*2.4. [18F]FDS PET Imaging*

Immediately after EB injection (<60 s), anesthetized mice were transferred to the microPET scanner (Inveon, microPET, Siemens Healthcare, Knoxville, TN, USA). [18F]FDS was administered intravenously (4.2 ± 0.76 MBq) using a microinjection pump at the rate of 0.2 mL·min−<sup>1</sup> during 60 s (*<sup>n</sup>* = 5 FUS; *<sup>n</sup>* = 3 sham). Dynamic PET acquisition (60 min) started with [18F]FDS injection.

PET images were reconstructed using the three-dimensional ordered subset expectation maximization with maximum a posteriori algorithm (3D OSEM/MAP) and corrected for attenuation, random coincidences and scattering. Volumes of interests (VOIs) were manually delineated using Pmod software (version 3.8, PMOD Technologies Ltd., Zurich, Switzerland). In the FUS group, extravasation of <sup>18</sup>F-FDS was obvious in the sonicated area on late PET images. The region with disrupted BBB was delineated and mirrored to the contralateral hemisphere. In sham animals, VOI<sup>s</sup> were drawn in each brain hemisphere. Another VOI was drawn on the aorta (blood-pool), obvious on early time-frames, to generate an image derived input function (IDIF).

Time activity curves (TACs) were corrected for radioactive decay, injected dose and animal weight and reported as standardized uptake value (SUV) vs. time. Area under the TAC (AUC) was calculated from either 0 to 15, 30 or 60 min (AUCbrain) as well as corresponding AUCbrain/AUCblood. Kinetic modeling of the 60 min PET data was performed using either the Logan graphical method [34] and a 1-tissue compartment (1-TC) model using IDIF to estimate the total volume of distribution (*V*T, mL·cm−<sup>3</sup> ) of [18F]FDS and describe its transport across the BBB. Parametric images were generated to visualize the distribution of *V*<sup>T</sup> (1-TC model) of [18F]FDS in the brain and peripheral tissues.

#### *2.5. Statistics*

Data are presented as mean ± S.D. Statistical analysis was performed using GraphPad Prism software version 9.1, La Jolla, CA, USA. Outcome parameters of kinetic modeling were compared using a two-way ANOVA with "hemisphere" and "group" as factors followed by Tukey's multiple comparison test. Ratio of radioactivity measured in the right to the left hemisphere, as well as blood data obtained in the FUS and the sham group were compared using the Mann–Whitney U test. The Pearson's test was used to test the correlation between different kinetic parameters obtained in the same brain regions. Statistical significance was set a *p* < 0.05. Intragroup variability of outcome parameters was estimated by the coefficient of variation (CV = 100 × S.D./mean).

#### **3. Results**

Chemical transformation of [18F]FDG into [18F]FDS was very effective (Figure 1). Radio-TLC analysis of the preparation revealed a single peak with retardation factor (Rf) = 0.6, consistent with the presence of [18F]FDS [29]. The Rf of untransformed [18F]FDG, tested using the same TLC conditions, was 0.9 (data not shown). No [18F]FDG could be detected using this method in the final preparation of [18F]FDS.

In FUS animals, strong EB staining was observed in the posterior part of the sonicated (right) hemisphere only (Figure 2A). This contrasted with the low EB staining still observed in brain vessels of the contralateral hemisphere, as well as in sham animals, because no

*Pharmaceutics* **2021**, *13*, x FOR PEER REVIEW 5 of 15

using this method in the final preparation of [18F]FDS.

**3. Results** 

washing out was performed to remove the circulating blood from the resected brain tissue. This confirmed that enhanced EB staining could be used as a positive control for effective BBB disruption. cause no washing out was performed to remove the circulating blood from the resected brain tissue. This confirmed that enhanced EB staining could be used as a positive control for effective BBB disruption.

