A New in Vivo Method to Study P-Glycoprotein Transport in Tumors and the Blood-Brain Barrier¹

Resistance of tumors to chemotherapeutic drugs is a major problem in the treatment of cancer patients. Important anticancer drugs, such as anthracyclines (daunorubicin and doxorubicin), Vinca alkaloids (vincristine, vinblastine), epipodophyllotoxins (etoposide), and taxanes (paclitaxel) are involved in the so-called MDR³ (1, 2, 3, 4, 5). Several mechanisms are responsible for MDR. One of them is the overexpression of the MDR1 gene, resulting in increased levels of the ATP-dependent M_r 170,000 P-gp drug efflux pump (2 , 6, 7, 8). Apart from the above-mentioned anticancer drugs, several other nonchemotherapeutic drugs are substrates for this pump (9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19). For instance, for several HIV-1 protease inhibitors such as indinavir, nelfinavir, and saquinavir, it is shown in vitro and in vivo that P-gp plays a role in the transport function of these HIV-1 protease inhibitors (19). Also cyclosporin A (immunosuppressive agent), domperidone (antiemetic agent), and loperamide (antidiarrheal drug) are transported by P-gp (18). To overcome MDR, several drugs have been identified that can inhibit transport by P-gp in P-gp-expressing tissues. Modulation with relatively nontoxic reversal agents such as cyclosporin A and its nonimmunosuppressive analogue PSC833 might increase the pharmacological effects of MDR involved chemotherapeutic drugs in different normal tissues (20, 21) and in P-gp-overexpressing solid tumors. However, until now, clinical studies in solid tumors combining reversal agents with MDR chemotherapeutic drugs are relatively disappointing (10, 22, 23). However, in these studies the P-gp functionality in the tumors is unknown. In these tumors P-gp may be present but does not have to be functional. Then modulation will be disappointing. A technique that would allow us to visualize blockade by modulators of the P-gp drug efflux pump might help to select the proper patients for treatment of MDR involved chemotherapeutic drugs in combination with a modulator. Until now, several studies have been described with ^99mTc-labeled substrates and single photon emission computed tomography to visualize a decreased uptake of ^99mTc-sestamibi and ^99mTc-Q-complexes in solid tumors and in the blood-brain barrier due to P-gp expression (24, 25, 26, 27, 28). However, we and others observed for example that ^99mTc-sestamibi is apart from P-gp, also a substrate for MRP (29, 30). In the present study, the pharmacokinetics of [¹¹C]verapamil is investigated with PET in tumor-bearing rats. For these studies, rats were inoculated with a human P-gp-negative small cell lung carcinoma cell line GLC₄ and its MDR1 gene-transfected, P-gp-overexpressing subline GLC₄/P-gp. The potential of imaging P-gp reversal effect by cyclosporin A on the tumor pharmacokinetics of [¹¹C]verapamil was analyzed by PET and validated with biodistribution studies with [¹¹C]daunorubicin and [¹¹C]verapamil. Namely, when [¹¹C]verapamil is combined with cyclosporin A, both compounds are actively transported by P-gp. Because of competition for P-gp, the amount of transported [¹¹C]verapamil will be diminished, resulting in increased radioactivity in the tumor.

RPMI 1640 and FCS were purchased from Life Technologies, Inc. (Paisley, United Kingdom) and 3-(4,5-dimethyl-thiazol-2-yl)-2,5-diphenyl tetrazolium bromide from Sigma Chemical Co. (St. Louis, MO). Ketamine (Ketalar; 50 mg/ml) was obtained from Parke-Davis (Münich, Germany), xylazine (Rompun solution 2%) was from Bayer (Leverkusen, Germany), cyclosporin A (50 mg/ml) in Cremophor EL (650 mg/ml; Sandimmune) was from Sandoz (Basel, Switzerland), Matrigel was from Becton Dickinson (Bedford, MA), and C219 was from Thamer Diagnostica (Uithoorn, the Netherlands). N-Nor-methyl verapamil was purchased from RBI (Natick, MA), and vincristine was from Eli Lilly (Indianapolis, IN). 9-Formyl trifluoroacetyl daunorubicin was a kind gift from Farmitalia Carlo Erba (Dr. Suarato, Milano, Italy).

