Tau‐PET correlates of mild behavioral impairment in preclinical and prodromal Alzheimer's disease: A head‐to‐head comparison of Flortaucipir and MK6240

Abstract

INTRODUCTION

We compared the association between mild behavioral impairment (MBI) and first‐ versus second‐generation tau‐positron emission tomography (PET) tracers in preclinical and prodromal Alzheimer's disease (AD).

METHODS

We assessed 110 individuals with head‐to‐head [¹⁸F]MK6240 and [¹⁸F]Flortaucipir tau‐PET. Voxel‐wise and region‐of‐interest analyses investigated relationships between tau‐PET and MBI Checklist (MBI‐C) scores, which were also compared across tau‐PET positivity and PET‐Braak stages.

RESULTS

For both tracers, MBI severity was associated with tau‐PET signal in all Braak regions and increased across PET‐Braak stages. [¹⁸F]MK6240 showed more extensive voxel‐wise associations with MBI‐C scores and additional correlations with affective symptoms in stratified analyses by diagnostic groups. Tau positivity was associated with a higher frequency of MBI. Individuals positive for [¹⁸F]MK6240 but negative for [¹⁸F]Flortaucipir exhibited greater MBI burden than individuals negative for both tracers.

DISCUSSION

MBI in early AD is associated with tau‐PET, with [¹⁸F]MK6240 providing greater sensitivity to detect MBI‐related symptoms.

Highlights

• Tau‐positron emission tomography (PET) signal in Alzheimer's disease (AD) ‐related regions was associated with mild behavioral impairment (MBI) severity across both tracers.

• [¹⁸F]MK6240 revealed more extensive voxel‐wise associations with global MBI severity.

• [¹⁸F]MK6240 detected MBI severity variation within clinical‐biological stages.

• Individuals positive only for [¹⁸F]MK6240 exhibit higher MBI burden than those negative for both tracers.

• Tau‐PET‐MBI associations depended on tracer, symptom, region‐of‐interest, and disease stage.

1. BACKGROUND

Mild behavioral impairment (MBI) refers to the emergence of persistent neuropsychiatric symptoms (NPS) in later life that may represent an early manifestation of Alzheimer's disease (AD). ¹ MBI encompasses apathy, mood and anxiety disturbances, impulse dyscontrol, social inappropriateness, and psychotic symptoms emerging prior to dementia, at any of the pre‐dementia cognitive stages (i.e., cognitively unimpaired [CU], subjective cognitive impairment [SCD], mild cognitive impairment [MCI]). ¹ Beyond negatively impacting quality of life, ² MBI is associated with an increased risk of progression to more advanced clinical stages and predicts future neuropathologically confirmed AD, suggesting that it may serve as an early prognostic marker. ³ , ⁴ , ⁵ , ⁶ Therefore, identifying MBI during the preclinical and prodromal stages of AD may improve early detection strategies, facilitate the development of preventive and therapeutic interventions, and provide a practical, low‐cost approach for identifying at‐risk individuals in routine clinical settings.

Incorporating biomarkers is a promising strategy to determine the neuroanatomical correlates and severity of MBI symptoms linked to AD, as MBI has been associated with core in vivo AD biomarkers, ⁷ , ⁸ including tau positron emission tomography (tau‐PET) using both first‐ and second‐generation tracers. ³ , ⁹ , ¹⁰ , ¹¹ However, evidence regarding the relationship between MBI and tau‐PET remains limited and heterogeneous across study populations, with some findings showing no significant associations in CU individuals. ¹² In this context, it remains unclear whether tau‐PET tracer choice contributes to the variability of reported findings and influences the detection of MBI‐related tau associations in AD. Specifically, first‐ and second‐generation tau‐PET tracers differ in their affinity for tau aggregates, off‐target binding characteristics, and binding sites, raising the possibility that tracer selection may affect the strength and detectability of associations between MBI symptoms and tau pathology. ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ , ¹⁹

Therefore, we aimed to investigate the association between the presence and severity of MBI and tau pathology measured head‐to‐head using first‐ and second‐generation tau‐PET tracers across the spectrum of healthy aging and preclinical and prodromal AD. To this end, we applied multiple analytical approaches to identify whether and where tracer‐specific differences emerge in the detection of associations between tau pathology and MBI symptoms. We hypothesized that MBI severity would be associated with greater tau pathology measured by both tracers and that the improved binding characteristics of the second‐generation tracer [¹⁸F]MK6240 would provide greater sensitivity for detecting these associations.

2. MATERIALS AND METHODS

2.1. Participants

This cross‐sectional study included CU and MCI individuals aged 50 years or older from the Translational Biomarkers of Aging and Dementia (TRIAD; https://triad.tnl‐mcgill.com/) cohort scanned at McGill University as part of the HEAD study, ¹⁷ a multicentric initiative across nine research centers throughout the United States and Canada. Participant selection followed the criteria for MBI, which may be diagnosed in CU and MCI individuals. ¹ Participants were recruited either from the community or from the McGill Research Centre of Studies on Aging outpatient memory clinic between December 2022 and November 2024.

Participants underwent clinical evaluations conducted by a multi‐professional team including neurologists, geriatricians, and neuropsychologists specialized in dementia care. To be classified as CU, participants had a Clinical Dementia Rating (CDR) score of 0 and presented no objective cognitive impairment in standardized neuropsychological testing. Participants with MCI had a CDR score equal to 0.5 and met the criteria for MCI due to AD according to the National Institute on Aging—Alzheimer's Association. ²⁰ We included CU amyloid‐β (Aβ) negative controls and participants with preclinical (i.e., CU Aβ+) or prodromal (i.e., MCI Aβ+) AD. MCI Aβ‐ individuals were excluded to restrict the sample to the AD continuum. ²¹

RESEARCH IN CONTEXT

1. Systematic review: We searched PubMed on June 22nd, 2026, by using the terms Tau and (positron emission tomography [PET]) and alzheimer* and “mild behavioral impairment” for studies investigating the association between mild behavioral impairment (MBI) and tau positron emission tomography (tau‐PET) across the Alzheimer's disease (AD) continuum. The search identified 14 records, of which six were relevant and therefore discussed in the present work. These studies were conducted in four different cohorts and used three tau‐PET tracers: [¹⁸F]MK6240, [¹⁸F]RO948, and [¹⁸F]Flortaucipir. None of them scanned participants head‐to‐head with more than one tau‐PET tracer.

