Breast Cancer Resistance Protein and P-Glycoprotein Influence In Vivo Disposition of ¹¹C-Erlotinib

DISCUSSION

Among several different radiolabeled reversible and irreversible TKIs tested for PET imaging of receptor tyrosine kinases, ¹¹C-erlotinib has emerged as a promising candidate tracer to distinguish responders from nonresponders to treatment with EGFR-targeted TKIs (e.g., gefitinib, erlotinib) (17). Preclinical as well as clinical data indicate that ¹¹C-erlotinib binds specifically to EGFR with activating mutations (e.g., delE746-A750, L858R), whereas it lacks specific binding to wild-type EGFR (2–6). Preclinical studies have shown that specific binding of ¹¹C-erlotinib to mutated EGFR is reduced when pharmacologic doses of erlotinib are coadministered, presumably due to saturation of EGFR-specific binding sites (5,6). Consequently, ¹¹C-erlotinib PET imaging in patients undergoing TKI treatment may prohibit the visualization of EGFR-specific binding in tumors. In addition to binding to mutated EGFR, the known interaction of erlotinib with ABC and solute carrier (SLC) transporters (8,9,18) may exert an influence on the distribution of ¹¹C-erlotinib to tumoral and nontumoral tissue. These transporters are widely expressed all over the body, such as at blood–tissue barriers (e.g., BBB or blood–testis barrier) and in excretory organs (liver, kidney, intestine), and may also become overexpressed in multidrug-resistant tumors in which they restrict intracellular distribution of anticancer agents (12,13). The transport of drugs by ABC and SLC transporters is of concern because this may lead to interindividual variability in drug pharmacokinetics due to genetic polymorphisms or to transporter-mediated drug–drug interactions potentially causing adverse reactions (12). Moreover, dose-dependent partial saturation of drug transporters may lead to nonlinear pharmacokinetics of transporter substrates, which may result in differences in organ distribution between a microdose and therapeutic dose.

In the present study, we assessed the effect of ABCG2 and ABCB1 on whole-body distribution and excretion of ¹¹C-erlotinib and investigated whether PET imaging with a microdose of ¹¹C-erlotinib is capable of predicting the organ distribution of a therapeutic erlotinib dose. We addressed these questions by performing small-animal PET/MR experiments with ¹¹C-erlotinib without and with coinjection of a pharmacologic dose of unlabeled erlotinib (10 mg/kg) in wild-type and Abcb1a/b or Abcg2 knockout mice. Previous work has shown that Abcb1a/b^(−/−), Abcg2^(−/−), and Abcb1a/b^(−/−)Abcg2^(−/−) mice lack compensatory alterations in expression levels of other ABC and SLC transporters and are useful tools to study the effects of ABCB1 and ABCG2 on drug disposition (19). A potential limitation of using mice is their small blood volume, which prohibits arterial blood sampling, which is needed for quantitative analysis of PET data. We overcame this problem by generating an image-derived blood curve from the left ventricle of the heart. The use of a previously described graphical analysis approach (integration plot) (Supplemental Fig. 2) (14,15) allowed us to estimate the rate constants for transfer of radioactivity from blood into different organs (brain, liver, kidney, lungs) (CL_uptake) as well as for transfer of radioactivity from liver into the intestine (CL_bile) to quantitatively evaluate the effects of Abcb1a/b and Abcg2 on ¹¹C-erlotinib disposition (Figs. 2C and 5; Supplemental Table 1).

We were able to confirm earlier data that brain distribution of erlotinib is restricted by Abcb1a/b and Abcg2 (8,10,11). It has now been documented for many dual ABCB1/ABCG2 substrates that these 2 transporters are capable of mutually compensating their function when 1 transporter is knocked out or inhibited (20,21). This phenomenon causes only small increases in brain distribution of ABCB1/ABCG2 substrates when only 1 transporter is absent and disproportionally large increases when both transporters are absent, as was the case for ¹¹C-erlotinib (Fig. 2C). Importantly, we could show that radioactivity in the brain after injection of ¹¹C-erlotinib is mainly in the form of unmetabolized ¹¹C-erlotinib (Supplemental Table 2) thus making ¹¹C-erlotinib PET suitable to assess the effect of ABCB1 and ABCG2 on ¹¹C-erlotinib brain distribution. It is now known that most currently available TKIs (e.g., sorafenib, gefitinib, sunitinib) are dual ABCB1/ABCG2 substrates resulting in low brain distribution (20,21). This is of concern as poor BBB penetration may render these drugs ineffective in the treatment of primary or secondary brain tumors (e.g., gliomas, NSCLC brain metastases) (11,22). Pharmacologic inhibition of ABCB1 and ABCG2 has been suggested as a promising approach to enhance brain distribution of TKIs and thereby increase their efficacy to treat brain tumors (20,22). However, a limitation in realizing this therapeutic concept is the current lack of clinically usable, marketed dual ABCB1/ABCG2 inhibitors. The most potent dual ABCB1/ABCG2 inhibitor known to date is elacridar, which has shown great promise in enhancing brain distribution of dual ABCB1/ABCG2 substrates in preclinical studies (20–22). We found that 10 mg of elacridar per kilogram given intravenously 20 min before ¹¹C-erlotinib injection can increase brain distribution of ¹¹C-erlotinib to an extent similar to that in Abcb1a/b^(−/−)Abcg2^(−/−) mice, indicating complete inhibition of Abcb1a/b and Abcg2 at the BBB (Figs. 1, 2A, and 2C). An oral formulation of elacridar has been used in clinical studies in cancer patients to increase oral bioavailability of concomitantly administered anticancer drugs (e.g., topotecan) by inhibition of ABCG2 and ABCB1 in the intestine (23). However, as the oral bioavailability of elacridar is low, resulting in low plasma concentrations, future use of elacridar as an inhibitor of ABCB1 and ABCG2 at the human BBB will most likely require an intravenous formulation of the drug (22).

