Test–retest variability and reference region-based quantification of 18 F-BCPP-EF for imaging mitochondrial complex I in the human brain

Abstract

Mitochondrial complex I (MC-I) is an essential regulator of brain bioenergetics and can be quantified in the brain using PET radioligand ¹⁸F-BCPP-EF. Here we evaluate the test–retest reproducibility of ¹⁸F-BCPP-EF in humans, and assess the use of a non-invasive quantification method (standardised uptake value ratio – SUVR). Thirty healthy volunteers had a 90-min dynamic ¹⁸F-BCPP-EF scan with arterial blood sampling, five of which received a second scan to be included in the test–retest analysis. Time-activity curves (TAC) were analysed using multilinear analysis 1 (MA1) and the two-tissue compartment model (2TC) to estimate volumes of distribution (V_T). Regional SUVR-1 values were calculated from the 70 to 90-min TAC data using the centrum semiovale as a pseudo reference region, and compared to kinetic analysis-derived outcome measures. The mean absolute test–retest variability of V_T ranged from 12% to 18% across regions. Both DVR-1and SUVR-1 had improved test–retest variability in the range 2%–7%. SUVR-1 was highly correlated with DVR-1 (r² = 0.97, n = 30). In conclusion, ¹⁸F-BCPP-EF has suitable test–retest reproducibility and can be used to quantify MC-I in clinical studies.

Keywords

Brain imaging positron emission tomography MC-I kinetic modelling neurodegeneration

Introduction

Mitochondria are the primary source of cellular energy in the brain. Mitochondrial complex I (MC-I) is an enzyme located in the inner mitochondrial membrane and is the first major entry point for electrons into the electron transport chain, making it the rate-limiting step in oxidative phosphorylation and therefore a critical component of cellular energy production. MC-I has additional roles in apoptotic signalling, regulation of calcium homeostasis and free radical production.¹ Altered MC-I function can be detrimental to brain bioenergetics and contribute to neurodegenerative and neurological disease pathology and as well as to accelerated aging.^2,3 Since its introduction by Tsukada et al. as a promising novel positron emission tomography (PET) probe for non-invasively imaging MC-I in the living brain, ¹⁸F-BCPP-EF has been used to successfully detect neuronal degeneration induced by ischemic insult, aging and neurodegenerative disease models in animals.^4–7 The specific binding of ¹⁸F-BCPP-EF to MC-I has been confirmed by a rotenone blockade in the non-human primate (NHP) brain, paving the way to translation to clinical human studies.⁵

Recent work by our group presented first time in human data and an initial evaluation of ¹⁸F-BCPP-EF kinetics in the brains of 12 healthy volunteers.⁸ This demonstrated that ¹⁸F-BCPP-EF possesses rapid metabolism and reversible kinetics that were well described by both graphical method multilinear analysis 1 (MA1) and a two-tissue compartmental (2TC) model. ¹⁸F-BCPP-EF showed a good specific signal in striatal regions, with significant contrast over the white matter region centrum semiovale which had ∼50% lower signal compared to grey matter regions. As a result of markedly higher energy cost associated with neuronal signalling, energy consumption in the white matter is approximately 20–25% of that in the grey matter, consistent with the low MC-I related signal seen in this region, and supporting the centrum semiovale as a good candidate for investigation as a pseudo reference region.^9,10 However, an assessment of the reproducibility of ¹⁸F-BCPP-EF signal in the human brain and a careful evaluation of whether the centrum semiovale is a useful pseudo reference region has yet to be performed.

The work presented here aims to further characterise ¹⁸F-BCPP-EF binding in the human brain, by evaluating its test–retest reproducibility, and assessing the suitability of the centrum semiovale as a pseudo reference region based on preclinical blocking data. In addition, quantification from simplified static acquisitions is evaluated in comparison to full kinetic analysis approaches.

Materials and methods

Subjects

Thirty healthy subjects (16 M/14 F; age: 51 ± 19 years, range: 22–78) were included in the study, five of which received a second ¹⁸F-BCPP-EF scan to be included in the test–retest analysis.

All subjects were screened and scanned at Invicro London and at the Imperial College neuroepidemiology and at the Aging Research unit as part of the MIND MAPS consortium (www.invicro.com/mindmaps). All procedures followed were in accordance with the ethical standards of East of England – Cambridge South Research Ethics Committee (REC reference number 17/EE/0028), ICH Good Clinical Practice (GCP) and with the guiding principles of the 2008 Declaration of Helsinki. Written informed consent was obtained from all subjects.

