Test-Retest Variability of Relative Tracer Delivery Rate as Measured by [11C]PiB

Purpose Moderate-to-high correlations have been reported between the [11C]PiB PET-derived relative tracer delivery rate R1 and relative CBF as measured using [15O]H2O PET, supporting its use as a proxy of relative CBF. As longitudinal PET studies become more common for measuring treatment efficacy or disease progression, it is important to know the intrinsic variability of R1. The purpose of the present study was to determine this through a retrospective data analysis. Procedures Test-retest data belonging to twelve participants, who underwent two 90 min [11C]PiB PET scans, were retrospectively included. The voxel-based implementation of the two-step simplified reference tissue model with cerebellar grey matter as reference tissue was used to compute R1 images. Next, test-retest variability was calculated, and test and retest R1 measures were compared using linear mixed effect models and a Bland-Altman analysis. Results Test-retest variability was low across regions (max. 5.8 %), and test and retest measures showed high, significant correlations (R2=0.92, slope=0.98) and a negligible bias (0.69±3.07 %). Conclusions In conclusion, the high precision of [11C]PiB R1 suggests suitable applicability for cross-sectional and longitudinal studies.


Background
Cerebral blood flow (CBF) is known to decline with age, and elderly individuals (75-80 years) may present with reductions in CBF of up to 25 % compared with young adults (±25 years) [1,2]. In the context of Alzheimer's disease (AD), additional reductions in CBF have been reported in several cortical brain regions such as the frontal, parietal and temporal cortices with both absolute reductions as well as relative to cerebellar grey matter reference tissue [3,4]. This pattern of CBF reductions is considered characteristic of AD pathology and may therefore be used as proxy for measuring disease severity or progression [5,6]. The gold standard technique for measuring CBF is [ 15 O]H 2 O positron emission tomography (PET) [7], but MR-based techniques such as arterial spin labelling (ASL) have also been introduced [5]. More recently, several studies have evaluated whether a valid proxy of cerebral perfusion can be obtained from the early frames of dynamic scans using currently available PET tracers (e.g. for measuring amyloid-β or tau burden) [3,8]. The relative influx rate (R 1 =K 1 /K 1 '), which can be calculated from these early frames, is an indirect measure of relative CBF as it is also affected by the extraction fraction (K 1 =E·CBF). In particular, for the amyloid tracer [ [8]. As longitudinal PET studies in AD become more common for measuring treatment efficacy or disease progression, it is important to know the intrinsic variability of R 1 in order to determine what magnitude of change in R 1 signifies an actual change. Therefore, the purpose of the present retrospective analysis of a previously reported test-retest (TRT) [ 11 C]PiB study was to assess the precision of [ 11 C]PiB R 1.

Subjects
Data from twelve participants belonging to a TRT study conducted within the Amsterdam UMC, location VUmc, were reanalysed as the original study only reported TRT variability for the non-displaceable binding potential [10]. This dataset consisted of five cognitively unimpaired (CU) subjects, one patient with mild cognitive impairment (MCI) and six with AD dementia, which in the present study was used to examine TRT variability for R 1 [10]. Before enrolment, written informed consent was obtained from all individual participants included in the study, and the Medical Ethics Review Committee of the Amsterdam UMC, location VUmc, had approved the study.

Image Processing
Structural T1-weighted MR images were co-registered to their corresponding PET image segmented into grey matter (GM), white matter (WM) and cerebrospinal fluid (CSF) using PVE-lab software [11]. Next, volumes of interest (VOIs) were delineated based on the Hammers atlas and a reference tissue time-activity curve (TAC) of the cerebellar grey matter was extracted [12,13].

Parametric Analysis
The PPET software tool [14] with the voxel-based implementation of the two-step simplified reference tissue model (SRTM2), as validated for [ 11 C]PiB, and cerebellar grey matter as reference tissue were used to compute relative tracer delivery (R 1 ) images [15][16][17]. For SRTM2, k 2 ' was determined across all voxels with a BP ND higher than 0.05 by taking the median k 2 ' from a first run using receptor parametric mapping (RPM) [18]. Regional R 1 values were obtained by superimposing the following grey matter VOIs on the parametric images: medial and lateral anterior temporal lobe, posterior temporal lobe, superior, middle a n d i n f e r i o r te m p o r a l g y r u s , f u s i f o r m g y r u s , parahippocampal and ambient gyrus, anterior and posterior cingulate gyrus, middle and orbitofrontal gyrus, gyrus rectus, inferior and superior frontal gyrus, pre-and postcentral gyrus, superior parietal gyrus and the (infero)lateral remainder of the parietal lobe and a global cortical composite region (i.e. volume-weighted average across all target regions).

