Influence of P-Glycoprotein Inhibition or Deficiency at the Blood–Brain Barrier on ¹⁸F-2-Fluoro-2-Deoxy-d-glucose (¹⁸F-FDG) Brain Kinetics

Abstract

The fluorinated d-glucose analog ¹⁸F-2-fluoro-2-deoxy-d-glucose (¹⁸F-FDG) is the most prevalent radiopharmaceutical for positron emission tomography (PET) imaging. P-Glycoprotein’s (P-gp, MDR1, and ABCB1) function in various cancer cell lines and tumors was shown to impact ¹⁸F-FDG incorporation, suggesting that P-gp function at the blood–brain barrier may also modulate ¹⁸F-FDG brain kinetics. We tested the influence of P-gp inhibition using the cyclosporine analog valspodar (PSC833; 5 μM) on the uptake of ¹⁸F-FDG in standardized human P-gp-overexpressing cells (MDCKII-MDR1). Consequences for ¹⁸F-FDG brain kinetics were then assessed using (i) ¹⁸F-FDG PET imaging and suitable kinetic modelling in baboons without or with P-gp inhibition by intravenous cyclosporine infusion (15 mg kg⁻¹ h⁻¹) and (ii) in situ brain perfusion in wild-type and P-gp/Bcrp (breast cancer resistance protein) knockout mice and controlled d-glucose exposure to the brain. In vitro, the time course of ¹⁸F-FDG uptake in MDR1 cells was influenced by the presence of valspodar in the absence of d-glucose but not in the presence of high d-glucose concentration. PET analysis revealed that P-gp inhibition had no significant impact on estimated brain kinetics parameters K₁, k₂, k₃, V_T, and CMR_Glc. The lack of P-gp effect on in vivo¹⁸F-FDG brain distribution was confirmed in P-gp/Bcrp-deficient mice. P-gp inhibition indirectly modulates ¹⁸F-FDG uptake into P-gp-overexpressing cells, possibly through differences in the energetic cell level state. ¹⁸F-FDG is not a P-gp substrate at the BBB and ¹⁸F-FDG brain kinetics as well as estimated brain glucose metabolism are influenced by neither P-gp inhibition nor P-gp/Bcrp deficiencies in baboon and mice, respectively.

INTRODUCTION

¹⁸F-2-Fluoro-2-deoxy-d-glucose (¹⁸F-FDG), a d-glucose analog, is the most widely prescribed radiopharmaceutical agent for positron emission tomography (PET) imaging in Nuclear Medicine. PET with ¹⁸F-FDG is a valuable diagnostic modality in various diseases. PET/CT (computed tomography) with ¹⁸F-FDG has become an important component of cancer imaging: ¹⁸F-FDG PET/CT is useful for the detection, staging and assessment of cancer recurrence (1). In some situation, PET/CT with ¹⁸F-FDG is also used to detect and/or predict response/resistance to therapy and thus impact treatment decision making and monitoring (2–4). PET with ¹⁸F-FDG is also an accurate method for the investigation of regional human brain metabolism in health and disease states (5). Alzheimer’s disease is characterized by a specific regional pattern of reduced ¹⁸F-FDG uptake, allowing early and preclinical diagnosis (6). Brain PET imaging using ¹⁸F-FDG is also performed routinely during presurgical evaluation in pharmacoresistant epilepsy. The locally reduced ¹⁸F-FDG uptake allows the noninvasive localization of the epileptic focus and is predictive of a seizure-free outcome after surgical resection (7).

