Isoflurane and ketamine-xylazine modify pharmacokinetics of [¹⁸F]SynVesT-1 in the mouse brain

Abstract

We investigated the effect of isoflurane and ketamine-xylazine anesthesia on the positron emission tomography (PET) tracer [¹⁸F]SynVesT-1 in the mouse brain. [¹⁸F]SynVesT-1 PET scans were performed in C57BL/6J mice in five conditions: isoflurane anesthesia (ANISO), ketamine-xylazine (ANKX), awake freely moving (AW), awake followed by isoflurane administration (AW/ANISO) or followed by ketamine-xylazine (AW/ANKX) 20 min post tracer injection. ANISO, ANKX and AW scans were also performed in mice administered with levetiracetam (LEV, 200 mg/kg) to assess non-displaceable binding. Metabolite analysis was performed in ANISO, ANKX and AW mice. Finally, in vivo autoradiography in ANISO, ANKX and AW mice at 30 min post-injection was performed for validation. Kinetic modeling, with a metabolite corrected image derived input function, was performed to calculate total and non-displaceable volume of distribution (V_T(IDIF)). V_T(IDIF) was higher in ANISO compared to AW (p < 0.0001) while V_T(IDIF) in ANKX was lower compared with AW (p < 0.0001). Non-displaceable V_T(IDIF) was significantly different between ANISO and AW, but not between ANKX and AW. Change in the TAC washout was observed after administration of either isoflurane or ketamine-xylazine. Observed changes in tracer kinetics and volume of distribution might be explained by physiological changes due to anesthesia, as well as by induced cellular effects.

Introduction

The synaptic vesicle protein 2 A (SV2A) can be found on neuronal vesicles and is expressed in the entire brain. For this reason, SV2A serves as a biomarker for synaptic density. The positron emission tomography (PET) tracer [¹¹C]UCB-J targets SV2A and its uptake reflects synaptic density. ¹ Due to the short half-life of carbon-11, fluor-18 SV2A tracers have been developed to improve practicality. [¹⁸F]SynVesT-1 (known also as [¹⁸F]SDM-8, and [¹⁸F]MNI-1126) is one of these tracers. ² It has been demonstrated to serve as a proxy for central nervous system synaptic density and presents high specific binding and excellent test-retest reproducibility in humans. ³ SV2A tracers have been used for example in research on epilepsy ⁴ and Alzheimer’s disease. ⁵

[¹⁸F]SynVesT-1 can be used in preclinical PET research in animal models of neurological diseases where changes in synaptic density might be expected, such as in epilepsy, Alzheimer’s disease, ⁶ Parkinson’s disease, and Schizophrenia. ⁷ Translation of results from preclinical PET studies to human subjects can be compromised due to the additional use of anesthesia to perform preclinical PET scans.^8,9 For example, anesthesia has been shown to change the uptake of the glucose analog [¹⁸F]FDG, as well as the binding of tracers targeting neurotransmitter receptors. ⁹ To circumvent the use of anesthesia in brain PET, motion correction methods have been developed to scan awake rodents. ⁹ Particularly for mouse brain PET scans in freely moving animals, point source tracking with motion correction can be performed. ¹⁰

In this study, we investigated the effect of isoflurane and ketamine-xylazine, two common anesthetics used in preclinical PET, on the binding of [¹⁸F]SynVesT-1 in the mouse brain. Radiometabolites in plasma were measured in isoflurane and ketamine-xylazine anesthetized, as well as in awake mice, to obtain an image derived input function corrected with a population based metabolite correction in the 3 conditions. Dynamic scans were performed in isoflurane, ketamine-xylazine anesthetized, and awake mice, and kinetic modeling was performed to estimate the regional brain volume of distribution based on the metabolite corrected image-derived input function (V_T(IDIF)). The non-displaceable binding was also calculated in anesthetized and awake mice using levetiracetam (LEV) to block the binding site. Finally, the timing to observe the effect of anesthesia on tracer washout was also investigated, and autoradiography was performed to validate the PET results.

Methods

Animals

For isoflurane followed by awake [¹⁸F]SynVesT-1 PET experiments, 16 female C57BL/6J (8 weeks old) mice (Charles River, Lyon France) were divided into 2 groups (Supplemental Figure 1). For ketamine-xylazine followed by awake [¹⁸F]SynVesT-1 PET experiments 18 female mice (8 weeks old) were divided in 2 groups (Supplemental Figure 1). For radiometabolite analysis 36 mice (16 weeks old) were considered, from which 9 mice also underwent ex vivo brain [¹⁸F]SynVesT-1 autoradiography. All animals were housed in a temperature-controlled room with a 12 h light-dark cycle (food and water available ad libitum). The experiments followed the European Ethics Committee recommendations (Decree 2010/63/CEE) and are reported in compliance with the ARRIVE guidelines. Experiments were approved by the Animal Experimental Ethical Committee of the University of Antwerp, Antwerp, Belgium (ECD 2020-71 and 2022-39).

