Against Repurposing Methadone for Glioblastoma Therapy

Methadone, which is used as maintenance medication for outpatient treatment of opioid dependence or as an analgesic drug, has been suggested by preclinical in vitro and mouse studies to induce cell death and sensitivity to chemo- or radiotherapy in leukemia, glioblastoma, and carcinoma cells. These data together with episodical public reports on long-term surviving cancer patients who use methadone led to a hype of methadone as an anti-cancer drug in social and public media. However, clinical evidence for a tumoricidal effect of methadone is missing and prospective clinical trials, except in colorectal cancer, are not envisaged because of the limited preclinical data available. The present article reviews the pharmacokinetics, potential molecular targets, as well as the evidence for a tumoricidal effect of methadone in view of the therapeutically achievable doses in the brain. Moreover, it provides original in vitro data showing that methadone at clinically relevant concentrations fails to impair clonogenicity or radioresistance of glioblastoma cells.


Introduction
Methadone is a synthetic opioid with an asymmetric carbon atom giving rise to two enantiomeric forms: R(−)-(L-)methadone (levomethadone) and S(+)-(D-)methadone (dextromethadone). It was developed in 1937 under the name Polamidon (Hoechst 10820) [1] by Gustav Ehrhart and Max Bockmühl, both German chemists at the IG Farben company [2]. The analgesic effect of methadone was discovered in 1942, and in 1945, methadone was finally approved as an analgesic drug in Germany [3]. After World War II, the US American company Eli Lilly acquired the rights for Polamidon and distributed the substance under the new name methadone [4]. In 1964, studies at the Rockefeller Hospital demonstrated that methadone is effective in maintenance treatment for long-term heroin addiction by "relieving the narcotic hunger and inducing sufficient tolerance to block the euphoric effect of an average illegal dose of diacetylmorphine" [5]. In the clinical setting and in laboratories, the racemic mixture of methadone usually is applied. R(−)-(L-)methadone has analgesic activity, and S(+)-(D-)methadone is discussed to be less active [6] or to have antagonistic activity to the N-methyl-d-aspartate (NMDA) receptor affecting morphine tolerance [7]. 0.34 L/min [24], whilst others found a remarkable variation of 0.023 up to 2.1 L/min in cancer patients, maybe due to drug-drug interactions [26].
One potential reason for these interactions is the extensive metabolism of methadone by several cytochrome P450 enzymes (for a review, see [30]); the major contributor to methadone metabolism is CYP3A4, and CYP2D6 plays a minor role. Several other CYP enzymes are also involved, such as CYP2B6, CYP2C8 and CYP2C9, some of which are implicated in contributing to the heterogeneity by polymorphisms but also by drug-drug interactions. Elimination of both methadone and its metabolites occurs via renal and fecal elimination, whereas the proportion, again, varies from person to person, even though renal elimination seems to be responsible to a larger extent more frequently [31]. Table 1 summarizes clinically or clinicopathologically observed methadone concentrations in blood, brain or cerebrospinal fluid.
The only clinical examination of methadone in one glioma patient so far measured a basal plasma concentration of 182 ng/mL (0.59 µM) after an oral uptake of 30 mg methadone [19]. In general, effective doses for its main use, opioid addiction, are 60-120 mg daily [32], with recommended plasma concentrations in the range of 150-600 ng/mL (0.5-2 µM) [33]. Concentrations above 700 ng/mL (2.3 µM) are assumed to be associated with toxic effects [34]; however, the median concentration of methadone in methadone-related deaths was only 500 ng/mL (1.6 µM) [35].