From a molecular imaging perspective, it is important to understand the factors that govern ¹⁸F-FDG kinetics in tissues and assess the biological correlates of ¹⁸F-FDG uptake in both the tumor and the brain (8,9). Basically, ¹⁸F-FDG is transported through the cell membrane into the cytosol by glucose transporters (GLUT), phosphorylated by hexokinase (HK), and metabolically trapped in the cell as ¹⁸F-FDG-6-phosphate. Thus, ¹⁸F-FDG accumulation depends on its rate of transport through the cell membrane (influx and potential efflux), the HK activity, and the (low) rate of dephosphorylation in the cell (9). Tumors generally exhibit increased transport of glucose by overexpression and/or expression of avid GLUT isoform and increased HK activity (10,11). In the brain, GLUT1 (SLC2A1), a member of solute carrier (SLC) transporter family, is the predominant glucose transporter at the blood–brain barrier (BBB) and is also responsible for the transfer of ¹⁸F-FDG from the blood to the brain (12).

The P-glycoprotein (P-gp; ABCB1) is a member of the ATP-binding cassette (ABC) transporter family, also known as multidrug-resistant protein 1 (MDR1). P-gp mediates the unidirectional efflux of many structurally and mechanistically unrelated drugs across the plasma membrane. P-gp overexpression in cancer cells is a key component of the multidrug resistance (MDR) phenotype, which is associated with chemotherapeutic agent treatments failure (13). Numerous in vitro studies, performed using different cancer cell lines, have revealed that MDR or P-gp-overexpressing cells incorporate ¹⁸F-FDG at lower levels (14,15). This inverse relationship between P-gp function and ¹⁸F-FDG uptake was also reported in vivo with selected cancer cells implanted as xenografts: P-gp-overexpressing tumors showed reduced ¹⁸F-FDG uptake (16,17), which was restored using pharmacological P-gp inhibition (17). In patients, an inverse relationship was found between ¹⁸F-FDG uptake and the level of P-gp expression examined in tumor resection (lung cancer, hepatocellular carcinoma, and intrahepatic cholangiocarcinoma) (18–22). Therefore, it was hypothesized that ¹⁸F-FDG may be a substrate of P-gp (15,22). Conversely, it was also reported an enhanced ¹⁸F-FDG uptake in various P-gp-expressing cancer cells compared to P-gp negative in vitro (23–25). No explanation was given so far for this discrepancy. Nevertheless, these controversial results suggest that P-gp function may be an additional modulatory factor for ¹⁸F-FDG accumulation in tissues (8).

P-gp is also physiologically expressed at the BBB and limits the distribution of many drugs into the central nervous system (CNS) (26). The level of expression and/or functionality of P-gp can be modulated by inhibition, up- or down-regulation. This may affect the brain distribution and subsequent efficacy or toxicity of many P-gp substrates (27). Recent PET studies have revealed that P-gp function is decreased in patients with Alzheimer’s disease (28). P-gp overexpression is also proposed as a possible explanation for pharmacoresistance in epilepsy: Up-regulation of P-gp and other efflux transporters in capillaries, as well as aberrant expression in glial and neuronal cells, has been reproducibly demonstrated in surgically resected brain specimens (29).

Both the P-gp and GLUT1 are physiologically coexpressed at the BBB (26). Few studies have addressed the relationship between P-gp function at the BBB and ¹⁸F-FDG brain kinetics in a physiological condition. Tariquidar is a potent inhibitor of both the P-gp and the breast cancer resistance protein (BCRP and ABCG2), two major efflux transporters expressed at the BBB (26). It was reported that tariquidar induces a 45% increase in the brain uptake of ¹⁸F-FDG in rats, suggesting that ABC-transporter function at the BBB may also modulate ¹⁸F-FDG uptake by the brain (30). This also questions whether P-gp deficiency observed at the BBB in some physiopathological states may also impact ¹⁸F-FDG brain kinetics.

In the present study, we used standardized Madin–Darby canine kidney II (MDCKII) cells transfected with the human MDR1 (multidrug resistance 1) gene, instead of cancer cell lines, to test the influence of human P-gp function on ¹⁸F-FDG uptake. The overall influence of P-gp inhibition on ¹⁸F-FDG brain kinetics was then assessed using PET imaging and kinetic modeling in nonhuman primates. Finally, we used in situ brain perfusion in mice to test the influence of combined P-gp/Bcrp deficiency on ¹⁸F-FDG brain uptake under controlled d-glucose exposure to the brain.

