Voxel-based correlation of ¹⁸F-THK5351 accumulation and gray matter volume in the brain of cognitively normal older adults

Neuropathological studies have revealed that neurofibrillary tangles (NFTs) mainly accumulate in the medial temporal lobe (MTL) with human aging. Although neurodegeneration is also a feature of aging, in addition to tau pathology, only a few imaging studies have specifically investigated correlations between tau accumulation and brain volume [1, 2]. Although ¹⁸F-THK5351 was originally developed as a tau-specific tracer, recent studies have clarified the off-target binding to monoamine oxidase B (MAO-B) [3]. Because ¹⁸F-THK5351 accumulation reflects the combination of astrogliosis, in addition to tau pathology, it is now considered the promising biomarker for detecting neuroinflammation [4]. Previously, using region of interest (ROI) analysis, we did not detect significant negative correlations between ¹⁸F-THK5351 accumulation and gray matter (GM) volume [5]. In this study, we aimed to re-examine the correlations between ¹⁸F-THK5351 accumulation and GM volume at the voxel level.

The participants’ demographics are shown in Table 1. Mean age ± standard deviation was 65.0 ± 7.9 years and mean cognitive scores were 29.3 ± 1.1 for MMSE and 13.4 ± 2.9 for WMSR LM-II.

Localized ¹⁸F-THK5351 accumulation was detected mainly in the basal ganglia, thalamus, MTL but slightly extended into the inferior temporal lobe, insula, posterior cingulate cortex/precuneus, and basal frontal lobe (Fig. 1). We found a significant negative correlation between ¹⁸F-THK5351 accumulation and GM volume in the bilateral MTL, right parahippocampal gyrus (cluster size, 40 voxels; Z-score, 4.68; MNI coordinate [x, y, z], [32, − 18, − 18]), right hippocampus, left parahippocampal gyrus (cluster size, 56 voxels; Z-score, 4.68; MNI coordinate [x, y, z], [− 26, − 8, − 22]), and left amygdala (Fig. 2).

This is the first study to investigate direct correlations between ¹⁸F-THK5351 accumulation and GM volume at the voxel level using BPM in cognitively normal older adults. We detected significant correlations between an increased ¹⁸F-THK5351 accumulation and reduced GM volume in the MTL. Our findings may contribute to a better understanding of the pathophysiology of human aging.

We found significant ¹⁸F-THK5351 accumulation mainly in the basal ganglia, thalamus, MTL slightly extending into the inferior temporal lobe, insula, posterior cingulate cortex/precuneus, and basal frontal lobe. Our finding corresponded to Braak stage III–IV, which is not consistent with previous neuropathological and tau PET studies that tau pathology was usually localized in the MTL in the cognitively healthy participants [8, 9]. However, a recent large cohort study [10] showed elevated ¹⁸F-THK5351 tau tracer retentions in Braak stage III–IV areas with normal amyloid status and raised the possibility of primary age-related tauopathy (PART) [11]. Our results also support their results and might reflect PART. PART is defined as NFT pathology mostly restricted to the MTL, basal forebrain, brainstem, and olfactory areas in the absence of β-amyloid in the aged brain. Cognitive function is usually normal to mildly impaired in PART, and severe cognitive decline is rarely seen. Although PART is suggested to be an age-related phenomenon distinct from early Alzheimer’s disease, its pathophysiology is still unclear [1]. Thus, further studies with tau PET are needed to better understand the pathogenesis of PART.

The BPM analysis demonstrated a significant voxel-wise negative correlation between ¹⁸F-THK5351 accumulation and GM volume in the MTL, which was not detected in the previous ROI-based analysis [5]. Although ROI analysis is a common approach, it may not be able to accurately assess localized accumulation due to dilution effects. Because the BPM toolbox enables a direct comparison across imaging modalities at the voxel level, this voxel-based analysis may be more reliable than traditional ROI-based analyses. Our findings are consistent with neuropathological studies of PART patients showing that a higher Braak NFT stage is associated with hippocampal head atrophy [1]. Similar findings have been reported in recent ¹⁸F-AV1451 tau PET studies, namely, that higher MTL tau is associated with MTL atrophy [2]. Although they managed to obtain these results using a predefined FreeSurfer ROI approach, it is possible that nonspecific ¹⁸F-AV1451 accumulation in the choroid plexus adjacent to the MTL may not have been eliminated.

This study has several limitations. First, the number of participants was relatively small. Second, the study lacked pathologic confirmation of tau pathology. Third, the high affinity of ¹⁸F-THK5351 to MAO-B [3, 4] may contribute to the relatively higher Braak stage of our findings compared to PART type pathology which is usually III or lower [11]. MAO-B concentration increases during ongoing astrogliosis, which is considered as neuroinflammation changes that occur in response to brain injury and neurodegenerative disease [12]. Previous PET study in healthy subjects reported global MAO-B increases in the whole brain even with human aging [13], and the astrocytes seem to contribute to low-grade inflammation in the aged brain [14]. Because ¹⁸F-THK5351 uptake reflects the combination of astrogliosis and tau pathology, the degree and extent of tracer retention could be higher than that of only tau pathology. Recently, it is reported that aggregation of misfolded proteins including TDP-43 and α-synuclein in addition to Aβ and NFTs observed in PART are common even in cognitively healthy elderly brain [15]. Since the mixed pathologies are frequently observed in the aged brain, they could evoke neuroinflammation and increase astrogliosis, resulting in accumulation of ¹⁸F-THK5351. The early phase of age-related TDP-43 accumulation, known as “limbic-predominant age-related TDP-43 encephalopathy (LATE),” which tends to extend to limbic areas including the amygdala, could be part of the reasons [16]. In addition, the stereological cell counting studies showed declining of neocortical neuronal populations but no changes of the total astrocyte numbers in the aged human brains [17]. Therefore, the numbers of astrocytes would tend to concentrate in the atrophied MTL regions, suggesting the contribution of MAO-B in addition to tau pathology.

In summary, we found significant voxel-wise negative correlations between ¹⁸F-THK5351 accumulation and GM volume in the MTL. These results may reflect the concept of PART and contribute to a better understanding of the neurobiology of aging. Further studies are needed to confirm whether our findings reflect PART pathology or not by taking an oral dose of MAO-B inhibitor selegiline [3] or using second-generation tau-specific tracers with much less off-target binding.