Regardless of the exact interpretation of these data, it is safe to assume that there is also profound variability in tolerability to methadone, especially contemplating the use in the patient population of glioma patients, differing significantly from drug addict patients. This danger is well depicted in the disproportionally high ratio of methadone in opioid pain reliever-associated deaths in the USA from 1999 to 2010 [36]. Of specific interest for the in vitro testing for an anti-glioblastoma activity are methadone concentrations in brain tissue, which were reported to be between 219 and 2652 ng/mL (0.7-8.6 µM) in methadone-related deaths [37], whereas other authors found even higher concentrations of 3700 ng/mL (12 µM) [38]. These data support the notion that many in vitro experiments-including parts of our work (see below)-use methadone concentrations, which are either clinically irrelevant or could lead to serious adverse events, such as respiratory depression [39,40], but also prolongation of the QTc interval, and torsade de pointes arrhythmias [41][42][43][44]. Whether a partially broken blood-brain barrier in GBM [45] will aggravate or alleviate this problem remains an unanswered question. General knowledge about the blood-brain barrier permeability of methadone is mainly based on studies in rats, showing an uptake into the brain of 42% of the methadone injected into the common carotid artery [49,50]. Several other groups reported the importance of the efflux transporter p-glycoprotein (MDR1) for the blood-brain barrier permeability of methadone in mice [51][52][53]. Whether polymorphisms in MDR1 may also contribute to different dosing requirements in humans is debated [54][55][56] (for a review of the pharmacogenetics of methadone in general, see [40]).
To conclude, methadone is a highly challenging drug for each clinician, probably necessitating increasing doses slowly, tapering at the end of the treatment, thorough evaluation of potential drug-drug interactions, regular drug concentration measurements and close monitoring of potential adverse effects. Beyond the above mentioned severe toxicities, methadone may diminish quality of life, starting from obstipation and nausea [19], and ending up with sedation, delirium and severe withdrawal symptoms [57]. These real challenges and accompanying risks should be balanced against the potential benefits [58]. The next paragraphs introduce putative molecular targets of methadone in glioblastoma.
In addition, methadone reportedly inhibits L-and T-type Ca 2+ channels, with estimated IC 50 values above 10 µM. Interestingly, it seems that there are different ways of how methadone interacts with the two subfamilies of voltage-gated Ca 2+ channels. The µ-opioid receptor antagonist naloxone partially reverses the inhibition of the T-type Ca 2+ channels, which suggests a µ-opioid receptor-dependent inhibition, but not the blockage of L-type Ca 2+ channels [66].
Beside the action of methadone on voltage-gated K + channels, a direct blockage of voltage-gated (Na v ) Na + channels (Na v 1.2, Na v 1.3, Na v 1.7, and Na v 1.8) has been demonstrated in mouse peripheral nerves, with IC 50 values ranging from 86 to 119 µM [71]. Considerably lower, and thus, in the therapeutically achievable plasma concentration range, is the reported methadone IC 50 value (100 nM) for the µ-opioid receptor-independent blockage of human Clchannel-2 (hClC2) in T84 intestinal epithelial cells or when stably expressed in HEK293EBNA cells [72]. Finally, methadone has been demonstrated to inhibit human p-glycoprotein (MDR1): methadone was found to block calcein efflux in p-glycoprotein-transfected HEK293 cells with an estimated IC 50 value above 25 µM and paclitaxel uptake in preparations of human placental inside-out vesicles with a K i of 18 µM [73,74]. To which extent methadone may exert tumoricidal activity via stimulation of the µ-opioid receptor and/or modulation of further target molecules (Table 2 and Figure 1B) is discussed in the next paragraphs. In addition, methadone reportedly inhibits L-and T-type Ca 2+ channels, with estimated IC50 values above 10 µM. Interestingly, it seems that there are different ways of how methadone interacts with the two subfamilies of voltage-gated Ca 2+ channels. The µ-opioid receptor antagonist naloxone partially reverses the inhibition of the T-type Ca 2+ channels, which suggests a µ-opioid receptordependent inhibition, but not the blockage of L-type Ca 2+ channels [66].