[¹¹C]d,l-daunorubicin and [¹¹C]d,l-verapamil (t_1/2 = 20.4 min) were prepared as described by Elsinga et al. (31). The specific activity of [¹¹C]daunorubicin was >370 GBq/mmol and [¹¹C]verapamil was 555 GBq/mmol at the end of the synthesis. The ¹¹C-labeled radiopharmaceuticals were purified by high-pressure liquid chromatography (31). After evaporation of the eluent, radioactivity was dissolved in sterile saline.

The human small cell lung carcinoma cell line GLC₄ and the human small cell lung carcinoma cell line GLC₄/P-gp were cultured in RPMI 1640/10% FCS in a humidified atmosphere with 5% CO₂ at 37°C. The P-gp-overexpressing cell line was obtained after infection of GLC₄ cells with a MDR1 gene-carrying retrovirus. The MDR1 gene was introduced into the retrovirus-producing packaging cell line GP+envAm12, and clones resistant to 50 nm vincristine were pooled (32). The virus titer was 1 × 10⁴ vincristine-resistant colony-forming units/ml on NIH-3T3 cells. Subsequently, GLC₄ tumor cells were infected with virus supernatant, and selection with 50 nm vincristine resulted in the GLC₄/P-gp subline. Stable resistance in the GLC₄/P-gp cell line was assured by culturing this line in the presence of vincristine (50 nm) twice a week. Daunorubicin resistance was tested in vitro with a cytotoxicity test, the microculture tetrazolium assay (33). Cells in logarithmic growth phase, 3,750 and 10,000 per well for GLC₄ and GLC₄/P-gp, respectively, were incubated for 4 days with daunorubicin. For GLC₄ cells, the concentration range was 10–90 nm and 50–1250 nm in 0.1 ml of culture medium for GLC₄/P-gp. On day 4, the percentage of cell survival was calculated as the mean of test samples/mean of untreated samples. Controls consisted of media without cells (background extinction) and cells incubated with medium instead of the drug. Three independent experiments were performed, each in quadruplicate. Results were expressed as mean ± SD. From the survival curves, the daunorubicin concentrations were determined that inhibited cell survival by 50% (IC₅₀).

Male nude rats (HSD Ham RNU rnu; Harlan, Horst, the Netherlands; 280–300 g; age, 6 weeks) were injected with GLC₄ and GLC₄/P-gp tumor cells (10⁷ cells/0.1 ml RPMI/FCS 10%). Before injection of tumor cells, GLC₄ and GLC₄/P-gp cells (10⁷/0.1 ml RPMI/FCS 10%) were mixed with 0.1 ml Matrigel. Subsequently, the GLC₄ or GLC₄/P-gp cell suspension was injected s.c. in the left and the right sides, respectively, of the nude rat. Ten to 14 days after injection of the cells, solid tumor nodules of ∼1 cm diameter were grown. Thereafter, the tumors were removed, and pieces of GLC₄ and GLC₄/P-gp tumors (∼2–3 mm in diameter) were transplanted in the left and the right sides, respectively, in other male nude rats (HSD Ham RNU rnu; Harlan). Fourteen days after transplantation of these GLC₄ and GLC₄/P-gp tumor pieces, tumors of 1–2 cm in diameter were developed. These rats were used various experimental settings. All experiments were carried out in compliance with the local guidelines.

To allow histological data of GLC₄ and GLC₄/P-gp tumors, the 1–2 cm-sized tumors were excised and frozen in liquid nitrogen. From the frozen tumor tissue, 4-μm slices were cut. The cell viability and the cell density were measured in H&E-stained slides. Cell density was scored by counting the cells in 10 microscopic fields of 2.5 × 10⁻⁴ mm³ (0.25 × 0.25 × 0.004 mm) microscopic fields on H&E-stained slides obtained from three different tumors for both tumor types (mean ± SD).

The P-gp expression in cell lines and tumor tissue was tested with the P-gp-directed monoclonal antibody C219. To check whether the MRP₁ expression was induced, MRP₁ was stained with monoclonal antibody MRPrl (kindly provided by Dr. R. Scheper, Free University, Amsterdam, the Netherlands) on cytospins and tumor tissue (34).

