The main findings of the present study are threefold: (1) despite new sets of criteria, diagnoses of FTD syndromes remained low in routine care in our regional memory clinic network; when diagnosed, bvFTD patients had longer referral delay and diagnostic wandering than AD patients; (2) the peak of incidence of bvFTD occurred between 75 and 79 years, clearly advocating against the conception of FTD as exclusively an early-onset dementia; (3) FTD syndromes differed from AD with regard to cognition and autonomy at baseline, cognitive decline, and disease duration; and (4) therapeutic strategies radically differed from the ones in AD.
Misconceptions about FTD lead to underdiagnosis
In this retrospective study, we calculated an FTD age-standardized incidence rate of 2.9/100.000 py in our region. Our results stand in-between the ones of two recent studies using updated FTD criteria that found an incidence of 1.6/100.000 py in the UK (Norfolk and Cambridgeshire counties) and 3.05/100.000 py in Italy (Leccia and Brescia provinces) [17, 29]. However, while we used the same European reference population as our British colleagues, Logroscino et al. used the Italian population for standardization. Standardization of their incidence rate with the same European population yields an FTD age-standardized incidence rate of 2.78/100.000 py, strikingly similar to ours (our unshown data).
We found that FTD syndromes represented 3% of the Méotis network caseload. Similar MC surveys in Netherlands  and Sweden  had 7% and 3.6% of FTD syndromes, respectively. However, all patients in the Dutch cohort were followed in the Alzheimer center of the VU University Medical Center (VUmc), a tertiary center where atypical dementias are likely to be addressed, possibly leading to an overrepresentation of FTD patients. Likewise, there was a 8.1% proportion of FTD patients in Lille tertiary center. A recent review on the epidemiology of FTD highlighted three studies with high methodological standards . In these publications using the Lund and Manchester  or Neary  criteria, FTD syndromes accounted for 1.1% , 3% , and 3.8%  of dementia cases, which is consistent with our findings.
In sharp contrast, consistent with the underdiagnosis of FTD, systematic neuropathology surveys show much higher figures. In UK brain banks from donors (of whom two thirds had dementia), FTD represented 5.1% of diagnoses  and up to 9.4% of elderly people participating in a community brain donation program were found to have some FTD lesions at autopsy .
The reasons for FTD underdiagnosis are manifold. First, late-onset FTD are often overlooked. FTD is historically considered as a major cause of early onset dementia , which probably contributes to FTD diagnosis being overlooked in late-onset dementia. Yet, in recent studies with pathological confirmation, one fourth of FTD cases had an age at onset > 65 years . In the recent literature, there is a trend toward an increase in the age at diagnosis of FTD syndromes, which may relate to the increasing age at dementia diagnosis in recent surveys . While older studies showed an age at diagnosis of 65.9 years , we found an age at diagnosis of 71.3 years, which compares to recent publications showing a mean age at diagnosis of 69.4 , 70.0 , or 71.3 years . Interestingly, the peak of incidence occurred between 75 and 79 years in our survey as in the aforementioned Italian and English studies [17, 29], reminding that FTD is not only a dementia of early onset.
Second, the positive diagnosis of bvFTD and its differentiation with primary psychiatric disorders is another diagnostic challenge  that is reflected by the increased time to presentation and time to diagnosis of the bvFTD variants as compared with the others [39, 40]. Prolonged diagnostic wandering in bvFTD, associated in our study with an increased reliance on diagnostic biomarkers, seems to be a universal finding [15,16,17] and suggests that many cases could remain misdiagnosed. Future studies should focus on the exact determinants of the delay in referral and in diagnosis. Third, all the possible clinical presentations of FTD have not been thoroughly described and some are not taken into account by the available clinical criteria. The amnestic variant of FTD, in particular, is difficult to differentiate from AD [20, 41] in particular in late-onset dementia . Another example is the right temporal variant of FTD—although a recent publication proposing clinical criteria will contribute to fill the gap .
Overall, our survey confirms that FTD are still probably overlooked despite the use of novel clinical criteria and incorporation of new phenotypes. While progress has been made in the recognition of late-onset forms, differential diagnosis between FTD and AD remains a challenge, particularly in the oldest old, and bvFTD cases are probably still mistaken for primary psychiatric disorders.
