Further evaluation of ¹⁸F-MK-6240 reference region kinetics

Introduction: ¹⁸F-MK-6240 is a second-generation positron emission tomography (PET) tau imaging agent that has demonstrated higher signal-to-noise, lower off-target binding, and in-vivo kinetics that are more favourable than previous tau PET radioligands. Reference tissue methods have been used to quantify ¹⁸F-MK-6240 binding, using cerebellar gray matter (CerGM) based reference regions^1-4. However, contamination of CerGM, by extra-axial signal, remains a challenge for the detection of emergent specific tau signal. In this work, we further explored alternative reference regions less prone to extra-axial contamination: eroded CerGM and pons^1,2,3.

Methods: Dynamic 120 min ¹⁸F-MK-6240 (180±10 MBq, GE Discovery-MI PET/CT) PET imaging was performed in 19 cognitively unimpaired control participants (CN, age 70±10 years) and 5 individuals diagnosed with mild cognitive impairment (4 MCI) and Alzheimer’s disease (age 64±6 years). Dynamic PET data were reconstructed with the ordered-subset expectation-maximization algorithm (5 iterations, 16 subsets) including time-of-flight information and point spread function modeling. Reconstructed PET data were motion corrected and coregistered to the corresponding T1-weighted magnetic resonance images with FSL Flirt. Time activity curves were generated using Freesurfer segmentations in 4 reference regions (CerGM, CerGM with 2mm and 3mm erosion from the outer edge to reduce contamination from adjacent high uptake areas, and pons) and 7 target regions (entorhinal, parahippocampal, amygdala, hippocampus, precuneus, inferior temporal and lateral occipital). Standardized uptake value tissue ratios (SUVR) were computed using 90-110 min of dynamic data. Distribution volume ratios (DVR) were computed by multilinear reference tissue modeling (MRTM2, t^*=30min, k₂’=0.04). Analysis of variance (ANOVA, p<0.05) and coefficients-of-variation were used to compare the mean and variability of SUVR and DVR values computed using different reference regions. Correlations between SUVR and DVR values were evaluated using linear regression models. Bonferroni correction for multiple comparisons was applied.

Results: SUVR_CerGM and SUVR_erodedCerGM curves exhibited most consistent plateaus across target regions after 90 min. SUVR_Pons curves were still increasing at 120 min in the target tau-binding regions-of-interest. The SUVR_Pons and DVR_Pons values were approximately 2-fold greater than SUVR_CerGM and DVR_CerGM (p < 10^-6). Coefficients-of-variation of SUVR_Pons (0.21±0.05) and DVR_Pons (0.21±0.06) were also greater (not significant - n.s.) than SUVR_CerGM(0.20±0.1) and DVR_CerGM (0.14±0.04)across all target regions. Mean SUVR_erodedCerGM and DVR_erodedCerGM values were slightly greater (n.s.) than mean SUVR_CerGM (16%) and DVR_CerGM (5%) values. The SUVR and DVR values obtained with different reference regions were highly correlated in all target regions (R² > 0.96, p < 10^-4), with positive bias of 1.34±0.13 in SUVR_CerGM and 1.34±0.18 in SUVR_Pons values.

Conclusions: Eroded CerCM reference regions performed similarly to CerGM, while potentially reducing contamination from extra-axial and cortical signal. Using pons as the reference region in ¹⁸F-MK-6240 MRTM2 analyses resulted in greater dynamic range (magnitudes) and variability similar to that for CerGM for both SUVR and DVR measures. Using reference regions that generate larger dynamic ranges of outcome measures (e.g. pons) may improve the sensitivity for longitudinal tracking of tau accumulation in subjects with low tau accumulation; however, further studies are needed to better characterize factors that impact the specific and non-specific ¹⁸F-MK-6240 kinetics and the bias in outcome measures in target regions.