PET scans

[¹⁸F]SynVesT-1 radiosynthesis was performed as previously described. ¹¹ Group 1 (N = 10) underwent three 2-hour dynamic [¹⁸F]SynVesT-1 scan sessions immediately following i.v. tail vein tracer injection under the following conditions: isoflurane anesthesia (ANISO, 3% for induction, 1.5% for maintenance, injected dose 11.9 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.5 MBq, injected mass 5.2 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.8 nmol/kg, body weight 18.8 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.85 g), awake (AW, 8.79 $\hspace{0.17em}\pm \hspace{0.17em}$ 4.32 MBq, 5.49 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.74 nmol/kg, 18.8 $\hspace{0.17em}\pm \hspace{0.17em}$ 1 g), and awake followed by anesthesia induction at 20 min post tracer injection (AW/ANISO, 7.91 $\hspace{0.17em}\pm \hspace{0.17em}$ 4.02 MBq, 5.26 $\hspace{0.17em}\pm$ 0.68 nmol/kg, 19.4 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.1 g). A minimum period of 2 days was considered between each animal scan. The [¹⁸F]SynVesT-1 tracer injection in awake mice was performed outside the scanner, rapidly placing the mouse inside the scanner after injection, effectively losing the initial 20 to 40 seconds of the tracer kinetics. Group 2 (N = 6) underwent 2 dynamic scans sessions after 200 mg/kg levetiracetam (Sigma-Aldrich, UK) intraperitoneal administration (full binding site blocking) ¹¹ 30 min before the start of the scan under the following conditions: ANISO (15.5 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.65 MBq, 4.89 $\hspace{0.17em}\pm$ 0.67 nmol/kg, 20.5 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.1 g), and AW (10.5 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.3 MBq, 6.3 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.2 nmol/kg, 21.0 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.52 g).

Group 3 (N = 10) underwent three 1-hour dynamic [¹⁸F]SynVesT-1 scan sessions immediately following i.v. tail vein tracer injection under the following conditions: ketamine-xylazine anesthesia (ANKX, ketamine 150 mg/kg, xylazine 15 mg/kg, injected dose 9.15 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.5 MBq, injected mass 6.4 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.23 nmol/kg, body weight 18.4 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.93 g), awake (AW, 9.84 $\hspace{0.17em}\pm \hspace{0.17em}$ 3.42 MBq, 6.73 $\hspace{0.17em}\pm$ 0.12 nmol/kg, 18.4 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.92 g), and awake followed by anesthesia induction at 20 min post tracer injection (AW/ANKX, 12.35 $\hspace{0.17em}\pm \hspace{0.17em}$ 3.45 MBq, 6.78 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.09 nmol/kg, 19.8 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.98 g). Group 4 (N = 8) followed ANKX (4.61 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.55 MBq, 6.45 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.16 nmol/kg, 18.5 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.4 g), and AW (5.94 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.2 MBq, 6.73 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.15 nmol/kg, 19.4 $\hspace{0.17em}\pm$ 1.6 g) scans, but with levetiracetam administration (200 mg/kg) 30 min before tracer injection.

A significant difference in injected dose was found between AN and AW but not in injected mass. For each of the 6 experimental protocols, the animals were randomly divided into 2 subgroups to allow cross-design in the anesthesia and awake scans sessions order (Supplemental Figure 1). Supplemental Figure 2 shows a diagram of the different scanning protocols.

All awake scans, and awake scans in combination with anesthesia, were performed using the point source tracking method, maintaining the mice inside the scanner (Inveon, Siemens Medical Solutions, Inc., Knoxville, USA) field of view within a plastic holder with a platform of 9  $\times \hspace{0.17em}$ 10 cm ^{10 , . 12} Seventy minutes (AW, AW/ANISO, AW/ANKX) before tracer administration mice were anesthetized with isoflurane (3% for induction, 1.5% for maintenance) to attach 4 point sources on the head as previously described. ¹⁰ In AW, AW/ANISO, and AW/ANKX, isoflurane was stopped after point source fixing to allow a 60 min recovery for anesthesia washout before tracer administration.

Radiometabolites analysis

Thirty-six mice were divided in 3 conditions: isoflurane (3% for induction, 1.5% for maintenance, administered 30 min before tracer injection, 7.32 $\hspace{0.17em}\pm \hspace{0.17em}$ 3.4 MBq, 21.8 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.2 g), ketamine-xylazine (ketamine 150 mg/kg, xylazine 15 mg/kg, administered 30 min before tracer injection, 7.25 $\hspace{0.17em}\pm \hspace{0.17em}$ 3.2 MBq, 22.1 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.96 g) anesthetized, and awake (7.03 $\hspace{0.17em}\pm \hspace{0.17em}$ 3.24 MBq, 22.1 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.5 g). For each condition, plasma tracer metabolites were measured at 4 time points (2, 10, 30 and 60 min), using 3 mice per time point. The procedure for radiometabolites processing was performed as previously described. ¹¹ Briefly, a reverse-phase (RP)-HPLC system equipped with a Waters Xbridge C18 4.6 × 150 mm, 5 µm column, and Phenomenex security guard pre-column, was equilibrated with NaOAc 0.05 M pH 5.5 and acetonitrile. RP-HPLC fractions were collected at 0.5 min intervals for 10 min and radioactivity was measured in a gamma counter. The radioactivity was expressed as a percentage of the total area of the peaks based on the radiochromatograms. A sigmoid fit was calculated on the data points to obtain the population based percentage of intact tracer in the 3 different conditions.

