Evaluation of Intraperitoneal [18F]-FDOPA Administration for Micro-PET Imaging in Mice and Assessment of the Effect of Subchronic Ketamine Dosing on Dopamine Synthesis Capacity

Positron emission tomography (PET) using the radiotracer [18F]-FDOPA provides a tool for studying brain dopamine synthesis capacity in animals and humans. We have previously standardised a micro-PET methodology in mice by intravenously administering [18F]-FDOPA via jugular vein cannulation and assessment of striatal dopamine synthesis capacity, indexed as the influx rate constant KiMod of [18F]-FDOPA, using an extended graphical Patlak analysis with the cerebellum as a reference region. This enables a direct comparison between preclinical and clinical output values. However, chronic intravenous catheters are technically difficult to maintain for longitudinal studies. Hence, in this study, intraperitoneal administration of [18F]-FDOPA was evaluated as a less-invasive alternative that facilitates longitudinal imaging. Our experiments comprised the following assessments: (i) comparison of [18F]-FDOPA uptake between intravenous and intraperitoneal radiotracer administration and optimisation of the time window used for extended Patlak analysis, (ii) comparison of KiMod in a within-subject design of both administration routes, (iii) test-retest evaluation of KiMod in a within-subject design of intraperitoneal radiotracer administration, and (iv) validation of KiMod estimates by comparing the two administration routes in a mouse model of hyperdopaminergia induced by subchronic ketamine. Our results demonstrate that intraperitoneal [18F]-FDOPA administration resulted in good brain uptake, with no significant effect of administration route on KiMod estimates (intraperitoneal: 0.024 ± 0.0047 min−1, intravenous: 0.022 ± 0.0041 min−1, p = 0.42) and similar coefficient of variation (intraperitoneal: 19.6%; intravenous: 18.4%). The technique had a moderate test-retest validity (intraclass correlation coefficient (ICC) = 0.52, N = 6) and thus supports longitudinal studies. Following subchronic ketamine administration, elevated KiMod as compared to control condition was measured with a large effect size for both methods (intraperitoneal: Cohen's d = 1.3; intravenous: Cohen's d = 0.9), providing further evidence that ketamine has lasting effects on the dopamine system, which could contribute to its therapeutic actions and/or abuse liability.

In more recent years, the development of micro-PET technology [20] has enabled the study of small animal models of human disease, allowing for translational investigation of underlying pathology and supporting the development of novel therapies. Mice and rats in particular offer the advantage of being experimentally tractable models, including the generation of phenotypes by pharmacological treatment, genetic models using protein knockout or overexpression, and functional manipulation using optogenetic or chemogenetic regulation [21][22][23]. Work from our group recently showed that subchronic ketamine treatment in mice resulted in elevated striatal dopamine synthesis capacity, resembling the dopaminergic alteration seen in patients with schizophrenia, and demonstrated that this measure could be reduced using pharmacological treatment [24].
In mice, radiotracers for PET imaging are routinely administered intravenously (i.v.), particularly via the superficial tail vein. However, repeated administration can lead to injury and, due the small bore of this vein, the injectable volume is limited, while partial paravenous injection is a risk that can lead to inadequate radiotracer delivery. Jugular vein cannulation, which enables fast delivery to the brain, is a challenging and invasive procedure in mice which is normally a terminal procedure as recovery after extended anaesthesia is challenging, thus preventing longitudinal assessments. In the case of longitudinal studies, chronic intravenous catheters are technically difficult to maintain and can be stressful to mice. L-3,4-Dihydroxyphenylalanine (L-DOPA) is absorbed in the rat peritoneum with less interindividual variability than oral dosing [25]. However, it is not known if this is the case in mice, where difference in size and metabolism may affect radiotracer uptake in the brain.
In the current study, we investigated intraperitoneal (i.p.) [

Subjects.
Male wild-type C57BL/6J mice, aged 6-7 weeks old upon arrival, were obtained from Charles River (Kent, UK) and housed in groups of up to 4 under a 12 : 12-hour light dark cycle with food and water provided ad libitum. Animals were habituated for a minimum of 1 week, before being used in procedures at a minimum age of 8 weeks; the overall weight range of animals used in this study was 19-34 g (26:0 ± 2:6 g), and animals were randomly assigned to different experiments and treatment groups. All animal experimental procedures were performed in accordance with the UK Animals (Scientific Procedures) Act 1986 and EU directive 2010/63/EU, and protocols were also approved by Invicro and Imperial College Animal Welfare and Ethical Review Body.
Experiment 3 evaluated the test-retest estimation of K i Mod in a within-subject design of i.p. radiotracer administration.