2. Interpretation: Here, greater MBI symptom burden was associated with higher tau‐PET signal measured by both [¹⁸F]MK6240 and [¹⁸F]Flortaucipir. MBI severity increased across PET‐derived Braak stages, and associations were primarily observed in Aβ‐positive individuals. Compared with [¹⁸F]Flortaucipir, [¹⁸F]MK6240 identified more extensive voxel‐wise associations and additional stage‐specific relationships with MBI symptoms, suggesting greater sensitivity to early behavioral manifestations of AD.

3. Future directions: Longitudinal studies are needed to determine whether distinct tau‐PET tracers differentially predict future MBI burden and enhance the prognostic value of MBI for predicting progression to more advanced clinical stages of AD.

Other exclusion criteria were visual and auditory impairments hampering cognitive and neuropsychiatric assessments, inability to speak English or French, inability to provide a reliable informant (e.g., close friend, family member) for neuropsychiatric symptoms (NPS) assessments, presence of other neurological or psychiatric disorders, uncontrolled systemic disorders, recent major surgery or traumatic brain injury, PET/magnetic resonance imaging (MRI) contraindications, and a Hachinski Ischemic Score > 4.

Approval for this study was granted by the Montreal Neurological Institute (MNI) PET working committee and the Douglas Mental Health University Institute Research Ethics Board (IUSMD 21–39). Participants were informed about all procedures involved in the research and signed a written informed consent form.

2.2. Procedures

We employed [¹⁸F]AZD4694 Aβ‐PET to evaluate the Aβ status of participants by calculating centiloid values based on published recommendations. ²² Individuals with a Centiloid value ≥ 20 were considered as Aβ+. ²³ All participants underwent brain MRI and head‐to‐head tau‐PET imaging with [¹⁸F]AV1451 ([¹⁸F]Flortaucipir) and [¹⁸F]MK6240 performed within a mean ± SD of 24 ± 24 days of each other.

Participants underwent extensive clinical evaluation, including neurological examination as well as assessments for comorbidities, function, behavioral changes, sleep, nutritional status, and comprehensive laboratory workup. Participants further underwent neuropsychological assessment including tests for global cognition, such as the CDR Sum of Boxes (CDR‐SB), and tools for specific cognitive domains following the National Alzheimer's Coordinating Center Uniform Data Set, either version 3.0 or 4.0. ²⁴

2.3. Neuroimaging acquisition and processing

Structural MRI was acquired on a 3T Siemens Magnetom scanner equipped with a standard head coil, following protocols aligned with ADNI4 guidelines. ¹⁷ , ²⁵ PET imaging was conducted using a brain‐dedicated Siemens High‐Resolution Research Tomograph. Tau‐PET scans were acquired with both [¹⁸F]MK6240 and [¹⁸F]Flortaucipir at the same timepoint, and Aβ‐PET with [¹⁸F]AZD4694. [¹⁸F]MK6240, [¹⁸F]Flortaucipir, and [¹⁸F]AZD4694 images were collected respectively between 90 and 110, 80 and 110, and 40 and 70 min following tracer administration. PET images were reconstructed with ordered‐subset expectation maximization (OSEM)‐based algorithms into four consecutive 5‐min frames using standardized reconstruction parameters and without post‐reconstruction smoothing. ¹⁷ Corrections for attenuation, motion, dead time, radioactive decay, scatter, and random coincidences were applied. ¹⁷

Using affine + SyN transformations (ANTs), T1‐weighted MRI scans were spatially normalized to the International Consortium for Brain Mapping (ICBM) 2020 Nonlinear Symmetric template. ¹⁷ , ²⁶ We also used ANTs to linearly co‐register PET images to native T1‐weighted MRIs. MRI‐derived spatial transformations were used to register PET images to the ICBM stereotaxic space.

Standardized uptake value ratios (SUVRs) were derived in native space, by using the inferior cerebellar gray matter as the reference region for tau‐PET and the whole cerebellar gray matter for Aβ‐PET. ¹⁷ , ²⁷ Spatial smoothing was applied to achieve a final resolution corresponding to an 8‐mm full width at half maximum (FWHM). Further details on imaging processing can be found elsewhere. ¹⁷

2.4. Tau‐PET measures

We derived regional average SUVR from previously validated Braak‐like anatomical masks. ²⁸ , ²⁹ , ³⁰ PET‐Braak regions were grouped as follows: I–II (transentorhinal cortex, entorhinal cortex, and hippocampus), III–IV (amygdala, fusiform gyrus, and parahippocampal gyrus, inferior parietal, inferior temporal, insula, lateral temporal, and posterior cingulate), V‐VI (anterior cingulate, cuneus, inferior frontal, lateral occipital, lingual gyrus, orbitofrontal, precuneus, rostromedial frontal, superior frontal, superior parietal, superior temporal, and supramarginal gyrus, paracentral, pericalcarine, postcentral, and precentral). ²⁸ , ²⁹ , ³⁰ Regions of interest (ROIs) were defined using the Desikan–Killiany–Tourville atlas. ³¹ The transentorhinal cortex was segmented to ensure no spatial overlap with the entorhinal cortex. ²⁹ Following prior validation with [¹⁸F]MK6240 and [¹⁸F]Flortaucipir, the Braak I–II mask was modified to include only the hippocampal CA1 subfield ³⁰ to reflect the neuropathological definition of Braak stage II ³² and avoid the off‐target binding in the choroid plexus characteristic of [¹⁸F]Flortaucipir. ¹⁴