Image reconstruction

Anesthesia PET images were reconstructed with in-house developed list-mode OSEM reconstruction, ¹³ considering a spatially variant resolution, with 16 subsets and 16 iterations. Attenuation correction was performed using the attenuation map calculated from a CT scan. Awake PET images were reconstructed with list-mode OSEM motion correction reconstruction, with motion-dependent spatially variant resolution, ¹⁴ with 16 subsets and 16 iterations. The awake attenuation map was calculated from the body outline activity ¹⁵ assigning a linear attenuation coefficient of soft tissue (0.096 cm⁻¹) to the entire body. Images had a size of 128 × 128 × 159 voxels (0.776 × 0.776 ×0.796 mm) in the $x$ , $y$ , and $z$ dimensions, respectively. Dynamic scans were reconstructed with framing of 12 frames × 10 sec, 3 × 20 s, 3 × 30 s, 3 × 60 s, 3 × 150 s, and 9 × 300 s.

Kinetic modelling

Using the integral image of the entire scan, the brain was first non-rigidly aligned to a previously calculated mouse brain [¹⁸F]SynVesT-1 template, where different brain regions are delineated. ¹¹ This transformation was then applied to the dynamic images, and the Cortex (CTX), caudate putamen (CP), thalamus (TH), hippocampus (HC), and cerebellum (CB) time activity curves (TACs) were extracted. For scans with LEV blocking where brain activity is minimal after a few minutes, it was observed that brain structures were visible in the initial 4 minutes of the scan. Therefore, a separate template considering only the initial 4 minutes of brain activity after tracer injection was used to perform the non-rigid alignment in LEV blocking scans. Image processing was performed with PMOD 3.6 (PMOD Technologies Ltd, Zurich, Switzerland).

Kinetic modeling was performed by measuring the image-derived input function (IDIF) in anesthesia scans as a proxy for the real input function. ¹ The whole blood TAC is measured in the heart VOI and is corrected for metabolites and plasma-whole blood fraction using a population-based correction factor. ¹¹ The IDIF is used in one (1TCM) and two-tissue (2TCM) compartmental model, as well as the Logan plot to calculate V_T(IDIF). In scans with LEV blocking, only Logan plot quantification is performed due to the poor fit of compartment models. Voxel-wise V_T(IDIF) maps of the brain are calculated with the Logan plot.

For kinetic modeling in awake scans, a scaled version of the IDIF calculated in the anesthesia scan of the same mouse is used. IDIF scaling is based on the injected activity. First, a linear relation between the IDIF integral, uncorrected for metabolites, and the injected activity in the anesthesia scans is determined. A linear relation was calculated separately for ANISO and ANKX scans (Supplemental Figure 4). From the awake scan injected activity together with the determined linear relation, the corresponding IDIF integral value is calculated and the anesthesia IDIF is scaled accordingly. Finally, the scaled IDIF is corrected for metabolites and plasma-whole blood fraction measured in the awake state. Kinetic modeling non-linear fit was performed in Matlab (The Mathworks, Inc. Natick, United States).

Considering animals of every group separately, average $\hspace{0.17em}\pm \hspace{0.17em}$ SD standardized uptake value (SUV) TACs were calculated for all respective conditions in each group (ANISO, ANKX, AW/ANISO, AW/ANKX, AW) and plotted together to qualitatively look for differences in tracer wash in/wash out.

Autoradiography

[¹⁸F]SynVesT-1 ex vivo autoradiography (ARG) was performed to quantify brain uptake at 30 min post-injection. Injected dose in the ARG experiment was reduced (compared to PET scan doses) to avoid saturation of the ARG imaging plate. Brain ARG was performed as previously described. ¹⁶ Briefly, isoflurane anesthetized (2.17 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.06 MBq), ketamine-xylazine anesthetized (2.15 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.08 MBq), and awake (2.06 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.04 MBq) animals were administered with [¹⁸F]SynVesT-1 (tail vein injection). Thirty minutes after tracer administration mice were sacrificed and brains were removed and snap-frozen (−35 $°C$ , 2 min). For each animal, 18 sagittal sections (20  $\mu$ m thickness) were collected with Paxinos and Watson coordinates (4th edition, 2013) as reference (1.56 mm lateral). A range of tracer dilutions was prepared to calculate the standard curve to transform from ARG gray values to kBq/µL. The air-dried brain slices, together with the tracer dilutions, were exposed on a phosphor imaging plate (BAS-IP MS2040 E, Fujifilm, Japan) in a light impermeable cassette (Hypercassette, Amersham Biosciences, UK) for 10 minutes. The plate was then scanned with a plate reader (Typhoon FLA7000, pixel size 25 µm, GE Healthcare, USA).