Statistical Analysis
Statistical analyses were performed in R (version 4.0.3; R Foundation for Statistical Computing, Vienna, Austria). First, global R 1 values were compared between CU and AD dementia groups using a non-parametric Mann-Whitney U test, separately for test and retest scans. Next, TRT variability was calculated for regional and global cortical R 1 values according to Eq. 1, where T represents the estimate of R 1 measured during test, and R the one measured during retest.
In addition, a correlation analysis was used to assess the relationship between TRT variability and regional volume.
Furthermore, to assess the relationship between test and retest R 1 measures, linear mixed effect models (LME) were fitted and correlation coefficients were calculated using the nlme and MuMIn packages, respectively [19,20]. Visual read (amyloid-β positive or negative) was used as a covariate and the analysis accounted for the within-subject correlation between regions. Finally, a Bland-Altman analysis was used to assess potential bias between test and retest R 1 using the blandr package [21,22].

Results
Participant characteristics are shown in Table 1. Relative tracer delivery measures (R 1 ) are reported in Table 2, with a significantly lower global R 1 in AD dementia patients compared with CU participants, for both test and retest scans (p G 0.01).
Regional and global cortical TRT variability values are presented in Table 3. TRT variability for the global cortical composite was low (1.70 %), while the range of regional TRT variability showed slightly higher values (range: 1.52-5.78 %). Furthermore, there was a trend effect towards smaller TRT variability for larger regions (R 2 =0.14, p=0.09).
LME analyses showed that test and retest R 1 values were strongly correlated and that the slope was not significantly different from 1 (R 2 =0.92, slope=0.98 C.I. [0.94-1.01], pG0.001). Furthermore, amyloid status as measured by visual read did not have a significant effect on this relationship. Finally, Bland-Altman analysis showed a negligible bias (0.69±3.07 %) between test and retest R 1 (Fig. 1). All analyses were also carried out using RPMderived R 1 which resulted in essentially identical results (data not shown).

Discussion
The present study assessed precision of the R 1 parameter, a measure of relative tracer delivery, through a retrospective analysis of a previously reported [ 11 C]PiB test-retest study [11]. Low test-retest variability was observed for SRTM2derived R 1 , and this was true for regions of different sizes.
Differences in R 1 between diagnostic groups were as expected, with lower average R 1 values in AD dementia patients compared with CU participants. This finding is in agreement with existing literature where decreases in (relative) perfusion related to AD pathology have been reported for both R 1 and [ 15 O]H 2 O PET studies [3,23,24]. Furthermore, by incorporating these two groups, the present study covered the entire range of R 1 values that would be expected in clinical studies across the AD spectrum.
Excellent TRT variability was observed for the global cortical composite (1.70 %) and only a slightly poorer TRT variability for some of the smaller regions (max 5.8 %). These findings were supported by the results of the LME analysis which showed a high correlation between test and retest R 1 measures (R 2 =0.92) and a slope that was close to identity. As expected, the results indicate that smaller TRT variability was associated with larger regions. This finding suggests that studies should consider looking at relatively larger regions with PET when their aim is to detect small (G5 %) changes. Despite distinct kinetics, the present findings were also comparable with results from a [ 18 F]florbetapir study that assessed TRT variability of SRTM-derived R 1 in a very similar population in terms of age and diagnosis (max. TRT variability of 6 %) [25]. Comparing [ 11 C]PiB R 1 TRT variability with TRT variability of absolute perfusion as    [8]. Nevertheless, a smaller change in R 1 , especially in elderly subjects, may be related to an increased extraction fraction [27]. However, longitudinal studies of R 1 in a younger population are needed to confirm whether the same difference is present earlier in life. Furthermore, using a relative parameter to measure CBF such as R 1 essentially assumes that there are no CBF changes in the reference tissue (R 1 =K 1 /K 1 '). In this regard, it should be noted that differences in whole cerebellum CBF (a commonly used reference tissue) have been reported when comparing AD dementia patients and age-matched controls [28]. In contrast, such differences have not been demonstrated for cerebellar cortex CBF by studies using a similar design in terms of technique and participants [24,[29][30][31]. This suggests that careful interpretation is required when comparing longitudinal R 1 measurement between AD dementia patients and controls or that alternative reference tissues, unaffected by CBF changes, should be considered. Yet, further research is required to understand whether such changes in cerebellar CBF also occur in early AD stages.

Conclusion
Relative tracer delivery rate R 1 of [ 11 C]PiB showed high global and regional precision in participants covering the AD spectrum. Therefore, [ 11 C]PiB R 1 appears to be a stable parameter for measuring cross-sectional differences and longitudinal changes in relative CBF.
Author Contribution. FH, JH, MY, ILA, AAL contributed to the study design, analyses, data interpretation and drafting of the manuscript. NT and BvB contributed to data acquisition. All authors critically reviewed, and approved the final version of the work.
Funding. This project received funding from the EU/EFPIA Innovative Medicines Initiative (IMI) Joint Undertaking (EMIF grant 115372) and the EU-EFPIA IMI-2 Joint Undertaking (grant 115952). This joint undertaking receives support from the European Union's Horizon 2020 research and innovation program and EFPIA. This communication reflects the views of the authors and neither IMI nor the European Union and EFPIA are liable for any use that may be made of the information contained herein.

Ethical Approval
All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national  Fig. 1. Relationship between SRTM2-derived test and retest R 1 . (a) The correlation between R 1 test and retest measures, with R 2 and slope parameters corresponding to the LME analysis and (b) a Bland-Altman plot, which indicates the bias between the two measures.
research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.