Radiotracer synthesis

¹⁸F-BCPP-EF was synthesised as previously described.⁴

Competition data from rhesus monkey

Previously acquired data were provided to us by Tsukada et al.⁵ from a study in rhesus monkeys (Macaca mulatta) which were scanned with ¹⁸F-BCPP-EF at baseline and following pre-administration of 0.1 mg/kg of rotenone (MC-I inhibitor).⁵ Dose-escalation was not possible due to the lethal effects of rotenone on cardiac function. Volume of distribution (V_T) data were provided for a set of nine regions of interest (ROIs) from four male monkeys (age: three to five years) (Supplemental Table 1). Full details on scanning procedures, rotenone administration and study protocol are provided in the manuscript by Tsukada et al.⁵

Data acquisition

Each subject had a T1-weighted MRI scan for coregistration with PET images. Scans were acquired on a Siemens 3 T Trio clinical MRI scanner (Siemens Healthineers, Erlangen, Germany) with a 32-channel phased-array head coil using a 3D MPRAGE sequence (TE = 2.98 ms, TR = 2300 ms, flip angle of 9°, voxel size = 1.0 mm × 1.0 mm × 1.0 mm).

All subjects underwent a 90-min dynamic PET scan with ¹⁸F-BCPP-EF with arterial sampling. Five of the subjects received a second scan approximately one month after the first scan (Supplemental Table 2). PET scans were acquired on either a Siemens Hi-Rez Biograph 6 or Biograph 6 TruePoint PET/CT scanner (Siemens Healthcare, Erlangen, Germany). The scanner model was kept consistent across visits for the test–retest subjects. A low-dose CT scan (30 mAs, 130 KeV, 0.55 pitch) was performed prior to each PET scan to estimate attenuation. An intravenous cannula was inserted into a cubital or forearm vein for radioligand administration, and a second cannula was inserted into the radial artery to enable the collection of arterial blood samples. The radioligand was administered as a slow bolus (over 20 s) in a volume of 20 mL at the start of the PET scan. Dynamic emission data were acquired over 90 min following radiotracer administration and were reconstructed into 26 frames (frame durations: 8 × 15 s, 3 × 60 s, 5 × 120 s, 5 × 300 s, 5 × 600 s). Corrections were applied for attenuation, randoms and scatter.

Whole blood activity was measured in all subjects using a continuous automatic blood sampling system (Allogg, AB, Marlefred, Sweden) acquired at a rate of 5 mL/min for the first 15 min post-injection. Discrete blood samples were taken at 10, 15, 20, 25, 30, 40, 50, 60, 70, 80 and 90 min after scan start to determine the fraction of plasma radioactivity constituted by unchanged parent radioligand (ppf) using high-performance liquid chromatography analysis. Total radioactivity concentration was evaluated in blood and plasma in a Perkin Elmer 1470 10-well gamma counter. The plasma-free fraction (f_p) was measured by ultrafiltration in triplicate using an arterial blood sample taken prior to tracer injection for each subject.

Image processing

All image data were analysed using the MIAKAT™ software package (version 4.3.7, http://www.miakat.org), in MATLAB (version R2016a; Mathworks Inc., Natick, MA, USA), using SPM12 (Wellcome Trust Centre for Neuroimaging, http://www.fil.ion.ucl.ac.uk/spm) for image segmentation and registration.

Brain extraction, grey matter segmentation and rigid body coregistration to a standard reference space (MNI152) was performed on each structural T1 image.¹¹ The template brain image and associated atlas (CIC) were then nonlinearly warped to the individual subject’s MRI image where the following ROIs were defined: brainstem, substantia nigra, thalamus, striatum, globus pallidus, ventral striatum, caudate, putamen, posterior cingulate cortex, anterior cingulate cortex, frontal cortex, insular cortex, hippocampus, amygdala, temporal lobe, parietal lobe and cerebellum.¹² The centrum semiovale was defined using the automated anatomic labelling (AAL) template as previously described.^13,14 PET images were registered to each subject’s MRI image and corrected for motion using frame-to-frame rigid-body registration. Regional time activity curves (TACs) were generated for each ROI.

Quantitative analysis

Tissue kinetic modelling

TACs were fitted with both the two-tissue compartmental model (2TC) and graphical model multilinear analysis 1 (MA1) with a fixed blood volume of 5% and a t* of 30 min as determined previously.⁸ V_T and V_T/f_p were assessed as outcome measures. In addition, distribution volume ratio (DVR) was calculated as V_T^ROI/V_T^{CENTRUM SEMIOVALE} to enable assessment of DVR-1 as an additional outcome measure. The term DVR-1 is used instead of non-displaceable binding potential as the centrum semiovale may only be a pseudo reference region. 2TC-derived K₁ values were also included in the analysis to investigate the potential susceptibility of ¹⁸F-BCPP-EF outcome measures to changes in tracer delivery rate.

