In order to maximise the utility of [11C]-PBR28 for use in longitudinal studies and clinical trials in Alzheimer’s disease (AD), there is a need to develop non-invasive metrics of tracer binding that do not require arterial cannulation. Recent work has suggested that standardised uptake value (SUV)-based methods may be sensitive to changes in translocator protein (TSPO) levels associated with neurodegeneration. However, the test-retest reliability of these approaches in AD over a time period relevant for clinical trials is unknown. In this study, the test-retest reliability of three SUV-based metrics was assessed in AD patients over 12 weeks.
Five patients with mild AD and the high-affinity binding TSPO genotype underwent two [11C]-PBR28 PET scans approximately 12 weeks apart. The test-retest reliability (TRR) of the unadjusted SUV, SUV relative to cerebellar grey matter (SUVRC) and SUV normalised to whole brain activity (SUVRWB) in nine cortical and limbic regions of interest was assessed using the absolute variability and the intraclass correlation coefficient.
Of the three measures, SUVRWB performed best overall, showing low absolute variability (mean −0.13 %, SD 2.47 %) and high reliability (mean ICC = 0.83). Unadjusted SUV also performed well, with high reliability (ICC = 0.94) but also high variability (mean −1.24 %, SD 7.28 %). By comparison, the SUVRC showed higher variability (mean −3.98 %, SD 7.07 %) and low reliability (ICC = 0.65).
In this AD sample, we found that SUV-derived metrics of [11C]-PBR28 binding showed high stability over 12 weeks. These results compare favourably with studies reporting TRR of absolute quantification of [11C]-PBR28. Pending further validation of SUV-based measures of [11C]-PBR28, semi-quantitative methods of [11C]-PBR28 analysis may prove useful in longitudinal studies of AD.