Methadone-Induced Hypoglycemia

Abstract

To determine if recent observations of hypoglycemia in patients receiving high-dose methadone extended to an animal model, we explored the effects of methadone and other mu-opioids on blood glucose levels in mice. Methadone lowered blood glucose in a dose-dependent manner with 20 mg/kg yielding a nadir in average glucose levels to 55 ± 6 mg/dL from a baseline of 172 ± 7 mg/dL, an effect that was antagonized by naloxone and mu selective antagonists β-funaltrexamine and naloxonazine. The effect was stereoselective and limited to only the l-isomer, while the d-isomer was ineffective. Despite the robust decrease in blood glucose produced by methadone, a series of other mu-opioids, including morphine, fentanyl, levorphanol, oxycodone or morphine-6β-glucuronide failed to lower blood glucose levels. Similar differences among mu-opioid agonists have been observed in other systems, suggesting the possible role of selected splice variants of the mu-opioid receptor gene Oprm1. This mouse model recapitulates our clinical observations and emphasizes the need to carefully monitor glucose levels when using high methadone doses, particularly intravenously, and the need for controlled clinical trials.

Introduction

There exist scattered reports in the literature that connects opioids, primarily methadone, with glucose dysregulation in both clinical (Ceriello et al. 1987, 1988; Howard et al. 2005; Tiras et al. 2006) and animal studies (Amirshahrokhi et al. 2008; Leibetseder et al. 2006; Rastelli et al. 1987; Sadava et al. 1997; Lee et al. 2007; Watts 1951). In several studies, intrathecal morphine induced a dose-dependent hypoglycemia in mice whose specificity for an opioid mechanism was confirmed by its reversal by the opioid antagonist naloxone (el Daly 1996; Lux et al. 1988, 1989). Similarly, patients on long-term intrathecal opioids for intractable pain have significantly lower peak cortisol and peak growth hormone responses to insulin-induced hypoglycemia compared to patients not on opioids (Abs et al. 2000). Methadone maintenance patients have blunted prolactin responses to insulin-induced hypoglycemia (Willenbring and Stevens 1997). For forty years, case reports have linked propoxyphene, an opioid drug structurally related to methadone, to hypoglycemia, especially in patients with renal failure (Wiederholt et al. 1967; Lee et al. 2007). While the mechanisms remain unknown, these investigations have raised the possibility of a spinally medicated central effect, although a more direct endocrine effect on the pancreas cannot be ruled out. In this regard, it is interesting that in rats methadone inhibited pancreatic exocrine secretion in a naloxone-reversible manner (Chariot et al. 1986). Thus, these observations emphasize the importance of opioids in endocrine actions and especially in glucose homeostasis. .

A recent case report emphasizes a potential role of methadone in hypoglycemia (Maingi et al. 2008). The patient was a 46-year-old non-diabetic man with rectosigmoid cancer and renal failure, who was receiving total parenteral nutrition (TPN) and who also needed parenteral opioids due to malabsorption. Upon changing his intravenous fentanyl to methadone, he developed symptomatic hypoglycemia that resolved only with the cessation of methadone. In this case, there appeared to be a correlation between the level of his methadone dose and the degree of the hypoglycemia. Endocrine evaluation revealed no etiology for his recurrent drops in blood glucose. Since then several other patients with both intravenous methadone and hypoglycemia have been encountered in our Institution, although none have been evaluated as extensively as the case described above. These observations led us to explore the potential role of methadone in hypoglycemia.

Methods

Male CD1 mice 4–6 weeks, weighing 20–30 g, were purchased from Charles River Laboratories. The mice had free access to water and ad libitum food, and were housed in a room with a 12-h light/dark cycle. All animal studies were reviewed and approved by the Institutional Animal Care and Use Committee (IACUC). The animal care systems of the MSKCC are fully accredited by the Association for Assessment and Accreditation of Laboratory Animal Care (AAALAC) and are in compliance with the Guide for the Care and use of Laboratory Animals. We are also in compliance with the Animal Welfare Act and agree to adhere to the Public Health Service “Principles for the Use of Animals” (NIH Manual Chapter 4,206).

All opioids were obtained from the Research Technology Branch of the National Institute of Drug Abuse of the National Institutes Health. Drugs were administered subcutaneously. Each cage of mice was kept on a heating pad (SoftHeat, Inc.) for the duration of the experiment to avoid hypothermia associated with the opioid. Animals were dosed based on weight. Blood glucose levels were determined by nicking the tail vein and blood glucose was determined using a One Touch UltraSmart Glucometer (Lifescan, Inc.) at the indicated times.

Results

We first examined the ability of methadone to lower blood glucose levels over time using a fixed dose of methadone (Fig. 1). Racemic methadone lowered glucose levels with a peak effect at 60 min. The average nadir value was close to 50 mg/dL, a marked and significant decrease compared to the saline and baseline levels.