Test–retest reproducibility

The test–retest reproducibility of PET of both MA1 and 2TC-derived outcome measures was evaluated by calculation of the relative and absolute test–retest variability (TRV, aTRV) and intraclass correlation coefficient (ICC) using a one-way random effects ANOVA model

TRV (%) = 200 \times \frac{retest value - test value}{(retest value + test value)}

(1)

aTRV (%) = 200 \times \frac{|retest value - test value|}{(retest value + test value)}

(2)

ICC = \frac{BSMSS - WSMSS}{BSMSS + WSMSS}

(3)

where BSMSS is between-subject mean sum of squares and WSMSS is within-subject mean sum of squares. Two-tailed paired t-tests were used in the statistical analysis of test and retest conditions and linear regression was used to compare the two conditions. All results are reported as mean ± standard deviation.

Quantification from simplified static acquisition

In addition to V_T, V_T/f_p and DVR-1, semi-quantitative standardised uptake value ratio (SUVR – 1) was assessed as a simplified outcome measure. The summed SUV in a target region determined over 70–90 min after ¹⁸F-BCPP-EF administration was normalized to that of the centrum semiovale to derive regional SUVR-1 values. SUVR-1 values were correlated with regional DVR-1 measures from full kinetic analysis using least-squares linear regression.

Estimation of non-displaceable ¹⁸F-BCPP-EF signal in humans

To provide an estimate for the non-displaceable volume of distribution (V_ND) in monkeys, regional V_T^BASELINE – V_T^ROTENONE data were plotted against V_T^BASELINE data to create an occupancy plot as described by Lassen et al.^15,16 A nonlinear least-squares estimator was used to derive fractional brain occupancy of MC-I by rotenone for each monkey, and a shared V_ND across all four monkeys.

Assuming the conservation of BP_ND across non-human primate and human ( ${B P}_{N D}^{NHP} ≅ {B P}_{N D}^{Human}$ ), yields the following equation

\frac{V_{T}^{Human}}{V_{T}^{NHP}} = \frac{V_{N D}^{Human}}{V_{N D}^{NHP}}

(4)

This allowed for translation of this V_ND estimate in monkey to humans, using the ratio of V_T derived from baseline ¹⁸F-BCPP-EF scans in monkeys to V_T in humans which was calculated for the cerebral white matter, cerebellum, frontal cortex, cingulate cortex, temporal cortex, parietal cortex, occipital cortex, striatum and hippocampus. The regional ratios were averaged to create a scaling factor which was applied to the estimated $V_{N D}^{NHP}$ to derive $V_{N D}^{Human}$ . The human BP_ND in the cerebral white matter (CWM) was predicted from the following equation

{B P}_{N D}^{CWM} = \frac{V_{T}^{CWM}}{V_{N D}}

(5)

Results

The radiochemical purity of ¹⁸F-BCPP-EF was greater than 99% for all scans. The mean injected radioactivity was 92 ± 3 MBq for the test condition and 91 ± 3 MBq for the retest condition. The mean injected mass for the test and retest PET scans was 0.08 ± 0.03 μg and 0.08 ± 0.02 μg, respectively. There were no statistical significant differences between test and retest injection parameters. The injection parameters for all subjects are summarised in Supplemental Table 2. See Supplemental Table 3 for individual injected activity and mass values for all subjects.

The average ¹⁸F-BCPP-EF ppf and the total parent radioactivity concentration in plasma over time for the test and retest conditions are shown in Figure 1(a) and (b) (see Supplemental Figure 1 for individual plots). The test–retest variability of parent clearance from plasma was 24% ± 18% with an ICC of −0.25 (Supplemental Table 3). The average f_p was 0.072 ± 0.012 for the test PET measurements and 0.065 ± 0.010 for the retest PET measurements (p < 0.05). The TRV and aTRV of f_p were −9.7 ± 6% and 9.7%, respectively, with an ICC of 0.74.

Figure 1.

¹⁸F-BCPP-EF radioactivity in plasma. (a) Unchanged ¹⁸F-BCPP-EF ppf over time. Data are shown as mean ± SD. (b) Total ¹⁸F-BCPP-EF concentration in plasma over time. First five minutes excluded from plot for clarity. Data are mean (solid line) and 95% confidence internal (dotted line) for test and retest condition (n = 5).

¹⁸F-BCPP-EF entered the brain readily, and demonstrated a heterogeneous distribution as previously described.⁸ Figure 2 shows individual subject SUV images derived from the test and retest conditions and corresponding structural images. Tracer uptake was rapid, reaching peak uptake values (SUV) ranging from 3.9 g/mL in the brainstem to 6.5 g/mL in the putamen between 5 and 20 min post injection across the 17 grey matter regions investigated. Uptake was between 36% and 62% lower in the centrum semiovale compared to grey matter regions and reached an average peak uptake value of 2.5 g/mL at 10 min post injection.

