CYP2D in the brain impacts oral hydrocodone analgesia in vivo

Abstract

Cytochrome P450 2D (CYP2D) metabolises many centrally-acting substrates including opioids. Hydrocodone, an opioid and CYP2D substrate, is metabolised to hydromorphone, an active metabolite. CYP2D in the brain is active in vivo and can alter drug response however, it is unknown whether metabolism by CYP2D in the brain alters oral hydrocodone induced analgesia. Propranolol, a selective CYP2D mechanism-based inhibitor, or vehicle, was administered into the right cerebral ventricle of male rats, (HAN Wistars, Envigo), 24 hours before testing for analgesia from oral hydrocodone (or hydromorphone, a non-CYP2D substrate). Hydrocodone and its CYP2D-mediated metabolites were simultaneously quantified using a novel LC-MS/MS assay. After propranolol vs vehicle pretreatment, there was significantly higher analgesia from oral hydrocodone, and a significantly lower brain CYP2D metabolic ratio (an in vivo phenotype of brain CYP2D activity that was derived from the molar sum of hydromorphone and its metabolites divided by hydrocodone). The brain CYP2D metabolic ratio correlated significantly with analgesia. There was no pretreatment effect on plasma hydrocodone concentrations, elimination rates, or metabolic ratio (an in vivo phenotype for hepatic CYP2D activity). The liver CYP2D metabolic ratio did not correlate with analgesia. Propranolol pretreatment had no impact on analgesia from oral hydromorphone. These data suggest that inhibited CYP2D activity in brain, causing reduced metabolism of brain hydrocodone, resulted in higher analgesia from oral hydrocodone, despite hydrocodone having a lower μ-opioid receptor affinity than hydromorphone. Thus, variation in CYP2D in the brain may be an important source of interindividual differences in response to CYP2D substrates, including oral hydrocodone.

Graphical Abstract

1. Introduction

Hydrocodone is a semi-synthetic opioid derivative used for the treatment of moderate to severe pain and as an antitussive for productive cough in adults (Cofano and Yellon, 2021). In 2020, hydrocodone containing products accounted for the most misused prescription opioid in individuals aged 12 and older in the USA (Substance Abuse and Mental Health Services Administration, 2021) emphasising the need for further investigation into hydrocodone’s pharmacology.

Hydrocodone is metabolised by the CYP2D enzyme family to hydromorphone, which is postulated to mediate hydrocodone’s analgesic effects, and by the CYP3A enzyme family to norhydrocodone, an analgesic with low μ-opioid receptor affinity (Navani and Yoburn, 2013; Stauble et al., 2014). The human CYP2D6 and CYP3A4 pathways assessed in vitro account for 52% and 47% of hepatic clearance of hydrocodone, respectively (Hutchinson et al., 2004; Otton et al., 1993). Hydromorphone is primarily glucuronidated to hydromorphone 3-glucuronide, which is not an analgesic (Wright et al., 2001; Zheng et al., 2004).

Human CYP2D6 is highly genetically polymorphic, resulting in metabolic phenotypes ranging from poor to ultrarapid metabolisers (Crews et al., 2021). CYP2D6 metabolises 25% of clinically used drugs, many of which are centrally-acting such as opioids, antidepressants, and antipsychotics (Ingelman-Sundberg, 2004). While CYP2D6 is found predominantly in the liver, it is also located in extrahepatic tissues, including the brain (Miksys et al., 2002; Miksys et al., 2000).

In addition to hydrocodone, several opioids are metabolised by CYP2D to active metabolites, including codeine to morphine, oxycodone to oxymorphone, and tramadol to O-desmethyltramadol. Each of the metabolites has a greater affinity for the μ-opioid receptor than their parent opioid (Gillen et al., 2000; Volpe et al., 2011). Hydromorphone has a 100-fold greater affinity for the μ-opioid receptor than hydrocodone (Volpe et al., 2011). However, after oral hydrocodone administration, there was no difference between CYP2D6 poor metabolisers (PMs) and CYP2D6 extensive metabolisers (EMs) in objective (pupil diameter) and subjective (drug-liking) effects, despite PMs having lower plasma hydromorphone to hydrocodone ratios than EMs (Kaplan et al., 1997). In contrast, in a separate study, following oral hydrocodone, pain scores were significantly associated with serum hydromorphone, but not hydrocodone or norhydrocodone concentrations, suggesting hydrocodone may be a prodrug (Crews et al., 2021; Stauble et al., 2014).

CYP2D in the brain is expressed in humans, rats, mice, and monkeys (Mann et al., 2008; Miksys et al., 2002; Miksys et al., 2000; Miksys et al., 2005) and is sensitive to environmental inducers such as smoking/nicotine and alcohol (Mann et al., 2008; Mann et al., 2012; Miller et al., 2014; Yue et al., 2008). Manipulating metabolism by CYP2D in the brain, but not liver, alters response to centrally-acting substrates such as opioids, antipsychotics, and neurotoxins (Miksys et al., 2017; Stocco et al., 2020; Zhou et al., 2013). CYP2D in the liver is regulated exclusively by genetics and is considered uninducible (Edwards et al., 2003). However, plasma drug concentrations are not always predictive of drug response (Spina et al., 1997). Hence, variation in CYP2D-mediated drug metabolism in the brain may contribute to interindividual variation in drug response.