Beside the action of methadone on voltage-gated K + channels, a direct blockage of voltage-gated (Nav) Na + channels (Nav1.2, Nav1.3, Nav1.7, and Nav1.8) has been demonstrated in mouse peripheral nerves, with IC50 values ranging from 86 to 119 µM [71]. Considerably lower, and thus, in the therapeutically achievable plasma concentration range, is the reported methadone IC50 value (100 nM) for the µ-opioid receptor-independent blockage of human Clchannel-2 (hClC2) in T84 intestinal epithelial cells or when stably expressed in HEK293EBNA cells [72]. Finally, methadone has been demonstrated to inhibit human p-glycoprotein (MDR1): methadone was found to block calcein efflux in p-glycoprotein-transfected HEK293 cells with an estimated IC50 value above 25 µM and paclitaxel uptake in preparations of human placental inside-out vesicles with a Ki of 18 µM [73,74]. To which extent methadone may exert tumoricidal activity via stimulation of the µ-opioid receptor and/or modulation of further target molecules (Table 2 and Figure 1B) is discussed in the next paragraphs.  Ca v channels: voltage-gated L-and T-type Ca 2+ channels, nAChRs: nicotinic acetylcholine receptors, NMDA receptors: N-methyl-d-aspartate receptors, Na v channels: voltage-gated Na + channels, KCNH2: hERG1 K + channel, CLCN2: ClC-2 Cl − channel, ABCB1: p-glycoprotein (MDR1).
Moreover, methadone reportedly decreases proliferation and increases apoptotic cell death of CEM and HL-60 leukemia cells with an IC 50 of~15-100 µM. The pro-apoptotic effect of methadone is probably mediated by upregulation of caspase activities and downregulation of anti-apoptotic proteins such as Bcl-xL, XIAP, Bcl-2 or p21 [17,78]. In addition, the cytotoxic effect of methadone in combination with cytostatic agents was tested in primary acute lymphoblastic leukemia (ALL) cells in vitro [18,79,80] and in an ectopic mouse model (20 mg/(kg·BW·d) methadone) [18]. In these studies, methadone (IC 50 <0.3-≥32 µM) enhanced the viability-attenuating (or cell death-promoting) effect of Bcl-2 targeting by ABT-737 [79], doxorubicin chemotherapy [18], and asparaginase therapy [80]. The doxorubicin-sensitizing action of methadone (20-120 mg/(kg·BW·d) in increasing doses) was also observed in mice ectopically xenografted with ALL cells [18]. Mechanistically, methadone induced the uptake and inhibited efflux of doxorubicin into/from the ALL cells. Doxorubicin, on the other hand, stimulated µ-opioid receptor upregulation [18]. Moreover, a genome-wide shRNA library-based screening identified the µ-opioid receptor as an asparaginase resistance gene [80]. In summary, these preclinical data suggest that clinically achievable concentrations of methadone may impair viability of SCLC and NSCLC lung cancer cells in monotherapy and may sensitize leukemia cells to, e.g., doxorubicin chemotherapy.
Doxorubicin only poorly penetrates the blood-brain barrier (BBB). Few clinical data from glioblastoma patients are available, e.g. after i.v. application of BBB-permeant pegylated liposomal doxorubicin [89,90]. These data do not prove high tumoricidal efficacy of this DNA-intercalating drug in glioblastoma, suggesting doxorubicin is not the best choice for second line chemotherapy in patients with recurrent glioblastoma or elevated temozolomide toxicity. Accordingly, doxorubicin is not applied routinely in glioblastoma. Nevertheless, the doxorubicin-sensitizing effect of methadone (3.2-32 µM) was studied in vitro in human A172 and U118MG glioblastoma cells, indicating indeed a doxorubicin-sensitizing action of methadone with IC 50 in the range of ≤3-~10 µM [16,84]. Additional experiments in A172 cells showed a cisplatin-, 5-fluoruracil-and paclitaxel-sensitizing action of methadone with IC 50 of about 4 µM, >>32 µM, and >>32 µM, respectively [84].
Combined, except the most recent findings from Shi et al. [82], these studies suggest very low anti-glioblastoma efficacy of methadone monotherapy. Accordingly, methadone-mediated growth delay of U87MG tumors ectopically xenografted into mice could be observed at very high doses (up to 240 mg/(kg·BW·d) p.o.), greatly exceeding commonly given doses for opioid addiction therapy, which are in the range around 1-2 mg/(kg·BW·d). In contrast to monotherapy, in vitro studies indicate that methadone at clinically achievable concentrations may enhance the efficacy of experimental therapies [16,82,84]. To analyze whether methadone may also enhance the efficacy of the standard anti-glioblastoma therapy [9], U87MG, U251 and primary human glioblastoma cells were treated with a combination of methadone (1-145 µM) and/or temozolomide (100-200 µM) and/or radiation (4 Gy), and cell survival/death was analyzed by crystal violet staining, annexin-V binding, dehydrogenase activity or ATP measurement. As a result, clinically achievable methadone concentrations are not able to sensitize to temozolomide (apparent methadone IC 50 >>30->>145 µM) [81,91], to radiation (apparent methadone IC 50 >>30µM) or to concomitant temozolomide-radiation therapy (apparent methadone IC 50 >>30 µM) [91].