Fig. 1.

Time action of methadone-induced hypoglycemia. Male CD-1 mice (n ≥ 5) were randomly divided and baseline glucose levels were determined at time zero after which each group received either saline, d-methadone, l-methadone, racemic d/l-methadone (all at 20 mg/kg, s.c.), and glucose levels determined as described in Methods. Analysis of variance showed that the curves were statistically significant from each other (p < 0.001). Tukey posthoc analysis showed that the saline and d-methadone curves were not different from each other and the racemic d/l-methadone and the l-methadone curves were not different from each other, but the saline and the d-methadone groups were different from both the racemic d/l-methadone and l-methadone groups

This response was dose-dependent, with only doses greater than 10 mg/kg, s.c. significantly lowering blood levels (Fig. 2). Note that the hypoglycemia seen in these mice required doses of methadone at least five-fold higher than its analgesic ED₅₀ value in the radiant heat tail flick assay (Bolan et al. 2002; Chang et al. 1998).

Fig. 2.

Dose-response of methadone-induced hypoglycemia. Male CD-1 mice (n ≥ 15) were randomly divided into groups and baseline values determined. Each group received saline, racemic d/l-methadone (5, 10, 15, or 20 mg/kg, s.c.). One group receiving d/l-methadone (20 mg/kg. s.c.) was administered naloxone (1 mg/kg, s.c.) after baseline glucose determination, but five minutes before receiving the methadone. This is the result from combining three independent experiments. The methadone doses and their baselines were analyzed using ANOVA and differed (p < 0.0001). Posthoc analysis using Tukey revealed that the 15 and 20 mg/kg methadone doses were significantly different from baselines, which did not differ from each other

Several approaches indicated that the hypoglycemia was mediated through mu-opioid receptors. The response was stereoselective, with only the l-isomer showing activity (Fig. 1). Indeed, the l-isomer was slightly more potent, with a peak effect below 50 mg/dL, although there were several animals whose glucose went as low as 20 mg/dL. The duration of action of the l-isomer was also longer than the d/l-racemate, appearing to plateau between 60 and 120 min. The d-isomer lacked any inhibition of glucose levels, with values indistinguishable from those seen in the saline controls. A similar stereo selectivity has been established for analgesia. The greater potency of the l-isomer was anticipated. Since only the l-isomer is active, the racemate is equivalent to only approximately half the dose of the l-isomer used.

The drop in glucose levels was sensitive to mu-opioid antagonists. Naloxone is an opioid-selective antagonist. Although it is most effective against mu receptors, it can also block delta and kappa sites at higher doses. Against the highest dose of methadone tested (20 mg/kg, s.c.), naloxone (1 mg/kg, s.c.) fully reversed the activity (Fig. 2). β-Funaltrexamine (β-FNA) is an irreversible and selective blocker of all mu-opioid receptors when given the day before. In this model, β-FNA pretreatment effectively reversed the hypoglycemia seen in saline pretreated controls (Fig. 3). Naloxonazine is another mu-selective, irreversible antagonist, but its actions are restricted to a subset of mu-opioid receptors. While it also reduced the methadone effect, the reversal was only partial. Together, these observations strongly implicate a mu-opioid receptor mechanism of action.

Fig. 3.

Antagonism of methadone-induced hypoglyemia. Male CD-1 mice were randomly divided into six groups (n = 15). All groups were pretreated 24 h before baseline testing with either saline, naloxonazine (35 mg/kg, s.c.) or β-FNA (40 mg/kg, s.c.). Glucose determinations were then made 60 min following methadone administration. Baseline glucose was determined at time zero. Groups one, three, and five received normal saline and groups two, four, and six receieved racemic methadone 20 mg/kg. This is the result from combining three independent experiments. ANOVA revealed that the groups were significantly different (p < 0.0001). Posthoc analysis using Tukey showed that the saline groups pretreated with either saline, β-FNA or naloxonazine were not significantly different. These groups were significantly different from the others, which were also different from each other

Molecular biology of the mu-opioid receptors is quite complex, with extensive splicing of the receptor generating a large number of splice variants (Pasternak and Pan 2013). In addition to the traditional mu-opioid receptors that contain the seven transmembrane domains associated with G-protein coupled receptors, a second set of splice variants has been identified associated with exon 11 of the gene that only contain six transmembrane domains (Majumdar et al. 2011; Pan et al. 2001, 2009). Although our understanding of how they might act is limited, studies with a knockout mouse lacking exon 11 and its associated variants clearly shows that these variants are important in the analgesic actions of several classes of analgesics, including mu drugs, while morphine retains full analgesic activity in these animals. We therefore examined the effect of methadone hypoglycemia in these knockout mice which are on the C57 background (Fig. 4). Again, methadone lowered glucose levels in wild-type mice as effectively as in the CD-1 mice. The response in the knockout mice was partially reversed.