PET Acquisition.
[ 18 F]-FDOPA was synthesised as described previously in the supplementary material published by Jauhar et al. [3]. One hour prior to administration of [ 18 F]-FDOPA, mice were anaesthetised with isoflurane to undergo cannulation, and were maintained under isoflurane anaesthesia at 1-2% in 1 L/min oxygen for the duration of the study, with respiration rate and body temperature being monitored continuously throughout (BioVet, m2m Imaging Corp, OH, USA). To allow radiotracer delivery, either the external jugular vein was cannulated 45 minutes prior to scanning or an intraperitoneal cannula was inserted 30 minutes before the scan. In order to prevent peripheral metabolism of the [ 18 F]-FDOPA, inhibitors of catechol-Omethyl-transferase (COMT) and aromatic amino acid decarboxylase (AADC), entacapone (40 mg/kg; Sigma-Aldrich, SML0654) and benserazide hydrochloride (10 mg/kg; Sigma-Aldrich, B7283), were given intraperitoneally, respectively, at 45 minutes and 30 minutes prior to administration of [ 18 F]-FDOPA [22][23][24]. To ensure optimal absorption of these inhibitors and the radiotracer and to avoid competition with uptake of other neutral amino acids in diet across the blood-brain barrier (BBB) [24,26], animals were fooddeprived for 45 minutes before being anaesthetised. After administration of the inhibitors, mice underwent a 10 min CT scan in an Inveon PET/CT scanner (Siemens, Surrey, UK) to allow for attenuation correction of the PET signal. A dynamic PET scan was started concomitantly with bolus injection of [ 18 F]-FDOPA, and data was collected for up to 140 minutes.
2.4. PET Image Processing. Following PET acquisition, data were histogrammed into 43 frames (10 × 3 s, 6 × 5 s, 8 × 30 s, 5 × 60 s, 6 × 300 s, and 8 × 600 s) for PET scans lasting 120 minutes. In the case of scans acquired for 140 minutes (experiment 4, i.p. injected group), two additional frames of 600 s were included. Data were reconstructed using filtered back projection with CT attenuation correction, adjusting for random noise, scatter, and radiotracer decay. Image processing was carried out using the Inveon Research Workspace (Siemens, USA), where the CT image was coregistered to the reconstructed PET image and checked for alignment. For each subject, the percentage-injected dose was normalised for body weight and injected activity to provide standardised uptake values (SUV). Regions of interest (ROI) with predefined shape and size were drawn manually on summation radioactivity images around the left and right striata (square; 0.07 cm 3 each) and the cerebellum (rectangular; 0.1 cm 3 ) to extract time-activity curves (TACs) guided by CT and mouse brain anatomy [27].

Extended Patlak Analysis.
Kinetic analyses were performed using an in-house pipeline in MATLAB version R2020a (MathWorks, MA, USA). Extended Patlak graphical analysis was used to determine the rate constant K i Mod (min -1 ), which provides an estimate for the rate of [ 18 F]-FDOPA striatal uptake and its subsequent conversion into Mod is a modified version of the influx rate constant K i Cer (min -1 ) and takes into account an estimation of k loss (min -1 ), the rate at which signal is removed from the system due to metabolism of [ 18 F]-FDA, as described previously [22,[28][29][30][31]. In brief, peripherally administered [ 18 F]-FDOPA is transported across the BBB to be taken up into striatal-projecting dopamine neurons and is converted by AADC into [ 18 F]-FDA, which then becomes irreversibly trapped in brain tissue [30,32]. The rate at which [ 18 F]-FDOPA is taken up and converted into [ 18 F]-FDA is used as an index of presynaptic dopamine synthesis capacity. This rate constant is referred to as K i Cer when the cerebellum is used as a reference region in place of an arterial input function. The ratio between irreversibly trapped radiotracer measured in the striatum versus free radiotracer in the reference cerebellum region takes on a linear relationship when plotted against the ratio of running time integral of cerebellum SUV and cerebellar SUV, a function referred to as stretch time. The slope of this linear plot as derived by graphical Patlak analysis [33] corresponds to the influx rate constant K i Cer . In humans, [ 18 F]-FDA trapping is considered irreversible for up to two hours after [ 18 F]-FDOPA administration [32]. Rodents have a much faster metabolic rate, and [ 18 F]-FDA is lost through metabolism during typical scan durations [21,22], resulting in a deviation from the linear relationship between striatal uptake and stretch time. This can be corrected for using an extended Patlak analysis, which provides a modified influx rate constant, K i Mod , that estimates the metabolism of [ 18 F]-FDA as rate of loss, k loss and adjusts for it to restore the linear relationship [24,29,30].