A total of 2 × 10⁶ GLC₄ and GLC₄/P-gp cells were incubated in polystyrene tubes for 60 min at 37°C with 5 μm [¹¹C]daunorubicin or 5 μm [¹¹C]verapamil in 5 ml of RPMI 1640/10% FCS. To study modulating effects, GLC₄ and GLC₄/P-gp cell lines were incubated simultaneously with cyclosporin A (50 μm) and daunorubicin (5 μm) or verapamil (5 μm). Thereafter, the cells were washed with 5 ml of ice-cold PBS (0.14 m NaCl, 2.7 m KCl, 6.4 mm Na₂HPO₄, and 1.5 mm KH₂PO₄, pH 7.4), followed by centrifugation (5 min of 180 g; 4°C). The supernatant was removed, and the cellular radioactivity was measured in water in a gamma counter (LKB Wallac, Turku, Finland). Correction for the extracellular adhesion of radioactivity to polystyrene tubes was performed by subtracting the results obtained after incubation with [¹¹C]daunorubicin or [¹¹C]verapamil for 60 min at 4°C. The cellular accumulation is expressed as % dose/10⁶ cells (mean ± SD).

Tumor rats were anesthesized with ketamine:xylazine (2:1; 1 ml/kg). Thereafter, 0.3 ml [¹¹C]daunorubicin (10 mg/kg) or [¹¹C]verapamil (0.1 mg/kg) was injected into the tail vein. After 60 min, the rats were sacrificed by extirpation of the heart, and several tissues were dissected. Heparin-plasma was obtained from the collected blood by centrifugation (3 min; 1000 g). Radioactivity was measured with a gamma counter. Modulation of [¹¹C]daunorubicin and [¹¹C]verapamil kinetics with cyclosporin A was tested in the nude rats. Cyclosporin A was injected into the penile vein (50 mg/kg). After 30 min, [¹¹C]verapamil or [¹¹C]daunorubicin was administered into the tail vein. Sixty min after the injection of [¹¹C]verapamil or [¹¹C]daunorubicin, the biodistribution of ¹¹C-labeled radioactivity was measured as described above. Results are expressed as ng/g tumor for [¹¹C]verapamil and as μg/g tumor for [¹¹C]daunorubicin.

Rats were anesthesized as described above, and the carotid artery was cannulated. Arterial blood samples (100–200 μl) were drawn at various time points from 0–60 min after injection of the ¹¹C-labeled radioactivity. Plasma and RBCs were separated by centrifugation (3 min; 3000 g). Plasma (50 μl) samples were counted in a calibrated gamma counter. Results were expressed as mean ± SD.

For analysis of [¹¹C]verapamil, 1 h after injection of radioactivity in the penile vein, the rats were sacrificed by extirpation of the heart. In the brain, the plasma and both tumors, metabolism of [¹¹C]verapamil was studied. These tissues were homogenized in acetonitrile (2 ml), using a Heidolph Diax 600 apparatus (Salm and Kipp b.v., Breukelen, the Netherlands). Thereafter, the extracts were centrifuged (3 min; 3000 g), and acetonitrile was evaporated under reduced pressure at 50°C. To the extracts of the brain and the tumors, 180 μl of sterile water, 20 μl of sodium hydroxide (0.1 m), and 1 ml of diaethyl-ether was added. The mixtures were vortexed for 1 min and centrifuged (5 min; 3000 g). The organic phase was separated and acidified with 300 μl of sulfuric acid (0.5 m). Subsequently, the mixtures were vortexed for 1 min and centrifuged (5 min; 3000 g). To the supernatant, 350 μl of sterile water was added, and this was applied onto a HPLC column [Novapack C18 4 μm, 150 × 3.9 mm; mobile phase, KH₂PO₄ (0.1 m):acetonitrile = 70:30; pH 2.5 by addition of ortho-phosphoric acid; flow rate, 1.0 ml/min]. Fractions (0.5 ml) of the eluate were collected during 10 min and counted in a calibrated gamma counter (LKB Wallac, Turku, Finland). The radioactive recovery was >80%. The recovery was calculated as follows. Before injection of radioactivity on HPLC, the radioactivity was counted in a gamma counter. After separation of fractions on a HPLC column, all of the fractions were counted, and the eluted percentage of radioactivity over the HPLC column was calculated. Results are expressed as percentage of parent compound (mean ± SD).