Average SUVR was also derived from a temporal meta‐ROI comprising the entorhinal cortex, amygdala, parahippocampal gyrus, fusiform gyrus, inferior temporal cortex, and middle temporal cortex. ³³ Regional tau abnormality thresholds were defined using a reference sample of 27 CU Aβ‐ young adults aged 19–27 years (mean [SD] age, 22.74 [1.83] years; 59% female) with head‐to‐head [¹⁸F]MK6240 and [¹⁸F]Flortaucipir scans available. Regions were classified as tau‐positive when SUVR values exceeded the reference‐group mean by at least 2.5 standard deviations. ²⁸ , ²⁹ Participants were assigned a PET‐based Braak stage according to the latest Braak ROI exhibiting abnormal tau‐PET signal. ³⁴ In addition, binary tau‐PET positivity status was determined based on Braak I–II and temporal meta‐ROI positivity, consistent with commonly used approaches for defining tau positivity in AD research. ³³ , ³⁵

The spatial extent of tau pathology (SEOT) was used to quantify the proportion of a given ROI affected by tauopathy. SEOT was derived using a voxel‐wise thresholding approach based on the same reference group and abnormality cutoff used for the regional SUVR analyses, as previously described. ³⁶ , ³⁷ , ³⁸ Individual binary maps of tau‐PET abnormality were generated, and SEOT was calculated as the percentage of abnormal voxels within the Braak I–II ROI and within a temporal meta‐ROI. SEOT values ranged from 0% to 100%, with higher values indicating a greater spatial extent of tau pathology.

2.5. MBI assessment

MBI symptoms were evaluated using the 34‐item MBI Checklist (MBI‐C). ³⁹ To meet MBI criteria, symptoms are required to be present for at least 6 months and to represent a change from the participant's longstanding pattern of behavior. ³⁹ Knowledgeable informants, such as family members or close friends, completed the MBI‐C and rated symptom severity when present using a three‐point scale ranging from 1 (mild) to 3 (severe). ³⁹ The instrument comprises five domains: decreased motivation (0–18), affective dysregulation (0–18), impulse dyscontrol (0–36), social inappropriateness (0–15), and abnormal thoughts and perception (0–15). Domain scores are summed to generate a total score ranging from 0 to 102, with higher scores indicating greater symptom burden.

For MBI status assignment, we applied previously validated cognitive status‐specific cutoffs. In CU participants, MBI was defined as an MBI‐C total score > 8, whereas in participants with MCI, as a score > 6. ⁴⁰ , ⁴¹

2.6. Statistical analyses

Statistical analyses were conducted using R software version 4.5.2. Statistical significance was defined as a two‐tailed p value < 0.05. Voxel‐wise linear regression analyses were conducted in R using the RMINC library with MBI‐C total score, age, sex, and years of education as predictors, and [¹⁸F]MK6240 and [¹⁸F]Flortaucipir PET images as outcomes of separate models. Random field theory correction for multiple comparisons was applied at a threshold of p < 0.001. Voxel‐wise analyses were conducted inside a mask including all regions from the CerebrA atlas. ¹⁷ , ⁴² We calculated the proportion of significant voxels following Random Field Theory (RFT) correction in the resulting t‐maps relative to the analysis mask. To compare the distribution of effect sizes, voxel‐wise standardized β coefficients were extracted from all voxels within the analysis mask from the β‐maps generated for each tracer model. Standardization of β coefficients was performed by z‐scoring the tau‐PET images using the same reference group as for regional and voxel‐wise thresholding.

Associations between regional tau burden and MBI‐C scores were examined using Spearman's rank correlation coefficients. Prior to correlation analyses, tau‐PET SUVR values and MBI‐C scores were separately residualized for age, sex, and years of education using linear regression models. Spearman correlations were then computed between the resulting residuals for Braak I–II, III–IV, and V–VI SUVRs and MBI‐C total and domain scores. Correlation coefficients were compared using Steiger's test for dependent overlapping correlations implemented in the cocor R package (version 1.1.4). ⁴³

To assess whether the observed associations were independent of global Aβ burden, sensitivity analyses were performed with additional adjustment for centiloid values. Furthermore, analyses were repeated after stratifying participants into CU Aβ−, CU Aβ+, and MCI Aβ+ groups, adjusting for age, sex, and years of education.

MBI‐C total scores were compared across PET‐derived Braak stages assigned separately for [¹⁸F]MK6240 and [¹⁸F]Flortaucipir using Kruskal–Wallis tests, followed by pairwise Wilcoxon rank‐sum tests with Benjamini–Hochberg correction for multiple comparisons. To visualize domain‐specific symptom trajectories across tau stages, scores were plotted across PET‐derived Braak stages using locally estimated scatterplot smoothing (LOESS) with a smoothing span of 1.4 and 95% confidence intervals. These analyses were intended for visualization of overall trends rather than statistical inference.

To evaluate the association between PET‐derived Braak stage classifications and MBI burden, Spearman's rank correlations were calculated between ordinal Braak stage classifications (0 = 0, I–II = 1, III–IV = 2, and V–VI = 3) and MBI‐C total and domain scores for each tracer. Differences in correlation strength were assessed using Steiger's test for dependent correlations.

To evaluate the ability of tau‐PET measures to identify clinically relevant MBI, receiver operating characteristic (ROC) analyses were performed using MBI+ status as the outcome. Area under the curve (AUC) values were calculated for Braak I–II and temporal meta‐ROI measures (SEOT and SUVR) for both [¹⁸F]MK6240 and [¹⁸F]Flortaucipir. AUCs were compared using DeLong's test implemented in the pROC R package (version 1.19.0.1). ⁴⁴

Finally, we compared the frequency of MBI+ participants between tau‐positive and tau‐negative individuals for each tracer using Fisher's exact test. MBI‐C scores were subsequently compared across four groups defined by tracer positivity status: tau‐positive for both tracers, tau‐negative for both tracers, tau‐positive only for [¹⁸F]MK6240, and tau‐positive only for [¹⁸F]Flortaucipir. Group differences in MBI‐C scores were assessed using the Wilcoxon rank‐sum test.