MATERIAL AND METHODS

Animals

All animal use procedures were in accordance with the recommendations of the European Community (86/809/CEE) and the French National Committees (law 87/848) for the care and use of laboratory animals. The experimental protocol was evaluated by a local ethic committee for animal use (CETA reference no. 10-004). PET studies were carried out in adult male Papio anubis baboons (3–4 years old, 12–15 kg in weight during the study). Adult male Fvb mice (20–30 g, 7–11 weeks old) were obtained from Janvier (France). The dual P-gp/Bcrp-deficient mouse strain Mdr_1a;Mdr_1b;Bcrp^{(−/−;−/−;−/−)} (derived from Fvb mice), also called P-gp/Bcrp^(−/−), was bred in-house from progenitors obtained from the laboratory of Dr. Alfred H. Schinkel (The Netherlands Cancer Institute, The Netherlands).

Chemicals and Radiochemicals

Cyclosporine (CsA) was administered using the commercial drugs Sandimmun® (Novartis, Basel, Switzerland). Valspodar (PSC833), a nonimmunosuppressant CsA derivative, was a gift from Novartis. ¹⁸F-2-fluoro-2-deoxy-d-glucose (¹⁸F-FDG, Glucotep®) was purchased from Cyclopharma (Saint-Beauzire, France). ^99mTc-sestamibi was prepared from the Cardiolite® kit (Iba Molecular, Gif-sur-Yvette, France).

Cell Culture and 18F-FDG Uptake Assay

The Stably Transfected Madin–Darby Canine Kidney Cell Lines

MDCKII-hMDR1 were a gift from Dr. Alfred H. Schinkel (The Netherlands Cancer Institute, The Netherlands). Culture and Western blot analysis of transporter expression was performed as previously described (31). Cells were seeded onto 24-well plates (3 × 10⁵ cells/well) and allowed to grow for 3 days to reach confluence. On the day of the experiment, the culture medium was removed and cell monolayers washed twice with PBS. Monolayers were then incubated with 200 μL incubation buffer containing tracer dose ¹⁸ F-FDG (∼3.7 MBq/mL) and the P-gp substrate ^99mTc-sestamibi (∼3.7 MBq/mL), as a positive control for P-gp function (32). Tracer uptake was measured in the presence of the P-gp inhibitor valspodar (5 μM) compared to vehicle [DMSO 0.5% (v/v) in incubation buffer]. The experiments were performed either in standard Dulbecco’s modified Eagle’s medium (DMEM) containing a 17.5-mM d-glucose concentration (“high d-glucose”) or in the absence of d-glucose. Plates were stored at 37°C under a humidified atmosphere. At selected times, ranging from 0 to 120 min, incubation medium was removed, the monolayers (n = 4–8 for each condition) were immediately washed twice with ice-cold PBS and cells lysed using 500 μL NaOH 1 N. Four hundred microliter samples from each well were counted using a calibrated γ-counter. The fluorine-18 energy channel was selected to estimate fluorine-18 decay corrected radioactivity expressed in Bq (half-life = 109.8 min). After 24-h decay, the samples were counted in the technetium-99m energy window to measure technetium-99m radioactivity (half-life = 6.02 h). Cell viability after 120-min incubation in glucose-free medium was assessed using trypan blue exclusion assay following the manufacturer instructions (Sigma-Aldrich, Saint-Quentin, France).

