Biomarcadores

[¹⁸F]MK6240 exhibits high affinity to Alzheimer's disease (AD) tauopathy, but meningeal or extracerebral off-target signal (Figure 1AB) complicates early tau detection. We previously demonstrated that [¹⁸F]MK6240 off-target signal follows irreversible kinetics, and an alternative compartmental model incorporating an extracerebral term (s) may quantify extracerebral contamination. Here, we further evaluate this model's performance and its impact on late-frame SUVR compared to partial volume correction (PVC) for early tau detection.

Thirteen participants (8 controls, CN; 5 mild cognitive impairment, MCI) underwent 2-hr [¹⁸F]MK6240 PET, MRI, and arterial sampling (Table 1). A within-reconstruction kernel-based PVC using MRI information was applied (n = 8). Time-activity curves were generated for extracerebral (Figure 1C), target, and reference regions. Regional uptake was modeled using: 1) 2TCM, standard two-tissue compartmental model, 2) 2TCM+s, alternative model incorporating an additive extracerebral term, sC_s (s = fractional contribution of extracerebral signal C_s) (Figure 2A), 3) 2TCM-kernel: 2TCM applied to kernel-PVC data. Outcomes included distribution volume (V_T), distribution volume ratio (DVR), and SUVR_90-110min. T-tests (p <0.05) and Akaike Information Criteria (AIC) compared model parameters and performance. Cohen's d assessed effect size for CNs vs. MCIs using SUVR_90-110min with/without PVC and with 2TCM4k+s correction.

The 2TCM+s model demonstrated the best AIC, followed by 2TCM-kernel, then 2TCM. The extracerebral term s (0.16-0.43, Table 1) reflected extracerebral contamination, being highest in regions near the meninges (e.g., entorhinal). Accounting for s led to lower V_T due to higher k₄, particularly in highly contaminated regions (Figure 2B-D). In early-tau target regions, compared to DVR_2TCM, DVR_2TCM+s was lower in CNs but similar in MCIs, while DVR_2TCM-kernel was similar in CNs but higher in MCIs. The SUVR_90-110min with 2TCM+s correction exhibited the largest effect size for differentiating CNs and MCIs, followed by kernel-PVC SUVR_90-110min, with the smallest effect size in non-PVC SUVR_90-110min (entorhinal Cohen's d, non-PVC: -0.58, kernel-PVC: -1.0, 2TCM+s: -1.2).

A compartmental modeling approach quantifies regional [¹⁸F]MK6240 extracerebral contamination, potentially improving V_T, DVR, and SUVR accuracy and sensitivity for early tau detection, even compared to PVC data. Future studies will include more participants and extend these corrections to [¹⁸F]MK6240 data without dynamic acquisition, broadening applicability in clinical and research settings.