3. RESULTS

3.1. Participant characteristics

We included 110 participants in the study (71 [64.5%] female; mean [SD] age 67.14 [7.26] years; Table 1). The sample comprised 52 CU Aβ‐, 27 CU Aβ+, and 31 MCI Aβ+ individuals. Groups differed in age, CDR‐SB scores, Centiloid values, MBI‐C total score, and percentage of APOE ε4 carriers (all p < 0.001), but not in sex distribution or years of education. CDR‐SB scores and percentage of APOE ε4 carriers were higher in MCI Aβ+ participants compared with both CU groups. Centiloid values differed across all three groups, increasing progressively from CU Aβ− to CU Aβ+ and MCI Aβ+ participants. MBI‐C total scores were significantly higher in MCI Aβ+ participants, whereas no differences were observed between CU Aβ− and CU Aβ+ individuals.

TABLE 1.

Clinical and demographic characteristics of the study population stratified by cognitive diagnosis and amyloid status.

Parameter	Overall (n = 110)	CU Aβ− (n = 52)	CU Aβ+ (n = 27)	MCI Aβ+ (n = 31)	p‐value
Age, years (mean ± SD)	67.14 ± 7.26	63.8 ± 7.1a	69.7 ± 6.6b	70.6 ± 5.5b	<0.001
Female, n(%)	71 (64.5%)	36 (69.2%)	19 (70.4%)	16 (51.6%)	0.205
Years of education (mean ± SD)	15.6 ± 3.3	15.2 ± 2.9	16.3 ± 3.3	15.7 ± 3.7	0.369
CDR‐SB (mean ± SD)	0.6 ± 1.1	0 ± 0.1a	0 ± 0a	1.9 ±  ± 1.3b	<0.001
Centiloid (mean ± SD)	46.62 ± 50.41	11.5 ± 5.4a	47.7 ± 34.6b	104.7 ± 50.2c	<0.001
MBI‐C (mean ± SD)	2.6 ± 4.1	1.3 ± 3.1a	1.3 ± 2.8a	5.9 ± 4.8b	<0.001
APOE ε4 carriers, n (%)	37 (33.6%)	11 (21.2%)a	7 (25.9%)a	19 (61.3%)b	<0.001

Note: Values are presented as mean ± SD or n (%). p‐values were calculated using Kruskal–Wallis tests for continuous variables and chi‐squared test for categorical variables. Superscript letters (a, b and c) indicate pairwise group comparisons: groups sharing the same letter are not significantly different, whereas groups with different letters differ significantly (adjusted p < 0.05). Pairwise comparisons were performed using Wilcoxon rank‐sum tests for continuous variables and Fisher's exact tests for categorical variables, with Benjamini–Hochberg correction for multiple comparisons.

Abbreviations: Aβ, amyloid‐β; CDR‐SB, Clinical Dementia Rating—Sum of Boxes; CU, cognitively unimpaired; MBI‐C, Mild Behavioral Impairment Checklist; MCI, mild cognitive impairment.

3.2. [18F]MK6240 demonstrates broader voxel‐wise associations between tau pathology and MBI severity

Voxel‐wise associations were primarily localized to bilateral medial temporal regions, with additional involvement of lateral temporal cortex, the temporoparietal junction and posterior temporo‐occipital areas (Figure 1A). Additional smaller clusters were observed in frontal association regions. [¹⁸F]MK6240 had a greater proportion of significant voxels than [¹⁸F]Flortaucipir (5.18% vs. 2.81% of voxels within the analysis mask, p < 0.001), while 1.75% of voxels were significant for both tracers (Figure 1B). Moreover, [¹⁸F]MK6240 demonstrated a broader distribution of standardized β values with a more pronounced rightward tail, indicating more widespread and stronger positive voxel‐wise associations with MBI‐C total score (Figure 1C). Consistent with these findings, 3.43% of voxels were uniquely associated with [¹⁸F]MK6240 compared with 1.06% for [¹⁸F]Flortaucipir, highlighting tracer‐specific spatial patterns despite substantial overlap between tracers.

FIGURE 1.

Comparison of voxel‐wise associations between MBI‐C total score and tau‐PET burden measured with [¹⁸F]Flortaucipir and [¹⁸F]MK6240. (A) Representative axial, sagittal, and coronal slices illustrating the spatial overlap of significant clusters. (B) Percentage of significant voxels identified in voxel‐wise analyses for the association between MBI‐C total score and tau‐PET burden for [¹⁸F]Flortaucipir, [¹⁸F]MK6240, and the spatial overlap between tracers. Significant voxels were considered as those surviving RFT correction with a threshold of p < 0.001. (C) Density plot of the distribution of voxel‐wise standardized‐β coefficients across the whole brain for each tracer. MBI‐C, Mild Behavioral Impairment Checklist; PET, positron emission tomography; RFT, Random Field Theory.

3.3. MBI symptom severity is associated with tau burden across Braak regions, with additional regional associations detected by [18F]MK6240

Associations between MBI‐C symptoms and tau‐PET SUVR across Braak ROIs were subsequently examined, adjusting for age, sex, and education (Figure 2). Higher MBI‐C total score, greater decreased motivation and greater affective dysregulation were significantly associated with higher tau load across all Braak ROIs for both [¹⁸F]MK6240 and [¹⁸F]Flortaucipir. Higher impulse dyscontrol was significantly associated with higher tau burden only in Braak I–II regions, whereas higher social inappropriateness showed significant associations with higher tau burden across all Braak ROIs for [¹⁸F]MK6240 but only in Braak I–II regions for [¹⁸F]Flortaucipir. The only association observed for abnormal thoughts and perception was that higher symptom scores were associated with higher [¹⁸F]MK6240 SUVR in Braak III–IV regions. Direct comparison of correlation coefficients revealed significantly stronger associations for [¹⁸F]MK6240 than [¹⁸F]Flortaucipir in Braak III–IV regions for social inappropriateness and abnormal thoughts and perception. Scatterplots illustrating the relationship between tau‐PET SUVR and MBI‐C scores for [^{1 8}F]MK6240 and [^{1 8}F]Flortaucipir across Braak regions are available in Figure S1.