FTD syndromes differ with AD in baseline characteristics and natural history
We found several key differences between FTD syndromes and AD at baseline. First, as we had previously shown , we confirmed that the MMSE score is higher in FTD. However, behavior, social cognition, and executive functions, the main domain impaired in bvFTD, are not properly assessed by the MMSE, which somewhat undermines the assumption that the general cognitive status is better preserved in FTD syndromes. The higher IADL-4 score in FTD compared to AD contrasted with past studies that retrieved either lower  or equal  autonomy. However, IADL-4 only assesses restriction in four activities (telephone, transportation, drug treatment, and finances) that are best associated with future dementia risk , thus preventing a direct comparison of our results with studies that employed the full ADL. The younger age and the better preservation of memory and visuo-motor functions may explain the lesser impairment found in FTD as compared to AD. Impaired functional capacity in bvFTD is primarily due to behavioral symptoms and impaired social cognition, and the routine (although complex) instrumental activities of the IADL-4 may not be the most representative of the loss of autonomy in FTD syndromes. Among the FTD syndromes, the lvFTD patients had the most preserved autonomy, as found in previous studies [44, 45].
Although FTD syndromes as a whole had a similar rate of MMSE decline to AD, lvFTD and mFTD variants specifically showed a higher rate of MMSE decline in time. Additionally, lvFTD had a slightly lower score at baseline than the other variants. Since the MMSE relies mostly on language, aphasia has likely impacted the score in lvFTD. In recent studies, patterns of longitudinal MMSE decline across the FTD phenotypes have already been studied, and semantic dementia cases were shown to decline the most . Regarding survival, we, as others , reported that mFTD had the more severe prognosis of FTD syndromes , followed by lvFTD and bvFTD. Despite similar MMSE decline rates between bvFTD and AD, mFTD patients had a significantly lower survival median.
Therapeutic strategies in FTD
The drug treatments used in FTD syndromes markedly differed from the ones used in AD. These observations should be interpreted with caution since differences may only reflect different customs, and not different responses to treatment. However, clinical guidance on the symptomatic treatment of FTD is limited , prompting physicians to use psychotropic drugs that may be used non-specifically in dementia, based on the medical needs and immediate efficacy. Hence, the prescription habits in FTD may also reflect the neuropsychiatric symptoms and treatment response of FTD patients.
A publication from the Boxer’s group showed that off-label use of AChEI and memantine in FTD was common in the USA in 2010 . In our region and in the 2010–2019 time span, we found that AChEI and memantine were used in only 12.0% and 5.7% of FTD syndromes, in accordance with recent data supporting lack of efficacy—or even deleterious effects in bvFTD and mFTD ([49, 50], reviewed in [51, 52]). The remaining prescriptions may reflect diagnostic hesitations with AD at the beginning of follow-up.
Antipsychotics and anxiolytics were more frequently used in FTD syndromes than in AD, and the difference with AD was driven by the bvFTD variant. Antipsychotics are prescribed to treat agitation in dementia whatever the etiology (AD, FTD, or others) , although their use is restricted to patients with severe symptoms (aggression, agitation, or psychosis) who fail to respond adequately to other pharmacological and nonpharmacological treatments. The use of anxiolytics and antipsychotics in 38.3% and 24.4% of bvFTD patients, as opposed to 23.6% and 9.3% in AD, is thus a reflection of the higher rate of productive behavioral symptoms (e.g., agitation, aggression, and psychosis) in this variant. However, the low rate of antipsychotics use in FTD demonstrated that physicians took into account the alerts on side effects [54, 55] and increased mortality rate  in FTD and dementia patients treated with antipsychotics. The black box warning from the Food and Drug Administration was followed by a similar warning from the French Haute Autorité de Santé in 2009 (https://www.has-sante.fr/jcms/c_885227/fr/limiter-la-prescription-de-neuroleptiques-dans-la-maladie-d-alzheimer) that had found a strong echo in the neurologic and geriatric communities.
