schizoaffective disorder (SZA) in bvFTD patients, and which kind of prior disorder, if any, is predominant in this group of patients; and 2) to evaluate if these patients have a different clinical, behavioral. or functional phenotype in comparison with patients without past medical history that fulfils diagnostic criteria for BAD/SCZ/SZA. We hypothesized that the prevalence of a past medical history of these selected psychiatric disorders would be considerable, and that this group of patients would be quite indistinguishable from the other one.

A prior diagnosis of severe unipolar non-psychotic major depressive disorder (MDD) was out of the scope of this study, since: 1) the relationship between depression and dementia in general has been extensively investigated [21, 22]; 2) a previous epidemiological survey in familial FTD have not suggested a relationship between these two disorders [23]; 3) except for apathy and a possible similar dysexecutive profile [24], depressive symptoms usually do not overlap with core behavioral manifestations of bvFTD. Finally, severe psychiatric disorders such as SCZ, SZA, and BAD were grouped together as a research goal not only because of their severity and functional impact, but also because they have shown to be biologically indistinguishable in large previous research that sought biomarkers for psychosis [25]. Neurobiologically distinct biotypes identified into the psychosis dimension include all three diagnoses (BAD/SCZ/SZA). Furthermore, within the concept of a psychosis dimension, part of these patients can present with significant cognitive impairment and neuroimaging abnormalities [26], akin to a neurodegenerative disorder.

MATERIALS AND METHODS

RESULTS

We were able to recruit 46 participants fulfilling probable bvFTD diagnosis, with 13 of them (28.3%) being referred from psychiatry/psychogeriatric services. All of them presented significant neuroimaging abnormalities ascertained independently by at least two of the teams’ researchers. According to SCID-I, bvFTD+ patients represented 36.9% (95% CI: 22.8%, 51.0%) of the sample (n = 17). Within this group, 15/17 patients were previously diagnosed with a psychiatric disorder and two of these 15 previously diagnosed subjects had the diagnosis changed after SCID-I application. Past medical history fulfilling BAD diagnostic criteria was ascertained in 76.5% of cases, distributed as follows: 64.7% BAD type I, 5.9% BAD type II, and 5.9% BAD not otherwise specified (NOS). Past medical history fulfilling diagnostic criteria for SZA and SCZ were verified in 17.6% and 5.9% of patients, respectively.

According to the results shown in Table 1, groups were similar regarding sociodemographic features, age at dementia onset, and dementia duration. Overall, 47.1% of the bvFTD+ patients were referred from psychiatric outpatient units, compared to only 17.2% of the bvFTD- group being referred from these units (p = 0.04). All patients referred from a psychiatric clinic in the bvFTD+ group (8/17) were lifelong attenders, whereas subjects coming from a psychiatric unit in the bvFTD- group had never been under regular psychiatric follow-up and were assigned to our unit exclusively because of the suspected diagnosis of dementia. Age at psychiatric disorder onset (33.2±8.9 years) confirmed that the bvFTD+ participants presented a lifetime mental illness. In general, the bvFTD+ group displayed a high prevalence of core bvFTD diagnostic features, and a statistically significant higher frequency of apathy/inertia (p = 0.03) and ritualistic/perseverative behavior (p = 0.01) in comparison with bvFTD- group. There were differences between groups: bvFTD+ higher than bvFTD-, regarding family history of suicide (17.5% versus 3.5%), and psychosis (64.7% versus 24.1%), the latter reaching statistical significance (p = 0.01). The Goldman scale scores were similar between both groups.

The bvFTD+ group had worse cognitive performance only on the FAB (p = 0.01), considering Z-score. The two groups had similar functional performance according to FRS and FAQ scores. Additionally, patients in the bvFTD+ group had a higher severity of neuropsychiatric symptoms as measured by global NPI-12 scores (p = 0.01), despite an absence of significant findings for comparisons between groups according to NPI sub-domains. Similarly, the bvFTD+ group had more individuals under use of antipsychotics (p = 0.01) and of other psychotropic drugs (p = 0.01). This group also presented a higher frequency of primitive reflexes (p = 0.04).