Optimal Window for Determining K i
Mod and k loss Parameters. Accurate estimation of the K i Mod parameter requires the acquisition of sufficient temporal data so as to capture the reversible nature of the kinetics present in the system. Subsequently, k loss is determined as the deviation from the linear regression over time. This makes estimation of these two parameters sensitive to the time range chosen to 3 Molecular Imaging carry out the analysis. To establish the optimal window for determining K i Mod and k loss parameters, individual TAC data, acquired following i.p. radiotracer administration in experiment 1 (N = 13, acquired up to 120 minutes) and experiment 4 (naïve group only, N = 12, acquired up to 140 minutes), was processed through the extended Patlak analysis using varying start (T * ) and end times (T end ). For each parameter, the standard deviation (S.D.) for K i Mod and k loss and the coefficient of variation (%CV) were calculated at the different time points to identify the analysis window that produced the least within-group variability. Using these measures, the optimal time range for analysis of K i Mod and k loss following i.v. administration was previously determined to be 15-120 minutes [23]; in the current study, the effect of altering time windows on both parameters was tested for i.p. administration.

Statistical Analysis.
Statistical tests were carried out using GraphPad Prism v9 (GraphPad Software, La Jolla, California, USA). Average numerical values in the text are expressed as mean ± S:D:; in graphs, data are shown as mean ± S:D: Data were tested for statistical outliers using ROUT test, Q = 0:5%, and tested for normal distribution using the Kolmogorov-Smirnov test to decide on the use of parametric (t-test) or nonparametric tests; no statistical outliers were identified, and all data showed a normal distribution. Slopes of SUVR curves were compared using simple linear regression analysis. Statistical tests used are specified in the text and figure legends. Comparison of Radiotracer Uptake. The weight of animals scanned did not significantly differ between groups (i.p., 26:8 ± 3:6 g; i.v., 24:7 ± 2:4 g; p = 0:12, t = 1:62, unpaired two-tailed t-test). However, the dose injected in the i.v. group was lower than that in the i.p. group due to the more restrictive injectable volume using i.v. administration (i.p., 0:45 ± 0:23 MBq/g; i.v., 0:17 ± 0:10 MBq/g; p = 0:0018, t = 3:574, unpaired two-tailed t-test).
All mice subjected to an [ 18 F]-FDOPA scan with i.p. injection of the radiotracer (N = 13) made an unremarkable recovery, enabling their reuse for longitudinal studies. Mod and k loss is lowest (Figures 2(a) and 2(b)), suggesting the greatest degree of consistency across samples. This is identical to the analysis window optimised for i.v. administration of [ 18 F]-FDOPA in previous work [23].
Due to the slower uptake kinetics following i.p. radiotracer administration, resulting in a shorter time window available for the estimation of k loss , we verified in a separate dataset (N = 12 naïve animals) whether extending the scan time and therefore the analysis window used for the estimation of K i Mod and k loss by the extended Patlak analysis to 140 minutes could further reduce the within-group variability of 4 Molecular Imaging

K i
Mod and k loss . Assessment of a limited set of T * values showed that at a T end of 140 minutes, the within-group variability and %CV are less sensitive to variation in T * than when using a T end of 120 minutes, and an analysis window of 20-140 minutes yielded slightly reduced variability compared to 15-120 minutes (%CV K i Mod , 19.1%; %CV k loss , 19.7%; Figures 2(e)-2(h)). Therefore, the analysis window of 20-140 minutes was chosen for analysis of scan data acquired up to 140 minutes following i.p. administration of [ 18 F]-FDOPA, whereas an analysis window of 15-120 minutes was selected as most optimal for datasets acquired up to 120 minutes.