The pharmacokinetics of [¹¹C]verapamil in rats were studied with a positron camera. After anesthesia as described above, the long axis of the rat was positioned in the PET camera (ECAT 951/31 positron camera; Siemens/CTI, Knoxville, TN) parallel to the transaxial plane of the tomograph to obtain sagittal sections. A transmission scan to correct for attenuation of photons by the body tissues was obtained immediately before the emission scan. Because of homogeneous diffusion of [¹⁵O]H₂O over the rat body, a PET-[¹⁵O]H₂O study is useful for a good anatomical localization of the brain and the tumors. Therefore, [¹⁵O]H₂O (37 MBq) dissolved in 0.3 ml of saline was injected into the tail vein. The following frames were acquired: 4 consecutive frames of 5 s, 1 frame of 10 s, 2 consecutive frames of 30 s, and 1 frame of 120 s. Twenty min after injection of [¹⁵O]H₂O, [¹¹C]verapamil (37 MBq, 0.1 mg/kg) was injected into the tail vein. The following frames were acquired: 12 consecutive frames of 10 s, 6 of 30 s, 5 of 60 s, 5 of 120 s, and 8 of 300 s. After finishing [¹¹C]verapamil kinetics, cyclosporin A (50 mg/kg, i.v.) was injected into the penile vein. Thirty min after injection of cyclosporin A, [¹¹C]verapamil (0.1 mg/kg) was injected in the tail vein, and an emission scan was obtained as described above. Therefore, in this experimental setting, the rats were their own controls when modulating effects of cyclosporin A on [¹¹C]verapamil kinetics in GLC₄ and GLC₄/P-gp tumors were investigated. In addition, to avoid carryover of the first [¹¹C]verapamil injection, there was a time interval of at least 100 min between both injections. After 100 min, only ±3% of [¹¹C] radioactivity could be present when [¹¹C]verapamil was injected for the second time. Radioactive decay was corrected automatically, and data analysis was performed using the Siemens ECAT software (V6.3 D) on a Sun workstation. Results are expressed as percentage of injected dose/ml tumor.

Differences in ¹¹C accumulation in GLC₄ and GLC₄/P-gp tumors and differences in ¹¹C accumulation with and without cyclosporin A modulation were analyzed using Wilcoxon’s test (two-sided). P < 0.05 were considered significant.

Cell survival curves of the GLC₄ and GLC₄/P-gp cell lines after exposure to daunorubicin are shown in Fig. 1. GLC₄ is 12.2-fold more sensitive for daunorubicin than GLC₄/P-gp. The IC₅₀s for daunorubicin are 36 nm for GLC₄ cell line and 440 nm for the GLC₄/P-gp cell line.

Cellular accumulation of [¹¹C]daunorubicin and [¹¹C]verapamil were 3.3-fold (P < 0.001) and 2.3-fold (P < 0.001) higher, respectively, in GLC₄ compared with GLC₄/P-gp. Coincubation with cyclosporin A increased the cellular content of [¹¹C]daunorubicin and [¹¹C]verapamil in GLC₄/P-gp 3.2-fold (P < 0.005) and 2.0-fold (P < 0.02), respectively. Cyclosporin A did not affect the cellular content of [¹¹C]daunorubicin and [¹¹C]verapamil in GLC₄ cells (Fig. 2). In addition, although the ratio of [¹¹C]daunorubicin accumulation in GLC₄ and GLC₄/P-gp is higher than the [¹¹C]verapamil ratio, only camera studies in vivo with [¹¹C]verapamil were performed. The reason for this was the fact that [¹¹C]verapamil production could be scaled up so that sufficiently high enough radioactivity allowed extensive in vivo studies in rats using the positron camera.

A semiquantitative assessment demonstrated that >80–90% of the cells in both tumor types were viable. The GLC₄ and the GLC₄/P-gp tumors consisted of tumor cells, blood vessels, and rat stroma. Histology revealed that in GLC₄ tumors, the cell density was lower than in GLC₄/P-gp tumors (P = 0.0008; n = 3). The cell density in GLC₄ tumors was 8.7 × 10⁹ ± 0.7 × 10⁹ cells/ml tumor and in GLC₄/P-gp tumors, 13 × 10⁹ ± 0.4 × 10⁹ cells/ml tumor. GLC₄ tumor tissue showed no P-gp expression, and GLC₄/P-gp stained clearly for P-gp. Furthermore, the GLC₄ and GLC₄/P-gp tumors showed an equally weak MRP₁ staining (data not shown).