DISCUSSION

In this study, 1/3 of bvFTD patients attending a tertiary outpatient unit presented past medical history fulfilling diagnostic criteria for BAD/SCZ/SZA. This feature was associated with a subtly different clinical profile, in accordance with our primary hypothesis. A few characteristics, such as frequency of antipsychotic drug use, primitive reflexes, apathy, stereotyped/compulsive/ritualistic behaviors, and family history of psychosis were statistically higher in the bvFTD+. This group also disclosed a higher severity of neuropsychiatric symptoms and worse performance in the FAB. However, many of these subtle findings could be reasonably explained by the very long-term (∼30 years) past psychiatric history itself.

Our finding of a past medical history compatible with diagnoses of BAD/SCZ/SZA in 36.9% of the bvFTD sample is different from a previous report that pointed to any past psychiatric disorder in only 8.7% of subjects with probable or definite bvFTD [15]. In that study, using a distinct set of inclusion criteria, Gossink and colleagues applied the structured psychiatric interview directly to bvFTD participants. Severe lack of insight is a core feature in bvFTD [39], and autobiographic memory deficits constitute another important underappreciated deficit in a subset of patients with this disorder [40]. Together, these caveats could critically jeopardize the psychiatric interview results. Interestingly, however, in their study no individual considered to have probable bvFTD had a severe prior psychiatric diagnosis as such we describe herein. This finding suggests that individuals with these severe lifetime psychiatric disorders were not considered suitable for a bvFTD diagnosis ascertainment, in a remarkable different strategy from our study design. Moreover, in their “psychiatric diagnosis” group of late-onset frontal lobe syndrome [6], 22.5% had either previous BAD (17.5%) or a psychotic disorder (5%), in close similarity with our findings. Unfortunately, the authors do not provide behavioral and cognitive measures from their participants, thus precluding comparison between their study and ours.

In accordance with a higher prevalence of previous psychiatric diagnosis in bvFTD, Woolley and colleagues conducted a retrospective research through blinded chart review and identified a psychiatric past history in approximately 28% of the subjects for any neurodegenerative disease, and in more than 50% for bvFTD, when considered the 10 years prior to diagnosis of dementia [10]. However, our findings stem from a distinct methodological approach, dealing with a remarkably different clinical profile of lifelong psychiatric patients. In their study, Wooley and colleagues found a mean duration of psychiatric diagnosis for less than ten years before a neurodegenerative disease could be ascertained. In opposition, our bvFTD+ group had around three decades of past medical history of severe psychiatric diagnoses and displayed an impressive decline in their cognitive and functional trajectory at a similar age than the bvFTD- group.

Both groups had similar global cognitive performance measured by the MMSE and had similar functional status based on FRS and FAQ evaluations. On cognitive tests related to prefrontal abilities, similarity between groups was again observed and the bvFTD+ performed significantly worse only on the FAB. However, factors such as higher NPI and higher use of psychoactive medications should be considered as possible confounders.

The bvFTD+ group displayed median z-scores far lower than 2.0 standard deviations (SD) below reference values for most of the cognitive tests. This performance is generally compatible with dementia according to DSM-5, and reinforces the severe cognitive impairment presented by this group [41]. Conversely, patients with BAD/SCZ/SZA displayed a performance standing between 1.0 and 1.5 standard deviations below healthy controls in specific cognitive tests in a previous large research, a pattern corresponding to mild neurocognitive disorder [42].

From a distinct perspective, in which only severely symptomatic psychiatric patients were included (e.g., major depressive disorder patients referred for electroconvulsive therapy), Vijverberg and colleagues have documented worse cognitive deficits in this group (Z-scores usually better than – 2.0 SD according to control group performance) when compared to patients with probable bvFTD. However, in their study there is no explicit report of cognitive performance adjustment for their sample’s heterogeneous educational background [43]. For instance, 12.3% of bvFTD patients were considered to have low education in opposition to MDD patients, where low education was found in 71.4%. Not surprisingly, MDD patients displayed the worst cognitive performances, especially in verbal memory, which is highly sensitive to education [44]. On the other hand, patients from our bvFTD+ group were unequivocally controlled for the underlying diseases, according to inclusion criteria, and the magnitude of their cognitive decline was adjusted for schooling.

Our findings reinforce the typical bvFTD phenotype identified in our sample, despite participants’ different past psychiatric history. Indeed, our study was not designed to disentangle subtle cognitive and behavioral differences between bvFTD and severe psychiatric disorders. In another initiative, Badiou and colleagues explored facial emotions recognition abilities in different clinical groups, including SCZ and FTD. Both of these latter groups performed poorly when compared to controls; however, worse outcomes were observed in the FTD group [45]. Interestingly, our bvFTD+ displayed a remarkable worse performance than bvFTD- in the Ekman’s facial emotion recognition test, despite attaining statistical difference only in crude scores. However, both groups were impaired in comparison to reference standards [46].