FIGURE 2.

Associations between MBI‐C scores and tau‐PET uptake in Braak regions. Bar plots display Spearman correlation coefficients (ρ) between MBI‐C total and domain scores and tau‐PET SUVR values measured with [18F]MK6240 and [18F]Flortaucipir in Braak I–II (A), Braak III–IV (B), and Braak V–VI (C) regions. Correlations were adjusted for age, sex, and years of education. Asterisks within bars indicate significant correlations. Brackets indicate significant differences between tracers as assessed using Steiger's test. *p < 0.05, **p < 0.01, ***p < 0.001. MBI‐C, Mild Behavioral Impairment Checklist; PET, positron emission tomography; RFT, Random Field Theory.

After additional adjustment for global Aβ burden, most associations between tau‐PET SUVR and MBI‐C domains were no longer significant, with only the association between affective dysregulation and [¹⁸F]MK6240 uptake in Braak I–II and V–VI regions remaining significant (Figure S2). Thus, most regional tau‐MBI associations were not statistically independent of global Aβ burden in this cross‐sectional sample.

3.4. Tau‐PET explains MBI symptom severity within Aβ+ clinical groups

Associations were then examined within diagnostic subgroups (Figure 3). Stratified scatterplots illustrating the relationship between tau‐PET SUVR and MBI‐C across CU Aβ−, CU Aβ+, and MCI Aβ+ participants across Braak regions using [^{1 8}F]MK6240 and [^{1 8}F]Flortaucipir are accessible in Figures S3‐S5.

FIGURE 3.

Associations between MBI‐C symptoms and tau burden across Braak regions stratified by diagnostic groups. Bar plots display Spearman correlation coefficients (ρ) between MBI‐C total and domain scores and tau‐PET SUVR values measured with [18F]MK6240 and [18F]Flortaucipir in Braak I–II (A), Braak III–IV (B), and Braak V–VI (C) regions. Correlations were adjusted for age, sex, and years of education. Correlations are shown separately for CU Aβ+, CU Aβ−, and MCI Aβ+ participants. Asterisks indicate statistically significant correlations. *p < 0.05. CU, cognitively unimpaired; MBI‐C, Mild Behavioral Impairment Checklist; MCI, mild cognitive impairment; PET, positron emission tomography; SUVR, standardized uptake value ratio.

Significant associations were only present in Aβ+ subgroups. In the CU Aβ+ group, greater affective dysregulation was significantly associated with higher [¹⁸F]MK6240 uptake in Braak I–II regions. In the same subgroup, decreased motivation was significantly associated with higher uptake of both [¹⁸F]MK6240 and [¹⁸F]Flortaucipir in Braak V–VI regions, although only two out of 27 participants presented this symptom (Figure S5). In MCI Aβ+ individuals, greater affective dysregulation was significantly associated with higher [¹⁸F]MK6240 uptake in both Braak I–II and III–IV regions. Neither the MBI‐C total score nor other MBI‐C domains demonstrated significant associations with tau‐PET SUVR in the stratified analysis. 3.5 MBI burden increases across PET‐based Braak stages

Furthermore, MBI‐C total scores increased progressively with advancing PET‐Braak stages for both tracers (Figure 4A, B ). Participants classified as Braak III–IV and Braak V–VI exhibited significantly higher MBI‐C total scores than those at Braak stage 0, regardless of tracer. Analysis of individual MBI‐C domains across Braak stages revealed distinct trajectories for different behavioral symptoms (Figure 4C, D ). While impulse dyscontrol scores tended to fluctuate across stages, affective dysregulation exhibited the most pronounced increase with advancing Braak stage for both tracers. Decreased motivation showed a more gradual increase, whereas social inappropriateness and abnormal thoughts and perception remained comparatively low throughout the Braak continuum. Ordinal Braak stage classifications derived from both tracers were associated with greater MBI symptom burden across domains, with the exception of abnormal thoughts and perception scores for [¹⁸F]MK6240 (Figure 4E).

FIGURE 4.

MBI‐C symptom burden across PET‐based Braak stages. (A–E) Violin plots showing the distribution of MBI‐C total scores across [¹⁸F]MK6240‐derived (A) and [¹⁸F]Flortaucipir‐derived (B) Braak stages. Individual participants are represented by dots, and significance brackets indicate pairwise group comparisons. (C–D) LOESS curves illustrating the progression of MBI‐C domain scores across [¹⁸F]MK6240‐derived (C) and [¹⁸F]Flortaucipir‐derived (D) Braak stages. Shaded areas represent 95% confidence intervals. (E) Spearman correlations between ordinal PET‐derived Braak stage classifications and MBI‐C total and domain scores for [^{1 8}F]MK6240 and [^{1 8}F]Flortaucipir. Bars represent correlation coefficients (ρ), and asterisks indicate statistically significant correlations (p < 0.05, p < 0.01, *p < 0.001). LOESS, locally estimated scatterplot smoothing; MBI‐C, Mild Behavioral Impairment Checklist; PET, positron emission tomography.

3.5. Tau‐PET discriminates MBI across tracers, with an additional benefit of SEOT for [18F]MK6240

ROC analyses evaluated the ability of tau‐PET Braak regions to discriminate individuals with MBI (Figure 5). For Braak I–II regions (Figure 5A), SUVR and SEOT derived with both tracers demonstrated comparable discriminative performance, with AUC values ranging from 0.78 to 0.80. No significant differences were observed in AUCs. In contrast, for the temporal metaROI (Figure 5B), discriminative performance varied across measures. For [^{1 8}F]MK6240, the SEOT measure demonstrated significantly better discrimination of MBI status than the corresponding SUVR measure (AUC = 0.78 vs. 0.69, p < 0.05). No significant differences were observed between [^{1 8}F]Flortaucipir measures or across tracers.

FIGURE 5.