Intriguingly, we could show that the coinjection of a pharmacologic dose of erlotinib results in increases in CL_uptake,brain of ¹¹C-erlotinib similar to those in elacridar-treated animals (Fig. 2C). This increase suggests that erlotinib may be of potential use as a clinically available inhibitor of ABCB1 and ABCG2 to enhance brain distribution of other dual ABCB1/ABCG2 substrates (24). After a comparably high initial brain uptake, washout of radioactivity was faster from the brains of erlotinib-coinjected wild-type mice than from Abcb1a/b^(−/−)Abcg2^(−/−) mice (Figs. 2A and 2B), which indicated that the Abcb1a/b/Abcg2 inhibitory effect of erlotinib is reversible. For future use of erlotinib as an ABCB1/ABCG2 inhibitor in human PET studies, it may therefore be preferable to administer erlotinib as a continuous intravenous infusion that is maintained for the duration of the PET scan, as we have done in a previous PET study with the ABCB1 inhibitor tariquidar (16). Importantly, we could demonstrate that brain distribution of ¹¹C-erlotinib was nonlinear between a microdose and a pharmacologic dose—that is, for the pharmacologic dose brain exposure (Fig. 2B) and CL_uptake,brain (Fig. 2C) were significantly higher.

We also found an influence of Abcg2 and Abcb1 on the distribution of ¹¹C-erlotinib to peripheral organs. Our data clearly indicated that ¹¹C-erlotinib or its radiolabeled metabolites undergo hepatobiliary excretion into the intestine mediated by Abcg2 and Abcb1a/b (Fig. 4C). The absence of both Abcg2 and Abcb1a/b caused a 2.6-fold reduction in CL_bile (Fig. 5C), suggesting that ¹¹C-erlotinib PET may be useful to study the functional activity of Abcg2 and Abcb1a/b in the liver. Interestingly, coadministration of a pharmacologic dose of erlotinib caused a 5-fold reduction in CL_bile in wild-type mice (Fig. 5C), indicating that the influence of Abcg2 and Abcb1a/b on hepatobiliary excretion of erlotinib may be more important when a microdose is administered. However, the determination of CL_bile was based on the assumption that excretion of radioactivity into the intestine occurs exclusively via bile and not by direct secretion from blood. To confirm this assumption, it would be necessary to examine bile duct–cannulated mice, which is technically challenging and which was not done in the present study.

In the absence of Abcg2 (i.e., in Abcg2^(−/−) and Abcb1a/b^(−/−)Abcg2^(−/−) mice), a shift from hepatobiliary to renal excretion of ¹¹C-erlotinib was observed, which otherwise showed negligible excretion into urine (Figs. 3 and 4D). This suggested that transporters other than Abcg2 and Abcb1a/b accounted for urinary excretion of radioactivity. As radio–thin-layer chromatography analysis showed that most radioactivity in urine was in the form of radiolabeled metabolites (Supplemental Table 2) it appears likely that not ¹¹C-erlotinib itself but one or several radiolabeled metabolites underwent active transport at the brush border membrane of kidney proximal tubule cells, with multidrug resistance–associated proteins 4 and 2 (Abcc4 and Abcc2) as the most likely candidate transporters.

Both in the liver and in the kidneys, CL_uptake was significantly lower for the pharmacologic dose than for the microdose (Figs. 4A, 4B, 5A, and 5B). This suggests partial saturation of basolateral uptake transporters in hepatocytes and kidney cells, such as organic anion transporting polypeptides (e.g., OATP2B1, SLCO2B1) and organic anion transporter 3 (SLC22A7). This assumption is in line with previous findings that TKIs including erlotinib are competitive inhibitors of these SLC transporters (18). Partial saturation of uptake transporters most likely also accounted for the significantly higher blood AUCs for the pharmacologic dose than the microdose. An increase in systemic exposure of drugs on inhibition of basolateral uptake transporters in hepatocytes has been observed for several different drugs (25).

Lung tissue can be considered as a target tissue for treatment of NSCLC with erlotinib. It is known that several different ABC and SLC transporters are expressed in lung tissue (e.g., multidrug resistance–associated protein 1 [ABCC1] and ABCG2), where they control pulmonary absorption of inhaled drugs as well as transport of drugs from epithelial cells to the circulating blood (26). We found no influence of Abcb1a/b and Abcg2 on the distribution of ¹¹C-erlotinib to the lung (Supplemental Fig. 3). However, the coinjection of a pharmacologic dose caused significant increases in CL_uptake,lung, pointing to partial saturation of other unknown efflux transporters. Interestingly, elacridar led also to an increase in CL_uptake,lung, but it is not known by which mechanism.

Our findings are clinically relevant in that they suggest that future therapy guidance with ¹¹C-erlotinib in tumor patients needs to take into account the influence of drug transporters on ¹¹C-erlotinib disposition. Concomitant treatment of tumor patients with therapeutic doses of erlotinib or other TKIs may lead to partial saturation of drug transporters, which could cause changes in ¹¹C-erlotinib PET signal in the tumor, which may be unrelated to treatment response or the mutational status of EGFR. However, because a standard oral therapeutic dose of erlotinib used in the clinic (150 mg) is lower than the presently used pharmacologic dose in mice, it appears likely that less pronounced effects on drug transporters will be observed in humans.