PET Imaging in Baboons

We studied the brain kinetics of ¹⁸F-FDG with or without P-gp inhibition using CsA in normoglycemic (12 h fasting) in three different adult baboons. Fasting animals were anesthetized, and PET data were acquired as previously described (33). Baboons were artificially ventilated, maintained, anesthetized with 66% N₂O/1% isoflurane. Experiments were performed on a HR+ Tomograph (Siemens Healthcare, Knoxville, TN, USA) for brain PET studies. CsA (15 mg kg⁻¹ h⁻¹, i.v.) was infused 30 min before beginning and throughout PET scanning. This P-gp inhibition protocol, based on previous report (34), was validated in-house using the same baboon species and ¹¹C-N-desmethyl-loperamide, a reliable PET probe for P-gp activity at the BBB (data not shown) (35). Heart rate, end-tidal pCO₂, and rectal temperature were continuously monitored. After a transmission scan for the purpose of subsequent attenuation correction, animals were then injected i.v. with 100–200 MBq ¹⁸F-FDG. During PET acquisition (60 min), arterial plasma samples were obtained to estimate the arterial input function (at 0, 0.25, 0.5, 0.75, 1, 1.25, 1.5, 1.75, 2, 3, 5, 7.5, 10, 15, 20, 30, 40, 50, and 60 min) and blood glucose measurement (Cobas® 6000 analyzer, Roche Diagnostics, France) in each animal. Dynamic PET data were acquired over the brain in three-dimensional mode with the animal supine. Emission data were acquired for 60 min after intravenous injection of ¹⁸F-FDG. Each scan consisted of a sequence of 24 frames (1 × 15 s, 3 × 30 s, 1 × 45 min, 3 × 1 min, 1 × 1.5 min, 6 × 2 min, 1 × 3.5 min, and 8 × 5 min). Measured attenuation and scatter corrections were applied to the emission data. Data were reconstructed using the reprojection algorithm with an axial and transaxial Hanning filter that had a 0.5 cutoff frequency. The resulting images had a spatial resolution of approximately 5–7 mm in full width at half maximum.

Control and CsA PET images were coregistered (rigid coregistration) for each animal using PMOD® software (version 3.5). Regions of interest (ROI; brain hemispheres and cerebellum) were drawn manually on control conditions PET images. ROIs were then applied to coregistered dynamic PET data obtained with CsA. Time–activity curves (TACs) were generated by calculating the mean radioactivity in the brain hemispheres and cerebellum. TACs were generated by correcting the radioactivity in each region for fluorine-18 decay, injected dose, and animal weight. TACs are expressed as standard uptake value (SUV; i.e., mean radioactivity/injected dose/weight) versus time. Kinetic modeling was then performed using a two-tissue compartment model considering the arterial plasma input function with PMOD® software. Outcome parameters K₁ (mL cm⁻³ min⁻¹), k₂ (L min⁻¹), k₃ (L min⁻¹), k₄ (L min⁻¹) and V_T (mL cm⁻³) were estimated in each condition from time–activity data in the brain and plasma (36). Cerebral metabolic rate of glucose utilization (CMR_Glc ; μmol/min/100 g) was estimated in each region from measured plasma glucose and a lumped constant set at 0.89 (37), assuming that P-gp inhibition has no influence on the lumped constant.

$${\mathrm{C}\mathrm{M}\mathrm{R}}_{\mathrm{G}\mathrm{l}\mathrm{c}}=\left(\frac{\mathrm{plasma}\mathrm{glucose}}{\mathrm{lumped}\mathrm{constant}}\right)\times \left(\frac{K_1\cdot k_3}{k_2+k_3}\right)$$

CMR_Glc parametric maps were then generated to allow visual comparison between tested conditions, taking arterial input function and plasma glucose into account (PMOD® software).