CTX, CP, TH, HC, and CB regions were manually delineated in all brain slices and the average gray value per region is calculated. The gray value was then converted to SUV using the standard curve.

Statistical analysis

In mice scans from group 1 and 3, a comparison between 1TCM and 2TCM fits was assessed by calculating the Akaike information criterion (AIC) weights model likelihood: ¹⁷

$$\mathit{AIC}=n\mathrm{\hspace{0.17em}}\mathit{log}\left(\frac{\mathit{RSS}}{n}\right)+2p$$

where $n$ is the number of frames, RSS is the residual sum of squares and $p$ is the number of parameters (4 for 2TCM, 2 for 1TCM). The difference between each model $\mathit{AIC}$ and the minimum $\mathit{AIC}$ considering all models ( $M$  = 2 in our case) is calculated as ${\Delta }_i={\mathit{AIC}}_i-{\mathit{AIC}}_{\mathit{min}}$ , to finally calculate the model likelihood $w_i$ :

$$w_i=\frac{\exp ({-\Delta }_i/2)}{{\sum }_{j=1}^M\exp ({-\Delta }_j/2)}$$

This metric compares different models fitting the data, calculating a value closer to 1 to the model(s) that best fit the data. In addition, V_T(IDIF) values calculated with 1TCM and 2TCM were compared using Bland-Altman plot analysis bias and standard deviation.

Statistical analysis for anesthesia and awake V_T(IDIF) in the different regions, as well as for ARG SUV, was achieved with a two-way ANOVA analysis. 2TCM K₁ values were compared between ANISO, ANKX and AW, pooling K₁ values from AW mice in group 1 and group 3. Only K₁ values with a standard error (SE) lower than 10% were considered. Non-displaceable binding potential (BP_ND) was calculated as the ratio k3/k4 from 2TCM fits, considering only ratios with an SE lower than 15%. Since for all AW mice k3/k4 ratios had an SE higher than 15%, the AW condition was not included in the BP_ND comparison. Statistical analysis was performed using GraphPad Prism 6.0 (GraphPad Software, California, USA).

Results

Metabolites analysis

Percentage of intact tracer for isoflurane, and ketamine-xylazine anesthetized mice, as well as for awake mice, is shown in Supplemental Figure 3. Percentage of intact tracer at 1 hour is higher for ketamine-xylazine (24.8 $\hspace{0.17em}\pm \hspace{0.17em}$ 3.4%) anesthetized compared with isoflurane (9.43 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.5%, p < 0.01), and awake (7.11 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.67%, p < 0.001) mice during the first hour after tracer injection. Metabolism of tracer in awake mice is similar (p = 0.072) to that in isoflurane anesthetized mice.

Validation of awake kinetic modeling

Supplemental Figure 4a shows a high correlation (r² = 0.937) between injected dose and anesthesia IDIF integral linear relation. The IDIF’s used for kinetic modeling are shown in Supplemental figure 4c. The bias between 1TCM and 2TCM V_T(IDIF) in ANISO can be observed in the Bland-Altman plot (Supplemental figure 5) (7.63%), which is lower for the corresponding AW case (2.63%). The ANKX bias (13.5%) is higher than ANISO, while AW bias is lower (2.34%). Supplemental Table 1 shows ANISO model likelihood of zero for 1TCM in all regions, with a model likelihood of one for 2TCM. For the awake condition TH, and HC have a larger 1TCM model likelihood (0.559, in both cases) while CTX, CP and CB have a larger 2TCM likelihood (0.618, 0.653 and 0.694, respectively). For ANKX, all regions have a model likelihood of zero for 1TCM and a model likelihood value of 1 for 2TCM. In the corresponding awake case all regions have a larger 1TCM model likelihood than for 2TCM.

Brain regional quantification

Volume of distribution V_T(IDIF)

The brain regional V_T(IDIF) in anesthetized and awake animals, without and with LEV administration is shown in Figure 1. V_T(IDIF) in the ANISO condition is significantly (p < 0.0001) higher than in awake mice for all regions (Table 1). On average, V_T(IDIF) in the ANISO vs AW condition is 27.6.1 $\hspace{0.17em}\pm \hspace{0.17em}$ 5.0 vs 18.3 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.8 mL/cm³, and 4.19 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.46 vs 3.25 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.65 mL/cm³ (F_(1,45) = 999, p < 0.0001, F_(1,25) =32.0, p < 0.0001), in animals without and with LEV blocking, respectively. In ANKX vs AW, V_T(IDIF) is 10.4 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.18 vs 14.4 $\hspace{0.17em}\pm \hspace{0.17em}$ 3.0 mL/cm³, and 2.51 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.34 vs 2.38 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.87 mL/cm³ (F_(1,45) = 83.0, p < 0.0001, F_(1,25) =0.381, not significant), in animals without and with LEV blocking, respectively. Largest regional V_T(IDIF) is found in TH, and lowest in CB for all conditions.

Figure 1.