Pharmacological paradigms have been developed to investigate the impact of metabolism by CYP2D in the brain on central drug concentration and drug response without affecting hepatic CYP2D or plasma drug concentrations (McMillan et al., 2019; Yue et al., 2008; Zhou et al., 2013). Inhibition of CYP2D in the rat brain is achieved through intracerebroventricular (icv) injection of propranolol (PRL, a mechanism-based inhibitor of CYP2D) 24 hours prior to administering a CYP2D substrate and assessing drug response (Arguelles et al., 2021; McMillan et al., 2019; Miksys et al., 2017; Stocco et al., 2021; Wang et al., 2015; Zhou et al., 2013). PRL has a half-life of 1-hour in rat brain (Elghozi et al., 1979) therefore after 24 hours it has been cleared from the brain which reduces the potential for effects other than inhibition of CYP2D in the brain (Elghozi et al., 1979).

Inhibition of CYP2D in the rat brain, using icv PRL, alters analgesic response to codeine, tramadol, and oxycodone (McMillan et al., 2020; McMillan et al., 2019; Zhou et al., 2013). Following propranolol inhibition of CYP2D in the rat brain, and oral oxycodone administration, oxycodone concentrations in the brain and resulting analgesia were higher in males and in females during estrus (Arguelles et al., 2021; McMillan et al., 2019). Thus, despite a 50-fold lower affinity of oxycodone compared to oxymorphone for the μ-opioid receptor, oxycodone within the brain contributes to analgesia (Arguelles et al., 2021; Arguelles et al., 2022; McMillan et al., 2019).This example indicates that manipulating CYP2D in the brain and then assessing response to centrally-acting substrates can 1) provide evidence for a role of central CYP2D-mediated metabolism in brain drug concentrations and drug response, 2) identify a potential source of interindividual variation in response to centrally-acting substrates beyond that of hepatic metabolism, and 3) provide some evidence for a role of a parent drug and/or metabolite in analgesic response.

Our objectives were to 1) construct oral hydrocodone and hydromorphone dose response curves and compare analgesia among orally administered CYP2D opioids, 2) develop a novel method to analyze hydrocodone and its five metabolites, and 3) assess the impact of inhibited metabolism by CYP2D in the brain on oral hydrocodone analgesia. The third objective was tested using an established inhibition paradigm, as well as the CYP2D metabolic ratio, an in vivo marker of CYP2D activity (Arguelles et al., 2021; McMillan et al., 2019; Miksys et al., 2017; Stocco et al., 2021; Zhou et al., 2013). Analgesia from oral hydromorphone, which is not further metabolised by CYP2D, was used as a control for potential effects of the inhibitor on analgesia other than inhibition of CYP2D.

2. Materials and methods

2.1. Materials

Analytical standards for hydrocodone, hydrocodone-d3, hydromorphone, hydromorphone-d3, norhydrocodone, norhydrocodone-d3, hydromorphone-3-glucuronide, HPLC-grade formic acid (Sigma-Aldrich, St. Louis, MO, USA), hydrocodol, 6-β-hydromorphol (Toronto Research Chemicals, Toronto, ON, CAN), and HPLC-grade solvents (Caledon Laboratory Chemicals, Georgetown, ON, CAN) were purchased to develop and validate a new assay for quantifying hydrocodone and its metabolites.

2.2. Animals

Adult male Wistar Han rats (Envigo, Indianapolis, IN, USA) were housed in groups of twos or threes with ad libitum access to water and were food restricted to maintain a maximum weight between 350–400 g. Rats were kept under a 12-hr light/dark cycle with experiments taking place in the light phase and were acclimatized to all experimental procedures to reduce stress. All animal procedures conformed to the ARRIVE guidelines (Percie du Sert et al., 2020), the recommendations of the National Research Council’s Guide for the Care and Use of Laboratory Animals, and the Canadian Council on Animal Care, and were approved by the Animal Care Committee at the University of Toronto.

2.3. Cannulation Surgeries

Stainless steel 22-gauge guide cannula (P1 Technologies, Roanoke, VA, USA) were implanted under isoflurane anaesthesia into the right cerebral ventricle (Bregma coordinates anteroposterior: −0.9 mm; lateral: −1.4 mm; dorsoventral: −3.6 mm) and secured with stainless steel anchor screws and dental cement. Animals were allowed to recover for a week before experiments started, with 5 mg/kg of meloxicam (CDMV, Brampton, ON, CAN) administered subcutaneously daily up to 72 hr post-surgery.

2.4. Drug Treatment

Hydrocodone phosphate (PCCA, London, ON, CAN) and hydromorphone hydrochloride (Toronto Research Chemicals) were dissolved in phosphate buffered saline (0.1 M, pH 7.4) and administered by oral gavage at 4 ml/kg. A range of oral doses of hydrocodone (2.5 mg/kg to 55 mg/kg) and hydromorphone (6.5 mg/kg to 12.5 mg/kg) was tested. PRL (Sigma-Aldrich), a CYP2D mechanism-based inhibitor, is metabolised by CYP2D to a reactive intermediate which binds covalently to the active site of CYP2D and inhibits its function (Narimatsu et al., 2001). PRL was dissolved in 20% (w/v) 2-hydroxypropyl-β-cyclodextrin (Sigma-Aldrich) in water (VEH); 20 μg PRL base in 4 μl VEH or 4 μl VEH alone was injected via the implanted cannula into the right ventricle over 3 min, with injector left in place for an additional min.