Discriminative performance of tau‐PET measures for identifying abnormal MBI‐C status. (A) ROC curves evaluating the ability of Braak I–II SUVR and SEOT measures to discriminate participants with abnormal MBI‐C scores using [¹⁸F]MK6240 and [¹⁸F]Flortaucipir. AUC values are reported for each measure. (B) ROC curves evaluating the ability of temporal metaROI SUVR and SEOT measures to discriminate participants with abnormal MBI‐C scores using [¹⁸F]MK6240 and [¹⁸F]Flortaucipir. AUC values are reported for each measure. *p < 0.05, **p < 0.01, ***p < 0.001. AUC, area under the ROC curve; MBI‐C, Mild Behavioral Impairment Checklist; PET, positron emission tomography; ROC, receiver operating characteristics curve; SEOT, spatial extent of tau pathology; SUVR, standardized uptake value ratio.

3.6. Tau‐PET positivity is associated with higher MBI frequency

Tau‐PET positivity defined with both tracers and index regions was associated with a higher frequency of MBI+ status (Figure 6A). In Braak I–II, MBI was present in 24% (11/46) of [¹⁸F]MK6240‐positive participants compared with 3% (2/64) of [¹⁸F]MK6240‐negative participants (p = 0.002). Similarly, 31% (9/29) of [¹⁸F]Flortaucipir‐positive participants were MBI+ compared with 5% (4/81) of [¹⁸F]Flortaucipir‐negative participants in Braak I–II (p < 0.001). Comparable findings were observed in the temporal meta‐ROI, where MBI+ frequency was higher among [¹⁸F]MK6240‐positive (29%, 6/21) than [¹⁸F]MK6240‐negative individuals (8%, 7/89; p < 0.05), and among [¹⁸F]Flortaucipir‐positive (30%, 6/20) than [¹⁸F]Flortaucipir‐negative individuals (8%, 7/90; p < 0.05).

FIGURE 6.

Tau‐PET positivity is associated with greater prevalence and severity of mild behavioral impairment symptoms. (A) Distribution of participants classified as MBI− and MBI+ according to tau‐PET positivity status for MK6240 and Flortaucipir in Braak I–II and temporal meta‐ROI regions. Bars represent the absolute number of participants, with labels indicating the number and percentage of MBI+ individuals within each tau‐PET group. p‐Values correspond to Fisher's exact tests comparing MBI prevalence across groups. (B) Mean MBI‐C domain and total scores across Braak I–II discordance groups: tau‐negative on both tracers (Neither), positive only on [¹⁸F]MK6240 ([¹⁸F]MK6240 only), and positive on both tracers (Both tracers). Error bars represent standard errors of the mean. (C) Pairwise Wilcoxon rank‐sum test p‐values comparing MBI‐C scores across Braak I–II discordance groups. MBI, mild behavioral impairment; MBI‐C, Mild Behavioral Impairment Checklist; PET, positron emission tomography; ROC, receiver operating characteristics curve.

To investigate whether tau‐PET positivity was associated with specific behavioral symptom profiles, we next examined tracer concordance and discordance patterns. In Braak I–II, 63 participants were negative on both tracers, 18 were positive only on [¹⁸F]MK6240, 1 was positive only on [¹⁸F]Flortaucipir, and 28 were positive on both tracers. Given the presence of a single [¹⁸F]Flortaucipir‐only participant, subsequent discordance analyses excluded this group. In contrast, discordant positivity was uncommon in the temporal meta‐ROI ([¹⁸F]MK6240‐only, n = 4; [¹⁸F]Flortaucipir‐only, n = 3), resulting in insufficient sample sizes for meaningful between‐group comparisons. Therefore, discordance analyses were limited to Braak I–II.

Within Braak I–II, participants positive on both tracers exhibited significantly higher scores for decreased motivation, affective dysregulation, impulse dyscontrol, social inappropriateness, abnormal thoughts and perception, as well as higher total MBI‐C scores, compared with participants negative on both tracers (all p < 0.05; Figure 6B, C ). In contrast, the [¹⁸F]MK6240‐only tau‐positive group exhibited higher affective dysregulation scores than those tau‐negative for both tracers (p < 0.05), whereas no differences were observed for the remaining domains. Compared with the [¹⁸F]MK6240‐only tau‐positive group, participants positive on both tracers exhibited higher decreased motivation scores and higher total MBI‐C scores (both p < 0.05), whereas no significant differences were observed for the remaining domains.

4. DISCUSSION

In this study, we compared the association of two widely used tau‐PET tracers, the United States Food and Drug Administration (FDA) ‐approved [¹⁸F]Flortaucipir and the second‐generation [¹⁸F]MK6240, with MBI burden in preclinical and prodromal AD. Tau‐PET signal measured with both tracers was associated with MBI burden, supporting a link between tau accumulation and early behavioral manifestations of AD. Both tracers showed comparable accuracy in identifying MBI+ individuals. However, [¹⁸F]MK6240 demonstrated more extensive voxel‐wise associations with MBI severity and intra‐clinical stage associations with symptoms not observed for [¹⁸F]Flortaucipir, suggesting that second‐generation tau‐PET tracers may provide greater sensitivity to subtle behavioral changes occurring early in AD.

Our findings corroborate previous studies linking tau‐PET signal to MBI using both [¹⁸F]MK6240 and [¹⁸F]Flortaucipir. ¹⁰ , ¹¹ Consistent with earlier reports, MBI severity increased across PET‐Braak stages, ¹¹ indicating that behavioral symptoms worsen alongside the cortical spread of tau‐PET signal and increasing biological disease severity, even in CU individuals. This reinforces tau‐PET staging as a marker of disease progression and MBI as a potential manifestation of transitional decline within the AD continuum, ²¹ which refers to the earliest clinically detectable manifestations in CU individuals. Supporting this, Johansson et al. ⁹ reported associations between MBI‐C scores and Braak I–II tau burden measured with [¹⁸F]RO948 in preclinical AD. ⁹ Here, Braak I–II [¹⁸F]MK6240 SUVR, but not [¹⁸F]Flortaucipir SUVR, was associated with mood symptoms at this clinical stage. Consistent with this finding, individuals classified as tau‐positive only by [¹⁸F]MK6240 exhibited higher affective dysregulation scores than those classified as tau‐negative by both tracers. Therefore, our findings extend this framework by suggesting that [¹⁸F]MK6240 may be more sensitive to early tau accumulation concurrent with subtle behavioral alterations in AD and may detect an intermediate clinical stage in which behavioral symptoms are present prior to concordant tau positivity across tracers. [¹⁸F]MK6240 likely has higher affinity for neurofibrillary tau than [¹⁸F]Flortaucipir. ¹³