Regional [¹⁸F]SynVesT-1 volume of distribution (V_T(IDIF)) in mice from (a) group 1 (ANISO, AW), (b) group 2 (ANISO, AW, with LEV blocking 200 mg/kg), (c) group 3 (ANKX, AW) and (d) group 4 (ANKX,AW, with LEV blocking 200 mg/kg), calculated with Logan plot. Plots y-axis scale is different.

Table 1.

Regional [¹⁸F]SynVesT-1 V_T(IDIF) values were calculated with Logan plot in mice from group 1 (without LEV) and group 2 (with LEV administration).

	VT(IDIF) (mL/cm3) Logan plot		Difference (mL/cm3)	VT(IDIF) (mL/cm3) Logan plot LEV (200 mg/kg)		Difference (mL/cm3)
	Isoflurane	Awake		Isoflurane	Awake
CTX	28.9 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.61	18.4 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.51	10.5****	4.13 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.34	3.48 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.742	0.65
CP	24.2 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.67	17.2 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.29	7.0****	3.95 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.40	3.49 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.62	0.46
TH	33.9 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.00	21.3 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.67	12.6****	4.55 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.51	3.30 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.87	1.25*
HC	30.3 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.60	20.0 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.59	10.3****	4.06 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.43	3.20 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.72	0.86
CB	20.6 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.38	14.4 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.05	6.2****	4.25 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.56	2.79 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.55	1.46**

p**** < 0.0001, p*** < 0.001, p** < 0.01, p* < 0.05.

Table 2.

Regional [¹⁸F]SynVesT-1 V_T(IDIF) values were calculated with Logan plot in mice from group 3 (without LEV) and group 4 (with LEV administration).

	VT(IDIF) (mL/cm3) Logan plot		Difference (mL/cm3)	VT(IDIF) (mL/cm3) Logan plot LEV (200 mg/kg)		Difference (mL/cm3)
	Ket-xyl	Awake		Ket-xyl	Awake
CTX	10.3 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.27	14.5 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.54	−4.2***	2.44 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.34	2.42 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.59	0.02
CP	9.23 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.21	13.5 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.10	−4.27***	2.43 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.29	2.37 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.60	0.06
TH	12.6 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.92	16.6 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.86	−4**	2.81 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.35	2.25 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.67	0.56
HC	11.7 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.49	16.0 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.70	−4.3***	2.49 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.32	2.24 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.64	0.25
CB	7.87 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.95	11.4 $\hspace{0.17em}\pm \hspace{0.17em}$ 2.06	−3.53**	2.34 $\hspace{0.17em}\pm \hspace{0.17em}$ 0.32	2.61 $\hspace{0.17em}\pm \hspace{0.17em}$ 1.63	−0.27

p*** < 0.001, p** < 0.01.

Average brain parametric V_T(IDIF) maps are shown in Figures 2 and 3, for ANISO vs AW, and ANKX vs AW, respectively, without and with LEV blocking. ANISO shows the largest overall brain V_T(IDIF), while AW and ANKX have similar V_T(IDIF), but a significant difference between AW and ANKX V_T(IDIF) is also present. Relative (to the whole brain) V_T(IDIF) images show, compared to awake, increased volume of distribution in the medulla for ANISO and medulla and cortex for ANKX (red arrows).

Figure 2.

Average brain [¹⁸F]SynVesT-1 V_T(IDIF) maps in isoflurane anesthetized and awake mice without (a group 1) and with (b, group 2) levetiracetam (LEV) blocking (200 mg/kg). (c) Relative (to the whole brain) V_T(IDIF) in mice without LEV blocking. The red arrow points to medulla region and (d) Magnetic resonance for anatomical reference.

Figure 3.

Average brain [¹⁸F]SynVesT-1 V_T(IDIF) maps in ketamine-xylazine anesthetized and awake mice without (a group 3) and with (b, group 4) levetiracetam (LEV) blocking (200 mg/kg). (c) Relative (to the whole brain) V_T(IDIF) in mice without LEV blocking. Red arrows points to cortex and medulla and (d) Magnetic resonance for anatomical reference.

2TCM K₁ exchange rate

Supplemental Figure 6a shows 2TCM K₁ for ANISO, ANKX and AW mice. The average brain regional K₁ is 2.12, 1.13 and 1.28 ml · cm⁻³ · min⁻¹, for ANISO, ANKX and AW, respectively. For all regions, there is a significant difference between ANISO vs ANKX, and ANISO vs AW, but not between ANKX and AW. Largest K₁ is found in TH (2.90, 1.35, and 1.54 ml·cm⁻³·min⁻¹, for ANISO, ANKX and AW, respectively), and lowest in CB (1.82, 1.04, and 1.11 ml·cm⁻³·min⁻¹, respectively) for all conditions. There is a high correlation between K₁ and V_T(IDIF) considering all regions in all conditions (Pearson’s r = 0.922, supplemental figure 6 b).

Non-displaceable binding potential BP_ND

Supplemental Figure 6c shows the BP_ND (2TCM k3/k4) for ANISO and ANKX. Largest BP_ND is found in HC (3.28) and CTX (2.02) for ANISO and ANKX, respectively, while lowest is found in TH (2.09) and CB (1.42).