Chemical transformation of [18F]FDG into [18F]FDS was very effective (Figure 1). Radio-TLC analysis of the preparation revealed a single peak with retardation factor (Rf) = 0.6, consistent with the presence of [18F]FDS [29]. The Rf of untransformed [18F]FDG, tested using the same TLC conditions, was 0.9 (data not shown). No [18F]FDG could be detected

In FUS animals, strong EB staining was observed in the posterior part of the sonicated (right) hemisphere only (Figure 2A). This contrasted with the low EB staining still observed in brain vessels of the contralateral hemisphere, as well as in sham animals, be-

**Figure 2.** Impact of focused ultrasound (FUS) on blood–brain barrier (BBB) integrity assessed using Evan's blue extravasation and positron emission tomography (PET) using [18F]2-fluoro-2-deoxy-sorbitol ([18F]FDS). Extravasation of Evan's blue was obvious in the right brain hemisphere of animals of the FUS group (**A**). Standardized uptake value (SUV)-normalized brain PET images (sum 30–60 min) of [18F]FDS uptake after hemispheric BBB disruption induced by focused ultrasound (FUS) or without (sham) are shown in (**B**). Corresponding parametric images describing the total volume of distribution (*V*T), estimated using the one-tissue compartment model from 0 to 60 min are shown in (**C**). In (**D**), the distribution of the fluorescence of the Evan's blue dye in a coronal slice of mouse brain after FUS protocol obtained ex vivo was overlaid to the corresponding slice of [18F]FDS PET image obtained in vivo in the same animal. **Figure 2.** Impact of focused ultrasound (FUS) on blood-brain barrier (BBB) integrity assessed using Evan's blue extravasation and positron emission tomography (PET) using [18F]2-fluoro-2-deoxysorbitol ([18F]FDS). Extravasation of Evan's blue was obvious in the right brain hemisphere of animals of the FUS group (**A**). Standardized uptake value (SUV)-normalized brain PET images (sum 30–60 min) of [18F]FDS uptake after hemispheric BBB disruption induced by focused ultrasound (FUS) or without (sham) are shown in (**B**). Corresponding parametric images describing the total volume of distribution (*V*T), estimated using the one-tissue compartment model from 0 to 60 min are shown in (**C**). In (**D**), the distribution of the fluorescence of the Evan's blue dye in a coronal slice of mouse brain after FUS protocol obtained ex vivo was overlaid to the corresponding slice of [18F]FDS PET image obtained in vivo in the same animal.

Similarly, no washing out of circulating blood radioactivity was performed to interpret in vivo PET images. PET images in sham animals confirmed the negligible baseline brain uptake of brain [18F]FDS across the intact BBB (Figure 2B), whereas a strong PET signal could be monitored in the sonicated brain hemisphere of FUS animals. Parametric PET images, expressed in *V*T-1TC;60min closely resembled the PET images in SUV units (Figure 2C). Images revealed a region with obvious increase in [18F]FDS uptake that was strictly localized in the right hemisphere, with very limited impact on the left hemisphere. Brain distribution of [18F]FDS PET signal was consistent with EB extravasation observed in the whole brain (Figure 2A).

Moreover, brain distribution of the fluorescent signal of EB observed ex vivo in the right brain hemisphere with higher spatial resolution was similar to the [18F]FDS PET signal obtained in the same animal of the FUS group (Figure 2D). It should be noted that although no washout was performed, no fluorescent signal was observed in the contralateral brain hemisphere underlining specific leakage of EB into the brain parenchyma. Interestingly, mapping of [18F]FDS brain distribution within the sonicated area displayed a gradient from the center to the periphery which was not observed using EB as a BBB integrity marker. teral brain hemisphere underlining specific leakage of EB into the brain parenchyma. Interestingly, mapping of [18F]FDS brain distribution within the sonicated area displayed a gradient from the center to the periphery which was not observed using EB as a BBB integrity marker. In hemispheres with intact BBB (sham animals or contralateral hemisphere), brain PET signal increased rapidly with maximal uptake at Tmax = 4 min, followed by slow de-

