Behavioral variant frontotemporal dementia (bvFTD): PET biomarker characterization of metabolism (18F‐FDG), amyloid (11C‐PIB) and tau (18F‐AV1451) and its clinical correlate ‐ analysis of a cohort from Argentina

Abstract

INTRODUCTION

Imaging biomarkers are fundamental in diagnosing neurodegenerative diseases, but their use in FTD remains limited. This study examines PET biomarkers in Argentine bvFTD patients.

METHODS

We studied a cohort of bvFTD patients (n = 20) and controls (n = 21) with three different PET radiotracers (18F‐FDG, 11C‐PiB, and 18F‐AV1451).

RESULTS

In bvFTD patients, 18F‐FDG PET showed significant hypometabolism in frontotemporal regions, along with hypermetabolism in the precentral gyrus, compared to normal controls. 11C‐PIB did not reveal a pattern typical of Alzheimer's disease, yet increased uptake was notably observed in the precentral region. We found 18F‐AV1451 uptake in frontal lobe, parietal, precuneus, cuneus, posterior cingulum, highly significant in bvFTD with respect to NCs.

DISCUSSION

PET biomarkers are a crucial tool in diverse real‐world clinical scenarios. However, their utility in revealing questions about the underlying pathology in FTD is still limited.

Highlights

• First bvFTD study using 18F‐FDG, 11C‐PIB, and 18F‐AV1451 PET in a Latin American cohort.

• Frontotemporal hypometabolism with compensatory precentral hypermetabolism due to amyloid.

• Amyloid deposits observed in the precentral gyrus without an Alzheimer's‐like pattern.

• 18F‐AV1451 shows limitations in specificity for bvFTD pathology.

• Study provides new insights into PET biomarker utility for bvFTD clinical assessment.

1. BACKGROUND

The use of biomarkers has revolutionized the field of neurodegenerative diseases, leading to earlier diagnosis and bringing us closer to unravelling of the underlying pathology in vivo with minimal invasiveness for patients. This has made it possible to analyze in greater depth the characteristics of the underlying biological phenomena in diseases such as Alzheimer's disease (AD). Similarly, biomarkers have become a major instrument in clinical trials of disease‐modifying drugs. However, the application of biomarkers for detecting proteinopathies in frontotemporal dementia (FTD) remains limited.

Frontotemporal lobar degenerations (FTLD) are a pathologically, genetically, and clinically heterogeneous group encompassing at least six clinical entities: behavioral variant frontotemporal dementia (bvFTD), primary progressive aphasia both semantic (svPPA) and non‐fluent (nfvPPA) variants, progressive supranuclear palsy (PSP), corticobasal degeneration (CBD), and frontotemporal dementia with motor neuron disease (FTD‐MND). Of these clinical entities, bvFTD accounts for about 60% of all FTD cases. ¹ This variant is characterized by behavioral changes, such as disinhibition, apathy, loss of empathy, compulsive behaviors, dietary changes, and a dysexecutive neuropsychological profile. ² At the neuropathological level, bvFTD, like all FTLDs, is characterized as a proteinopathy—a condition involving abnormal protein accumulation or dysfunction, which disrupts neuronal function and results in clinical symptoms. FTLDs affect the frontal and/or temporal brain structures and are classified based on the involved protein: FTLD‐tau, FTLD‐TDP (TDP‐43), and FTLD‐FET. Tau and TDP‐43 are implicated in 90%–95% of cases. ³

There are multiple lines of research exploring potential biomarkers to detect the underlying proteinopathies in FTD. The role of pathological forms of tau as a crucial protein in AD, an entity far more prevalent than FTD, accelerated the development of biomarkers for its detection to a much more advanced stage. ⁴ Understanding the unique characteristics of tau becomes a necessary task to understand the scope and limitations of currently available biomarkers for its detection. Tau exists in six isoforms generated by alternative splicing of the MAPT gene, which affects tau's microtubule‐binding affinity and aggregation propensity. ⁵ FTLD includes both 3R and 4R tauopathies, with specific conditions like Pick's disease and PSP representing 3R and 4R tauopathies, respectively. ⁶ Unlike FTLD, AD is characterized by amyloid plaques and neurofibrillary tangles (NFTs) comprising both 3R and 4R tau.