Anesthesia effect on tracer wash-out

Figure 4(a) shows the mean SUV TACs for mice in ANISO, AW, and AW/ANISO conditions. ANISO shows higher SUV magnitude than AW TACs during all the scan. In AW/ANISO TACs, during the initial 20 min (awake state) TACs show the same magnitude as AW TACs. Immediately after isoflurane administration at 20 min post-injection, wash-out becomes slower compared to AW TACs. After 2 hours AW/ANISO TACs magnitude is higher than ANISO TACs magnitude. Figure 4(b) shows the equivalent TACs for ketamine-xylazine anesthesia. Compared to AW TACs, washout in ANKX TACs is slower, but magnitude is lower in the initial 20 to 30 min, and higher afterwards. In AW/ANKX TACs, magnitude is the same as in AW TACs during the initial 20 min. After ketamine-xylazine administration at 20 min post-injection, wash-out becomes faster than in AW TACs.

Figure 4.

(a) Mean regional SUV time activity curves for ANISO, AW, and AW/ANISO conditions during 2 hours scans. (b) Mean regional SUV time activity curves for ANKX, AW, and AW/ANKX conditions during 1 hour scans. Time axis is different in column (a) and (b). CTX: cortex; CP: caudate putamen; TH: thalamus; HC: hippocampus; CB: cerebellum.

Autoradiography

Figure 5(a) shows the regional quantification after 30 min tracer uptake in the autoradiography brain images.On average for all regions, uptake in isoflurane anesthetized mice is higher than in ketamine-xylazine (38% higher), and awake (50% higher) mice. Compared to PET brain quantification at 30 min tracer post-injection (Figure 5(b)) isoflurane anesthetized mice have 29%, and 43% higher uptake than in ketamine-xylazine and awake mice, respectively. There is a high correlation (Pearson’s r = 0.912) between regional brain quantification in ARG and PET (Figure 5(c)).

Figure 5.

(a) Autoradiography regional brain quantification in animals under isoflurane, and ketamine-xylazine anesthesia, as well as in awake animals, in animals sacrificed 30 min tracer post-injection. (b) Regional brain quantification in brain PET images in ANISO, ANKX, and AW conditions, at 30 min tracer post-injection. (c) Correlation, and regression line, between brain mean regional quantification values (every point is a region), in ARG and PET at 30 min post-tracer injection, considering animals in the 3 conditions and (d) Example of brain autoradiography image and the different regions used for quantification.

Discussion

In this study we compared the brain regional quantification of [¹⁸F]SynVesT-1, in isoflurane, ketamine-xylazine anesthetized, and awake mice. Main findings include: i) V_T(IDIF) is significantly higher in isoflurane compared to ketamine-xylazine anesthetized, and awake mice, while V_T(IDIF) is significantly lower in ketamine-xylazine anesthetized compared with awake mice. ii) The non-displaceable V_T(IDIF) is significantly higher in isoflurane compared with awake mice, but not significantly different between ketamine-xylazine and awake mice. iii) Change in [¹⁸F]SynVesT-1 uptake can be observed immediately after isoflurane or ketamine-xylazine administration, compared with uptake in the awake state.

Metabolite profile of the tracer was slower in ketamine-xylazine anesthetized mice compared with isoflurane anesthetized mice. On average during the first hour after tracer injection, there was 20% more intact tracer in ketamine xylazine, compared with isoflurane anesthetized mice. These 2 anesthetics produced different changes in the animal physiology. For example, ketamine/xylazine reduces heart rate, ¹⁸ while isoflurane depresses respiration. ¹⁹ In addition, the pharmacodynamic profile of the anesthetic might change due to the different route of administration (intraperitoneal injection for ketamine-xylazine, and inhalation for isoflurane). These differences might be involved in the different metabolic profile of the tracer. Overall, awake mice present more similar metabolic profile to isoflurane anesthetized mice compared with ketamine-xylazine metabolism. Although there was a higher amount of intact tracer in plasma in ketamine-xylazine compared with isoflurane anesthetized mice, brain uptake (SUV) was larger in isoflurane compared with ketamine-xylazine mice. Moreover, the higher availability of intact tracer over time might explain the more stable TAC brain profile in ketamine-xylazine anesthetized mice, compared with isoflurane anesthetized and awake mice. Since an increase in cerebral blood flow has been reported in isoflurane anesthetized mice compared with ketamine-xylazine anesthetized and awake mice,^20,21 the higher brain uptake in isoflurane anesthetized mice compared to the other conditions might be explained by this increase. Kinetic parameters might also be affected as discussed below.

For both ANISO and ANKX mice, the Akaike weights model likelihood showed that 2TCM describe better the kinetics of the tracer, compared with 1TCM. On the other hand, for AW mice the 1TCM model likelihood was larger than 0.5 or close to 0.5 for all brain regions. The Bland-Altman bias also suggests difference between 2TCM and 1TCM values for ANISO and ANKW, while bias between 2TC and 1TCM is lower for AW mice.