Similarly, no washing out of circulating blood radioactivity was performed to interpret in vivo PET images. PET images in sham animals confirmed the negligible baseline brain uptake of brain [18F]FDS across the intact BBB (Figure 2B), whereas a strong PET signal could be monitored in the sonicated brain hemisphere of FUS animals. Parametric PET images, expressed in *V*T-1TC;60min closely resembled the PET images in SUV units (Figure 2C). Images revealed a region with obvious increase in [18F]FDS uptake that was strictly localized in the right hemisphere, with very limited impact on the left hemisphere. Brain distribution of [18F]FDS PET signal was consistent with EB extravasation observed

Moreover, brain distribution of the fluorescent signal of EB observed ex vivo in the

right brain hemisphere with higher spatial resolution was similar to the [18F]FDS PET signal obtained in the same animal of the FUS group (Figure 2D). It should be noted that although no washout was performed, no fluorescent signal was observed in the contrala-

*Pharmaceutics* **2021**, *13*, x FOR PEER REVIEW 6 of 15

in the whole brain (Figure 2A).

In hemispheres with intact BBB (sham animals or contralateral hemisphere), brain PET signal increased rapidly with maximal uptake at Tmax = 4 min, followed by slow decrease of the radioactivity. FUS selectively enhanced the brain PET signal and Tmax was achieved later at ~13.5 min, followed by a plateau (Figure 3A,B). SUV values in the sonicated (right) hemisphere at 15, 30 and 60 min were significantly higher than in the nonsonicated (left) hemisphere (*p* < 0.01, Figure 3A). Brain exposure (AUC0–60min) of [18F]FDS in the sonicated volume was 2.2 ± 0.5-fold higher compared with the contralateral volume (*p* < 0.01) and 1.8 ± 0.4-fold higher than the mean AUC0–60min measured in the corresponding (right) hemisphere of the sham group (*p* < 0.05, Figure 4A). The ratio of the PET signal in the sonicated volume to the contralateral region increased from 0 to 15 min to reach a plateau (Figure 3C). FUS did not impact the kinetics [18F]FDS in the blood-pool with significant difference in neither SUV values nor AUCblood between the FUS and the sham group (*p* > 0.05, Figure 3D). crease of the radioactivity. FUS selectively enhanced the brain PET signal and Tmax was achieved later at ~13.5 min, followed by a plateau (Figure 3A,B). SUV values in the sonicated (right) hemisphere at 15, 30 and 60 min were significantly higher than in the nonsonicated (left) hemisphere (*p* < 0.01, Figure 3A). Brain exposure (AUC0–60min) of [18F]FDS in the sonicated volume was 2.2 ± 0.5-fold higher compared with the contralateral volume (*p* < 0.01) and 1.8 ± 0.4-fold higher than the mean AUC0–60min measured in the corresponding (right) hemisphere of the sham group (*p* < 0.05, Figure 4A). The ratio of the PET signal in the sonicated volume to the contralateral region increased from 0 to 15 min to reach a plateau (Figure 3C). FUS did not impact the kinetics [18F]FDS in the blood-pool with significant difference in neither SUV values nor AUCblood between the FUS and the sham group (*p* > 0.05, Figure 3D).

**Figure 3.** Brain kinetics of [18F]2-fluoro-2-deoxy-sorbitol ([18F]FDS). BBB disruption was obtained in the right brain hemisphere using focused ultrasound (FUS group, *n* = 5, in (**A**), but not in the sham group (*n* = 3, in (**B**)). The time course of the right/left ratio of the PET signal in FUS and sham animals is shown in (**C**). Corresponding image-derived input functions are shown in (**D**). Data are mean ± **Figure 3.** Brain kinetics of [18F]2-fluoro-2-deoxy-sorbitol ([18F]FDS). BBB disruption was obtained in the right brain hemisphere using focused ultrasound (FUS group, *n* = 5, in (**A**), but not in the sham group (*n* = 3, in (**B**)). The time course of the right/left ratio of the PET signal in FUS and sham animals is shown in (**C**). Corresponding image-derived input functions are shown in (**D**). Data are mean <sup>±</sup> S.D Statistical comparison of values obtained at 15, 30 and 60 min after injection of [18F]FDS is reported with \* *p* < 0.05, \*\* *p* < 0.01, n.s = nonsignificant.