Discussion
In the current study, we evaluated the use of i.p. administration of [ 18 F]-FDOPA in comparison to i.v. administration in mice for the assessment of dopamine synthesis capacity in the striatum. Our data show that i.p. administration is a method that allows excellent striatal radiotracer uptake, with moderate test-retest reliability [34], a similar range of outcome values as compared to i.v. administration, and fast and unremarkable recovery of the animals following imaging, enabling longitudinal studies and within-animal comparison. Importantly, using i.p. [ 18 F]-FDOPA administration, we were able to reproduce previously measured enhanced dopamine synthesis in a mouse model of hyperdopaminergia [24]. Our results are in agreement with previous studies in rodents where 18 F-sodium fluoride ( 18 F-NaF) and 18 F-fludeoxyglucose ( 18 F-FDG) radiotracers showed comparable results between i.v. and i.p. administration routes [35,36].
We applied an extended Patlak analysis for the determination of dopamine synthesis capacity following i.p. [ 18 F]-FDOPA administration in mice, showing that both k loss and the modified influx rate constant K i Mod are sensitive to the start and end times of the analysis window. The derivation of K i Mod is dependent on the estimation of k loss , and so using the analysis range that provided the most consistent estimate for k loss also yielded the least variable estimates of K i Mod . Previously, Holden et al. demonstrated in nonhuman primates that with a longer scan duration (end time of 240 minutes), the k loss parameter is insensitive to the start time [29]. With longer scan duration, there are more data available with which to derive an accurate measure of k loss . Likewise, we found that extending the scan time from 120 to 140 minutes reduced sensitivity of K i Mod and k loss for variation in T * (Figure 2). Due to the complexity of maintaining mice under anaesthesia for an extended period, further lengthening of scan time is not desirable, particularly when recovering the animal after imaging. In the present study, we found that the 20-140-minute analysis window provided the lowest within-group variability over the time range tested (%CV K i Mod = 19%, %CV k loss = 20%, Figures 2(g) and 2(h)). This is a slight increase as compared to 11% in rats at 180-minute scan duration, where the radiotracer was injected i.v. in the tail [22]. As expected, with shorter scan lengths (up to 120 minutes), the within-group variability in our study increased to over 40% (Figures 2(c) and 2(d)); such high variability would potentially restrict the use of this method to studies with only very large expected between-group effect sizes.
With similar intergroup variability as reported here, future mouse studies using the extended Patlak analysis would require group sizes of at least 10 to detect effect sizes of the same magnitude as the effect we found for ketamine (d = 1:4) with >90% power. This can be reduced to N = 7 for studies where two within-subject measurements are possible; thus, i.p. administration of [ 18 F]-FDOPA could support the reduction of subject numbers in rodent PET imaging studies.
In the present study, we have used a group-housed inbred strain of mouse to reduce within-group individual variability and provide greater resolution of the impact of different analysis start and end times, as well as support reproducibility in the test-retest reliability study and the within-subject comparison of i.p. and i.v. administration of [ 18 F]-FDOPA. The within-group variability obtained in our study (19%) is somewhat higher than that seen in human [ 18 F]-FDOPA PET imaging studies, where the variability in K i Cer was 10% in the whole striatum, with a range of 6-12% in striatal subregions [37]. Likewise, the test-retest reliability (ICC = 0:52 and %VAR = 18%) in our study is 8 Molecular Imaging somewhat lower than that reported in humans (ICC = 0:84 and %VAR = 4:5% in human striatum) [37]. These differences probably reflect the far larger size of the striatum in humans, which results in higher resolution images and makes correct placement of the ROI more reliable and reproducible than in mice. Nevertheless, the ICC value in mice indicates moderate reliability [34], supporting the use of this approach. To our knowledge, we are the first group to apply the extended Patlak analysis to mice scanned with [ 18 F]-FDOPA, although others have used the radiotracer in mice to measure overall radiotracer binding at a predefined time point after radiotracer injection [38][39][40]. Although these studies revealed increased radiotracer binding in the striatum as compared to other brain regions, reliability of this parameter has not been assessed in those studies. A major advantage of kinetic modelling using Patlak analysis is that outcome values correspond directly to those used in human [ 18 F]-FDOPA PET imaging [10,41]. SUVR is an alternative proxy measure that has been used to measure dopamine synthesis capacity in humans [42]. Our data provide a preliminary indication that this output value may also be used as an alternative to K i Mod in the assessment of dopamine function using [ 18 F]-FDOPA PET imaging and i.p. radiotracer administration in mice ( Figure 5(c)).
The k loss values determined by our approach, both the i.v. and i.p. methods, are similar to comparable parameters determined in rats [43] and humans [44,45] in previous studies, but it has been determined to be an order lower in nonhuman primates [30,31]. There are no comparator studies in mice with the exception of an earlier publication from our group, and they are most likely related to the species in which they are measured [23]. A methodological consideration for [ 18 F]-FDOPA PET scanning in rodents is that administration of COMT and AADC inhibitors is required to prevent peripheral metabolism of the radiotracer and that the between-subject variability in the uptake of these inhibitors underlies the variability in the data. The intraperitoneal 9 Molecular Imaging injection of these inhibitors just prior to intraperitoneal radiotracer injection may introduce additional variability in radiotracer uptake as compared to intravenous injection. The large surface area of the peritoneal membrane aids rapid absorption; however, the radiotracer is predominantly expected to be transported to systemic circulation through the hepatic portal pathway [46]. Other drug-metabolising enzymes do not play a significant role in the breakdown of DOPA [47], and hence, its brain kinetics are suited our study. Peripheral inhibitors are also used in human [ 18 F]-FDOPA imaging, supporting the translational value of this approach in mice [3,6].
Importantly, using i.p. administration of [ 18 F]-FDOPA and in an independent cohort using i.v. administration, we were able to reproduce the previously observed increased dopamine synthesis rate in mice receiving subchronic administration of ketamine [24]. One potential limitation for the study on the effects of ketamine is that the control animals did not receive control injections, in contrast to the ketamine-treated animals. Thus, the design cannot disambiguate whether it is the ketamine or injection that resulted in the group differences. However, the study by Kokkinou et al. [24] used saline injections as the control condition, finding a significant difference with the ketamine group, indicating that the lack of control injections in our control group is unlikely to explain our finding. These results provide further evidence that ketamine has effects on the regulation of dopamine signalling. Furthermore, the high reproducibility of the effect supports the use of subchronic ketamine in mice to create a model of hyperdopaminergia as a proxy of the dopamine signalling abnormality associated with psychotic disorders [15]. Also, the lower SUVR to the cerebellum with i.p. than i.v. might mean that the i.p. approach is less sensitive to detect models of reduced dopamine synthesis capacity, although this effect may be offset by the lower variability with i.p. relative to i.v. Studies using models of reduced dopamine synthesis capacity are needed to evaluate this. It is also interesting to note that the mean K i values in experiment 2 are similar to those in the ketamine exposed group in experiment 4. This may be an artefact of the smaller sample size in experiment 2 but could also reflect cohort differences, which highlights the value of longitudinal imaging.