In human brain scans, [¹⁸F]SynVesT-1 has been reported to be well described using 1TCM² whereas our study suggest 1TCM can only be used in the awake state. Brain tissue tracer exchange therefore appears to change due to the administration of isoflurane or ketamine-xylazine anesthesia in mice. The rapid equilibrium between compartments tracer exchange, necessary to simplify the tracer kinetic model from several compartments to a reduced number of compartments, might be altered by the administration of anesthesia.

The kinetics of [¹¹C]UCB-J (the most applied SV2A tracer until now) in the mouse brain under anesthesia are also well described with a 1TCM¹, in contrast to 2TCM for [¹⁸F]SynVesT-1 using isoflurane ¹¹ or ketamine-xylazine in our study. Although [¹⁸F]SynVesT-1 and [¹¹C]UCB-J have the same target, the molecular structure of the tracers is different, ² possibly causing different interaction with the anesthetics. Moreover, difference in kinetics under anesthesia between [¹⁸F]SynVesT-1 and [¹¹C]UCB-J makes a comparison between studies performed with these tracers difficult since different assumptions are considered for each model (e.g. negligible non-displaceable to specific binding compartment exchange rate for 1TCM). In addition, microparameters from 1TCM and 2TCM might not be comparable due to the usually larger standard error of 2TCM microparameters compared to 1TCM¹.

The total volume of distribution was estimated using a surrogate image-derived input function.^1,11 Although the IDIF could not be measured in awake animals, a good correlation between the IDIF integral and the injected dose was observed in anesthetized animals (r² = 0.937). Moreover, after scaling IDIFs by their integral, there was good similarity in magnitude and shape between all mice IDIFs. Considering SUV values, peak and tails activity in ANKX IDIFs was slightly higher compared with ANISO IDIFs. This pattern was also translated to the scaled IDIFs of AW mice from group 1 (isoflurane group) and group 3 (ketamine-xylazine group), i.e. scaled IDIFs in AW mice from group 3 had slightly higher activity than scaled IDIFs in AW mice from group 1. This could explain the higher V_T(IDIF) values in AW mice from group 1 compared with AW mice group 3. Since plasma to whole blood ratios and metabolites correction specific for awake mice were applied to the scaled IDIFs in both groups, the differences in awake scaled IDIFs between group 1 and 3, and therefore in V_T(IDIF) values, are due to relative differences in whole blood IDIFs between ANISO and ANKX mice. From whole blood samples in the metabolite study, we observed a minimal difference in activity between conditions (ANISO vs AW difference: 4%, ANKX vs AW difference: 7%). Nevertheless, these differences might increase when an IDIF is considered due to contamination from surrounding tissue (e.g. myocardium, right ventricle, lungs etc.). This is a well know issue when IDIFs are used. ²² Investigation of these differences in [¹⁸F]SynVesT-1 IDIF activity are currently being investigated.

Significantly higher V_T(IDIF) was found in ANISO compared with AW mice, while V_T(IDIF) in ANKX was significantly lower than in AW mice. Since V_T is proportional to the 2TCM K₁ parameter, which in turn is proportional to blood flow ²³ and tracer extraction, we investigated if changes in cerebral blood flow caused by isoflurane and ketamine-xylazine anesthetsia^20,21 were correlated with changes in V_T. A significantly high correlation between V_T(IDIF) and K₁ for the conditions under anesthesia and in the awake state suggests that the total volume of distribution is affected by the change of cerebral blood flow caused by the anesthesia. Since the percentage of intact tracer in plasma is similar between AW and ANISO mice, but higher V_T(IDIF) is present in ANISO mice, the higher perfusion to the brain under isoflurane anesthesia cannot be explained by difference in available tracer. Therefore, increase in cerebral blood flow might explain the increase in brain V_T(IDIF) in ANISO compared with AW.

The overall brain distribution of [¹⁸F]SynVesT-1 is similar in ANISO, ANKX and AW mice, as seen in the relative V_T(IDIF) parametric maps. This normalization also allows to observe relative brain uptake changes, with less influence of change in tracer delivery (e.g. due to change in cerebral blood flow). In the relative V_T(IDIF) parametric maps, uptake in the medulla region was lower in AW compared with both ANKX and ANISO. The medulla is involved in autonomic functions, such as cardiovascular control and respiration. ²⁴ Considering the possibility that binding of the tracer is dependent on the functional state of the synapse, ²⁵ the difference in relative medulla V_T(IDIF) between AW and ANISO or ANKX might be related to a change in the medulla functional state due to anesthetics. As previously mentioned, isoflurane and ketamine-xylazine change the cardiovascular function and/or respiratory rate. Nevertheless, more studies are needed to confirm/reject this hypothesis.

There was a trend for higher BP_ND in isoflurane compared with ketamine-xylazine anesthetized mice, but only in HC there was a significant difference. Similarly for [¹¹C]UCB-J in humans, BP_ND was found to be a stable parameter unaffected by changes in cerebral blood flow. ²⁶ Although we considered only BP_ND values with a standard error lower than 15% as a compromise to increase the sample size, results in BP_ND values need to be taken carefully.