2.5. Analgesia assessment

Antinociception from hydrocodone and hydromorphone in the animals is described as analgesia throughout the manuscript. Analgesia was assessed as the tail-flick latency using a tail-flick meter (Columbus Instruments, Columbus, OH, USA) and was expressed as % maximal possible effect (MPE) using the formula: %MPE = (post-injection latency – baseline latency)/ (maximum latency – baseline latency) × 100%. The temperature of the tail-flick meter was adjusted to give baseline latencies between 3–4 s and a maximum latency cut-off of 12 s was used to prevent tissue damage.

2.6. Analgesia experimental design

Rats were pretreated with icv injections of PRL or VEH 24 hrs prior to oral opioid administration (hydrocodone at 45 mg/kg or hydromorphone at 10 mg/kg) and tested for analgesia. This paradigm selectively and irreversibly inhibits CYP2D in rat brain, but not liver (McMillan et al., 2019; Miksys et al., 2017; Stocco et al., 2021; Zhou et al., 2013). Twenty-three hours after icv pretreatment, rats were placed in the behaviour room to acclimatize and were food deprived for at least 30 min prior to oral opioid administration. An average of two baseline tail-flick latencies was taken prior to opioid gavage and analgesia was assessed at the following times post gavage: 5, 10, 15, 20, 25, 30, 40, 50, 60, 75, 90 and 120 min. Emphasis was placed on data generated during the first 60 min of analgesia as previous work showed a maximal effect of central drug metabolism during this time (McMillan et al., 2020; Zhou et al., 2013). Blood was collected from the saphenous vein in lithium heparin-coated microvettes (Sarstedt AG & Co. KG, Nümbrecht, Germany) at 30 min (during analgesia testing) and 130 min (after analgesia testing) post gavage to assess any PRL pretreatment effects on plasma opioid concentrations. Plasma was separated by centrifugation at 5000 g for 10 min, collected into new tubes, and stored at −80°C until analysis. One week following analgesia tests, animals were pretreated icv with PRL or VEH 24 hours prior to oral hydrocodone at 45 mg/kg and were euthanized by decapitation 30 min later. Drug concentrations were assessed ex vivo from plasma and brain tissue. Each experiment was run with cohorts of animals due to Covid 19-related restrictions on animal use. A description of cohorts (1–6) is found in the results; each cohort contained approximately equal number of PRL and VEH pretreated animals.

2.7. Analytical procedures

Concentrations of hydrocodone and its metabolites hydromorphone, hydromorphone-3-glucuronide, 6-beta-hydromorphol, norhydrocodone and hydrocodol were measured in brain tissue and plasma samples using a novel liquid chromatography – mass spectrometry (LC-MS/MS) method for simultaneous determination of these compounds. The sample extraction procedure and the LC-MS/MS conditions were adapted from previously published procedures (Barakat et al., 2014; Li et al., 2013). The methods were expanded to include additional metabolites of interest and to measure them in both plasma and in brain tissue. Brain tissue was homogenized 1:3 (w/v) in water; 50 μl of internal standard solution (combined hydrocodone-d3, norhydrocodone-d3 and hydromorphone-d3 at 20 ng/ml of each in water) was added to 50 μl of brain homogenate, or to 50 μl of plasma, followed by the addition of 1 ml of acetonitrile. After vortexing, the samples were centrifuged at 16,000 g for 10 min, then the supernatant was dried under nitrogen. The samples were each reconstituted in 100 μl of water and 40 μl of each sample were injected onto the LC-MS/MS.

The analysis was carried out on an Agilent 1260 LC system connected to a 6430 Triple Quadrupole system (Agilent Technologies, Santa Clara, CA, USA). Analytes were separated on an Agilent Zorbax SB C18 column, 150 × 2.1 mm 5 μm particle size (Agilent Technologies). The separation was performed at ambient temperature using a gradient elution with 0.1% formic acid in water (mobile phase A) and 0.1% formic acid in acetonitrile (mobile phase B) at a flow rate of 0.3 ml/min. The gradient was as follows: 0 – 0.5 min 5% B, 0.5 – 10 min linear increase to 20% B, 10 – 11 min linear decrease to 5% B, 11 – 17 min held at 5% B.

The mass spectrometer was operated under positive electrospray ionization mode using multiple reaction monitoring. The gas temperature, flow, and nebulizer pressure were optimized to 350°C, 10 l/min and 35 psi, respectively, and the capillary voltage was 4.0 kV. The nebulizer gas was high-purity nitrogen and was acquired from an in-house source. Two ion transitions were monitored for each analyte; the collision energies (CE) and fragmentor voltages (FV) are listed in Table 1.

Table 1:

Mass spectrometry parameters used for hydrocodone, metabolites, and internal standards.