Interestingly, Johansson et al. ⁹ found associations between Braak I–II tau burden and the overall MBI‐C score, although these relationships were largely driven by the impulse dyscontrol and mood domains, the latter also emerging in our preclinical AD subset. Therefore, our findings partially replicate and extend these observations, suggesting that the relationship between tau‐PET signal and MBI may be stronger for specific behavioral domains. We also found that uptake of both tracers in Braak V–VI regions was associated with decreased motivation in CU Aβ+ individuals. However, only two participants exhibited this symptom, and this finding should therefore be interpreted with caution. Nevertheless, the consistency of the association across both tracers is compatible with potential regional specificity across behavioral domains, although the small number of symptomatic individuals precludes firm conclusions. More broadly, distinct spatial patterns of tau deposition may contribute to different MBI symptoms, analogous to the proposed contribution of tau topography to domain‐specific cognitive profiles. ⁴⁵ , ⁴⁶ Similar regional heterogeneity has been proposed for overt neuropsychiatric symptoms in AD ⁴⁷ and may reflect the involvement of distinct neural networks underlying different mild behavioral manifestations. ⁴⁸ , ⁴⁹ This potential topographic heterogeneity may explain why previous studies failed to detect significant associations between tau‐PET and MBI when MBI was operationalized as a unified clinical construct or when analyses were restricted to anatomically limited ROIs. ¹² , ⁵⁰ , ⁵¹ Aggregating domains into a single score may dilute associations because different behavioral manifestations are likely supported by distinct neurobiological substrates, an issue that may be further exacerbated when MBI is operationalized without considering symptom persistence, ⁵¹ an important feature for MBI case ascertainment. ⁵²

Another explanation for heterogeneous tau‐PET–MBI associations relates to the clinical‐biological composition of the study population. In our study, tau‐PET was associated with MBI symptoms among Aβ+ individuals. However, most tau–MBI associations were attenuated after adjustment for global Aβ burden, indicating that these relationships were largely not independent of Aβ pathology. This attenuation may reflect the close biological coupling between Aβ and tau in early AD, when increasing Aβ burden is closely associated with the emergence and spread of tau pathology, ⁵³ making their respective associations with early behavioral manifestations difficult to disentangle cross‐sectionally. Nevertheless, associations between affective dysregulation and [^{1 8}F]MK6240 uptake in Braak I–II and V–VI remained significant after Aβ adjustment, suggesting that some specific tau–MBI relationships may be partly independent of global Aβ burden. Limited statistical power, particularly for low‐frequency MBI domains, may also have contributed to the loss of statistical significance for some correlations that retained a similar direction and magnitude after Aβ adjustment. This interpretation is consistent with our inclusion criteria, which were restricted to the healthy aging‐AD continuum. By contrast, MCI is a clinically heterogeneous syndrome encompassing multiple pathological etiologies, several of which may contribute to NPS independently of AD pathology. ⁵⁴ , ⁵⁵ , ⁵⁶ Moreover, MBI is frequently observed among individuals with MCI regardless of underlying etiology, with a pooled prevalence of 52% reported in a recent meta‐analysis. ⁵⁷ Consequently, inclusion of Aβ− individuals with MCI may weaken tau‐PET–MBI associations by introducing MBI symptoms driven by non‐AD pathological processes, whereas both [^{1 8}F]Flortaucipir and [^{1 8}F]MK6240 primarily index AD‐type tau pathology. ⁵⁸ , ⁵⁹ , ⁶⁰ This framework may help explain why previous studies including mixed MCI Aβ+ and Aβ− samples did not detect significant associations between tau‐PET and MBI. ⁵⁰ , ⁵¹ Together, these findings suggest that the detectability of tau‐PET–MBI associations depends partly on the biological composition of the study population and on the extent to which non‐AD pathologies contribute to MBI symptoms, an important question for future research.

Notably, mood symptoms showed a relatively consistent pattern of association with advancing PET‐Braak involvement, especially with [¹⁸F]MK6240. Importantly, these associations were observed both within and across clinical‐biological stages. While Braak I–II [¹⁸F]MK6240 SUVR was associated with mood symptoms in preclinical AD, [¹⁸F]MK6240 tracked affective symptoms in both Braak I–II and Braak III–IV regions among individuals with prodromal AD. This within‐stage consistency contrasts with other MBI domains, for which associations observed in the overall sample were not consistently reproduced within clinical–biological groups and occasionally differed in direction. Such differences are not necessarily contradictory, as pooled associations may partly reflect between‐group differences in tau and MBI burden, whereas stratified analyses capture relationships within more restricted ranges of disease severity. Moreover, affective dysregulation increased more consistently across PET‐Braak stages with [¹⁸F]MK6240 than with [¹⁸F]Flortaucipir, suggesting that the tracer may better capture affective symptom correlates of progressive tau spread across successive Braak stages. Together, these findings suggest that affective symptoms may reflect differences in tau burden both among individuals at similar clinical stages and across advancing stages of AD. Furthermore, the involvement of Braak III–IV regions is particularly noteworthy, as these tau‐PET stages have been linked to the emergence of broader clinical manifestations and more substantial cognitive decline, consistent with the transition to MCI. ⁶¹ Indeed, MBI itself has been linked to current and emergent cognitive impairment. ⁶² , ⁶³ Thus, MBI‐related mood symptoms may provide information about underlying AD pathological severity beyond cognitive status alone, potentially helping identify individuals with greater tau burden despite similar levels of cognitive impairment.