One h after injection, the [¹¹C]daunorubicin content was 159 ± 28% higher (n = 4; P < 0.05) in GLC₄ tumors than in GLC₄/P-gp tumors. Results for individual animals with their paired tumors are shown in Fig. 3,A. Modulation with cyclosporin A (50 mg/kg) increased the [¹¹C]daunorubicin levels in the GLC₄/P-gp tumor to the same levels as compared with GLC₄ tumors (Fig. 3,B). In addition, after modulation with cyclosporin A, [¹¹C]daunorubicin levels were not significantly increased in any other tissue than the GLC₄/P-gp tumor (data not shown). After injection of [¹¹C]verapamil, the verapamil content was 184 ± 53% higher (n = 5; P < 0.05) in GLC₄ tumors than in GLC₄/P-gp tumors (Fig. 4,A).The tissue levels obtained after injection of [¹¹C]verapamil are summarized in Table 1. After modulation with cyclosporin A (50 mg/kg), the [¹¹C]verapamil content in the GLC₄/P-gp tumor was raised to the level of the GLC₄ tumor (Fig. 4,B). In addition, the results of the [¹¹C]verapamil content observed in other tissues are shown in Table 1. One h after injection of [¹¹C]verapamil, [¹¹C]verapamil was present as parent compound for 90 ± 7% (n = 3) in GLC₄ and GLC₄/P-gp tumor tissue, in plasma for 83 ± 5% (n = 5), and in the brain for 93 ± 5% (n = 3).

Plasma clearance of ¹¹C was rapid after [¹¹C]daunorubicin injection (10 mg/kg; data not shown). The curve showed a biphasic pattern. The first phase had a half-life of 0.70 min, and the second phase had a half-life of 69.3 min. After [¹¹C]verapamil injection (0.1 mg/kg), a biphasic plasma curve was shown (data not shown). The first phase had a half-life of 0.50 min, and the second phase had a half-life of 69.3 min. Furthermore, no significant differences in [¹¹C]daunorubicin and [¹¹C]verapamil arterial plasma kinetics were observed after modulation with cyclosporin A (data not shown).

Experiments were performed with [¹⁵O]H₂O). PET-H₂O images enable separate localization of the tumor and the head region (data not shown).

Time activity curves after [¹¹C]verapamil injection (0.1 mg/kg) were obtained from PET images (Fig. 5) in three different rats. Fig. 6,A demonstrates different pharmacokinetics in GLC₄ and GLC₄/P-gp tumors. In both tumor types, the maximum radioactivity per tumor volume was reached within 5 min after injection of the radioactivity. In the GLC₄/P-gp tumor, the maximum was 23% lower than in GLC₄ tumor 5 min after injection of [¹¹C]verapamil. After pretreatment with cyclosporin A (50 mg/kg), the accumulation of [¹¹C]verapamil was increased to an equal level as in the GLC₄ tumor (Fig. 6,B). In the rat brain, 20 min after injection of [¹¹C]verapamil, the accumulation was 350% higher in the presence of cyclosporin A (50 mg/kg) than without cyclosporin A (Fig. 7). This result demonstrates visualization of a P-gp blockade by [¹¹C]verapamil-PET in the GLC₄/P-gp tumor and the brain after modulation with cyclosporin A.

The present study demonstrates that PET enables the measurement of P-gp mediated pharmacokinetics of [¹¹C]verapamil in solid tumors and in the blood-brain barrier. To study these kinetics, a rat model was developed. As tumor model, both a MDR1 gene-transfected, P-gp-overexpressing human small cell lung carcinoma GLC₄/P-gp and its P-gp negative small cell lung carcinoma counterpart GLC₄ were used. Because in each rat both a P-gp-negative and a P-gp-positive tumor were xenographed, equal systemic pharmacokinetics were guaranteed. For validation of the cell lines, in vitro studies were performed. In these studies, it was shown that in the GLC₄ cell line the accumulation of [¹¹C]daunorubicin and of [¹¹C]verapamil was increased compared with the GLC₄/P-gp cell line. Preincubation with cyclosporin A increased [¹¹C]daunorubicin and [¹¹C]verapamil levels in GLC₄/P-gp but not in GLC₄. Furthermore, in the cytotoxicity assay, the GLC₄/P-gp line was found to be 12-fold cross-resistant for daunorubicin compared with GLC₄. These results demonstrated that GLC₄ and GLC₄/P-gp cell lines were an elegant model to study P-gp functionality in the tumor-bearing rat model. Biodistribution studies demonstrated that in the P-gp-overexpressing GLC₄/P-gp tumors, the levels of [¹¹C]daunorubicin and [¹¹C]verapamil were lower than in GLC₄ tumors. Pretreatment with cyclosporin A increased the levels of [¹¹C]daunorubicin and [¹¹C]verapamil in GLC₄/P-gp tumors to the level in the GLC₄ tumors. The effects of cyclosporin A in the GLC₄/P-gp tumors were most likely due to blockade of the P-gp drug efflux by cyclosporin A. This is underscored by the fact that cyclosporin A had no effect on plasma kinetics of [¹¹C]daunorubicin and [¹¹C]verapamil in the rat.