In Situ Brain Perfusion in Mice

The intrinsic initial transport rate (or brain uptake clearance) at the luminal BBB membrane for ¹⁸F-FDG was measured in wild-type mice and in P-gp/Bcrp^(−/−) mice. Animals perfused with protein-free Krebs-carbonate buffer physiological saline containing a controlled and physiological concentration of d-glucose (9 mM), ¹⁸F-FDG (3.7 MBq/mL), and ^99mTcO₄⁻ (3.7MBq/mL) used as a vascular integrity marker. The solution was gassed with 95% O₂/5% CO₂ for pH control (7.4) and warmed to 37°C. Once anesthetized with ketamine (140 mg/kg, i.p.) + xylazine (8 mg/kg, i.p.), the right common carotid artery was exposed, and the external carotid artery was ligated at the level of the bifurcation of the common and internal carotid arteries. The right common carotid artery was catheterized and the catheter connected to a syringe containing the perfusion fluid. At this time, the perfusion fluid replaces all the blood in the right brain vasculature. The thorax was then opened, the heart was cut, and perfusion started immediately at a flow rate of 2.5 mL/min (38). Each mouse was perfused for 120 s. The intrinsic transport rate (K_in; μL g⁻¹ s⁻¹, n = 5) for ¹⁸F-FDG was calculated as previously reported (38). The brain intravascular volume was taken into account using the distribution of ^99mTcO₄⁻, which does not measurably cross the BBB (39). The distribution volume of ¹⁸F-FDG (V_brain; μL/g) and its brain uptake clearance (K_in = V_brain/120 s ; μL g⁻¹ s⁻¹) were calculated from decay corrected ¹⁸F radioactive count in the right brain hemisphere.

Statistical Analysis

Results are shown as mean ± standard deviation. In vitro data were log transformed and compared using a two-way ANOVA with “valspodar” and “d-glucose presence” as two main factors followed by Tukey’s post hoc test. In vivo data were compared using the Student’s t test. Statistical significance was set at p < 0.05 (R software version 3.1.2 (40)). The size of the different groups was set from a previously reported study using similar methods in both the mice and baboons (33).

RESULTS

In Vitro¹⁸F-FDG Uptake Assay

In the absence of d-glucose, ¹⁸F-FDG uptake by MDCKII-MDR1 cells appeared linear in the first 60 min. Then, in a second phase, ¹⁸F-FDG uptake increased more slowly. In the presence of the P-gp inhibitor valspodar, ¹⁸F-FDG uptake increased linearly from 0- to 120-min incubation. Valspodar significantly reduced ¹⁸ F-FDG uptake from 0 to 60 min. Conversely, ¹⁸F-FDG uptake measured after 120-min incubation was ∼14% higher in the presence of valspodar (Fig. 1).

Fig. 1.

Time course of ¹⁸F-FDG and ^99mTc-sestamibi accumulation in MDCKII-MDR1 cells. The uptake of ¹⁸F-FDG (a, b) and ^99mTc-sestamibi (c, d) in MDCKII-MDR1 monolayers was measured in the presence and the absence of the P-gp inhibitor valspodar (mean ± SD; n = 4−8 vial for each condition). ^99mTc-sestamibi, a P-gp substrate, is used as a positive control for P-gp activity. Data obtained in the absence of d-glucose in the incubation medium (a, c) were compared to those obtained in the presence of high d-glucose concentration (17.5 mM; b, d). ANOVA followed by followed by Tukey’s post hoc test was performed to assess statistical difference between groups: (i) absence compared to presence of valspodar, ***p < 0.001; *p < 0.05; n.s = nonsignificant and (ii) absence compared to presence of high d-glucose concentration ^# p < 0.001 when significant

The presence of a high d-glucose concentration in the incubation medium resulted in a significant ∼95% decrease in ¹⁸F-FDG uptake by MDR1 cells. ANOVA unveiled an interaction between the “d-glucose” and “valspodar” factors. P-gp inhibition using valspodar had no significant impact on ¹⁸F-FDG uptake in the presence of high d-glucose concentration in incubation medium over 120 min.

Valspodar significantly increased the accumulation of the known P-gp substrate ^99mTc-sestamibi (p < 0.001) in the presence and the absence of d-glucose in the medium. When P-gp is inhibited, the presence of d-glucose significantly increased ^99mTc-sestamibi cellular accumulation, while it was decreased in the presence of functionnal P-gp  (p < 0.001) (Fig. 1c, d).