Analyte	Transition	Collision Energy (eV)	Fragmentor Voltage (FV)
Hydromorphone 3-glucuronide	462.2→ 286.2 → 185.0	29 57	150 150
Hydrocodol	302.2→ 199.1 → 128.0	33 60	121 212
Hydrocodone	300.2→ 199.0 → 171.1	29 45	121 121
6-β-Hydromorphol	288.2→ 157.1 → 165.0	57 49	121 121
Norhydrocodone	286.2→ 199.0 → 171.0	29 37	150 150
Hydromorphone	286.2→ 185.0 → 157.0	33 45	121 121
Hydrocodone-d3	303.2→ 199.0	29	121
Norhydrocodon-d3	289.2→ 202.1	25	121
Hydromorphone-d3	289.2→ 185.1	33	121

The first transition was for a quantifier ion and was used to calculate concentration, while a second transition was for a qualifier ion and was used to ensure accurate target compound identification by monitoring the ratio of the qualifier ion to the quantifier ion. To meet criteria, the ratio deviation of ±20% of the average ion ratios for calibrators was accepted.

The standard curves were calibrated with linear function and 1/x weighing. The accepted accuracy of calibrators was +20% of the target value, and the R² was accepted when ≥ 0.99. Hydrocodone was quantified using hydrocodone-d3 as an internal standard, hydromorphone, hydromorphone-3-glucuronide and 6-β-hydromorphol were quantified using hydromorphone-d3 as an internal standard, and hydrocodol and norhydrocodone were quantified using norhydrocodone-d3 as an internal standard. The calibration curves were prepared in the same biological matrix as the samples (i.e. plasma or brain homogenate) and contained at least four calibration points and ranged from 0.5 to 1000 ng/ml for all compounds. The method was validated for intra-day and inter-day accuracy and precision at three different concentrations and met the criteria of % error and % CV that ranged from ± 15% to ± 20% at all concentrations tested. The limit of quantification was 0.5 ng/ml for all analytes; analytes below the limit of quantification were replaced for using the formula: (limit of quantification/square root(2)) (Kalkbrenner et al., 2010).

2.8. Derivation of brain and liver enzymatic metabolic ratios

Brain and plasma concentrations of hydrocodone and its metabolites hydromorphone, hydromorphone-6-glucuronide, 6-β-hydromorphol, norhydrocodone and hydrocodol, were quantified after 45 mg/kg of oral hydrocodone. The sum of the molar concentrations of hydrocodone’s CYP2D-mediated metabolite hydromorphone and its downstream metabolites (i.e., hydromorphone-6-glucuronide and 6-β-hydromorphol) was divided by the molar concentration of the substrate hydrocodone to create brain and liver CYP2D metabolic ratios, a phenotypic measure of in vivo CYP2D activity. The brain and liver CYP3A metabolic ratios were created similarly by dividing the CYP3A-mediated metabolite norhydrocodone by hydrocodone molar concentrations, and the reduction pathway metabolic ratios were created by dividing the metabolite hydrocodol by hydrocodone molar concentrations.

2.9. Calculations and Statistical Analyses

Data were analysed with GraphPad Prism v.6.01 (GraphPad Software Inc., La Jolla, CA, USA), with and without outliers; outliers were determined using a Grubbs test set at alpha = 0.05. The number of n’s in each group was determined using the means and common standard deviation from our pilot study; the online Power/Size Calculator from https://www.stat.ubc.ca/~rollin/stats/ssize/n2 was used to calculate the sample size required at 80% power. Two-way ANOVAs were used to analyse analgesia by time curves for hydrocodone and hydromorphone (variables were pretreatment and time) followed by Sidak post hoc testing adjusted for multiple comparisons. The theoretical analgesia AUC _0–60 maximal response (4500 %MPE.min) was created by finding the area under the curve, using the linear trapezoidal method, of a steady incline in analgesia response at zero time to a maximal response at 30 min, followed by the maximal response at 30–60 min; this pattern of response was frequently observed with the oral opioids. The ED₅₀ for analgesia from each opioid was determined by solving the equation of the linear regression at half the maximal analgesia response; for analgesia AUC _0–60, the half maximal response value was 2250 %MPE.min while for the %MPE, this was 50%. Differences between treatment groups for the analgesia AUC _0–60, plasma concentrations at 30 and 130 min, slopes for plasma concentrations, brain and liver metabolic ratios were assessed using an unpaired Student T-test, with a Welch’s correction if groups had unequal variance. Relationships between groups were assessed by Pearson’s (r) or by Spearman’s (r_s) correlation analyses, for normally or non-normally distributed data, respectively. Significance was set at 5% and data are presented as means plus and minus the standard error of the mean (SEM).

3. Results

3.1. Analgesia dose response curves were created and compared to other orally administered opioids

A range of oral doses of hydrocodone and hydromorphone was tested in male rats and the resulting analgesia was compared to that from other oral opioids, where the data were previously collected in our laboratory (Figure 1A and 1B) (McMillan et al., 2020; McMillan et al., 2019; McMillan and Tyndale, 2017). The rank order of potency (with lower ED₅₀ values representing greater potency) was similar for both measures of ED₅₀, with hydrocodone being more potent than tramadol, of similar potency to codeine, and less potent than morphine, hydromorphone, oxymorphone and oxycodone (Figure 1C). For each pair of CYP2D substrates and metabolites, the parent was less potent than the metabolite; hydrocodone, codeine, and oxycodone were less potent than hydromorphone, morphine, and oxymorphone, respectively. A hydrocodone dose of 45 mg/kg was chosen for subsequent experiments as it produced analgesia of approximately 60% MPE (Figure 1B). A hydromorphone dose of 10 mg/kg (used as a negative control as it is not a CYP2D substrate) was chosen as it produced analgesia of approximately 40% MPE (Figure 1B). Both doses chosen enabled detection of increases or decreases in analgesia following propranolol pretreatment.