Beyond regional tau burden, we evaluated whether the spatial distribution of tau‐PET signal provided additional information regarding the behavioral manifestations of AD. We found that tau‐PET index measures showed good accuracy for identifying MBI+, with both tracers demonstrating similar performance when using Braak I–II and temporal meta‐ROI measures. Despite broader associations with [^{1 8}F]MK6240, this did not translate into superior cross‐sectional discrimination of MBI, possibly because greater sensitivity to early tau deposition may capture pathology before clinically detectable behavioral manifestations emerge. ⁶⁴ Longitudinal studies are needed to determine whether this broader sensitivity translates into improved prediction of subsequent behavioral or clinical progression. Moreover, adapting the Braak I–II ROI to minimize [^{1 8}F]Flortaucipir off‐target binding may have reduced tracer‐specific differences in this region. Nonetheless, SEOT improved the predictive performance of [¹⁸F]MK6240 compared with conventional SUVR measures in the temporal meta‐ROI, consistent with previous studies demonstrating stronger associations of SEOT with AD clinical and biological severity. ³⁶ , ³⁷ , ³⁸ , ⁶⁵ Notably, SEOT has also been shown to correlate more strongly than tau load with cognitive impairment in early AD, ³⁸ , ⁶⁶ potentially because it is more sensitive to the early spatial propagation of tau‐PET signal across vulnerable cortical regions. ³⁸ Our findings indicate that extent‐based tau‐PET metrics may improve the identification of early behavioral manifestations of AD.

A strength of our study is its head‐to‐head design, which enables direct comparison of tracer‐specific associations between tau pathology and MBI within the same individuals, thereby minimizing between‐cohort variability. This study also has several limitations. First, its cross‐sectional design precludes conclusions regarding causality or the temporal relationship between tau accumulation and the emergence of MBI symptoms. Second, subgroup analyses by disease stage resulted in relatively modest sample sizes, which may have limited statistical power to detect additional associations, particularly for infrequent MBI domains. For example, although abnormal thoughts and perception symptoms have been associated with a high hazard of incident dementia, ⁶⁷ their low frequency among CU individuals may reduce statistical power to detect significant associations with tau‐PET. ⁵ Moreover, significant associations involving low‐frequency symptoms may also be disproportionately influenced by a small number of symptomatic individuals. For instance, the association between Braak V–VI uptake and decreased motivation in CU Aβ+ individuals was based on only two participants exhibiting this symptom and should therefore be considered exploratory. Replication in larger samples is needed to establish the robustness of these stage‐ and domain‐specific findings. However, the current sample represents a unique dataset combining MBI assessments with large‐scale head‐to‐head tau‐PET imaging and, to our knowledge, no comparable cohort is currently available for direct external replication. Third, MBI symptoms were assessed using informant‐reported ratings and, therefore, subject to reporting bias, despite the value of informant ratings when insight is limited. ⁶⁸ , ⁶⁹ Moreover, our cohort comprised predominantly highly educated North American participants, limiting generalizability. In addition, we did not assess the presence of co‐pathologies, which may contribute to behavioral symptoms. ⁵⁴ , ⁵⁵ , ⁵⁶ Future studies incorporating biomarkers of common co‐pathologies may clarify their potential impact on these relationships.

In conclusion, our findings strengthen the evidence linking tau pathology to MBI and demonstrate that this relationship is influenced by tracer choice, clinical profile, behavioral domain, region of interest, and quantification method. Although both [¹⁸F]MK6240 and [¹⁸F]Flortaucipir captured the association between tau pathology and MBI, the second‐generation tracer consistently provided greater sensitivity to subtle behavioral correlates of early tau accumulation, particularly within clinically homogeneous stages. These results support the use of second‐generation tau‐PET tracers in studies of early neuropsychiatric manifestations of AD, reinforce MBI as a clinically meaningful behavioral marker of early AD, and highlight the potential value of combining sensitive behavioral assessments with next‐generation tau‐PET measures identifying individuals in the earliest symptomatic phases of the disease.

CONFLICT OF INTEREST STATEMENT

J.T. has served as a consultant for the Neurotorium educational platform, outside of the scope of the submitted work. E.R. Zimmer has served on the scientific advisory board, as a consultant or speaker for Nintx, Novo Nordisk, Biogen, Lilly, Magdalena Biosciences and masima. He is also a co‐founder and minority shareholder of masima. T.C. has served on the scientific advisory board for Eisai outside the scope of this manuscript. V.J.L. reports grant support to his institution from Siemens Medical Imaging and Avid Radiopharmaceuticals; honoraria from PeerView Institute for Medical Education; research materials or services provided to his institution by Avid Radiopharmaceuticals; and grant support to his institution from the National Institutes of Health. D.S.M. receives research support from the NIH. He serves on a data monitoring committee for Eli Lilly and Company and performs grant review for the Alzheimer's Drug Discovery Foundation. He has received non‐financial support from the Society for Nuclear Medicine and Molecular Imaging and Alzheimer's Association outside the submitted work. The other authors declare that they have no competing interests. T.A.P. is a member of the CIHR‐CCNA Canadian Consortium of Neurodegeneration in Aging. Colin J. Adair Charitable Foundation. P.R.N. provides consultancy services for Roche, Cerveau Radiopharmaceuticals, Lilly, Eisai, Pfizer, and Novo Nordisk. P.R.N. also serves as a clinical trials investigator for Biogen, Novo Nordisk, and Biogen. Author disclosures are available in the Supporting Information.

CONSENT STATEMENT

Written informed consent was obtained from all participants prior to enrollment in the study. The study was conducted in accordance with the ethical standards of the Declaration of Helsinki and was approved by the Montreal Neurological Institute PET Working Committee and the Douglas Mental Health University Institute Research Ethics Board (IUSMD 21‐39).

Supporting information

Supporting Information

Supporting Information

Bizzi I, Macedo AC, Trudel L, et al. Tau‐PET correlates of mild behavioral impairment in preclinical and prodromal Alzheimer's disease: A head‐to‐head comparison of Flortaucipir and MK6240. Alzheimer's Dement. 2026;22:e71861. 10.1002/alz.71861