Correcting [ ¹⁸ F]MK‐6240 <i>SUVR</i> for non‐equilibrium effects using tissue clearance correction method does not improve its performance

Abstract Background Honhar et al. (PMCID:10993874) reported a method for bias and noise correction in standardized‐uptake value ratios ( SUVR , simplified outcome), relative to distribution volume ratios ( DVR , quantitative gold‐standard), without the need for dynamic data outside SUVR time‐window. The method performed well for rapidly‐reversible radiotracers, [ 18 F]FE‐PE2I and [ 11 C]LSN3172176 (i.e., radiotracers kinetics described by 1‐tissue compartment or simplified reference tissue models). Herein, we tested this method on human [ 18 F]MK‐6240 PET data. Methods Dynamic [ 18 F]MK‐6240 PET data (0‐120 min) in 22 human participants (age: 61±18 y, 8 females, 16 cognitively unimpaired controls [HC], 5 mild cognitive impairment [MCI], 1 Alzheimer's disease [AD]) were reanalyzed. Metabolite‐corrected arterial input functions were used to compute regional 2‐tissue compartment model DVR values (reference: eroded cerebellar gray matter). Regional SUVR (80‐100 min and 90‐110 min) values were corrected as: SUVR C = SUVR / [1‐ β ref / k 2 ,ref + β tar SUVR /( R 1 k 2 ,ref )], where SUVR C : corrected SUVR , k 2 ,ref : population‐based rate constant for efflux of the radiotracer from reference region, [ β ref , β tar ]: clearance rates of the radiotracer from the reference and target tissues during SUVR time‐window, R 1 =0.75 (population‐averaged value for ratio of target:reference radiotracer delivery rate). The primary outcome measures were the percent bias (%bias) in SUVR (and SUVR c) relative to DVR , and the standard deviation of %bias. Results Across subjects and selected regions of interest (ROIs, Tables 1‐3), the mean %bias was 0.9% for SUVR 80‐100 min and 1.2% for SUVR 90‐110 min. The variability in %bias was 13.1% for SUVR 80‐100 min and 11.9% for SUVR 90‐110 min. The mean %bias in SUVR c across subjects and ROIs was comparable for SUVR c 80‐100 min (‐1.3%) and SUVR c 90‐110 min (‐0.1%), with similar variability (∼11%) for SUVR c 80‐100 min and SUVR c 90‐110 min. Regional measures for (averaged across subjects) are listed in Tables 1‐2. Similar performance metrics were observed for a subcohort analysis of 5 high‐binding individuals with highest average regional DVR values (1 AD,3 MCI,1 HC). Conclusions [ 18 F]MK‐6240 SUVR in this cohort was already minimally biased; SUVR c did not significantly decrease the variability in SUVR bias. This SUVR correction method might not be beneficial for radiotracers that deviate considerably from its assumption of 1‐tissue compartment model kinetics.