1.1. Positron‐emission tomography imaging biomarkers

Neuroimaging, particularly positron‐emission tomography (PET) has been instrumental in the diagnosis and study of neurodegenerative diseases, including FTD. Current diagnostic criteria for probable bvFTD require imaging findings indicative of frontotemporal impairment, detectable via structural (computed tomography [CT] or magnetic resonance imaging [MRI]) or functional (PET or SPECT) studies. ² PET using [18F] fluorodeoxyglucose (18F‐FDG), is preferred over SPECT, ² and has been approved by the US Food and Drug Administration (FDA) to assist in the FTD diagnosis. In FTD, decreased 18F‐FDG uptake in the frontal and/or temporal lobes indicates tissue hypometabolism, aiding in differential diagnosis. ⁷ , ⁸

One of the most important differential diagnoses of FTD is with AD, especially with overlapping symptoms and the advent of research criteria for behavioral variant AD. ⁹ Thus, the use of in vivo biomarkers for AD, such as amyloid PET, has become an essential tool. Amyloid PET uses tracers like ¹¹C‐Pittsburgh Compound‐B (11C‐PIB) to detect amyloid deposits. However, 11C‐PIB's short half‐life limits its use to facilities with an on‐site cyclotron.

A definitive diagnosis of bvFTD requires pathological confirmation or the identification of a known genetic mutation. ² This leads to the fact that diagnostic confirmation in sporadic forms can only be performed post mortem, leaving most cases as probable FTD. New PET radiotracers offer promise for in vivo studies. For instance, 18F‐flortaucipir (18F‐AV1451) binds to PHF forms of 3R/4R tau and has been used to study AD in vivo. ¹⁰ However, its application in FTLD has produced inconsistent results across different cohorts and syndromes. ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ Authors like Tsai et al. ¹¹ described the use of 18F‐AV1451 across the FTD spectrum. In bvFTD, 5 of 10 patients showed increased frontal and temporal uptake, but no significant group‐level differences were found. In nfvPPA, uptake was prominent in the left inferior frontal gyrus, while in svPPA, retention varied from the left anterior temporal pole to bilateral orbital frontal and temporal regions. Smith et al. ¹⁶ reported elevated ¹⁸F‐AV‐1451 uptake in the hippocampus and surrounding temporal lobe regions in patients with MAPT mutations. In contrast, Bevan‐Jones et al ¹⁴ highlight the uptake of 18F‐AV1451 in a case of a patient with svPPA associated with C9ORF72 expansion, a mutation strongly associated with TDP‐43 type B pathology.

The objective of this study is to assess the in vivo performance of three radiotracers (18F‐FDG, 11C‐PIB, and 18F‐AV1451) in a cohort of patients clinically diagnosed with bvFTD at a center specializing in cognitive neurology and neuropsychiatry in Argentina. The study will also examine the clinical and potentially pathological implications of the results obtained.

2. METHODS

2.1. Participants

A total of 20 participants (mean age 64 ± 6.86 years) with a diagnosis of probable bvFTD were studied using PET with 18F‐FDG, 11C‐PIB, and 18F‐AV1451 (Table 1). Participants were recruited from the Fleni frontotemporal dementia clinic between 2018 and 2022. All cases were evaluated by cognitive neurologists and/or neuropsychiatrists and clinical diagnosis of bvFTD was reached after a comprehensive neurological, psychiatric, neuropsychological, and MRI/CT evaluation. Genetic testing was performed in patients where there was a family history of FTD or other early‐onset dementia, or late‐onset psychiatric disorders. The current diagnostic criteria were used to diagnose bvFTD. ² Additionally, a participant who carried a MAPT mutation (P301L) was included in the study. This patient was asymptomatic at the time of participating in the protocol but began to exhibit progressive symptoms compatible with bvFTD months after enrollment.

TABLE 1.