Fig 1.

Analgesia dose response curves were created and compared to other orally administered opioids

The oral dose response curves for hydrocodone (large grey circles), a CYP2D substrate, and its CYP2D mediated metabolite hydromorphone (large grey triangles) were created and then contrasted to the oral dose response curves for five additional opioids tested in our laboratory (McMillan et al., 2020; McMillan et al., 2019; McMillan and Tyndale, 2017). Data are shown as the analgesia AUC_0–60 (area under the curve) (A), as the %MPE (percent maximal possible effect) (B) and ED₅₀s (C).

3.2. Propranolol icv pretreatment increased oral hydrocodone analgesia

Analgesia from 45 mg/kg oral hydrocodone was assessed 24 hr after PRL or VEH pretreatments (n = 30, cohorts 1 and 2). Data are presented as the mean ± SEM. Propranolol (n = 16) compared to VEH (n = 14) pretreatment resulted in higher analgesia from 0 to 60 min, (pretreatment F (1,28) = 4.843, p = 0.0362), that was significantly 1.71-fold higher at 30 min (78 ± 6 vs 46 ± 8 % MPE, p < 0.05, Figure 2A). Additionally, PRL compared to VEH pretreatment resulted in a significant 1.36-fold increase in the analgesia AUC _0–60 (3346 ± 321 vs 2469 ± 273 %MPE.min, t (28) = 2.0481, p = 0.0500, Figure 2B). There was no PRL pretreatment effect on baseline tail flick latencies that were measured immediately prior to oral hydrocodone (t (28) = −0.0801, p = 0.9368).

Fig 2.

Propranolol icv pretreatment increased oral hydrocodone analgesia

Propranolol (n = 16) compared to VEH (n = 14) pretreatment resulted in significantly higher analgesia after 45 mg/kg oral hydrocodone (pretreatment F (1,28) = 4.843, p = 0.0362) (A), which was significant at 30 min using a Sidak post hoc test (p < 0.05). The PRL (n = 16) compared to VEH (n = 14) pretreatment also resulted in significantly higher AUC _0–60, t (28) = 2.0481, p = 0.0500 (B). Individual animals are shown by circles in B.

3.3. Propranolol icv pretreatment did not affect plasma hydrocodone concentrations

Blood samples (n = 30, cohorts 1 and 2) were taken at 30 min and 130 min after hydrocodone oral gavage. Two plasma samples (one in each cohort with different pretreatments) were contaminated and thus excluded. Propranolol (n = 15) compared to VEH (n = 13) pretreatment resulted in similar plasma hydrocodone concentrations at 30 min (88.6 ± 18.0 vs 76.4 ± 17.4 ng/ml, t (26) = 0.4712, p = 0.6414) and (75.1 ± 13.8 vs 76.4 ± 17.4 ng/ml, t (25) = 0.0594, p = 0.9531) with all animals and with one outlier removed from the PRL group respectively. PRL, compared to VEH, pretreatment also resulted in similar plasma hydrocodone concentrations at 130 min (25.4 ± 4.4 vs 26.2 ± 3.7 ng/ml, t (26) = −0.1238, p = 0.9024) and (22.1 ± 3.4 vs 26.2 ± 3.7 ng/ml, t (25) = 0.8023, p = 0.4300 with one outlier removed from the PRL group respectively, Figure 3A). Plasma hydrocodone elimination was estimated for each animal from the slopes of plasma hydrocodone concentrations from 30 to 130 min. Propranolol (n = 15) compared to VEH (n = 13) pretreatment had no effect on plasma hydrocodone elimination with all animals (−0.6 ± 0.2 vs −0.5 ± 0.2 ng/ml/min, t (26) = −0.5543, p = 0.5841), and even after removing one outlier from each pretreatment group (−0.5 ± 0.1 vs −0.4 ± 0.1 ng/ml/min, t (24) = −0.6827, p = 0.5014, Figure 3B). These results indicate that PRL did not cross into the periphery at levels which altered hepatic CYP2D sufficiently to impact systemic hydrocodone concentrations.

Figure 3.

Propranolol icv pretreatment did not affect plasma hydrocodone concentrations

Propranolol (n = 14) compared to VEH (n = 13) pretreatment resulted in similar hydrocodone concentrations at 30 min t (25) = 0.0594, p = 0.9531) and at 130 min, t (25) = 0.8023, p = 0.4300 (A). Elimination slopes for hydrocodone were estimated for each animal and PRL (n = 14) compared to VEH (n = 12) pretreatment resulted in similar slopes (t (24) = −0.6827, p = 0.5014) (B). Individual animals are shown by circles in A and B while bars represent the average for each pretreatment group.