To conclude, preclinical data suggest that methadone might be effective as monotherapy in lung cancer or might sensitize to established therapies in certain types of cancer, such as pediatric ALL. In glioblastoma, however, evidence for a tumoricidal and standard therapy-sensitizing action of methadone in clinically achievable concentrations is very weak and missing, respectively. Moreover, the preclinical studies on glioblastoma mainly used methadone-induced cell death or impairment of total dehydrogenase activity (measure of cell viability) as endpoints (see Table 3). From our radio-oncological point of view, it is not so decisive whether a given anti-cancer treatment kills a lower or higher percentage of tumor cells or delays tumor cell proliferation more or less. A much more meaningful measure of therapy efficacy is how many tumor cells maintain their clonogenicity and are capable to regrow the tumor after the end of therapy. Clonogenic survival of tumor cells results in tumor relapse and therapy failure. We, therefore, performed further in vitro experiments that addressed the effect of "supratherapeutic" and clinically relevant concentrations of methadone alone and in combination with ionizing radiation on clonogenic survival in different human glioblastoma cell lines. Since our previous work suggests that blockage of putative methadone targets may decrease clonogenic survival of irradiated tumor cells by impairment of cell cycle arrest [95], we also tested the effect of methadone and ionizing radiation on cell cycle regulation. Prior to that, we analyzed the expression of putative methadone target molecules in glioblastoma, as described in the following paragraphs.

Methadone in "Supratherapeutic" Concentrations May Modify Cell Cycle but Fails to Impair Clonogenic Survival or Radioresistance of Human Glioblastoma Cells in Clinically Relevant Concentrations In Vitro
Querying the glioblastoma database of The Cancer Genome Atlas (TCGA) for the mRNA abundances of putative methadone targets indicates undetectably low OPRM1 (µ-opioid receptor) mRNA abundance in the majority of glioblastoma specimens ( Figure 1A,B, first box plot). In contrast to OPRM1 mRNA, GalR1 mRNA encoding for the galanin receptor 1 is abundant in most glioblastoma specimens ( Figure 1A,B, second box plot). This might suggest that the µ-opioid receptors in glioblastoma predominately form heteromeric complexes with the galanin receptor 1 that exhibit low methadone affinity (see above). Figure 1B shows the mRNA abundances of further putative methadone targets in glioblastoma specimens.
To test for tumoricidal effects of methadone in glioblastoma, we first analyzed expression of the µ-opioid receptor (OPRM1) and the NMDA receptor subunit ζ1 (GRIN1) in our human glioblastoma cell lines A172, T98G and U251 by RT-PCR and in part, by western blotting. As shown in Figure 2A, OPRM1-and GRIN1-specific mRNA could be detected in all three glioblastoma lines. In addition, weak immunoreactive bands at the expected migration range of 47-65 kDa were apparent in blots of T98G and U251 total lysates probed against the µ-opioid receptor ( Figure 2B). Together, this suggests expression of µ-opioid receptors in T98G, A172, and U251 glioblastoma cells. Protein expression of µ-opioid receptors has been reported previously for A172 [16,82] and U251 [81], as well as for U87MG [81] human glioblastoma cells (U87MG cells were also used in the present study; see below).
Beyond µ-opioid and ζ1-NMDA (GRIN1) receptors (Figure 2A), expression of further putative molecular methadone targets, such as L-type and T-type voltage-gated Ca 2+ channels (see Table 2), has been reported previously by our group for U251 and T98G cells [97]. In addition, the glioblastoma dataset of the TCGA (see Figure 1) suggests expression of further methadone target molecules (see Table 2), such as the cardiac voltage-gated K + channels hERG1 (KCNH2) or nicotinic acetylcholine receptors CHRNA3/B4.