**0.0**

x = y

**0.2**

**0.4**

**VT** or **AUCbrain/AUCblood**

**0.6**

**0.8**

*Pharmaceutics* **2021**, *13*, x FOR PEER REVIEW 7 of 15

S.D Statistical comparison of values obtained at 15, 30 and 60 min after injection of [18F]FDS is re-

ter to describe the brain distribution of [18F]FDS in all tested conditions.

ported with \* *p* < 0.05,\*\* *p* < 0.01, n.s = nonsignificant.

S.D Statistical comparison of values obtained at 15, 30 and 60 min after injection of [18F]FDS is re-

Brain TACs of [18F]FDS were accurately described with acceptable fit (error < 3%) by *V*T estimated using the 1-TC model (Figure 4B). The Logan graphical method provided similar estimation of brain *V*T (correlation *p* < 0.001, R2 = 0.99, Figure 5). A significant increase in the *V*T-1TC of [18F]FDS was observed in the sonicated region compared with the contralateral region (2.43 ± 0.8-fold increase, *p* < 0.01) or the brain of sham animals (1.9 ± 0.2-fold increase, *p* < 0.05) (Figure 4). In the FUS group, the variability of *V*T-1TC values was lower in the sonicated brain (CV = 10.1%) compared with the nonsonicated brain (CV = 29.7%). The influx and efflux transfer rates K1 and k2 were estimated with acceptable fit (error < 7%). However, K1 values were associated with a high intragroup variability in the nonsonicated brain (CV = 37.6%) compared with the sonicated brain (CV = 18.5%), which questions the relevance of K1 and k2 to describe the extremely low BBB penetration of

ported with \* *p* < 0.05,\*\* *p* < 0.01, n.s = nonsignificant.

**Figure 4.** Comparison of [18F]FDS brain PET data. Ultrasound-induced BBB disruption (FUS+) was obtained in the right brain hemisphere of the FUS group (*n* = 5) while BBB in the contralateral brain hemisphere was intact (FUS−). The brain exposure of [18F]FDS was estimated by the area under the time-activity curve (AUCbrain, in (**A**). Brain distribution was estimated by *V*T-1TC (1-tissue compartment model, in (**B**) using an image-derived input function. Data were also compared with a sham **Figure 4.** Comparison of [18F]FDS brain PET data. Ultrasound-induced BBB disruption (FUS+) was obtained in the right brain hemisphere of the FUS group (*n* = 5) while BBB in the contralateral brain hemisphere was intact (FUS−). The brain exposure of [18F]FDS was estimated by the area under the time-activity curve (AUCbrain, in (**A**). Brain distribution was estimated by *V*T-1TC (1-tissue compartment model, in (**B**) using an image-derived input function. Data were also compared with a sham group (no FUS, *n* = 3). Data are reported as mean ± S.D with \* *p* < 0.05 and \*\* *p* < 0.01.

group (no FUS, *n* = 3). Data are reported as mean ± S.D with \* *p* < 0.05 and \*\* *p* < 0.01. **0.0 0.2 0.4 0.6 0.8** 0-60 min 0-15min VT-Logan;60min 0-30min **Slope** 1.09 0.78 **Parameter** 0.74 0.43 VT-1TC;30min 0.89 VT AUCbrain / AUCblood Brain TACs of [18F]FDS were accurately described with acceptable fit (error < 3%) by *V*<sup>T</sup> estimated using the 1-TC model (Figure 4B). The Logan graphical method provided similar estimation of brain *V*<sup>T</sup> (correlation *p* < 0.001, R<sup>2</sup> = 0.99, Figure 5). A significant increase in the *V*T-1TC of [18F]FDS was observed in the sonicated region compared with the contralateral region (2.43 ± 0.8-fold increase, *p* < 0.01) or the brain of sham animals (1.9 ± 0.2-fold increase, *p* < 0.05) (Figure 4). In the FUS group, the variability of *V*T-1TC values was lower in the sonicated brain (CV = 10.1%) compared with the nonsonicated brain (CV = 29.7%). The influx and efflux transfer rates K<sup>1</sup> and k<sup>2</sup> were estimated with acceptable fit (error < 7%). However, K<sup>1</sup> values were associated with a high intragroup variability in the nonsonicated brain (CV = 37.6%) compared with the sonicated brain (CV = 18.5%), which questions the relevance of K<sup>1</sup> and k<sup>2</sup> to describe the extremely low BBB penetration of [18F]FDS across the intact BBB. *V*T-1TC was therefore retained as the gold-standard parameter to describe the brain distribution of [18F]FDS in all tested conditions. **Figure 4.** Comparison of [18F]FDS brain PET data. Ultrasound-induced BBB disruption (FUS+) was obtained in the right brain hemisphere of the FUS group (*n* = 5) while BBB in the contralateral brain hemisphere was intact (FUS−). The brain exposure of [18F]FDS was estimated by the area under the time-activity curve (AUCbrain, in (**A**). Brain distribution was estimated by *V*T-1TC (1-tissue compartment model, in (**B**) using an image-derived input function. Data were also compared with a sham group (no FUS, *n* = 3). Data are reported as mean ± S.D with \* *p* < 0.05 and \*\* *p* < 0.01.