Non-displaceable V_T(IDIF) was not significantly different between ANKX and AW mice, and only significantly different in thalamus and cerebellum between ANISO and AW mice. Similar to the total V_T(IDIF), the increase in cerebral blood flow under isoflurane anesthesia might have caused the differences in non-displaceable V_T(IDIF) as well. Overall, the difference between non-displaceable V_T(IDIF) in anesthetized and awake mice was small, suggesting that the significant differences observed in V_T(IDIF) are mainly due to specific binding.

Autoradiography SUV served to validate uptake in PET images, eliminating any possible difference due to variation in image spatial resolution between awake and under anesthesia scans. Autoradiography SUV at 30 minutes post-injection was significantly lower in awake compared to isoflurane anesthetized animals, but not significantly different between ketamine-xylazine and awake mice. These trends are also observed in the PET SUV values, but with significant differences in some regions between ketamine-xylazine and awake mice. Nevertheless, there was a high correlation between autoradiography and PET SUV values considering all conditions. Possible methodological differences between autoradiography and PET, as well as differences in noise (larger in ARG), could explain the small differences.

The effect of isoflurane and ketamine-xylazine on tracer wash-out was also investigated by administering anesthesia after tracer injection. Since ketamine-xylazine maintains the anesthesia effect during approximately 90 min, its effect was investigated during 1 hour scans, but longer maintenance of anesthesia using isoflurane allowed to perform 2 hour scans. Tracer washout was decreased in animals administered with isoflurane at 20 min after tracer injection compared with awake animals, and activity concentration at 2 hours was higher in scans with isoflurane administered at 20 min after scan start than in scans starting with animals under anesthesia. This could be explained by the fact that animals anesthetized at 20 min after scan start receive a higher induction isoflurane dose (3%) followed by a maintenance dose (1.5%), while animals already starting under anesthesia were kept with a maintenance dose (1.5%) at the start of injection. Therefore, we could expect the effect to be isoflurane dose-dependent.

For animals administered with ketamine-xylazine at 20 min after scan start, tracer washout was increased compared with awake animals, but brain uptake was similar at 1 hour tracer post-injection. Unlike the isoflurane uptake profile showing a similar peak time to that in awake scans (7-9 min tracer post-injection) followed by similar tracer washout profile, ketamine-xylazine mice show a faster peak time (2–3 min post-injection) followed by a slower washout compared with isoflurane and awake mice. As mentioned above, differences in tracer metabolism as well as changes in physiological parameters might have caused these differences.

Overall, a clear effect of isoflurane and ketamine-xylazine anesthesia on [¹⁸F]SynVesT-1 brain uptake was observed. In addition to physiological effects, such as an increase in cerebral blood flow,^20,21,27 previously reported effects of isoflurane and ketamine-xylazine at the cellular level might explain some of these differences. For example, it has been reported that isoflurane anesthesia increases the blood-brain barrier (BBB) permeability in a dose-dependent manner by changing the expression of tight junction proteins.^28–30 Permeability surface area, and therefore tracer extraction, can be affected by these changes in BBB function. This effectively affects the diffusion of the tracer into the brain tissue, or in our case, of the PET tracer, observed as increased brain tracer activity concentration immediately after isoflurane administration. Moreover, isoflurane increases cell membrane fluidity, ³¹ which in turn can affect membrane permeability.^32,33 The exchange of tracer in the brain tissue might be affected by this change in cell membrane fluidity, affecting the kinetics of tracer uptake under anesthesia. Ketamine has been shown to also affect the BBB tight junctions function, ³⁴ and to be a substrate of blood-brain barrier transporters,^35,36 which can also affect the delivery of tracers in the brain. Finally, isoflurane has been shown to change the free fraction in plasma of certain drugs, ³⁷ therefore changing the availability of tracer in plasma in our case, and in consequence changing the washout and binding of the tracer in the brain.

Conclusions

Isoflurane and ketamine-xylazine anesthesia affect [¹⁸F]SynVesT-1 kinetics and uptake compared to unanesthetized mice. The total volume of distribution, estimated with an image-derived input function, was significantly lower in awake compared to isoflurane, but higher in awake compared with ketamine-xylazine anesthetized mice. In the awake state, the tracer kinetics could be well modeled with a one tissue compartment model, similar to humans [¹⁸F]SynVesT-1 scans, while in both anesthesia cases the two-tissue compartment model was necessary to describe kinetics.

Authors’ contributions

AM, DB, and JV were involved on the experimental design, and data analysis. AM, DB, and CdeW were involved in the experiment data acquisition. AM was involved in the software writing. AM, DB, JV, and SS were involved in drafting and editing the manuscript and figures. All authors assisted in reviewing the manuscript and approved the final version of this manuscript.

Material availability

All data associated with this study are presented in the paper or the Supplementary Materials. Any request for material reported in this study will be available through a material transfer agreement (MTA).

ORCID iDs

Alan Miranda https://orcid.org/0000-0002-5381-015X

Daniele Bertoglio https://orcid.org/0000-0003-4205-5432

Supplemental material

Supplemental material for this article is available online.