The most remarkable difference however regarded the prescription of antidepressants, which was twice as important in bvFTD (55.2%) as in AD (27.0%). Indeed, although results are mixed, comprehensive reviews of the evidence from clinical trials favored the use of selective serotonin reuptake inhibitors to treat behavioral symptoms [47, 51, 52, 57]. Our team in particular demonstrated that trazodone, a serotonin antagonist and reuptake inhibitor, reduced irritability, agitation, and depressive symptoms in FTD . The much better tolerance profile and apparent efficacy of serotonin-acting drugs logically imposed them as the mainstay of FTD treatment in our network.
Strengths and limitations
This naturalistic study of a 6-year period is rooted in 23 years of data sharing and harmonization across a regionwide memory clinic network . It allowed to analyze the trends of real-life FTD diagnosis and care over time. We reached a considerable number of new patients per year, equivalent to the one of nation-wide MC networks. Analyzing the characteristics of consecutive FTD patients first referred between 2010 and 2016 allowed us to focus on patients in which the diagnosis was made using the new criteria for bvFTD and lvFTD  and strengthened by follow-up. By considering a wide spectrum of FTD variants, we included patients that are often withdrawn from FTD cohorts.
Our survey confirmed many previously published data, which reinforces of the quality and validity of our database. We showed that approximately two thirds of FTD patients had a behavioral variant (bvFTD), and 17% had a language variant, which matches other databases [9, 59, 60]. We, as others, found a sex ratio of approximately 1:1 in FTD . Thirty-five percent of our FTD patients had a family history of neuropsychiatric disease, in agreement with the literature [14, 61]. Only our rate of mutations was lower than previously reported since a mutation was identified in C9ORF72, MATP, or GRN in only 6% of the FTD patients that had a genetic analysis, against 10–15% in the literature . Last but not the least, pathological diagnoses when available matched the clinical diagnoses, confirming the high accuracy of the clinical diagnoses made in a structured regional network and confirmed by a prolonged follow-up.
Our survey has however a few limitations. First, important data are not systematically populated in our database. We still lack accurate cognitive, functional, or disease-specific scales to assess disease progression. Furthermore, the mean follow-up of ~ 2 years precludes a comprehensive overview of FTD progression in many of our patients. We also acknowledge a selection bias due to the different networks involved in movement disorders and dementia care in our region, an issue that had been acknowledged in similar studies. Patients with overt motor neuron disease at presentation were not included because they were referred to a specialized regional care pathway rather than to memory clinics. Likewise, the PSP and CBS patients that were referred to our memory clinics were probably the ones presenting early behavioral and/or cognitive changes. Conversely, PSP and CBS with prominent motor symptoms were likely to be followed in movement disorders clinics, where secondary referral to MCs is not systematic. Still, our incidence rate compares to the ones of two regional cohorts including the full FTD spectrum [17, 29].
Conclusion and outlook
Overall, our study showed that FTD syndromes have specific clinical features, different progression patterns, and therapeutic strategies. Yet, even in a region with an organized memory clinic network, FTD is still overlooked and diagnosis wandering remains longer than in AD. Psychiatric, amnestic, and/or late-onset presentations of FTD are particularly treacherous, and the overlap between cognitive/behavioral and motor presentations leads to an underestimation of the motoric presentations of FTD in memory clinics.
There is an obvious need of accurate FTD biomarkers to improve FTD diagnosis. Until and even after the avenue of such biomarkers, neuropsychology has and will have a role to play at a limited cost. The development of novel tests exploring new domains of social cognition beyond mentalization and emotion recognition is a steppingstone in this direction. Social cognition deficits have been found to be a reliable and effective cognitive marker of FTD, especially in patients with a psychiatric  or amnestic [64, 65] presentations. Social cognition deficits are probably underestimated in mFTD as well , advocating for a more systematic assessment of social cognition in memory, geriatric, movement disorders, and psychiatry clinics. In order to improve FTD diagnosis, the classical boundaries between specialties should be broken. Indeed, it is only through a harmonization in diagnostic procedures and databases involving geriatricians, movement disorders specialists, old-age psychiatrists, neuropsychologists, speech-therapists, and memory clinics that the real scope of FTD will be thoroughly apprehended. The gathering of these different disciplines into consortiums such as the Centers of Excellence in Neurodegeneration (CoEN) responds to this objective.
Additionally, initiatives are needed to raise awareness on FTD in the general population. At the eve of disease-modifying therapies, misdiagnosis of FTD may already be a loss of opportunity for patients.