In the blood-brain barrier, P-gp is abundantly expressed (35), and because in this study [¹¹C]verapamil levels were increased after pretreatment with cyclosporin A, we were able to show P-gp function and its reversal effects ex vivo and in vivo. This important role of P-gp in the blood-brain barrier was also shown earlier in the mouse model. In biodistribution studies in the brain of mdr1a(−/−) P-gp knock-out mice, drug levels of several drugs, such as verapamil, digoxin, and cyclosporin A, were found to be increased compared with mdr1a(+/+) wild-type mice (16, 17). Furthermore, in wild-type mice, the brain levels of digoxin were increased after blockade of P-gp in the blood-brain barrier by PSC 833 (20). The [¹¹C]daunorubicin and [¹¹C]verapamil biodistribution data are in agreement with results obtained with the radiolabeled P-gp substrate colchicine, confirming that it is possible to distinguish P-gp-expressing tumors from sensitive tumors in vivo by ¹¹C-radiolabeled drugs (36, 37, 38). In contrast to verapamil, the daunorubicin levels in the brain were not increased after modulation with cyclosporin A. Daunorubicin is a substrate for P-gp as well as MRP₁. Because of expression of MRP₁ in many tissues in the body, including the blood-brain barrier (39), it might be possible that after modulation with cyclosporin A, daunorubicin is still transported by the MRP₁ drug efflux pump.

Ex vivo data demonstrated the usefulness of P-gp-mediated [¹¹C]verapamil kinetics in the P-gp-expressing tumor and in the central nervous system. For [¹¹C]verapamil production, it was possible to scale up so that sufficiently high enough radioactivity was available for in vivo studies in rats and potentially for humans with a positron camera. Therefore, in vivo studies were performed with [¹¹C]verapamil. After injection of [¹¹C]verapamil in the rat, a fast radioactivity transport was observed in the P-gp-overexpressing tumor, and 10 min after injection, a steady-state level was reached (Fig. 6). Possibly, due to the high lipophilicity, verapamil may be bound in the tumor, preventing transport from the tumor back into the blood. Furthermore, it might be possible that other routes such as passive efflux may play an important role in the cellular transport of verapamil. It is speculated that the time-activity curve of [¹¹C]verapamil might be the result of partly passive efflux of verapamil over tumor membranes and partly by P-gp-mediated pharmacokinetics. This is supported by the observed rapid cellular verapamil passive diffusion of 50% within 5 min for P-gp-negative and P-gp-positive ovarian carcinoma cell lines (40, 41).

Several experiments with ^99mTc-sestamibi and other ^99mTc-labeled substrates, such as ^99mTc-Q-complexes and ^99mTc-tetrofosmin, demonstrated in rats and in patients an increased efflux of ^99mTc-sestamibi in P-gp-expressing tumor cells and modulation effects with reversal agents (24, 25, 26, 27, 28, 42, 43, 44). However, with single photon emission computed tomography, quantitative information about pharmacokinetics in the tumor is less accurate than with PET. Recently, we have shown that ^99mTc-sestamibi is also a substrate for MRP₁ (29, 30). In contrast to ^99mTc-sestamibi, [¹¹C]verapamil is a relatively poor substrate for MRP₁. This makes [¹¹C]verapamil a more specific tracer for P-gp function than ^99mTc-sestamibi. Comparison of sensitivity between ^99mTc-sestamibi and [¹¹C]verapamil demonstrates that ^99mTc-sestamibi has the ability to image P-gp in a tumor that is 2–3-fold more resistant for doxorubicin compared with the P-gp-negative tumor (25). In contrast, [¹¹C]verapamil showed 2-fold less radioactivity in a P-gp-expressing tumor, which is 12-fold resistant for doxorubicin compared with a P-gp-negative tumor. Because of high lipophilicity of [¹¹C]verapamil, this compound most likely binds aspecificly to cellular membranes. Additional studies in cancer patients have to be performed to elucidate whether it is possible to measure low resistance factors with this PET approach in human tumors.

In conclusion, P-gp kinetics and its reversal can be visualized in vivo with a positron camera. This can be a new clinical tool to study the P-gp function noninvasively and to predict the effect of modulators on treatment with MDR involved chemotherapeutic drugs.