Figure 2.

Individual MR and SUV Add-images (10–90 min) of ¹⁸F-BCPP-EF during test and retest condition.

MA1 and 2TC both produced good fits to all TAC data and produced V_T values that were highly correlated with each other (r² = 0.99), consistent with our previous report.⁸ Given the similarity between the results derived by the two methods, we focus most of the subsequent discussion on MA1-derived results (Figure 3). Mean V_T values were between 9.0 ± 1.4 mL/cm³ in the centrum semiovale and 26.7 ± 4.8 mL/cm³ in the putamen for the test condition, which was consistent with previously reported human ¹⁸F-BCPP-EF V_T values (Supplemental Table 4). Corresponding V_T values were slightly higher for the retest condition, ranging from 10.3 ± 1.5 mL/cm³ in the centrum semiovale and 29.0 ± 5.3 mL/cm³ in the putamen.

Figure 3.

Regional time activity curves and MA1 model fits in subject 5 for test (solid line) and retest (dotted line) condition. CS: centrum semiovale; BST: brainstem; CAU: caudate; HIP: hippocampus; TL: temporal lobe; FTCX: frontal cortex.

V_T values derived from the test and retest PET scans were well correlated (r² = 0.81), with retest values being 9 ± 2% higher on average compared to test values (Figure 4(a); Supplemental Figure 2(a)). The differences in V_T estimates derived from the test and retest scans did not reach statistical significance in any of the regions. The mean regional TRV and aTRV for V_T were 9% ± 2% and 13% ± 2%, respectively (Table 1). ICC values for V_T were greater than 0.60 for all regions with the exception of the centrum semiovale, globus pallidus and ventral striatum which had ICC values of 0.30 and 0.57 and 0.54, respectively. The global mean ICC for V_T was 0.68 ± 0.13. 2TC-derived V_T had test–retest reproducibility measures that were comparable to MA1-derived values (aTRV 14% ± 2%, ICC 0.65 ± 0.11, Supplemental Table 5).

Figure 4.

Correlation between regional (a) V_T (b) V_T/f_p (c) DVR-1 derived under test and retest conditions. All regions and subjects are included in the plots (18 ROIs, 5 subjects).

Table 1.

Test–retest reproducibility of ¹⁸F-BCPP-EF outcome measures.

ROI	V_T (mL/cm³)			V_T/f_p (mL/cm³)			DVR-1			SUVR-1
ROI	TRV (%)	aTRV (%)	ICC	TRV(%)	aTRV(%)	ICC	TRV (%)	aTRV (%)	ICC	TRV (%)	aTRV (%)	ICC
CS	13 ± 16	18	0.30	21 ± 21	26	0.39	–	–	–	–	–	–
BS	9 ± 12	13	0.64	17 ± 16	21	0.64	−13 ± 25	18	0.31	−12 ± 14	12	0.83
SN	9 ± 13	12	0.61	17 ± 16	20	0.62	−10 ± 19	17	0.15	−6 ± 9	8	0.86
THA	8 ±12	13	0.78	16 ± 17	17	0.74	−9 ±9	11	0.84	−7 ± 5	7	0.94
STR	8 ± 12	13	0.79	16 ± 17	21	0.72	−8 ± 8	9	0.87	−10 ± 5	10	0.91
GP	10 ± 13	14	0.57	18 ± 18	22	0.64	−5 ± 10	9	0.83	−8 ± 6	8	0.89
VSTR	11 ± 14	15	0.54	20 ± 18	23	0.59	−2 ± 9	8	0.81	−7 ± 7	7	0.89
CAU	6 ± 13	12	0.91	14 ± 19	20	0.83	−15 ± 8	15	0.94	−16 ± 7	6	0.94
PUT	8 ± 12	13	0.71	16 ± 18	21	0.69	−7 ± 10	9	0.75	−9 ± 4	9	0.89
PC	11 ± 15	17	0.66	19 ± 20	25	0.58	−2 ± 2	3	0.97	−1 ± 3	3	0.98
AC	8 ± 13	13	0.72	17 ± 17	21	0.64	−8 ± 10	10	0.78	−7 ± 5	7	0.94
FTCX	8 ± 14	14	0.69	16 ± 19	22	0.63	−9 ± 6	9	0.86	−8 ± 2	8	0.94
INS	8 ± 11	12	0.71	16 ± 15	20	0.67	−8 ± 12	11	0.51	−8 ± 6	8	0.86
HIP	6 ± 9	10	0.83	14 ± 14	18	0.77	−15 ± 16	19	0.63	−10 ± 9	11	0.89
AMY	7 ± 12	12	0.71	15 ± 17	20	0.66	−11 ± 12	13	0.57	−10 ± 8	11	0.79
TL	8 ± 11	11	0.67	16 ± 15	19	0.65	−9 ± 15	13	0.16	−7 ± 7	8	0.81
PL	10 ± 15	16	0.67	18 ± 21	24	0.64	−5 ± 6	5	0.88	−3 ± 8	7	0.94
CER	8 ± 11	11	0.71	16 ± 15	20	0.70	−9 ± 14	12	0.31	−8 ± 6	8	0.82
Mean	9	13	0.68	17	21	0.66	−9	11	0.66	−8	9	0.89
SD	2	2	0.13	2	2	0.09	4	4	0.37	3	3	0.05