Demographics (bvFTD and NC)

Diagnosis	Female	Mean ± SD	18F‐FDG	11C‐PIB	18F‐AV1451
bvFTD (n = 20)	7 (35%)	64.05 ± 6.86	16 (80%)	17 (85%)	20 (100%)
NC (n = 21)	16 (76.19%)	66.9 ± 5.95	21 (100%)	8 (38.1%)	21 (100%)
NC—Fleni	6 (75%)	69.12 ± 6.71	8 (100%)	8 (100%)	8 (100%)
NC—ADNI	10 (76.92%)	65.54 ± 5.24	13 (100%)	0 (0%)	13 (100%)

Abbreviations: bvFTD, behavioral variant frontotemporal dementia; NC, normal control; PIB, 11C‐Pittsburgh Compound‐B SD, standard deviation.

RESEARCH IN CONTEXT

1. Systematic review: An extensive search was conducted using PubMed and Google Scholar. Previous studies on behavioral variant frontotemporal dementia (bvFTD) have extensively evaluated [18F] fluorodeoxyglucose (18F‐FDG) as a biomarker for brain metabolism, while ¹¹C‐Pittsburgh Compound‐B (11C‐PIB) has been used to distinguish Alzheimer's disease cases. Few studies use 18F‐AV1451 in bvFTD, with inconsistent results. This study's strength lies in utilizing all three radiotracers in a Latin American bvFTD cohort.

2. Interpretation: Our findings confirm frontotemporal hypometabolism in bvFTD and reveal unexpected 11C‐PIB uptake in the precentral region, which aligns with hypermetabolism on 18F‐FDG. We also observed widespread 18F‐AV1451 uptake, raising further questions about its specificity in FTD.

3. Future directions: This study provides insights into the clinical utility of positron emission tomography (PET) biomarkers and uncovers phenomena such as hypermetabolism and amyloid deposition in bvFTD. Further research is needed to develop radiotracers targeting bvFTD's underlying pathology.

Study participants underwent 18F‐FDG to measure neuronal metabolism and as part of the diagnostic process for FTD. Similarly, 11C‐PIB was performed as part of the differential diagnosis with AD, thus excluding any case where the 11C‐PIB uptake pattern was compatible with AD. All participants underwent 18F‐AV1451. Consequently, 85% of the participants (18 cases) underwent 11C‐PIB and 80% (16 cases) 18F‐FDG. The individual with a MAPT mutation (58 years old) did not undergo 11C‐PIB, as the differential diagnosis with AD was not considered a relevant clinical scenario.

Twenty‐one normal controls (NCs) were included in the study, in accordance with the Alzheimer's Disease Neuroimaging Initiative (ADNI) normality criteria (age range 66.9 ± 5.9 years). All NCs underwent paired 18F‐FDG and 18F‐AV1451 imaging, with 38% also undergoing 11C‐PIB imaging. To increase the number of 18F‐AV1451 normal controls, 13 subjects were included from the ADNI base, provided that they fell within the age range of normals and possessed 18F‐FDG imaging to spatially normalize to a template.

2.2. PET acquisition

In the participants, PET was initially performed with 11C‐PIB, following a 70‐min uptake period of the radiotracer. Images were acquired for 20 min. After 70 min from the end of the study, 18F‐FDG was administered. In patients where the 11C‐PIB qualitative report was negative for a pattern compatible with AD, 18F‐AV1451 was subsequently performed after at least 70 min had elapsed between both tracers. Clarifying that fluorinated radiotracer studies are always performed in two separate visits.

The PET data were acquired on a PET/CT Discovery 690 scanner (GE Healthcare, Milwaukee, USA) in 3D mode (full 3D iter, 47 parallel planes) with an axial field of view (FOV) of 25.6 cm, a resolution of 1.33 mm full‐width at half‐maximum (FWHM), and a slice thickness of 3.27 mm. The data were reconstructed using four iterations and 24 subsets (unfiltered) and corrected for decay. The attenuation and scatter corrections were made from a previously acquired transmission study (CT). The final resolution of the reconstructed images was approximately 6 mm FWHM.