Conclusion
In conclusion, we have demonstrated that the intraperitoneal administration of [ 18 F]-FDOPA provides a reliable measure of presynaptic dopamine synthesis capacity with similar values to those obtained with intravenous injection. We also show that ketamine elevates dopamine synthesis capacity, extending previous findings to show this in an independent cohort of animals. Subchronic glutamate receptor antagonism using other pharmacological interventions has also shown to elevate dopamine synthesis capacity and elimination of [ 18 F]-FDOPA in rodent and humans thus supporting our findings [44,48]. Our i.p. administration model finding is consistent with elevated dopamine synthesis capacity while the high control group k loss in the i.p. model may have obscured differences in catabolism, and so, further studies are needed. However, it must be mentioned that the ketamine model mimics findings seen with the catabolism and elimination of [18F]-FDOPA in the brain of patients with schizophrenia [45]. Application of this method in mice will allow for longitudinal studies in mouse models that assess the effect of genetic, pharmacological, and environmental changes on the dopamine system. The possibility of within-animal comparison allows for a reduction in animal numbers required, thereby supporting ethical standards in animal research to reduce usage of animals where possible. The method allows for translational measurement of presynaptic dopamine function, thereby providing an important tool in understanding the factors that may ultimately play a role in several neurological and psychiatric disorders as well as supporting the development of novel pharmacological treatments.

Data Availability
The datasets generated and/or analysed in this study, as well as all codes used in the present study, are available from the corresponding author upon reasonable request.