3.4. Analgesia correlated with in vivo CYP2D activity in the brain but not in the liver

Brain and plasma drug and metabolite concentrations were assessed ex vivo (n = 20 remaining from cohorts 1 and 2 due to attrition) and used to create metabolic ratios. Analgesia AUC_0–60 (illustrated in Figure 2B) significantly inversely correlated with the brain CYP2D metabolic ratio within animal (r = −0.4684, p = 0.0373, Figure 4A). As CYP2D activity in the brain increased, analgesia decreased (Figure 4A). Analgesia did not correlate with the liver CYP2D metabolic ratio (r_s = −0.0707, p = 0.7672, Figure 4B). Additionally, there was no correlation between brain and liver CYP2D metabolic ratios (r_s = 0.3057, p = 0.1899 (Figure 4C). This was expected, as 50% of the animals had CYP2D inhibited in the brain, but not in the liver. Analgesia AUC _0–60 (illustrated in Figure 2B) did not correlate with brain (r = - 0.2516, p = 0.2845) or liver (r = 0.0714, p = 0.7649) CYP3A metabolic ratios or with brain (r_s = −0.1368, p = 0.5651) or liver (r_s = 0.0782, p = 0.7431) reduction pathway metabolic ratios.

Fig 4.

Analgesia correlated with in vivo CYP2D activity in the brain but not in the liver

Analgesia AUC_0–60 resulting from 45 mg/kg oral hydrocodone (n = 20) significantly inversely correlated with the CYP2D metabolic ratio in the brain (r = −0.4684, p = 0.0373) (A), but not in the liver (r_s = −0.0707, p = 0.7672) (B). Additionally, as expected, the brain and plasma metabolic ratios were not significantly correlated (r_s = 0.3057, p = 0.1899) (C).

3.5. Propranolol icv pretreatment decreased in vivo CYP2D metabolic ratio in the brain but not the liver

A third cohort was added to further assess PRL pretreatment effects on the brain metabolic ratio (n = 28, cohorts 1, 2 and 3). PRL (n = 15) compared to VEH (n = 13) pretreatment resulted in a significantly lower brain CYP2D metabolic ratio (0.4 ± 0.1 vs 0.7 ± 0.1, t (26) = 2.2681 , p = 0.0319, Figure 5A). As expected, PRL compared to VEH pretreatment resulted in no difference in the brain CYP3A (t (26) = −1.1534, p = 0.2593) or reduction pathway (t (26) = −0.1615, p = 0.8730) metabolic ratios. A fourth cohort was added to further assess any pretreatment effect on the liver metabolic ratio (n = 36, cohorts 1, 2, 3 and 4). This was to increase confidence in the expected lack of pretreatment effect. PRL (n = 18) compared to VEH (n = 18) pretreatment resulted in no difference in the liver CYP2D metabolic ratio (6.8 ± 1.9 vs 8.1 ± 2.2, t (34) = −0.4713, p = 0.6405), and even following the removal of one outlier from both the PRL and VEH pretreatment groups (5.2 ± 1.2 vs 6.6 ± 1.6, t (32) = 0.6930, p = 0.4933, Figure 5B). Propranolol compared to VEH pretreatment also resulted in no difference in liver CYP3A (t (29) = −0.0600, p = 0.9525) or reduction pathway (t (34) = −0.1811, p = 0.8574) metabolic ratios. These results with those from section 3.4 and Figure 4 further support that PRL did not cross into the periphery at levels which altered hepatic CYP2D sufficiently to impact systemic hydrocodone concentrations.

Fig 5.

Propranolol icv pretreatment decreased in vivo CYP2D metabolic ratio in the brain, but not the liver

Propranolol (n = 15) compared to VEH (n = 13) pretreatment resulted in a significantly lower brain CYP2D metabolic ratio t (26) = 2.2681, p = 0.0319 (A). In contrast, PRL (n = 17) compared with VEH (n = 17) pretreatment resulted in no difference in the plasma CYP2D metabolic ratio t (32) = 0.6930, p = 0.4933 (B). Individual animals are shown by circles in A and B while bars represent the averages in each pretreatment group.

As the PRL, compared to VEH, group had higher analgesia (Figure 2) and lower brain CYP2D activity (Figure 5A) we examined the individual analytes in the brain. There was no difference in total analyte recovery between the two pretreatment groups (t (20) = 1.0223, p = 0.3189 and t (26) = 0.0281, p = 0.9778) with all animals and with outliers removed respectively. However, as there was wide variation in the CYP2D activity and total analyte recovery among the animals, we normalized each compound (in nmoles/g) to the molar sum of all the analytes in the brain of each animal. PRL (n = 15) compared to VEH (n = 13) pretreatment had non-significantly higher hydrocodone (0.48 ± 0.04 vs 0.42 ± 0.04, t (26) = 1.1029, p = 0.2802) and (0.47 ± 0.02 vs 0.42 vs 0.04, t (26) = 1.1417, p = 0.2644) in the brain with all animals in the data and with outliers removed respectively. There was also non-significantly lower hydromorphone (0.08 ± 0.01 vs 0.11 ± 0.02, t (26) = −1.8335, p = 0.0782) and (0.10 ± 0.01 vs 0.11 ± 0.02, t (26) = −1.0305, p = 0.3123) with all animals in the data and with outliers removed respectively. Thus, in PRL compared to VEH pretreatment hydrocodone modestly increased (12–14%) while hydromorphone modestly decreased (9–27%), suggesting that the increase in analgesia in PRL compared to VEH may have been due to increased hydrocodone.