2.3. 18F‐FDG

Brain metabolism acquisitions were performed by injecting ∼370 MBq of intravenous 18F‐FDG, in 3 mL (approx.) of solution. The administration and uptake of the radiopharmaceutical were carried out in a controlled environment (light and temperature). The biodistribution time prior to acquisition in the scanner was 35 min.

Data were acquired in list mode for 20 min using 4 × 300 s frames for motion correction. The frames were smoothed using a filter with a Gaussian kernel of 8 × 8 × 8 mm and realigned using SPM12 (Wellcome Trust Center for Neuroimaging). The realigned images were then averaged to obtain the glucose metabolism (18F‐FDG) uptake image. The images of all subjects were spatially normalized to an FDG template using the diffeomorphic algorithm SyN (symmetric normalization) of ANT (Advanced Neuroimaging Tools) to an MNI (Montreal Neurological Institute) space of 91 × 109 × 91 2 × 2 × 2 mm voxels.

2.4. 11C‐PIB

The amyloid image (11C‐PIB) is acquired 40 min post injection of 370 MBq list mode for 20 min using 4 × 300 s frames for motion correction. The frames were filtered (smoothed) using a filter with a Gaussian kernel of 8 × 8 × 8 mm and realigned using SPM12. The realigned frames were then averaged to obtain the amyloid cortical uptake image. This image is then co‐registered to the FDG and normalized by applying its transformation to the template.

2.5. 18F‐AV1451

The tau image (18F‐AV1451) is acquired 80 min post injection of 370 MBq in list mode for 20 min using 4 × 300 s frames for motion correction. The frames were filtered (smoothed) using a filter with a Gaussian kernel of 8 × 8 × 8 mm and realigned using SPM12. The realigned frames were then averaged to obtain the 18F‐AV1451 cortical uptake image. This image is then co‐registered to the FDG and normalized by applying its transformation to the template.

2.6. Image processing

2.6.1. Spatial normalization

Spatial normalization must ensure that all brain structures are in the same space and adequately compensate for variations in cortical thickness caused by age or pathophysiological alterations. In line with this, a high‐efficiency non‐linear deformation algorithm (SyN ANTs) ¹⁷ , ¹⁸ , ¹⁹ was used to generate an FDG template using 16 subjects uniformly distributed in the cognitive spectrum studied.

All FDG acquisitions were normalized to this template and then the same deformations calculated in the FDG were applied to 11C‐PIB and 18F‐AV1451 (motion correction, smooth, co‐registration to the FDG).

2.6.2. Intensity normalization

For voxel‐wise parametric differences, all acquisitions (18F‐FDG, 11C‐PIB, and 18F‐AV1451) were normalized to cerebellar gray matter (Cg).

Given that the images undergo intensity normalization, it is imperative to acknowledge that group statistical comparisons (at both the voxel and volumes of interest [VOIS] level) are inherently relative changes in the context of the analyzed comparisons. This consideration is crucial, as any observed increases or decreases in uptake of the respective radiopharmaceuticals are ultimately relative changes in the context of the analyzed comparisons.

2.6.3. Parametric voxel‐wise statistics

To evaluate the parametric differences between groups at the voxel level, for each of the measures of bvFTD against normal aging (NC), a multifactorial analysis analysis of covariance (ANCOVA) (SPM 12) was carried out with two factors: diagnostic category and measurement (18F‐FDG, 11C‐PIB, and AV‐1451), assuming independence between the levels in the measurements and unequal variance. The threshold for statistical significance was set at p < 0.05 (corrected for multiple comparisons by false discovery rate method). Comparisons were made in the sense: bvFTD‐NC.

The resulting images were created using BrainNet Viewer 1.6 software 27. SPM t‐maps were overlaid on the ICBM152 (smoothed) brain mesh provided in the software. The ranges of the color scale were relative to the maximum and minimum of each type of measurement.

2.6.4. VOIS

VOIS were selected using the Boggle DKT (Desikan‐Killiany‐Tourville) atlas. ²⁰ This atlas has 31 cortical regions for each hemisphere and 14 subcortical structures. To use the VOIS defined in this atlas, a dilation with a 3 × 3 × 3 (6‐connected) kernel was performed by mathematical morphology.