Fig. 4.

Methadone-induced hypoglycemia in exon 11 MOR-1 knockout mice. Groups of wild-type male C57 mice (n = 10) or Exons 11 MOR-1 knockout mice (n = 10) (Pan et al. 2009) received racemic d/l-methadone (20 mg/kg, s.c.) and glucose levels were determined 60 min later. The results differ using ANOVA (p < 0.001) with all three differing from each other using Tukey posthoc testing

Despite a role of mu-opioid receptors in methadone-induced hypoglycemia, a variety of other opiates also acting through mu receptors did not lower glucose levels, despite the high doses tested, which were 10-fold greater than their analgesic ED₅₀ values in the radiant heat tail flick assay (Bolan et al. 2002; Zelcer et al. 2005) (Fig. 5). These included morphine, its metabolite morphine-6β-glucuronide, fentanyl, oxycodone, and levorphanol. Thus, in our system, opioid-induced hypoglycemia does not extend to all opiates.

Fig. 5.

Effect of mu-opioids on glucose levels. Male CD1 mice were randomly divided into four groups. After determining baseline glucose levels, each group was administered the indicated drug at approximately ten times their respective analgesic ED₅₀ CD-1 mice as determined in the radiant heat tali flick assay (Bolan et al. 2002; Zelcer et al. 2005). Groups received morphine (50 mg/kg, s.c.; n = 15), M6G (40 mg/kg, s.c.; n = 10), levorphanol (10 mg/kg, s.c.; n = 15) oxycodone (20 mg/kg, s.c.; n = 15), or fentanyl (0.2 mg/kg, s.c.; n = 15), and glucose levels measured again after 60 min. ANOVA revealed no significant difference among any of the groups (p = 0.34)

Discussion

Opioids are widely used clinically and particularly in the treatment of cancer pain, where their intravenous use is common and often at high doses. Clinical observations suggested that methadone occasionally might be responsible for hypoglycemia, a possibility that deserves controlled clinical trials. This was strongly suggested by the patient recently reported (Maingi et al. 2008). In this patient, blood glucose levels were not problematic while he was on intravenous fentanyl, whereas he encountered episodes of hypoglycemia while on methadone that resolved when he was placed back on fentanyl. Although anecdotal, this case report suggests that intravenous methadone may be associated with hypoglycemia while other opiates may not. This would be consistent with our current observation that methadone, but not a series of other mu-opioids, produced hypoglycemia in our mouse model. It should be noted that these situations appear to be uncommon and associated with higher doses of the drug.

In the current study, we show that methadone at high doses clearly lowers glucose levels in a dose-dependent manner in mice. The effect is reversed by mu-selective opioid antagonists and is restricted to the analgesic l-isomer. Similar effects are seen in several strains of mice, including CD-1, C57, and Balb-C. Yet, a variety of other analgesic drugs active at mu-opioid receptors were ineffective in our model despite testing them at equivalently high doses. This lack of effect of other mu-opioid drugs is intriguing. Methadone is a potent mu-opioid analgesic. Its analgesic actions are totally lost in an exon 1 mu-opioid receptor (MOR-1) knockout mouse lacking all the traditional 7 transmembrane domain receptor splice variants and its analgesic actions are reversed by the mu-selective antagonists. Yet, its pharmacology in our model differs from a number of mu-opioids, including morphine. This is not the first situation in which the actions of these two mu-opioids can be distinguished. The CXBK strain of mouse is notable for its insensitivity toward morphine (Baron et al. 1975; Reith et al. 1981; Pick et al. 1993). However, a number of other mu-opioids, including methadone, retain full activity in this mouse strain (Chang et al. 1998; Pasternak and Pan 2013).

Several important questions remain regarding the mechanism of the effect. One involves the actions of the drugs at the receptor. Why these two mu drugs should differ is still not entirely clear. However, the complexity of the splicing of the mu-opioid receptor gene raises the possibility that the differences may reside in their differential interactions with various splice variants. Alternatively, it may result from biased agonism in which the functional activation of different drugs on a receptor can differ (Dewire et al. 2013). A second question is how activation of these receptor produces this effect. It is not clear whether this reflects a direct action on the pancreas or is mediated through the central nervous system. In addition, how the glucose is lowered has not yet been determined. Thus, many questions regarding this action remain to be determined.

Opiates have long been known to have a diverse range of endocrinological actions (Cicero 1980; Elliott et al. 2011a; Timmers et al. 1986; Spiegel et al. 1982; Howlett and Rees 1986; Zhou et al. 2006), with much attention in recent years focused upon their modulation of testosterone levels (Tenhola et al. 2012; Elliott et al. 2011b; Dev et al. 2011; Vuong et al. 2010). Our current studies raise the possibility of an additional role that may be more common with methadone and associated with higher doses. These are intriguing questions that need further study clinically in view of the increased use of high opioid doses in cancer patients.