There was no significant difference regarding family history of dementia or Goldman Scale scores between the groups, despite a higher frequency of family history of psychotic symptoms in the bvFTD+ subjects. There have been repeated reports of BAD as first manifestation of slowly progressive bvFTD defined by a known pathological mutation (C9ORF72 and GRN) [14, 47]. In those reports, patients developed bvFTD after a premorbid clinical history of BAD that lasted approximately 20 years, with no overt cognitive symptoms, and with a classical bipolar clinical profile with a relapsing-remitting course and good response to standard drug therapy. Nevertheless, the possibility of a monogenic inheritance does not fit with our family history data, which suggests a polygenic inheritance as a more reasonable explanation for our findings [48].

It is noteworthy that current staging and neuroprogression models of severe psychiatric disorders [19, 20] admit stages compatible with the diagnosis of dementia late in these disorders. Nevertheless, those theoretical models do not clarify the mechanisms underpinning this striking inflection in disease trajectory, failing to contemplate the hypothesis that another superimposed disease might be in progress. Moreover, those models do not explain why many individuals do not progress to dementia as an end-stage bipolar disorder. Interestingly, Dols and colleagues have described four patients with lifetime BAD who resembled the bvFTD+ group herein reported [40]. Their patients evolved to a bvFTD phenotype, disclosing a behavioral profile very different from these individuals’ previous levels of functioning. These authors recognize that their observations could not be better explained by patients’ primary psychiatric disease course, nor should be attributed to a “neurotoxic” effect— as BAD staging models suggest— from patient’s few prior severe mood disorder episodes. Nevertheless, differently from our sample, their patients had cognitive performance regarded as only mildly impaired, had no neuroimaging abnormalities, and did not show cognitive decline during follow-up. Thus, the authors did not propose a neurodegenerative condition as a reliable explanation for their findings.

Our study has some limitations. The number of participants and the proposed split-group study certainly compromise the statistical power of the analysis. However, it must be pointed out that bvFTD constitutes an uncommon condition, and our sample size is compatible with several other published studies. The lack of a reliable biomarker, including genetic information, to diagnose and to stage the conditions herein studied renders our findings overly dependent on clinical judgement. The C9ORF72 expansion, which has strong association with psychiatric phenotype and long-lasting disease, was not tested. Nevertheless, the family history investigation did not present a pattern that suggests this mutation as a good explanation for our findings. We must also acknowledge that our tertiary unit has its main referrals from other specialized services, which could lead to an overrepresentation of these overlapping conditions and compromise the external validity of the results. In our sample, 28.3% (bvFTD+: 47.1% and bvFTD-: 17.2%) of the patients were referred from the psychiatry/psychogeriatric services. Additionally, previous neuroimaging data from the patients were not available for comparison and neither quantitative or semi-quantitative analyses were performed. A recent study found that some patients with late-onset primary psychiatric disorders (major depression, BAD, personality disorders, and SCZ) may present neuroimaging changes in frontotemporal areas, particularly hypometabolism in FDG-PET [49]. If confirmed in future studies, this limitation could have biased the inclusion of patients with past medical history of psychiatric disorders under the diagnostic designation of probable bvFTD. Further, in the absence of neuropathological confirmation, the bvFTD+ may not have underlying frontotemporal lobar degeneration pathology and the changes may represent late degeneration in psychiatric disease.

Concluding, our results point to a high frequency of previous clinical pictures fulfilling diagnostic criteria for BAD/SCZ/SZA in patients with bvFTD in a tertiary outpatient unit. The presence of specific past psychiatric history within this group (36.9%) was 12 times higher than the lifetime prevalence of BAD, SCZ, and SZA (2.9%) in the general population [50]. Despite challenging diagnostic issues, we were able— through careful cognitive, psychiatric, and functional evaluations— to ascertain the diagnosis of bvFTD in these patients. We may speculate that these individuals are generally considered to have an equivalent stage to dementia resulting from their primary psychiatric disorder. However, the scarcely distinguishable clinical picture between bvFTD+ and bvFTD- groups should open the venue for further studies discussing if this phenomenon is indeed a trivial stage stemming from the primary disease, or otherwise represents a superimposing condition, which may have critical prognostic and therapeutic implications.