3.6. Propranolol icv pretreatment did not impact analgesia from oral hydromorphone

Hydromorphone, the CYP2D-mediated metabolite of hydrocodone that is not a CYP2D substrate, was given at 10 mg/kg as a negative control for the PRL pretreatment effect (n = 19, cohorts 5 and 6). Propranolol (n = 10) compared to VEH (n = 9) pretreatment resulted in no difference in analgesia over time from 0 to 60 min, (pretreatment F (1,17) = 0.0944, p = 0.7624, Figure 6A). There was also no difference in the analgesia AUC _0–60 for all animals (1088 ± 246 vs 1044 ± 207 %MPE.min, t (17) = −0.1341, p = 0.8949) and even after the removal of one outlier from the PRL pretreatment group (877 ± 141 vs 1044 ± 207 %MPE.min, t (16) = 0.6678, p = 0.5138, Figure 6B). One animal from the PRL group had a contaminated blood sample and was excluded from slope analyses. Propranolol (n = 9) compared with VEH (n = 9) pretreatment resulted in no difference in the slopes of plasma hydromorphone concentrations from 30 to 130 min (−0.2 ± 0.1 vs −0.2 ± 0.2 ng/ml/min, t (16) = −0.0755, p = 0.9408, Figure 6C). These results indicate that PRL has no effects on analgesia beyond its expected impact on CYP2D in brain in the hydrocodone experiments.

Fig 6.

Propranolol icv pretreatment did not impact analgesia from oral hydromorphone

Analgesia from 10 mg/kg of hydromorphone, which is not a CYP2D substrate, was assessed 24 hours after PRL or VEH pretreatment. Propranolol (n = 10) compared to VEH (n = 9) pretreatment resulted in no difference in analgesia (pretreatment F (1,17) = 0.0944, p = 0.7624) (A), or in analgesia AUC_0–60 (t (16) = 0.6678, p = 0.5138) (B). Furthermore, PRL (n = 9) compared to VEH (n = 9) pretreatment resulted in no difference in hydromorphone elimination slopes (t (16) = −0.0755, p = 0.9408) (C). Individual animals are shown by triangles in B and C, while bars represent the averages in each pretreatment group.

4. Discussion

This is the first study to demonstrate a role for metabolism by CYP2D in the brain in oral hydrocodone analgesia. Inhibition of CYP2D in the brain resulted in higher analgesia from oral hydrocodone, as well as lower brain, but not liver, CYP2D activity in vivo (phenotyped using the metabolic ratio). Further, analgesia correlated with brain, but not liver, CYP2D activity. Together, our findings indicate that reducing CYP2D activity in the brain increased analgesia from oral hydrocodone and suggest that the brain hydrocodone concentrations may play a role in analgesia. There was no inhibitor pretreatment impact on baseline analgesia, plasma hydrocodone concentrations, the rate of removal of hydrocodone from plasma, or brain or hepatic CYP3A or reduction enzymatic pathways. Additionally, the inhibitor pretreatment had no impact on hydromorphone induced analgesia as hydromorphone is not a CYP2D substrate. This suggests that the changes in analgesia following inhibitor pretreatment and oral hydrocodone were likely due primarily to inhibitor effects on central metabolism by CYP2D in the brain, and not to potential alternative impacts of the inhibitor on opioid response at, or downstream of, the μ-opioid receptor.

It has been unclear whether hydrocodone, or the more potent CYP2D metabolite hydromorphone, contributes to hydrocodone induced analgesia. Our data suggest that brain hydrocodone concentrations may contribute meaningfully to hydrocodone induced analgesia. There are several factors that influence analgesia from the parent and/or the metabolite. These include affinity of the parent vs metabolite for the μ-opioid receptor, drug concentrations at the site of action, and their rate of influx vs efflux across the blood brain barrier (Boström et al., 2005; Mercer and Coop, 2011). Hydrocodone is more lipophilic (Peckham and Traynor, 2006) than hydromorphone making it more likely to cross the blood brain barrier. In the vehicle pretreatment groups in our study, the hydrocodone concentration in rat brain at 30 min was 4.6-fold higher than the hydromorphone concentration following oral hydrocodone administration. Thus, some of the lack of clarity about the relative roles of hydrocodone vs hydromorphone in oral hydrocodone analgesia may be due to relative hydrocodone and hydromorphone influx/efflux and the impact of metabolism by CYP2D on hydrocodone in the brain.

Species differences in substrate specificity of CYP2D isoforms may be a limitation when extrapolating our findings to humans. However, we have previously used a transgenic CYP2D6 mouse model that expresses both mouse CYP2D and human CYP2D6 and demonstrated that the human CYP2D6 in brain is active in vivo and able to alter drug response (Stocco et al., 2020; Tolledo et al., 2020). Chronic smokers, heavy alcohol users, and the elderly, each of which has been associated with higher CYP2D6 in the human brain, but not the liver, may experience less analgesia from hydrocodone due to increased metabolism of hydrocodone in the brain (Mann et al., 2012; Miksys et al., 2002). In agreement with this, current smokers prescribed hydrocodone for 30 days had less pain relief, indicated by higher pain scores, compared to non-smokers (Ackerman, 2013).