**Figure 5.** Sensitivity of kinetic parameters to describe the impact of FUS on the brain distribution of [18F]FDS in mice. Mice (*n* = 5) underwent blood–brain barrier disruption induced by focused ultrasound in the right hemisphere only. Kinetic modeling was performed to estimate the total brain distribution (*V*T) using the one-tissue compartment model (*V*T-1TC) estimated from 0 to 60 min after injection of [18F]FDS. *V*T-1TC;0–60min values were correlated with *V*T-1TC estimated from 0 to 30 min (VT-1TC;0–30min), *V*<sup>T</sup> estimated using the Logan graphical method from 0 to 60 min (*V*T-Logan;60min) or the ratio **Figure 5.** Sensitivity of kinetic parameters to describe the impact of FUS on the brain distribution of [18F]FDS in mice. Mice (*n* = 5) underwent blood-brain barrier disruption induced by focused ultrasound in the right hemisphere only. Kinetic modeling was performed to estimate the total brain distribution (*V*T) using the one-tissue compartment model (*V*T-1TC) estimated from 0 to 60 min after injection of [18F]FDS. *V*T-1TC;0–60min values were correlated with *V*T-1TC estimated from 0 to 30 min (VT-1TC;0–30min), *V*<sup>T</sup> estimated using the Logan graphical method from 0 to 60 min (*V*T-Logan;60min) or the ratio of the time-activity curves (AUC) measured in brain regions and blood (AUCbrain/AUCblood) from 0 to 15, 30 or 60 min. The slope of the linear correlation is reported.

Other quantification methods were tested to either reduce the length of PET acquisition or simplify analysis of the PET data. *V*T-1TC values estimated from 0 to 30 min (*V*T-1TC;30min), as well as AUCbrain/AUCblood measured from either 0 to either 30 or 60 min, was significantly correlated with *V*T-1TC;60min (*p* < 0.001, R<sup>2</sup> = 0.99) with limited loss in the sensitivity to detect enhanced brain distribution of [18F]FDS induced by FUS (Figure 5). However, further reduction of the length of PET frame tended to underestimate the brain distribution of [18F]FDS, especially in in the sonicated hemisphere. Nevertheless, AUCbrain/AUCblood from 0 to 15 min was still significantly correlated with VT-1TC values (*p* < 0.001, R<sup>2</sup> = 0.94) but the slope of the linear regression was <0.5. (Figure 5)

#### **4. Discussion**

In the present study, [18F]FDS PET imaging was validated for the first time as a translational and quantitative marker of BBB permeability to estimate the impact of spatiallycontrolled FUS on BBB integrity in mice.

Unlike [18F]FDG, [18F]FDS is poorly taken up by mammalian cells because it lacks transporters and/or enzymes for specific cell uptake and retention [29]. [18F]FDS PET has been previously validated in animals and humans to study renal function [35] or detect/estimate bacterial burden in tissues because sorbitol is a specific metabolic substrate of some strains of Gram-negative bacteria [36]. [18F]FDS therefore benefits from extensive clinical pharmacology and safety data as an experimental radiopharmaceutical drug [30]. However, to our knowledge, the impact of BBB integrity on the brain kinetics of [18F]FDS has never been evaluated so far.