The Cg VOI used for intensity normalization was manually created by minimizing white matter absorption and spillover effects due to proximity to the occipital cortex. ²¹

A quantification of the mean value in the 18F‐FDG, 11C‐PIB, and 18F‐AV1451 images in each of the VOIS was generated (supplementary material).

3. RESULTS

In the quantitative uptake analysis of 18F‐FDG, 11C‐PIB, and 18F‐AV1451, the group statistical comparison at the voxel‐wise level between bvFTD and NC (bvFTD‐NC) shows the patterns of cortical involvement (Figure 1). It is important to interpret this statistical subtraction in the sense that negative values indicate regions where the uptake of the radiopharmaceutical is significantly lower in the bvFTD group than in NC (bvFTD < NC; p < 0.05, blue in Figure 1). Conversely, positive differences represent regions where the uptake is higher in bvFTD than in NC (bvFTD > NC, red in Figure 1).

FIGURE 1.

Voxel‐wise statistical maps showing significant differences (threshold p < 0.05) for comparisons between normal control groups versus bvFTD, in measurements of metabolism with FDG, cortical amyloid load with PIB and 18F‐AV151 uptake, each subject was normalized to cerebellar gray matter. bvFTD, behavioral variant frontotemporal dementia; FDG, [18F] fluorodeoxyglucose; PIB, 11C‐Pittsburgh Compound‐B

The mean values and standard deviations for 18F‐FDG, 11C‐PIB, and 18F‐AV1451 uptake in VOIs for both groups are provided in the supplementary materials section. The statistical significance of the p‐values is reported due to the t‐student differences between them. It should be noted that the results of the statistical analysis of the general linear model with voxel‐wise SPM may vary in relation to the VOIS statistic. This is not only due to the difference in size between a voxel and a VOI, but also because the general linear model assigns estimates to the modeled variables.

3.1. 18F‐FDG

The voxel‐wise parametric statistical analysis of cortical metabolism measured with 18F‐FDG shows a significant decrease in uptake (p < 0.05) in the fronto‐opercular, supraorbital, middle and inferior temporal, anterior cingulate, and thalamus regions. Additionally, a significant increase in uptake was observed in relation to the NCs at the level of the bilateral perirolandic convexity (precentral, postcentral, paracentral, and precuneus sulcus).

Group parametric statistical analysis of VOIS reveals a reduction in cortical metabolism in the bvFTD group with respect to NCs (bvFTD < NC) in the following regions: left lateral orbitofrontal (p <0.0314), left medial temporal (p <0.0231), left pars orbitalis (p <0.0116), left rostral anterior cingulate (p <0.0102), right caudal anterior cingulate (p <0.0224), right medial orbitofrontal (p <0.0052), and right rostral anterior cingulate.

The regions exhibiting an increase in cortical metabolism in the bvFTD group relative to the NC (bvFTD > NC) are the left medial orbitofrontal (p <0.0063), left postcentral (p <0.034), right postcentral (p <0.0208), right precentral (p <0.0434), and right superior parietal (p <0.0271) regions.

3.2. 18F‐AV1451

Voxel‐wise analysis shows a diffuse AV‐1451 uptake in frontal lobe, parietal, precuneus, cuneus, posterior cingulum, highly significant in bvFTD with respect to NCs.

The VOIS show increased uptake in the bvFTD group with respect to normals (bvFTD > NC) in the regions: left caudal middle frontal (p < 0.0001), left cuneus (p < 0.0332), left inferior parietal (p < 0.0183), left lateral orbitofrontal (p < 0.326), left paracentral (p < 0.0007), left pars triangularis (p < 0.0326), left postcentral (p < 0.0013), left precentral (p < 0.0017), left perecuneus (p < 0.0034), left rostral middle frontal (p < 0.0005), left superior frontal (p < 0.0008), left superior parietal (p < 0.0004), right caudal middle frontal (p < 0.0003), right cuneus (p < 0.031), right inferior parietal (p < 0.0202), right paracentral (p < 0.0001), right pars triangularis (p < 0.0254), right postcentral (p < 0.0012), right posterior cingulate (p < 0.0457), right superior frontal (p < 0.0001), right superior parietal (p < 0.017), and right supramarginal (p < 0.0261). There was no difference in any VOI in the bvFTD < NC direction.

