Relationship Between ¹⁸F-Flortaucipir Uptake and Histologic Lesion Types in 4-Repeat Tauopathies

Participant Cohort

All participants were prospectively recruited by the Neurodegenerative Research Group at Mayo Clinic, Minnesota, as part of a National Institutes of Health–funded grant and met 2 inclusion criteria: completion of antemortem volumetric head MRI, ¹⁸F-flortaucipir PET, and Pittsburgh compound B PET; and a brain autopsy with a pathologic diagnosis of PSP or CBD between April 7, 2016, and April 6, 2021.

The study was approved by the Mayo Clinic Institutional Review Board, and all patients gave written informed consent to participate in the study.

Neuropathologic Evaluation and Diagnosis

All patients had a standardized neuropathologic evaluation in accordance with accepted published methodologies (9), including immunohistochemistry to detect 4R tau immunoreactive lesions using an antibody to phospho-tau (CP13; 1:1,000; IgG1 to phospho-serine 202, gift from the late Peter Davies, Feinstein Institute). PSP and CBD pathologic diagnoses were rendered according to published criteria (1,2). PSP was diagnosed if there were globose neurofibrillary tangles, tufted astrocytes, and oligodendroglial coiled bodies in cardinal nuclei (2), whereas CBD was diagnosed if there were balloon neurons, as well as tau-specific lesions of pretangles, astrocytic plaques, and threads in cardinal nuclei (1). Additional pathologic findings were documented as previously described (8).

Semiquantitative 4R Tau Lesion Count

We assessed the burden (lesion count) of tau-specific lesions with anti-CP13 antibodies and immunohistochemistry. Lesions were counted by a single neuropathologist in the cerebellar dentate, midbrain tegmentum, substantia nigra, red nucleus, subthalamic nucleus, ventral thalamus, globus pallidus, striatum, superior frontal cortex, and precentral (motor) cortex. For each of these 10 regions, we determined lesion count for pretangles/neurofibrillary tangles, coiled bodies, tufted astrocytes (PSP only), and astrocytic plaques (CBD only) (Fig. 1) scored using a semiquantitative 4-point scale as follows: 0, absent lesions; 1+, few lesions (1–3 per ×200 field); 2+, moderate number of lesions (4–6 per ×200 field); and 3+, frequent number of lesions (7 or more per ×200 field). The density of neuropil threads was also graded on the same 0–3 scale (Fig. 1). This semiquantitative scale was published in 2006, in an article that displays specific exemplars or representative images for each score, for 3 of the lesions analyzed in this study (10).

• Download figure
• Open in new tab
• Download powerpoint

FIGURE 1.

Different types of 4R tau lesions in PSP and CBD. Bar at lower right of each image indicates magnification level.

PET Analysis

All participants underwent a standardized MRI protocol at 3 T that included a magnetization-prepared rapid gradient-echo (MPRAGE) sequence. ¹⁸F-flortaucipir and Pittsburgh compound B PET scans were acquired using a PET/CT scanner (Discovery MR750; GE Healthcare) (11). All MPRAGE scans were normalized to the Mayo Clinic Adult Lifespan Template and segmented via unified segmentation with Mayo Clinic Adult Lifespan Template priors or settings. ¹⁸F-flortaucipir images were coregistered to the MPRAGE using 6°-of-freedom registration in SPM12. Region-level data were calculated for the MPRAGE-space ¹⁸F-flortaucipir PET to allow for correlations with histopathology. Our analysis was performed on the hemisphere to match the hemisphere sampled at autopsy. To assess for correlations between ¹⁸F-flortaucipir uptake and histology lesion counts, we selected regions of interest that matched the regions assessed at autopsy. The Mayo Clinic Adult Lifespan Template atlas was used to assess uptake in the superior frontal, motor cortex, thalamus, globus pallidus, and striatum; the Deep Brain Stimulation Intrinsic Template (12) atlas for uptake in the substantia nigra, subthalamic nucleus, and red nucleus; and an in-house–developed atlas for the midbrain and cerebellar dentate, as previously described (5). All regional uptake values were divided by median uptake in the cerebellar crus gray matter to calculate SUV ratio (SUVR). A 2-compartment partial-volume correction was performed (13). The analyses were also repeated using PET data without partial-volume correction. Regional volumes were also calculated from the MPRAGE for the same 10 regions to allow for the assessment of potential confounding effects due to atrophy. Global Pittsburgh compound B SUVRs were calculated as previously described (14) and converted to centiloid units as follows: centiloid = 100 × [(−0.1620 + 0.9467 × global Pittsburgh compound B SUVR) − 1.009]/1.067 (15).

Statistical Analysis

For each region, we fit a linear regression model with log-transformed tau PET SUVR as the response and semiquantitative tau lesion burden scores as the predictors of interest. We incorporated the lesion count scores in several ways to have a more complete understanding of their effects. In our first approach, we included the scores for the 4 lesion types as 4 separate predictors in the model. In the second approach, we included the sum of the scores of the preneurofibrillary or neurofibrillary tangles plus tau threads as a first predictor and the sum of the scores of coiled bodies and astrocytic lesions as a second predictor. In the last approach, we summed all 4 semiquantitative scores to form a single numeric composite. We used the natural logarithm as the response in these models to reduce skew, maintain the regression assumption of constant variance, and allow an interpretation of the effect of tau burden in terms of percentage difference in PET uptake. All models included time from PET scan to death, age at death, and total intracranial volume as covariates.