Assessing THK523 selectivity for tau deposits in Alzheimer’s disease and non–Alzheimer’s disease tauopathies

Five months before death, a seventy-nine-year-old patient diagnosed with PSP underwent an Aβ imaging PET scan with ¹⁸F-florbetaben and a tau imaging scan with ¹⁸F-THK523. Approval of the study was granted by the Austin Health Human Research Ethics Committee, and written informed consent was obtained from all participants and caregivers before the study. The patient was recruited, reviewed and diagnosed on the basis of clinical and neuropsychological assessment by consensus of a neurologist and a neuropsychologist.

As part of the imaging protocol, we performed magnetic resonance imaging (MRI) using a three-dimensional magnetization-prepared rapid acquisition gradient echo sequence and T2-weighted fast spin echo and fluid-attenuated inversion recovery sequences. Both ¹⁸F-florbetaben and ¹⁸F-THK523 were synthesized at the Centre for PET, Austin Health, as previously described [37‐39]. PET scans were acquired using a Philips Allegro PET scanner (Philips Healthcare, North Ryde, Australia) at the Austin Health Centre. A transmission scan using a rotating Cs-137 source was taken for attenuation correction immediately prior to obtaining the emission scan. A 60-minute list-mode emission acquisition, followed by a 90- to 120-minute acquisition using 10-minute frames, was performed in three-dimensional mode after injection of 300 MBq of ¹⁸F-florbetaben. A 90-minute list-mode emission image acquisition was performed in three-dimensional mode after injection of 200 MBq of ¹⁸F-THK523. Images were reconstructed using a three-dimensional row action maximum likelihood algorithm.

PET images were processed using a previously described semiautomatic region of interest (ROI) method [40]. Briefly, coregistration of the patient’s MRI scans with the PET images was performed with Statistical Parametric Mapping 8 (SPM8) software [41]. A narrow cortical ROI template was placed on the coregistered MRI scanner by an operator (VLV) who was blinded to the participant’s clinical status, then it was transferred to the coregistered PET images. The ROI template covered cortical and subcortical GM structures as well as the midbrain and pons. Subcortical white-matter ROIs were placed at the centrum semiovale, and the cerebellar regions were placed over the cerebellar cortex, taking care to avoid white matter. Standardised uptake values (SUVs), defined as the decay-corrected brain radioactivity concentration normalized for injected dose and body weight, were calculated for all regions. In order to avoid arterial blood sampling, a simplified approach was applied using the cerebellar cortex as the reference region. SUVs were used to derive SUV ratios (SUVRs) referenced to the cerebellar cortex soon after the ratio of binding in neocortex to that in the cerebellar cortex reached an apparent steady state. Regional THK523 SUVRs were obtained for all regions sampled. Global tau burden was expressed as the average THK523 SUVR for the following cortical ROIs: frontal (consisting of the dorsolateral prefrontal, ventrolateral prefrontal and orbitofrontal regions), superior parietal, lateral temporal, lateral occipital, and anterior and posterior cingulate. Partial volume correction accounting for both GM atrophy and white-matter spillover was performed using a three-compartment approach with PMOD version 3.1 software (PMOD Technologies, Zurich, Switzerland). To establish whether either ¹⁸F-florbetaben or ¹⁸F-THK523 retention in the PSP patient was different from age-matched controls, a Z-score was generated for both global and regional retention. The respective Z-scores were generated against ten healthy controls who had ¹⁸F-florbetaben studies and ten healthy controls who had ¹⁸F-THK523 studies. Conservative Z-scores greater than 1.5, indicating just 1.5 standard deviations (SDs) from the mean of the control participants, were considered abnormal.

The demographics of the patients whose postmortem human brain tissue was utilized for these studies, expressed as mean ± SD, are presented in Table 1. All participants assessed were of similar age with comparable postmortem intervals for tissue collection. The patient with PSP who underwent PET was 79 years old and had 11 years of formal education. A neuropsychological examination revealed the patient had a Mini Mental State Examination score of 26, a Clinical Dementia Rating Scale score of 1 a Clinical Dementia Rating Scale–Sum of Boxes score of 5.5, an episodic memory composite score of -3.22 and a nonmemory composite score of -3.40. The PSP participant died 5 months after the PET scans were taken.

The results of our previous postmortem studies [29] have indicated that THK523 labels AD tau lesions, namely, NFTs in the hippocampus of patients with AD (Figure 1). To determine whether THK523 would also bind to non-AD tau lesions, brain sections from non-AD tauopathies were evaluated. For these studies, fixed contiguous serial sections from the striatum (CBD and PSP samples), the frontal cortex (PiD sample) and the pons (PSP sample) were either immunostained with a polyclonal tau antibody for the detection of tau lesions or incubated with the fluorescent compound THK523 to determine whether THK523 bound to non-AD tauopathy aggregates. Immunohistochemical staining of brain tissue regions rich in tau immunoreactivity was detected by light microscopy, and the same tissue region within the adjacent serial section was assessed by fluorescence microscopy to compare and determine whether the immunoreactive tau lesion colocalised with THK523 binding, indicated by a fluorescent signal. Immunohistological assessment of brain sections from CBD and PiD patients revealed the characteristic presence of globose tangles, coiled bodies (as indicated by the arrowheads in the figures; see top left panel of Figure 2) and Pick’s bodies (arrowheads in bottom left panel of Figure 2) in the striatum and frontal cortex. Nonetheless, examination of the same region within the adjacent serial section exhibited no signs of THK523 fluorescence, indicating that THK523 does not bind to these tau lesions. Likewise, immunohistological evaluation of the pons (left, top panel of Figure 3) and striatum (left, bottom panel of Figure 2) in PSP patients revealed that, despite the presence of tau globose tangles, there was no detectable THK523 fluorescence signal in the same region of the adjacent serial brain section, again suggesting that THK523 does not bind to globose tangles. It is noteworthy that one of the three PSP patients evaluated (Figure 3) had had ¹⁸F-THK523 and ¹⁸F-florbetaben PET scans 5 months prior to death. There was low ¹⁸F-florbetaben cortical retention (Figure 4), correlating with postmortem results showing absence of Aβ deposits. There was also low cortical ¹⁸F-THK523 retention as well as low ¹⁸F-THK523 retention in the basal ganglia, midbrain, pons and cerebellar white matter (Figure 4), which was indistinguishable from age-matched controls and in contrast to the relatively high density of tau lesions observed in the brain, confirming the absence of THK523 binding to globose tangles. Analysis of the PSP patient PET scans showed global and regional Z-scores less than 1.0 for ¹⁸F-THK523 and ¹⁸F-florbetaben, confirming the visual inspection of the images.

To further test the selectivity of THK523, we evaluated its ability to bind to Lewy bodies composed of α-synuclein and ubiquitin aggregates sharing a similar β-sheet secondary structure. For these studies, serial sections of the substantia nigra from a PD patient were either immunostained with antibodies raised to α-synuclein, or treated with a fluorescent compound, THK523. Evaluation of these stained serial sections demonstrated that, whilst the presence of Lewy bodies could be clearly identified by immunohistochemistry (Figure 5, left panel), the adjacent serial section was devoid of THK523 fluorescence (Figure 5, right panel), implying that THK523 did not bind to Lewy bodies.