Reportedly, electrosignaling and interdependent Ca 2+ signaling contribute to the stress response of cancer cells by regulating cell cycle progression and promoting their clonogenic survival (forreview, see [98]). Since methadone has been proposed to target hERG1 K + channels and ionotropic receptors (see Table 2) and since hERG1 reportedly [95] and NMDA-as well as nicotinic acetylcholine receptors most probably interfere with electrosignaling, we studied the effect of methadone on cell cycle progression in control and irradiated T98G, A172 and U251 cells by flow cytometry. In particular, we monitored the effect of a "supratherapeutic" methadone concentration (20 µM) in order to modulate the ionotropic nicotinic α 3 β 4 acetylcholine and NMDA receptors, as well as hERG1 and L-and T-type Ca 2+ channels (IC 50 s in Table 2) in addition to the µ-opioid receptor.
For cell cycle analysis in flow cytometry, the cellular DNA content was visualized by propidium iodide staining of permeabilized cells, 24 and 48 h after irradiation with 0 and 4 Gy ( Figure 3A). As endpoints, the percentage of cells residing in G 1 , S, and G 2 phase of cell cycle was calculated. Ionizing radiation (4 Gy) induced an increase in G 1 population and a decrease in S and G 2 population at 24 and 48 h after radiation in A172 cells ( Figure 3B, open bars), suggestive of a radiation-induced G 1 arrest. In T98G cells, in contrast, radiation (4Gy) stimulated a transient G 2 /M cell cycle arrest that became evident 24 h after irradiation ( Figure 3C, open bars) and in U251 cells, a transient G 1 arrest after 24 h followed by an S and G 2 /M arrest 48 h after irradiation ( Figure 3D, open bars).
Methadone (   We have previously shown that experimental interference with electrosignaling and cell cycle control decreases the clonogenic survival of irradiated tumor cells [95,[99][100][101][102][103]. We, therefore, tested whether the observed methadone (20 µM)-mediated modulation of cell cycle control was associated with an impairment of clonogenic survival and radioresistance.
To determine clonogenic survival of T98G, U251 and A172 glioblastoma cells by delayed plating colony formation assay, cells were pretreated for 1 h with a "supratherapeutic" concentration of methadone (20 µM) or vehicle alone (ethanol), irradiated with 0, 2, 4, 6 or 8 Gy and post-incubated (24 h) in the presence (20 µM) and absence (vehicle) of methadone, before washing and reseeding the cells for colony formation in the absence of the drug. As a result, methadone increased the plating efficiency of T98G cells but not of U251 and A172 cells ( Figure 4A,E,I, upper row, and Figure 4B,F,J). In addition, methadone had little or contrary effects on survival fractions of the individual irradiated glioblastoma lines tested ( Figure 4A,E,I, lower row, and Figure 4C,G,H). In A172 ( Figure 4C) and T98G ( Figure 4G) cells, methadone (20 µM) slightly radiosensitized the cells at some but not all of the applied radiation doses. In U251, in contrast, methadone promoted radioresistance in 4 Gy-irradiated cells. The survival fraction at 2 Gy (SF 2 Gy ), which is clinically relevant because 2 Gy are applied per daily fraction in normofractionated protocols, was reduced by methadone only in T98G cells ( Figure 4D,H,L). Combined, the data indicate that a "supratherapeutic" concentration of methadone may have both clonogenic survival-promoting and impairing effects in control and irradiated glioblastoma cells. In particular, in T98G cells, the methadone-induced decrease in radioresistance (decline of SF 2 Gy ) was probably compensated by the methadone-stimulated clonogenicity (plating efficiency), suggesting that methadone even at a very high concentration (that is assumed to modulate several molecular target proteins beyond the µ-opioid receptor) does not exert clinically relevant beneficial effects alone or concomitant to radiotherapy.