TRV: Test–retest variability; aTRV: absolute test–retest variability; ICC: intra-class correlation; CS: centrum semiovale; BS: brainstem; SN: substantia nigra; THA: thalamus; STR: Striatum; GP: globus pallidus; VSTR: ventral striatum; CAU: caudate; PUT: putamen; PC: posterior cingulate cortex; AC: anterior cingulate cortex; FTCX: frontal cortex; INS: insular cortex; HIP: hippocampus; AMY; amygdala; TL: temporal lobe; PL: parietal lobe; CER: cerebellum.

Note: Data are mean± SD. n = 5. Kinetic measures derived from MA1.

V_T/f_p values derived from the test scans ranged from 130 ± 31 mL/cm³ in the centrum semiovale to 388 ± 113 mL/cm³ in the putamen for the test condition (Supplemental Table 4). The corresponding retest values were 160 ± 41 mL/cm³ and 455 ± 135 mL/cm³, respectively. Test and retest-derived V_T/f_p values were moderately correlated (r² = 0.75) and there were no significant differences between the groups (Figure 4(b)). V_T/f_p had poorer test–retest variability, with a TRV and aTRV of 17% ± 2% and 21% ± 2%, respectively (Table 1). Regional average V_T/f_p ICC values were very similar to those of V_T, with a global mean ICC of 0.66 ± 0.09. There was a negative mean bias of 18% for the retest condition (Supplemental Figure 1(b)). V_T/f_p derived from 2TC had a similar mean aTRV of 22% ± 2% and mean ICC of 0.64 ± 0.08 (Supplemental Table 5).

DVR-1 values ranged from 0.51 in the brainstem to 1.95 in the putamen. Test and retest-derived DVR-1 values were well correlated (r² = 0.92), with Bland–Altman plots showing a consistent negative bias of ∼7% for the retest condition (Figure 4(c); Supplemental Figure 1(c)). DVR-1 had good reproducibility with a mean TRV of −9% ± 4% and aTRV of 11% ± 4%, and an ICC of 0.66 ± 0.37. 2TC-derived DVR-1 had similar test–retest reproducibility with a mean aTRV of 10% ± 4% and mean ICC of 0.69 ± 0.28 (Supplemental Table 5).

SUVR-1 computed over 70–90 min post injection was in good agreement with DVR-1 (n = 30, Figure 5(a), r² = 0.93), with SUVR-1 values ranging from 0.5 in the brainstem to 1.9 in the putamen. Test and retest-derived SUVR-1 values were well correlated (r² = 0.97), with Bland–Altman plots showing a consistent negative mean bias of ∼7% for the retest condition (Figure 5(b); Supplemental Figure 1(d)). SUVR-1 had global mean TRV of −8% ± 3% and aTRV of 9% ± 3%. The average ICC for SUVR-1 was 0.89 ± 0.05. Corresponding reproducibility and reliability results for DVR and SUVR are given in Supplemental Table 6 and Supplemental Figure 3.

Figure 5.

Correlation between regional (a) regional SUVR-1 and DVR-1 (n = 30). (b) SUVR-1 values derived under test and retest conditions. All regions and subjects are included in the plots (18 ROIs, 5 subjects).

To assess if SUVR was affected by radioligand delivery, the relationship between K₁ with DVR and SUVR, as well as between changes in K₁ estimates derived under test and retest conditions (ΔK₁) with changes in DVR (ΔDVR) and SUVR (ΔSUVR) were explored. As shown in Supplemental Figure 4, there was no obvious effect of K₁ on either outcome measure when looking at Δ relationships. In addition, the relationship between SUVR and K₁ was very similar to that of DVR and K₁ providing further confidence that the static SUVR measure was not affected by radioligand delivery.