PET Study in Baboon

PET images revealed a similar distribution of  ¹⁸F-FDG in brain regions with or without P-gp inhibition (Fig. 2). Representative TACs obtained in a monkey in the presence and the absence of CsA are shown in Fig. 3. PET quantification and kinetic analysis performed in several brain regions in baboon allowed the estimation of outcome parameters that were shown to reflect ¹⁸F-FDG kinetics and metabolism (36,41). Kinetic analysis revealed that P-gp inhibition with CsA does not influence ¹⁸F-FDG the transport rate from the blood to the brain (K₁). Neither was the brain to blood efflux rate (k₂) significantly affected by CsA. We found no difference in the k₃ value that was shown to mainly reflect brain HK activity. We obtained much better fits to the data when k₄ was included in the compartment model (i.e., k₄ not set to zero). Nevertheless, the existence of significant dephosphorylation of ¹⁸F-FDG-6-P to free ¹⁸F-FDG is still a source of much debate (42). TACs unveiled a surprisingly high ¹⁸F-FDG uptake in the cerebellum (Fig. 3). CMR_Glc was also higher in the cerebellum than in the brain hemispheres (Table I, Fig. 3).

Fig. 2.

Representative CMR_Glc parametric maps obtained in the brain of control (a) and cyclosporine-treated (b) baboons anesthetized using isoflurane. CMR_Glc reflects the local glucose metabolism (μmol/min/100 g). Parametric map was generated using PMOD® software (version 3.5) starting from ¹⁸F-FDG dynamic PET images, arterial input function, plasma glucose, and a lumped constant set at 0.89

Fig. 3.

Representative time–activity curves obtained in brain hemispheres and cerebellum of control (a) and cyclosporine-treated (b) baboons. Data are expressed as standard uptake value (SUV) versus time

Table I.

¹⁸F-FDG Kinetic Parameters Measured in Brain Regions of the Baboon Using the ¹⁸F-FDG 2-tissue Compartment Model in Control Conditions (Mean ± S.D; n = 3) and in the Presence of the P-gp Inhibitor Cyclosporine 15 mg kg⁻¹ h⁻¹ (CsA, Mean ± SD; n = 3). Student’s t Test Analysis Revealed No Difference Between Control and CsA

		K 1 (mL cm−3 min−1)	k 2 (L min−1)	k 3 (L min−1)	k 4 (L min−1)	CMRGlc (μmol/min/100 g)	V T (mL cm−3)
Brain hemispheres	Control	0.137 ± 0.031	0.263 ± 0.078	0.0485 ± 0.0037	0.0217 ± 0.0038	11.4 ± 0.9	1.73 ± 0.18
	CsA	0.113 ± 0.016	0.202 ± 0.048	0.0367 ± 0.0139	0.0174 ± 0.0108	10.8 ± 2.7	2.06 ± 0.86
Cerebellum	Control	0.164 ± 0.050	0.295 ± 0.103	0.0594 ± 0.0068	0.0233 ± 0.0053	14.9 ± 3.5	2.03 ± 0.17
	CsA	0.136 ± 0.026	0.225 ± 0.053	0.0446 ± 0.0102	0.0190 ± 0.0035	14.4 ± 3.7	2.04 ± 0.24

¹⁸F-FDG in situ Brain Perfusion

We used the in situ brain perfusion technique and a perfusion fluid containing a controlled and physiological concentration of unlabeled d-glucose, so that ¹⁸F-FDG brain distribution is not likely to be influenced by peripheral blood glucose. ¹⁸F-FDG initial transport BBB rate measured in wild-type mice was 1.76 ± 0.24 μL g⁻¹ s⁻¹ (n = 5). The brain uptake clearance of d-glucose measured under similar conditions was K_in = 1.5 μL g⁻¹ s⁻¹ (38). This confirms previous data showing that ¹⁸F-FDG accumulates at a higher rate than d-glucose itself (42). We found no difference with the K_in of ¹⁸F-FDG measured in P-gp/Bcrp-deficient mice (1.71 ± 0.15 μL g⁻¹ s⁻¹; n = 5).