Finally, we adjusted our protocol for analysis of clonogenic survival towards a clinically more relevant setting: T98G, A172, U251 and a further human glioblastoma cell line, U87MG, were plated and irradiated in five consecutive days with 0 or 2 Gy each and post-incubated for a further 2-3 weeks until formation of colonies. This pre-plating colony formation assay was carried out in the continuous absence (vehicle alone, ethanol) or presence of methadone (2 µM, Figure 5A). As a result, the clinically relevant methadone concentration of 2 µM did neither decrease plating efficiency ( Figure 5B-F, left) nor the survival fraction (SF 5 × 2 Gy, Figure 5B-F, right) indicating that methadone in clinically achievable concentrations neither impairs clonogenicity nor radioresistance of glioblastoma cells in vitro.  (20 µM). Then, cells were replated in methadone-free medium and grown for a further two weeks. Thereafter, colonies were defined as cell clusters of ≥50 cells and colony numbers were counted. Plating efficiency (PE) was defined by the ratio between counted colonies (N) and plated cells (n) under control conditions (PE = N/n). The survival fraction (SF) was calculated by normalizing in both arms (methadone and vehicle control) separately the plating efficiency after irradiation (PExGy) to that of the corresponding unirradiated control (PE0Gy) by the formula SF = PExGy/PE0Gy. SFs were fitted by the use of the linear quadratic equation (SF = e -(α·D + β·D^2) with D = radiation dose and α and β cell-specific constants).  (20 µM). Then, cells were replated in methadone-free medium and grown for a further two weeks. Thereafter, colonies were defined as cell clusters of ≥50 cells and colony numbers were counted. Plating efficiency (PE) was defined by the ratio between counted colonies (N) and plated cells (n) under control conditions (PE = N/n). The survival fraction (SF) was calculated by normalizing in both arms (methadone and vehicle control) separately the plating efficiency after irradiation (PE xGy ) to that of the corresponding unirradiated control (PE 0Gy ) by the formula SF = PE xGy /PE 0Gy . SFs were fitted by the use of the linear quadratic equation (SF = e -(α·D + β·Dˆ2) with D = radiation dose and α and β cell-specific constants).

Concluding Remarks
Recommended plasma methadone concentrations for patients in a methadone maintenance program are in the range of 0.5-2 µM (150-600 ng/mL), which are usually achieved by daily doses of up to 120 mg. Importantly, data from patients who died from methadone overdose suggest an overlap of therapeutic and toxic concentrations and, thus, a narrow therapeutic window. Therefore, methadone dose escalation strategies are not feasible. The µ-opioid receptor is not detected in the majority of glioblastoma specimens. Other molecular methadone targets such as NMDA receptors, in contrast, are frequently expressed in glioblastoma and are assumed to become modulated by therapeutic methadone concentrations. Nevertheless, preclinical data strongly suggest that methadone at clinically relevant concentrations in combination with radiotherapy or temozolomide chemotherapy, or temozolomide-radiochemotherapy does not exert any anti-glioblastoma effect. In addition, clinical data from patients under methadone maintenance therapy or glioblastoma patients with methadone prescriptions for other reasons, do also not suggest anyanti-glioblastoma activity of methadone. Therefore, patients need to be informed about the lack of evidence, and outside of clinical trials, the use of methadone to enhance anti-glioblastoma efficacy should be discouraged.

Concluding Remarks
Recommended plasma methadone concentrations for patients in a methadone maintenance program are in the range of 0.5-2 µM (150-600 ng/mL), which are usually achieved by daily doses of up to 120 mg. Importantly, data from patients who died from methadone overdose suggest an overlap of therapeutic and toxic concentrations and, thus, a narrow therapeutic window. Therefore, methadone dose escalation strategies are not feasible. The µ-opioid receptor is not detected in the majority of glioblastoma specimens. Other molecular methadone targets such as NMDA receptors, in contrast, are frequently expressed in glioblastoma and are assumed to become modulated by therapeutic methadone concentrations. Nevertheless, preclinical data strongly suggest that methadone at clinically relevant concentrations in combination with radiotherapy or temozolomide chemotherapy, or temozolomide-radiochemotherapy does not exert any anti-glioblastoma effect. In addition, clinical data from patients under methadone maintenance therapy or glioblastoma patients with methadone prescriptions for other reasons, do also not suggest anyanti-glioblastoma activity of methadone. Therefore, patients need to be informed about the lack of evidence, and outside of clinical trials, the use of methadone to enhance anti-glioblastoma efficacy should be discouraged.