[ <sup>18</sup>F]FDS benefits from the characteristics of an "ideal" marker of BBB integrity [21]. This includes, metabolic stability, low binding to plasma proteins (<0.1%) and low baseline brain uptake when the BBB is intact [30,31]. Interestingly, Li et al. reported that [18F]FDS visually accumulated in orthotopic brain tumor xenografts in mice, despite negligible uptake by implanted glioma cells in vitro [29]. Given the substantial sensitivity of [18F]FDS to the integrity of the BBB, it can be hypothesized that the local BBB leakage in the tumor environment, rather than uptake by cancer cells, may explain the enhanced [18F]FDS PET signal observed by Li and colleagues in the tumor area [29]. Further experiments are however needed to validate the use of [18F]FDS to quantitatively monitor the permeability of the blood-tumor barrier. Similarly, we advocate that the high sensitivity of [18F]FDS to BBB integrity may be considered for correct estimation of bacterial burden in brain lesions using [18F]FDS PET [37].

Pharmacokinetic modeling of brain [18F]FDS PET data is relatively simple. Estimation of brain *V*T, using either the graphical method (Logan plot) or compartmental modeling (1-TC) provided similar outcome parameters to describe the distribution of [18F]FDS in both the sonicated and the non-sonicated brain, with low intragroup variability. In the FUS group, variability was strikingly lower (CV ~10%) in the sonicated brain regions compared with regions with the intact BBB (CV ~30%). This may be explained by the lower PET signal in the brain with intact BBB, associated with higher signal-to-noise ratio compared with the sonicated brain in which the PET signal is higher. Altogether, this suggests that *V*T, which takes any change in the plasma kinetics of [18F]FDS into account, provides a reliable and sensitive outcome parameter to describe the BBB permeation of [18F]FDS across the BBB. Estimation of *V*T-1TC from 30 min PET data offers a compromise to observe and delineate the blood pool in aorta at early time-frames and allow for correct estimation of the brain distribution of [18F]FDS while reducing the total length of PET acquisition (Figure 5).

Simplified parameters were tested to quantitatively estimate the impact of FUS on the brain uptake of [18F]FDS. AUCbrain/AUCblood measured from 0 to 30 or 60 min after injection accurately predicted *V*<sup>T</sup> and can be used as a surrogate parameter to estimate BBB permeation with limited impact on the sensitivity of the method compared with 60 min scan (Figure 5). *V*<sup>T</sup> or AUCbrain/AUCblood should be used, rather than AUCbrain, in situations where change in the blood kinetics of [18F]FDS is expected, such as renal disorders to correctly estimate the blood to brain distribution of [18F]FDS [38].

The FUS method reported in this project allows for localized BBB disruption in the right hemisphere while keeping the contralateral hemisphere intact, as confirmed using the post mortem EB extravasation test. In this situation, the contralateral hemisphere can serve as a convenient reference region to accurately estimate the impact of FUS on enhanced [18F]FDS brain uptake (AUCbrain, Figure 3C). However, in many situations, BBB disruption induced by FUS or other methods may affect the whole brain or cannot be localized a priori [39]. In the absence of such reference region, increase in BBB permeability can be quantified by comparing [18F]FDS *V*<sup>T</sup> or AUCbrain/AUCblood obtained in a control (sham) group or in the same animals before any intervention, following a longitudinal design. This illustrates the added value of absolute quantitative PET compared with other neuroimaging techniques for noninvasive determination of different levels of BBB permeation in various situations [26].