Among the participants in this study there was one patient with a P301L mutation of MAPT, associated with tau pathology. In this case, AV 1451 radiotracer uptake was found in the prefrontal region, mainly medial, anterior cingulate, and bilateral polar temporal (Figure 2).

FIGURE 2.

A 58‐year‐old female patient with confirmed mutation in MAPT (P301L) asymptomatic at the time of PET imaging with 18F‐AV1451. Six months after the study patient exhibited deficits in executive tasks and behavioral symptoms (apathy, loss of empathy, and behavioral disinhibition). PET, positron‐emission tomography

3.3. 11C‐PIB

Voxel‐wise analysis of 11C‐PIB uptake evidence cortical PIB retention in the convexity of the paracentral lobe, pre and postcentral sulcus and bilateral cuneus in the bvFTD group. These differences are observed diminished in the VOIS analysis where statistically significant variations are only differentiated in the left postcentral (p < 0.032), right postcentral (p < 0.0265), and right precentral (p < 0.0322) regions showing an increase in cortical incorporation in the bvFTD group. In the NC group, isolated 11C‐PIB uptake was found without a specific pathological pattern. Following the quantitative analysis, a qualitative evaluation of 11C‐PIB uptake was perform by two nuclear medicine specialists with extensive experience in amyloid imaging. None of the cases demonstrated a pattern of uptake consistent with AD.

4. DISCUSSION

This paper presents the analysis of a cohort of patients diagnosed with bvFTD in which PET studies were performed with different radiotracers to evaluate brain metabolism (18F‐FDG), cortical amyloid deposition (11C‐PIB), and uptake of the first‐generation tracer for tau AV‐1451 (Figure 3). This study has both significant strengths and a few limitations. To the best of our knowledge, this is the first study to date in a cohort of patients with bvFTD to employ all three radiotracers. Furthermore, it also allowed us to explore the clinical and radiological characteristics of an often underrepresented population such as the Latin American population, ²² in this case all the patients being Argentinean. Similarly, this work required a significant multidisciplinary collaboration between neurologists, neuropsychiatrists, molecular imaging specialists, and physicists. It should be noted that the study is not without |ations. The number of patients recruited is limited, as recruiting participants willing to undergo the three biomarkers mentioned above can be an arduous task and may generate reluctance from participants. Furthermore, the lack of pathological confirmation of the underlying proteinopathy is another pitfall of this study.

FIGURE 3.

A 49‐year‐old male patient with a diagnosis of bvFTD who underwent the three radiotracers, showing 18F‐AV1451 uptake at the prefrontal region, mainly medial, anterior cingulate, and temporal polar, lateral and mesial, bilaterally; 18F‐FDG showed severe ventro‐medial prefrontal, anterior cingulate, polar temporal and mesial prefrontal hypometabolism, caudate nucleus, bilateral, with surrounding areas of moderate to mild hypometabolism; 11C‐PIB negative for cortical amyloid deposition. bvFTD, behavioral variant frontotemporal dementia; PIB, 11C‐Pittsburgh Compound‐B

The use of 18F‐FDG PET is one of the most prevalent functional imaging techniques employed in the diagnostic process of FTD As expected in FTD cases, we anticipated observing a pattern of hypometabolism localized in the frontal and/or temporal regions of the brain, corresponding to the neurodegeneration characteristic of the disease. In alignment with prior findings, ² , ⁷ , ²³ , ²⁴ , ²⁵ our study also demonstrated significant impairment in these brain structures, which holds considerable diagnostic value when determining the likelihood of FTD in clinical settings. Our group analysis revealed a more pronounced hypometabolism in the left temporal lobe, consistent with the frequent observation of hemispheric asymmetry in FTD. ²⁴ A salient finding of this study is the evidence of hypermetabolism at the precentral region. This finding is particularly noteworthy when considered in conjunction with the concordance of this hypermetabolism with 11C‐PIB uptake in the same area, which reflects amyloid deposition.