Baseline and post-rotenone competition data from rhesus monkeys along with the associated occupancy plot analysis are displayed in Figure 6. A V_ND of 8.8 mL/cm³ was estimated as the common intercept of these occupancy data and the occupancies for monkeys 1, 2, 3 and 4 were estimated as 55.7%, 55.1%, 50.2% and 55.8%, respectively. The monkey to human scaling factor for baseline V_T values was calculated as 1.07 ± 0.14, resulting in an estimated V_ND^Human of 8.2 mL/cm³ which is similar to the mean V_T estimate in the human centrum semiovale of 10.9 ± 1.9 mL/cm³ (n = 30, Supplemental Table 1). The ${B P}_{N D}^{NHP}$ in cerebral white matter was calculated as 0.54, which is indicative of a small amount of specific binding in this region.

Figure 6.

Occupancy plots of monkeys 1–4, where V_ND is given by the x intercept (n = 4, 9 ROIs).

Discussion

We examined the test–retest reproducibility of the MC-I PET radioligand ¹⁸F-BCPP-EF and evaluated the suitability of the centrum semiovale as a pseudo reference region. We used blocking data acquired in the non-human primate to estimate the V_ND. The applicability of a simplified outcome measure SUVR for ¹⁸F-BCPP-EF quantification was also assessed.

In line with previous work in the non-human primate and our recent report on¹⁸F-BCPP-EF uptake in the human brain, the highest uptake was observed in striatal regions, followed by cortical areas and was lowest in the hippocampus and temporal regions.

We observed a fairly good test–retest variability of ¹⁸F-BCPP-EF V_T in all grey matter regions examined (global mean aTRV 13% ± 2%). Test–retest variability was higher in the centrum semiovale (aTRV 18% ± 6%) as compared to grey matter regions. Overall reliability was the best in the caudate (ICC = 0.91), partly as a result of the relatively high inter-subject variability (∼30%) in this region. ICC results were variable across regions and poor in the substantia nigra, globus pallidus and ventral striatum, consistent with the lower signal-to-noise arising from the small size of these regions.

The relative difference in test and retest f_p measurements was 9% arising from the high coefficient of variance (∼18%) in such a small group. Correcting V_T for f_p slightly increased the test–retest variability and decreased the reliability of the results due to the high within-subject variability between test and retest measurements of f_p. Correction by f_p becomes especially important for longitudinal studies, or cross-sectional studies where differences may arise in f_p between groups or following therapeutic interventions. The reproducibility of DVR-1 was comparable to V_T, though ICC was slightly more varied across regions. Both the variability and reliability of 2TC-derived kinetic outcome measures were similar to those derived using the MA1 model, which further justifies the use of both models for quantification of ¹⁸F-BCPP-EF.

SUVR-1 values were calculated over the 70–90-min window of 30 ¹⁸F-BCPP-EF scans and assessed as an outcome measure that would be more suited for wide-spread application in large scale studies. SUVR-1 results were highly correlated with DVR-1 and had the best reliability (mean ICC 0.89 ± 0.05) out of all outcome measures assessed. This highlights the potential of SUVR-1 as a non-invasive and simpler measure for ¹⁸F-BCPP-EF quantification that would allow for a shortened scan time and obviate the need for arterial blood acquisition and kinetic modelling.

To provide confidence in the use of the centrum semiovale as a pseudo reference region, and consequently the outcome measures DVR-1 and SUVR-1, we analysed previously acquired ¹⁸F-BCPP-EF rhesus monkey data involving competition with the MC-I inhibitor rotenone to derive a V_ND in monkeys and then translate it to human using an appropriate inter-species scaling factor. This provided an estimated V_ND in human cerebral white matter of 8.2 mL/cm^3. This estimate was ∼ 25% lower than the average human centrum semiovale V_T value of 10.9 ± 1.9 mL/cm³. Part of this difference is likely explained by the fact that the cerebral white matter region will be more affected by partial volume as compared to the centrum semiovale and so the specific signal in this region will be less.

We additionally compared our V_ND result to those estimated using the SIME method which derives a brain-wide estimate of V_ND by fitting TAC data for multiple regions simultaneously¹⁷ (see supplemental materials for detailed explanation of application of SIME method to dataset). SIME estimated a mean V_ND of 5.2 ± 1.3 mL/cm³ (n = 30) (range 3.2–8.8 mL/cm³) compared to our human V_ND estimate of 8.2 mL/cm³. SIME, by its nature, will have a tendency to underestimate V_ND because a 2TC model is typically required to describe data even in the absence of specific binding (thus the estimated V_ND (=K₁/k₂) will be less than the true V_ND (=K₁/k₂ (1+k₃/k₄)) for that region). This underestimation for SIME has been observed recently when comparing to direct blocking data in humans.¹⁸ Given all these factors, our opinion is that the V_ND derived from primate blocking data is likely to be the best estimate in the absence of human blocking data.