Mapping of enhanced [18F]FDS PET signal induced by FUS in the brain was consistent with EB extravasation observed ex vivo and in vitro, although exact co-registration of PET images with brain slices is very challenging. This suggests a limited diffusion of [ <sup>18</sup>F]FDS from the sonicated to the nonsonicated brain with intact BBB. Interestingly, brain distribution within the sonicated area displayed a gradient from the center to the periphery (Figure 2). It may be hypothesized that heterogeneity in intensity of delivered ultrasound may occur within the sonicated area as a consequence of loss of ultrasound transmission relative to the transducer angulation at the skull surface. Interestingly, such a phenomenon could not be detected using high-MW markers of BBB integrity such as gadoterate or using EB extravasation in the same animals (Figure 2).

Several recent examples illustrate the added value of PET imaging to investigate the impact of FUS on BBB structure and function in vivo [18]. First, brain uptake of [ <sup>18</sup>F]FDG, which is actively transported by glucose transporter 1 (GLUT1), was shown to be lower during FUS-induced BBB disruption, consistent with the local decrease in GLUT1 expression in the sonicated brain. Baseline [18F]FDG uptake and GLUT1 expression were fully restored 24 h after FUS [40]. P-glycoprotein (P-gp) and breast cancer resistance protein (BCRP) are the main efflux transporters expressed at the BBB [2]. We have shown that FUS-induced BBB disruption did not significantly impact the efflux transport of the P-gp substrate [11C]*N*-desmethyl-loperamide and the dual P-gp/BCRP substrate [11C]erlotinib in rats [33]. These particular PET probes may however lack the sensitivity to detect subtle changes in transporter function [41]. Decrease in P-gp expression has been reported 24–48 h after FUS [42,43]. However, long-term impact on efflux transport function at the BBB remains to be investigated in details. In this framework, [18F]FDS may offer a quantitative tool to estimate the dynamics of BBB disruption and restoration after FUS with suitable temporal resolution [24]. Altogether, this suggests that [18F]FDS enriches the PET imaging toolbox to investigate the short- and long-term impact of FUS on BBB permeability with respect to its molecular environment, including regulation of tight junction expression [18].

Simple production of [18F]FDS from commercial [18F]FDG makes it an appealing radiopharmaceutical candidate for determination of BBB permeability using quantitative PET. [18F]FDS is not a "ready-to-inject" preparation compared with [18F]FDG, although radiopharmaceutical production and shipment could theoretically be ensured by an external manufacturer. However, most radiopharmaceuticals used in nuclear medicine and molecular imaging departments require handling by a radiopharmacy team, chemical reaction and quality control [44]. Efforts are moreover being made to develop kit formulation for radiosynthesis of [18F]FDS to simplify the production of this radiopharmaceutical [45]. Radiosynthesis of [18F]FDS for animal or clinical use can therefore be achieved in most molecular imaging departments, including those not equipped with radiochemistry facilities [37]. From a clinical perspective, the radiation dosimetry of this radiopharmaceutical was shown suitable for human use, although radiation exposure can be a limitation for the repeated use of [18F]FDS PET in humans [30].

This study has some limitations. The evaluation of the impact of FUS using [18F]FDS PET was evaluated in healthy animals only. As a consequence, FUS induced a large in-

crease in the BBB penetration of [18F]FDS, starting from a low baseline brain uptake. In this situation, statistical significance was reached using a small number of individuals. However, a large body of translational research supports that BBB integrity is compromised in many CNS pathological conditions including multiple sclerosis, hypoxic/ischemic insult, traumatic injury, Parkinson's and Alzheimer's diseases, epilepsy and brain tumors [1]. Different non-clinical models have shown that BBB leakage is a common feature of the neuroinflammatory cascade and contributes to disease-associated brain damage [3,46]. PET imaging is expected to offer more quantitative insight into BBB permeability in such pathophysiological conditions [27,47–49]. Nevertheless, further experiments are needed, probably in a larger number of animals, to test whether [18F]FDS has the sensitivity and appropriate test-retest variability to be useful as permeability marker to investigate the dynamics of disease-associated changes in BBB integrity in longitudinal studies. NMRI nu/nu mice are widely used as a model for cancer treatment, including orthotopic brain tumors, which will be the logical next step for the evaluation of the impact of FUS using [18F]FDS PET imaging [50]. We assume that the use of female individuals of this immunodeficient mouse strain in the present study is not likely to bias the immediate impact of FUS-induced BBB disruption compared with male and/or immunocompetent mouse strains. Nevertheless, immunodeficiency was shown to improve the long-term integrity of the BBB in a model of intracerebral hemorrhage [51]. Longitudinal studies using quantitative [18F]FDS PET imaging may be useful to further investigate the time-dependent cross-talk between the immune system and BBB permeability in pathophysiological situations [46].