The effects of inhibition of CYP2D in the brain on oral hydrocodone analgesia was demonstrated following an acute administration. However, chronic pain was estimated to affect 50 million people in 2016 in the USA and is often treated with chronic administration of opioids (Busse et al., 2018; Dahlhamer et al., 2018). Of note, a role for CYP2D in brain under chronic dosing has been demonstrated in rats for haloperidol side effects (approximately 4 months haloperidol dosing) (Miksys et al., 2017) , methamphetamine behavioural sensitization (seven day methamphetamine treatment) (Stocco et al., 2021), and codeine tolerance (7 day codeine treatment) (McMillan and Tyndale, 2017). Together this suggests that there may be a role for CYP2D activity in the brain in chronic dosing of hydrocodone, as we have shown here for acute dosing.

Sex differences in opioid analgesia are evident, but the findings are inconsistent in the literature (Craft, 2003). This variation may be due to differences in pain types being measured, the route of administration, analgesia testing methods, the specific opioid and perhaps also due to variation in drug metabolism in the brain. We have previously observed rodent sex and cycle differences in response to oral oxycodone (Arguelles et al., 2021; Arguelles et al., 2022). Males and females in estrus had lower brain oxycodone and analgesia compared to females in diestrus; analgesia correlated with brain, but not plasma, oxycodone concentrations (Arguelles et al., 2021). Furthermore, following ovariectomy, analgesia was higher and insensitive to CYP2D inhibitor pretreatment. Following subsequent acute estradiol treatment, analgesia from oral oxycodone was reduced and was again sensitive to CYP2D inhibitor pretreatment (Arguelles et al., 2022). Hence, compared to females in diestrus and ovariectomized females, males, females in estrus, and ovariectomized rats treated with estradiol have lower brain oxycodone and lower analgesia from oxycodone due to higher CYP2D in the brain; the same may be seen with hydrocodone. This also suggests that women, at certain phases of the menstrual cycle, or those taking estrogen-containing hormone replacement therapy, may experience less analgesia at standard doses of hydrocodone due to higher levels of CYP2D6 in the brain.

Oral dose response curves for hydrocodone and hydromorphone were compared to other orally administered opioids. In order of decreasing potency, the rank order was oxymorphone, oxycodone, hydromorphone, morphine, hydrocodone, codeine, and then tramadol. This order of potency did not align with previously published data on subcutaneous administration in rats or oral administration in humans using the morphine milligram equivalence (MME) factor (Dowell et al., 2016; Peckham and Traynor, 2006). Differences in route of administration or pharmacokinetics, such as oral bioavailability, may account for lack of concordance in potency. In evaluating the opioids further, we compared the oral and subcutaneous analgesia ED₅₀ (%MPE) for each opioid to their affinities for the rat and human μ-opioid receptor and their MME factor (Peckham and Traynor, 2006; Volpe et al., 2011). We excluded tramadol as its pharmacodynamics differ from the other opioids, that mediate their analgesia primarily through μ-opioid receptors, and also as tramadol subcutaneous ED₅₀ data were not available (Peckham and Traynor, 2006; Subrahmanyam et al., 2001). Using linear regression, compared to the oral ED₅₀ (% MPE), the rat (Peckham and Traynor, 2006) and human μ-opioid receptor (Volpe et al., 2011) affinity provided variance (R²) values of 0.45 and 0.47, respectively. This suggests that the affinity of an orally administered opioid for the μ-opioid receptor predicts approximately 50% of analgesia. An R² value of 0.62 was estimated when comparing the human oral MME factor to the oral analgesia ED₅₀, which dropped to R² of 35% when comparing human oral MME factor to the subcutaneous ED₅₀ data. This indicates that our oral administration captures a reasonable amount of the pharmacokinetic factors that influence the MME factor in humans.

We have considered the potential limitation of relating brain CYP2D and supraspinal drug concentrations to analgesia from the tail-flick assay as it primarily measures spinal reflex to a noxious thermal stimulus. However, in a prior publication, we have compared the results from the hotplate, an assay that measures supraspinal response to a noxious thermal stimulus (Le Bars et al., 2001), to the tail flick assay and both showed a similar impact of altering CYP2D in the brain in codeine-induced analgesia (McMillan and Tyndale, 2015). Additionally, intracerebral injections of μ-agonists morphine and sufentanil produced robust and sustained analgesia for at least an hour in the tail- flick assay suggesting that the supraspinal opioid system does impact the tail-flick latency(Jensen and Yaksh, 1986; Yoshikawa et al., 2007).

5. Conclusion

In conclusion, selective inhibition of CYP2D in the brain results in higher analgesia from oral hydrocodone with no impact on peripheral hydrocodone concentrations. Like oxycodone and tramadol, we speculate the parent compound hydrocodone in brain contributes to opioid analgesia, despite having a lower μ-opioid receptor affinity compared to its CYP2D-mediated metabolite. Together, variation in CYP2D metabolism within the brain may be an important source of interindividual differences in response to oral hydrocodone and other centrally-acting CYP2D substrates.

Abbreviations:

CYP2D

Cytochrome P450 2D

icv

Intracerebroventricular

PRL

Propranolol

VEH

Vehicle

PM

CYP2D poor metabolisers

EM

CYP2D extensive metabolisers

sc

Subcutaneous

MME

Morphine milligram equivalence