Overall, ¹⁸F-BCPP-EF showed acceptable test–retest reliability. V_ND derived from monkey data suggests that the centrum semiovale is a useful pseudo reference region with minimal specific binding, and lends support to the suitability of DVR-1 and SUVR-1 as outcome measures to quantify specific ¹⁸F-BCPP-EF signal in humans. It should be noted that the centrum semiovale is not a perfect reference region in the strictest sense of being completely devoid of MC-I. Further, the good test–retest reliability of a ligand alone is not sufficient evidence to obviate the need for full kinetic quantification. Therefore, careful consideration is required before eliminating arterial blood sampling in clinical studies, particularly in cross-sectional studies involving disease cohorts where there could be changes to mitochondrial function or MC-I expression in the white matter.

Although SUVR has practical advantages over the need for full dynamic acquisition, it is important to note that this can lead to bias in cross-sectional studies where clinical populations may have variations in reference region physiology, the amount of nonspecific binding and susceptibility to changes in global or regional blood flow.^19,20 Although our results show that that blood flow is not an issue for ¹⁸F-BCPP-EF, the use of SUVR-1 as an outcome measure requires further validation in clinical populations.

In conclusion, the current work showed that ¹⁸F-BCPP-EF V_T has suitable test–retest variability allowing it to be used in human PET studies of the living brain. Importantly, the centrum semiovale is a useful pseudo reference region for ¹⁸F-BCPP-EF in human studies, opening up the possibility of quantifying MC-I in settings where arterial blood acquisition is not possible. Whilst the SUVR-1 results are encouraging, their application should be considered carefully in relation to physiological differences in the conditions under investigation. ¹⁸F-BCPP-EF holds great promise for the investigation of impaired MC-I function.

Supplemental Material

sj-pdf-1-jcb-10.1177_0271678X20928149 - Supplemental material for Test–retest variability and reference region-based quantification of ¹⁸F-BCPP-EF for imaging mitochondrial complex I in the human brain

Supplemental material, sj-pdf-1-jcb-10.1177_0271678X20928149 for Test–retest variability and reference region-based quantification of ¹⁸F-BCPP-EF for imaging mitochondrial complex I in the human brain by Ayla Mansur, Eugenii A Rabiner, Hideo Tsukada, Robert A Comley, Yvonne Lewis, Mickael Huiban, Jan Passchier and Roger N Gunn in Journal of Cerebral Blood Flow & Metabolism

Footnotes

Acknowledgements

The authors thank Claire Power, Elbert Perez, Ryan Janisch, Daniela Ribeiro, Mari Lambrechts and Mark Tanner for their expert assistance. The MIND MAPS consortium consists of the following members: Laurent Martarello, Biogen; Robert A. Comley, AbbVie; Laigao Chen, Pfizer, Adam Schwarz, Takeda; Karl Schmidt, Celgene; Paul Matthews, Imperial College London; Marios Politis, King’s College London; Jonathan Rohrer, University College London; David Brooks, Newcastle University; James Rowe, University of Cambridge.

Authors’ contributions

AM conducted the experiments, analysed the data and wrote the manuscript. RG, ER and RC designed the study, aided in data interpretation and manuscript preparation. HT provided the rhesus data and aided in manuscript preparation. MH and JP oversaw radioligand production, YL oversaw the clinical conduct of the study and all contributed to manuscript preparation.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: The project was funded by the MIND MAPS consortium.

Declaration of conflicting interests

The author(s) declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: AM, ER YL, MH, JP and RG are employees of Invicro LLC; RC is an employee of AbbVie; RG is a consultant for AbbVie, Biogen & Cerveau. HT is an employee of Hamamatsu Photonics. No potential conflicts of interest relevant to this article exist.

ORCID iD

Ayla Mansur

Supplemental material

Supplemental material for this article is available online.

References

Sazanov

LA.

A giant molecular proton pump: structure and mechanism of respiratory complex I.

Nat Rev Mol Cell Biol 2015; 16: 375–388.

Schmitt

Grimm

Kazmierczak

, et al. Insights into mitochondrial dysfunction: aging, amyloid-β, and Tau – a deleterious trio. Antioxid Redox Signal 2012; 16: 1456–1466.

Grimm

Eckert

Brain aging and neurodegeneration: from a mitochondrial point of view. J Neurochem 2017; 143: 481–431.