DISCUSSION

The relationship between ¹⁸F-FDG uptake and P-gp function was initially studied in cancer cells. MDR phenotype was mostly obtained by culturing drug-sensitive parental cells in the presence of cytotoxic drugs to promote selection. These in vitro studies concluded to a reduced (14,16) or enhanced (23–25) uptake of ¹⁸F-FDG in MDR cells as compared to MDR negative. First, it may be hypothesized that chemotherapeutic agents used for the selection of MDR clones may differentially affect glucose metabolism and affect ¹⁸F-FDG uptake. In addition to the variety of cell lines and culture condition (25), we show that these controversies may result from different incubation duration used in the different studies that ranged from 10 to 120 min. Our results show that, in the absence of d-glucose, P-gp inhibition using valspodar, a CsA derivative, enhanced ¹⁸F-FDG uptake after 120-min incubation, while it was reduced at earlier incubation times (Fig. 1). This time-dependent shift was also found in cancer cells (17,25). Interestingly, CsA was previously shown to abolish the difference in ¹⁸F-FDG uptake between MDR-positive and MDR-negative cell lines (23). In both the MDR-positive and MDR-negative cells, neither ATP consumption nor lactate production was influenced by CsA (43). This suggests that the effect of CsA, and probably valspodar, on ¹⁸F-FDG uptake is related to P-gp function and does not reflect a potential effect of these compounds on cellular glucose metabolism.

It was recently shown that high concentration FDG does not impact ATP hydrolysis in P-gp-overexpressing cells (25). However, inconsistencies for P-gp substrates to impact the P-gp ATPase activity have been reported, and ATPase assay appeared insufficient to conclude to a lack of P-gp mediated transport (44). Standardized transfected epithelial cell line such as MDCKII-MDR1 including a positive control are recommended to test the transport function of chemical entities by P-gp and bidirectional transport assay is usually preferred (44). In this model, the fast and extensive metabolism and trapping of ¹⁸F-FDG into cells would undoubtedly limit ¹⁸F-FDG recovery from the donor to the receiver chamber. We used uptake assay instead, also allowing comparison to previous studies performed using different cancer cell lines.

In the absence of d-glucose and in contrast to ^99mTc-sestamibi, ¹⁸F-FDG uptake was found lower during the first 60-min incubation, when P-gp was inhibited. At 120 min, the effect of P-gp inhibition on ¹⁸F-FDG uptake in MDCKII-MDR1 cells (14% increase) was found significant but much lower than that previously obtained in breast carcinoma Bcap37-MDR1 cells (50% increase) (17). We may nonetheless hypothesize that this result may be biased by cell metabolism shift due to prolonged incubation in a d-glucose-free medium, although cell viability was >95% after 120-min incubation. MDCKII cells express GLUT1 and GLUT8 (45). We used a high concentration of d-glucose to unveil the potential transport of ¹⁸F-FDG by P-gp, apart from the GLUT activity. Unsurprisingly, ¹⁸F-FDG uptake was dramatically decreased in the presence of high d-glucose (Fig. 1). ¹⁸F-FDG cellular accumulation increased with time, suggesting significant ¹⁸F-FDG uptake, even in the presence of high d-glucose. In parallel, high d-glucose medium abolished the effect of P-gp inhibition on ¹⁸F-FDG uptake without impairing the ability of P-gp to mediate the efflux of ^99mTc-sestamibi.

It was noticed that d-glucose deprivation decreased the cellular/mitochondrial retention of ^99mTc-sestamibi in MDCKII-MDR1 cells when P-gp was inhibited (Fig 1d). ^99mTc-sestamibi uptake is known to depend on the mitochondrial metabolic activity and cell viability (46). In the present study, trypan blue exclusion assay did not detect any additional cell death in the absence of d-glucose after 120-min incubation. This suggests that cellular metabolic shift induced by d-glucose deprivation may be detected in vitro using ^99mTc-sestamibi.