While not a prerequisite for diagnosis, the use of biomarkers for cortical amyloid deposition is of paramount importance in several clinical scenarios where the differential diagnosis between FTD and AD is challenging. In this study, we included 11C‐PIB PET imaging not only as part of the diagnostic process but also due to concerns regarding the potential non‐specific binding of 18F‐AV1451 to amyloid plaque. ²⁶ None of the cases demonstrated a pattern of uptake consistent with AD, further supporting our findings

Interestingly, we found statistically significant bilateral uptake of 11C‐PIB in the precentral region among bvFTD patients, when compared to normal controls (NC), suggesting that this finding is not incidental and is unlikely to be attributable to other confounding factors such as age. The precentral gyrus is not typically implicated in the pathogenesis of bvFTD, nor is it a common site for amyloid deposition in AD. However, there are reports where amyloid deposition is evidenced at the precentral and postcentral regions in 15%–25% of AD cases. ²⁷ In our experience, the finding of 11C‐PIB in this area is a relatively frequent phenomenon in patients with clinical syndromes and uptake patterns not compatible with AD. Likewise, there are reports in the literature of similar phenomena in FTLD cases. ²⁸ This result is even more interesting when considered in conjunction with the findings obtained with 18F‐FDG, as an increase in tracer uptake, indicative of hypermetabolism, was observed in the same region. However, the interpretation of this phenomenon is constrained by several factors, both technical and theoretical. One possible interpretation for the presence of hypermetabolism in the same area as amyloid deposition is that the former phenomenon is compensatory for the latter. Several authors have proposed that this same mechanism reflects compensatory neuronal recruitment to maintain normal cognitive function as an adaptive response in the early pathological stages of AD. ²⁹ , ³⁰ , ³¹ , ³² , ³³ In contrast, it has also been proposed that increased regional metabolism may lead to increased amyloid deposition. ³¹ , ³⁴ Although the cause‐effect discussion between these phenomena is open, there is evidence for a relationship between patterns of hypermetabolism and amyloid deposition in AD. The findings from our study suggest that similar mechanisms may be at play in bvFTD.

Once a clinical diagnosis of bvFTD is reached, specialists in behavioral neurology and neuropsychiatry often speculate about the underlying pathology of the disease. While the answer to this question is reserved for post mortem pathological confirmation, the advent of new biomarker techniques raises hopes of finding a way to unveil these phenomena in vivo and in a minimally invasive manner. It is estimated that 45% of FTD cases are due to FTLD‐Tau, with Pick's disease (Tau 3R) being the pathology in bvFTD. ³ , ⁶ In this study, the 18F‐AV1451 was used to detect tau deposits, with tracer uptake patterns observed in the frontal and temporal regions, consistent with the expected areas of neurodegeneration in bvFTD. In our cohort, 18F‐AV1451 uptake was found in all cases. This finding can rapidly lead to the interpretation that all cases are tauopathies. However, this interpretation is unlikely to be correct as it strongly defies the reported prevalence of tau in FTD. An alternative hypothesis is that the radiotracer is binding nonspecifically to another molecular target, most likely TDP‐43. This hypothesis is supported by previous studies which described 18F‐AV1451 binding in clinical syndromes and genetic mutations known to be associated with TDP‐43 pathology, such as C9ORF72 and GRN mutations. ¹¹ , ¹⁴ , ³⁵ Moreover, given the potential for non‐specific binding to amyloid, ²⁶ the inclusion of 11C‐PIB imaging in our study was critical to exclude AD as a confounding factor. In cases where amyloid deposition was observed but deemed non‐specific for AD, the possibility of non‐specific 18F‐AV1451 binding to amyloid could be ruled out. Therefore, in the absence of amyloid, the observed 18F‐AV1451 uptake, particularly when correlated with hypometabolism on 18F‐FDG PET, on a well‐defined clinical syndrome such as bvFTD, may indicate the presence of tau or possibly TDP‐43 pathology. Despite this theoretical interpretation, the lack of pathological confirmation in this cohort precludes the advancement of this hypothesis.