Restoration of BBB structure and function is increasingly the subject of investigations as a therapeutic target to prevent or limit the outcome of neurological disorders [52]. Conversely, temporary FUS-induced BBB disruption is also regarded as a potential therapeutic strategy in various CNS diseases. FUS alone was shown to achieve promising results at reducing the amyloid load in Alzheimer's disease or at inducing neurogenesis [53,54]. This questions the long-term impact of the temporary FUS-induced BBB disruption in a situation where BBB integrity and function are already compromised [3]. This complex situation illustrates the crucial need for quantitative PET imaging techniques to untangle the impact of FUS on BBB permeability in relation to brain function in animal models and patients [18].

#### **5. Conclusions**

[ <sup>18</sup>F]FDS PET imaging presents essential properties to become an effective and quantitative marker of BBB permeability, which include availability, safety, low MW, low distribution across the intact BBB and low diffusion from the sonicated volume to the non-sonicated brain with intact BBB. This is demonstrated here for the first time using local induction of BBB permeability by FUS. [18F]FDS PET offers a very instrumental imaging marker to finely and dynamically measure BBB permeability variations over time scales of minutes.

**Author Contributions:** Conceptualization, G.H., A.N. and N.T.; methodology, S.G., L.B., B.L., A.W. and N.T.; software, G.H., S.G. and B.L.; validation, L.B., A.W., A.N. and N.T.; formal analysis, G.H., and A.D.; investigation, G.H., S.G., A.D., A.W. and A.N.; resources, A.N., B.L. and N.T.; data curation, N.T.; writing—original draft preparation, G.H. and N.T.; writing—review and editing, G.H, S.G., A.W., B.L., A.N. and N.T.; visualization, S.G., A.D. and L.B.; supervision, N.T. and A.N.; funding acquisition, A.N., A.W. and N.T. All authors have read and agreed to the published version of the manuscript.

**Funding:** Ambre Daubat received grant from Agence Nationale de la Recherche (grant DROPMUT, ANR-19-CE19-0011). Gaëlle Hugon received a PhD grant from the CEA (French Atomic and Alternative Energy commission) and the DGA (Direction Générale des Armées). Louise Breuil received a grant from CEA/AP-HP. This work was performed on a platform member of France Life Imaging network (grant ANR-11-INBS-0006).

**Institutional Review Board Statement:** All animal experiments were performed in accordance with the recommendations of the European Community (2010/63/UE) and the French National Committees (law 2013-118) for the care and use of laboratory animals. The experimental protocol was approved by a local ethics committee for animal use (Comité d'éthique pour expérimentation animale Ile-de-France n◦ 44) and by the French ministry of agriculture (APAFIS 16292-2018072609593031 v7, accepted the 1 June 2017).

**Informed Consent Statement:** Not applicable.

**Data Availability Statement:** Data is contained within the article or Appendix A.

**Acknowledgments:** We thank Maud Goislard and Philippe Gervais for helpful technical assistance.

**Conflicts of Interest:** The authors declare no conflict of interest.

#### **Appendix A**

Individual time-activity curves (SUV) units in the brain and arterial blood-pool of mice in the FUS group (*n* = 5) and the sham group (*n* = 3).





**Table A2.** *Cont.*

**Table A3.** SUV values—Blood (aorta).


#### **References**


MDPI St. Alban-Anlage 66 4052 Basel Switzerland Tel. +41 61 683 77 34 Fax +41 61 302 89 18 www.mdpi.com

*Pharmaceutics* Editorial Office E-mail: pharmaceutics@mdpi.com www.mdpi.com/journal/pharmaceutics

Academic Open Access Publishing

www.mdpi.com ISBN 978-3-0365-7698-5