Harada

Nishiyama

Kanazawa

, et al. Development of novel PET probes, [18F]BCPP-EF, [18F]BCPP-BF, and [11C]BCPP-EM for mitochondrial complex 1 imaging in the living brain. J Label Compd Radiopharm 2013; 56: 553–561.

Tsukada

Ohba

Kanazawa

, et al. Evaluation of 18F-BCPP-EF for mitochondrial complex 1 imaging in the brain of conscious monkeys using PET. Eur J Nucl Med Mol Imaging 2014; 41: 755–763.

Tsukada

Ohba

Nishiyama

, et al. PET imaging of ischemia-induced impairment of mitochondrial complex i function in monkey brain. J Cereb Blood Flow Metab 2014; 34: 708–714.

Tsukada

Kanazawa

Ohba

, et al. PET imaging of mitochondrial complex I with 18F-BCPP-EF in the brains of MPTP-treated monkeys. J Nucl Med 2016; 57: 950–953.

Mansur

Rabiner

Comley

, et al. Characterization of 3 PET tracers for quantification of mitochondrial and synaptic function in healthy human brain: 18 F-BCPP-EF, 11 C-SA-4503, 11 C-UCB-J. J Nucl Med 2020; 61: 96–103.

Harris

Attwell

The energetics of CNS white matter. J Neurosci 2012; 32: 356–371.

10.

Herman

Rothman

, et al. Evaluating the gray and white matter energy budgets of human brain function. J Cereb Blood Flow Metab 2018; 38: 1339–1353.

11.

Grabner

Janke

Budge

, et al. Symmetric atlasing and model based segmentation: an application to the hippocampus in older adults. Med Image Comput Comput Assist Interv 2006; 9(Pt 2): 58–66.

12.

Tziortzi

Searle

Tzimopoulou

, et al. Imaging dopamine receptors in humans with [11C]-(+)-PHNO: dissection of D3 signal and anatomy. Neuroimage 2011; 54: 264–277.

13.

Tzourio-Mazoyer

Landeau

Papathanassiou

, et al. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 2002; 15: 273–289.

14.

Finnema

Nabulsi

Eid

, et al. Imaging synaptic density in the living human brain. Sci Transl Med 2016; 8: 348ra96.

15.

Lassen

Bartenstein

Lammertsma

, et al.. Benzodiazepine receptor quantification in vivo in humans using [11C]flumazenil and PET: application of the steady-state principle. J Cereb Blood Flow Metab 1995; 15: 152–165.

16.

Cunningham

Rabiner

Slifstein

, et al. Measuring drug occupancy in the absence of a reference region: the Lassen plot re-visited. J Cereb Blood Flow Metab 2010; 30: 46–50.

17.

Todd Ogden

Zanderigo

Parsey

RV.

Estimation of in vivo nonspecific binding in positron emission tomography studies without requiring a reference region. Neuroimage 2015; 108: 234–242.

18.

Plavén-Sigray

Schain

Zanderigo

, et al. Accuracy and reliability of [11 C]PBR28 specific binding estimated without the use of a reference region. Neuroimage 2019; 188: 102–110.

19.

Barker

Ashby

Wellington

, et al. Pathophysiology of white matter perfusion in Alzheimer’s disease and vascular dementia. Brain 2014; 137(Pt 5): 1524–1532.

20.

Sander

Mandeville

Wey

, et al. Effects of flow changes on radiotracer binding: simultaneous measurement of neuroreceptor binding and cerebral blood flow modulation. J Cereb Blood Flow Metab 2019; 39: 131–146.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

0.00 MB

0.51 MB

Test–retest variability and reference region-based quantification of 18 F-BCPP-EF for imaging mitochondrial complex I in the human brain

Abstract

Keywords

Introduction

Materials and methods

Subjects

Radiotracer synthesis

Competition data from rhesus monkey

Data acquisition

Image processing

Quantitative analysis

Tissue kinetic modelling

Test–retest reproducibility

Quantification from simplified static acquisition

Estimation of non-displaceable 18F-BCPP-EF signal in humans

Results

Discussion

Supplemental Material

sj-pdf-1-jcb-10.1177_0271678X20928149 - Supplemental material for Test–retest variability and reference region-based quantification of 18F-BCPP-EF for imaging mitochondrial complex I in the human brain

Footnotes

Acknowledgements

Authors’ contributions

Funding

Declaration of conflicting interests

ORCID iD

Supplemental material

References

Supplementary Material

Estimation of non-displaceable ¹⁸F-BCPP-EF signal in humans

sj-pdf-1-jcb-10.1177_0271678X20928149 - Supplemental material for Test–retest variability and reference region-based quantification of ¹⁸F-BCPP-EF for imaging mitochondrial complex I in the human brain