Among the study participants was an individual with the P301L mutation in the MAPT gene, known to be associated with 4R tau pathology. At the time of the PET scan, this patient was asymptomatic, but subsequently developed behavioral changes characteristic of bvFTD, including apathy, loss of empathy, and disinhibition. In this patient, 18F‐AV1451 uptake was observed in the prefrontal region, mainly medial, anterior cingulate, and bilateral polar temporal. Mutations involving MAPT exon 10, such as P301L, are associated with 4R forms of tau pathology, whereas those falling outside this exon are associated with mixed 3R/4R forms of tau. ³⁶ , ³⁷ Similarly, previous studies in carriers of this mutation have reported that 18F‐AV1451 uptake is lower in mutations carriers such as P301L, as the tracer does not bind with as much affinity to 4R pathology. In contrast, this does not occur in mutations located outside exon 10, where 3R/4R forms of tau are prevalent and 18F‐AV1451 uptake is higher. In exon 10, where 3R/4R forms of tau are prevalent, the pathology is more similar to that observed in Alzheimer's disease (AD), and 18F‐AV1451 uptake is higher. ¹¹ , ³⁷

An important consideration when interpreting 18F‐AV1451 binding in bvFTD cases is its lack of specificity for PHF forms of 3R/4R tau, for which it was originally designed. Although this tracer was designed and approved to detect neurofibrillary tangles in AD, it exhibits uptake across various FTD phenotypes, regardless of whether they correspond to tauopathies or TDP‐43 pathology. ¹¹ , ¹⁶ , ³⁵ This raises concerns about its reliability for differential diagnosis, particularly when used in isolation without complementary biomarkers such as amyloid‐PET, and has important clinical implications. If 18F‐AV1451 exhibits off‐target binding, its ability to accurately differentiate AD from other neurodegenerative conditions becomes uncertain.

PET biomarkers have become an invaluable tool in the study of neurodegenerative disorders such as FTD. Each of the radiotracers described plays a pivotal role in the diagnostic process. The 18F‐FDG allows for the identification of hypometabolism in the frontal and temporal regions of the brain, which is a hallmark of FTD. In cases where AD is a differential diagnosis, 11C‐PIB PET imaging helps to confirm or exclude amyloid pathology. Once the symptoms of the disease, clinical progression, and imaging findings have been identified, a diagnosis of probable FTD can be made, which is the final instance for most patients in the context of routine clinical practice due to limitations in access to more complex studies, this being particularly true in middle‐ and low‐income countries. However, a key question that a specialist in behavioral disorders often considers at this juncture is the underlying pathological etiology of the disease. This question, which may initially appear to be merely a scientific curiosity, becomes essential with the realization that clinical prognosis may vary depending on whether the pathology is FTLD‐tau or FTLD‐TDP. This is exemplified by the likelihood of motor neuron disease in TDP‐43 pathology. Moreover, the emergence of novel clinical trials and molecules targeting specific pathological processes has highlighted the need to identify such processes in vivo. However, developing a radiotracer to detect the underlying proteinopathy in vivo in FTD remains a significant challenge. In this context, studies such as this one, which examine various biomarkers, are of particular importance to improve diagnosis and facilitate the development of effective treatments for patients with FTD.

CONFLICT OF INTEREST STATEMENT

The authors declare that they have no conflicts of interest to disclose in relation to this manuscript. Author disclosures are available in the supporting information.

CONSENT STATEMENT

All human subjects provided written informed consent.

Supporting information

Supporting information

Supporting information

Magrath Guimet N, Falasco G, Bergamo Y, et al. Behavioral variant frontotemporal dementia (bvFTD): PET biomarker characterization of metabolism (18F‐FDG), amyloid (11C‐PIB) and tau (18F‐AV1451) and its clinical correlate ‐ analysis of a cohort from Argentina. Alzheimer's Dement. 2025;21:e70187. 10.1002/alz.70187