Altogether, these results confirm that ¹⁸F-FDG is not transported by P-gp (25). Phosphorylation of ¹⁸F-FDG by HK, as well as P-gp-mediated transport, may competitively modulate d-glucose and intracellular ATP concentration. ^99mTc-sestamibi is a hydrophilic molecule (log P = 0.7) and was incubated at same tracer concentration (approximately pM) in each vial (47). Therefore, the presence of ^99mTc-sestamibi is not likely to influence ATPase activity (48). Despite differences in ¹⁸F-FDG uptake, it was reported that intracellular ATP content (14,49), HK activity (14), and GLUT transporter expression (25) are similar in MDR-positive and MDR-negative cells. Therefore, further experiments are needed to explain, at a molecular level, how P-gp function indirectly modulates the GLUT/HK-dependant ¹⁸F-FDG cellular distribution.

High level of P-gp and GLUT-1 expression were found in the endothelial cells of brain microvessels forming the BBB (26). In baboon, CsA did not impact K₁, suggesting that ¹⁸F-FDG uptake by the brain is not influenced by this drug even at high dosing used for P-gp inhibition in primate (34). We found no influence on estimated k₃ values, suggesting that CsA has no consequence on HK activity in the brain. CsA pretreatment did not either impact CMR_Glu in the cerebellum and other regions of the brain (Table I). Altogether, these results, obtained in primates, suggest that ¹⁸F-FDG kinetics and glucose metabolism are influenced by neither P-gp inhibition nor CsA administration in the healthy brain. It was observed a surprisingly high ¹⁸F-FDG uptake in the cerebellum compared to the brain hemispheres. This phenomenon was observed in all tested animals, regardless of CsA exposure. It was reported that anesthetics may differentially influence ¹⁸F-FDG brain kinetics (50,51). In the present study, baboons received the same anesthetic protocol (ketamine followed by isoflurane). Further investigations are needed to test the potential influence of this anesthetic protocol on regional ¹⁸F-FDG brain kinetics.

In rat, it was shown that administration of tariquidar, a combined P-gp/Bcrp inhibitor, resulted in a ∼45% increase in CMR_Glc (30). P-gp-unrelated effects of this molecule on peripheral and brain glucose metabolism was proposed as an explanation for its unexpected effect on the brain uptake of ¹⁸F-FDG (30). Therefore, we used transporter-deficient mice instead of pharmacological inhibition and performed in situ brain perfusion, so that ¹⁸F-FDG uptake by the brain is not influenced by peripheral blood glucose. This invasive method is an accurate approach to measure the influence of carrier-mediated transport on the initial distribution (e.g., brain uptake) of compounds at the BBB. In P-gp/Bcrp-deficient mice, the brain clearance parameter (K_in) of the P-gp substrate ^99mTc-sestamibi was shown to be 2.7-fold greater than in wild-type mice (52). As P-gp and Bcrp seem to work together in a synergistic fashion, dual P-gp/Bcrp inhibition or deficiencies could sometimes be a better strategy to reveal the carrier-mediated transport of their common substrates at the BBB (53,54). Our results show that combined P-gp/Bcrp deficiency had no influence on ¹⁸F-FDG distribution by the healthy brain and that ¹⁸F-FDG is not either likely to be a substrate of the rodent P-gp or Bcrp.

CONCLUSION

In vitro assay confirmed the modulatory effect of valspodar on ¹⁸F-FDG uptake in P-gp-overexpressing cells, although ¹⁸F-FDG is not a P-gp substrate. In vivo, ¹⁸F-FDG brain kinetics was not measurably influenced by P-gp inhibition and/or depletion, starting from a physiological P-gp expression at the BBB. Clinical ¹⁸F-FDG PET provides useful information regarding the brain metabolism in several pathophysiological states in human. This study suggests that P-gp decline at the BBB is not a major factor of ¹⁸F-FDG brain kinetics and subsequent estimation of brain glucose metabolism.