Evaluating early vs. late static SUVR windows of [¹⁸F]MK-6240 tau PET in Alzheimer disease: a head-to-head comparison...

[¹⁸F]MK-6240 is a widely used second-generation tau PET tracer in Alzheimer disease (AD) and is under FDA review, making it important to refine practical static imaging windows for clinical use. Prior dynamic studies have supported late static windows (~ 90-110 min) because they improve agreement between standard uptake value ratios (SUVR) and kinetic models in high-binding regions, but extracerebral skull and meningeal uptake becomes more prominent at later times and may confound visual interpretation and SUVR quantification. In this study, we aim to determine whether an earlier static window (40-60 min) provides comparable discrimination of tau burden while reducing extracerebral spill-in and SUVR instability.

In this retrospective within-subject study, 67 participants across the AD spectrum (normal controls (cognitively unimpaired) [NL], mild cognitive impairment [MCI], and AD) underwent [¹⁸F]MK-6240 PET with early (40-60 min) and late (90-120 min) frames. SUVRs were computed in the cerebral cortex (CTX) and Braak ROIs using cerebellar cortex as reference. Early-late agreement was assessed using correlation, regression, and Kolmogorov-Smirnov (KS) tests. Diagnostic and tau-status classification performance was evaluated with ROC analyses. Skull/meningeal uptake was graded visually and quantified, and PET-MRI misregistration simulations assessed SUVR robustness. Time-resolved analyses (30-120 min) evaluated temporal changes in stability and discrimination.

Early and late SUVRs were highly correlated (r ≥ 0.97; p < 0.01) and showed similar cohort-level distributions (CTX KS D = 0.060; p > 0.99). Discrimination was comparable for diagnosis and tau status (e.g., CTX AUC 0.88 vs. 0.85 for NL vs. MCI/AD). Late frames showed greater skull/meningeal uptake with increased spill-in and reference-region sensitivity; in tau-negative participants, skull SUVR exceeded entorhinal SUVR after ~ 60 min. Misregistration simulations showed greater variability for late versus early entorhinal SUVR (σ = 0.071 vs. 0.017). Time-resolved analyses showed that in this cohort, later acquisition did not materially improve diagnostic discrimination and was associated with more extracerebral contamination and instability.

The 40-60-minute window provides discrimination comparable to 90-120 min while reducing extracerebral contamination and improving robustness, supporting shorter static [¹⁸F]MK-6240 protocols for clinical and research applications.