Analgesic and Respiratory Depressant Effects of R-dihydroetorphine: A Pharmacokinetic–Pharmacodynamic Analysis in Healthy Male Volunteers

Despite their many side effects, opioids are the cornerstone of contemporary treatment of moderate to severe pain in patients with cancer and perioperative patients.¹  Respiratory depression is one of the more serious opioid side effects. Commonly used opioids, such as morphine and fentanyl, produce respiratory depression via activation of μ-opioid receptors expressed on pontine respiratory neurons with the cessation of breathing activity (apnea) at high doses.^2,3  Until recently, few studies addressed the (positive or negative) contribution of the other opioid receptors, such as κ- and δ-opioid receptors, to respiratory depression. We previously showed both in rodents and humans that buprenorphine, a partial agonist at the μ-opioid receptor, antagonist at κ-opioid receptor, and agonist at the nociception/orphanin FQ receptor, produces an apparent maximum in respiratory effect, even at full μ-opioid receptor occupancy.^4,5  This may be a major advantage over other opioids and suggests some protective effect at high dose. The molecular mechanism of this phenomenon (also called “ceiling” or “plateau” effect) has not yet been elucidated. Animal data suggest that apart from partial agonism at the μ-opioid receptor, the apparent maximum in respiratory depression may be related to the activation of the nociception/orphanin FQ receptor.⁶  On the other hand, there are data to suggest that activation of δ-opioid and/or κ-opioid receptors has some respiratory protective effect.⁷  For example, DPI-125, a mixed agonist acting at μ-opioid, δ-opioid, and κ-opioid receptors, has a high respiratory safety profile in rodents (as measured by the ratio carbon dioxide concentration increase/ED50, where ED50 is the dose causing an increase in antinociception by 50%), which is attributed to high δ-opioid receptor potency;⁸  and the κ-opioid receptor agonist U50,488H antagonizes μ-opioid receptor–induced respiratory depression in the rat.⁹  These findings suggest that opioid agonists acting at nociception/orphanin FQ, δ-opioid, and κ-opioid receptors may counteract, at least in part, the respiratory depression induced by the activation of the μ-opioid receptor.

In this study in healthy volunteers, we assessed the respiratory and analgesic effects of R-7,8-dihydro-7α-[1-(R)-hydroxy-1-methylbutyl]-6,14-endo-ethanotetrahydro-oripavine (R-dihydroetorphine; fig. 1), an opioid with full agonism and high affinity for the μ-opioid (Ki 0.10 nM), κ-opioid (Ki 0.74 nM), and δ-opioid receptors (Ki 1.5 nM), and low affinity for the nociception/orphanin FQ receptor (Ki 120 nM; Mundipharma Research Ltd., unpublished observation).¹⁰  We performed a population pharmacokinetic–pharmacodynamic modeling study, which allowed us to construct concentration-effect relationships for the wanted and adverse effects of the drug and construct safety or utility surfaces, which give information on the probability of drug harm in the light of its benefit.^11–13  Given the R-dihydroetorphine receptor affinity profile, we hypothesize that R-dihydroetorphine has an apparent maximum in respiratory effect and a favorable utility surface compared to full μ-opioid receptor agonists, such as fentanyl.^11–13  Although R-dihydroetorphine is currently not available in the western world for human clinical use, this specific opioid may serve as an example of the influence of combined μ-opioid, κ-opioid, and δ-opioid receptor agonism on ventilatory control. Additionally, it may become available in the near future given its possible advantageous over other “classical” opioids.

This study was performed in the Anesthesia and Pain Research Unit of the Department of Anesthesiology of the Leiden University Medical Center (Leiden, The Netherlands) from March 2011 until November 2011. The protocol was approved by the Institutional Review Board (Commissie Medische Ethiek) and the Central Committee on Research Involving Human Subjects in The Hague. The study was registered at the EU Clinical Trials register identification No. 2009-010880-17 on January 25, 2010. Before enrollment and after being informed about the study, all participants gave written informed consent. All study procedures were conducted according to good clinical practice guidelines and adhered to the tenets of the Declaration of Helsinki. The study protocol consisted of experiments on R-dihydroetorphine, fentanyl, and placebo, and was performed in 92 volunteers. The descriptive results of the complete data set have been published before without disclosure of the nature of the opioid.¹⁴  Here, we present the R-dihydroetorphine pharmacokinetic–pharmacodynamic analysis.

Forty healthy male volunteers successfully completed the R-dihydroetorphine part of the study. Inclusion criteria were: age, 18 to 45 yr; weight, 60 to 100 kg; body mass index, 18 to 30 kg/m²; forced expired lung volume in 1 s, greater than 85% of predicted; and no history of major medical disease, alcohol abuse, or illicit drug use. The use of medication was not allowed in the 7 days before the study (including vitamins); the use of opioids was not allowed in the 3 months before the study. The volunteers were asked to fast for 8 h before drug administration. The volunteers were recruited by the study team using flyers distributed on the university campus. Subjects were enrolled after passing a physical examination performed by an independent physician. Following screening (and passing the physical examination), all subjects were familiarized with the experimental setup and some test runs were performed.

After arrival in the research unit at 8:00 am, the subject received a venous line in the arm or hand (for drug infusion) and an arterial line in the radial artery of the nondominant arm (for blood sampling). Twenty subjects participated in the respiratory part of the study, twenty others in the analgesia part. Each subject participated once in the study in which he received one of four possible R-dihydroetorphine (Mundipharma Research Ltd., Cambridge, United Kingdom) doses: 12.5 ng/kg (n = 4 in the respiratory study, 4 others in the analgesia study), 75 ng/kg (n = 6 and 6 others), 125 ng/kg (n = 6 and 6), and 150 ng/kg (n = 4 and 4). After obtaining baseline data, R-dihydroetorphine was infused during 10 min using a syringe pump (Beckton Dickinson, St. Etienne, France).

The study was randomized and double blind. Randomization was done by the pharmacy using a computer-generated randomization list. The subjects were randomized 4:6:6:4 (R-dihydroetorphine dose 12.5, 75, 125, and 150 ng/kg) and allocated on the day before the experiment. The pharmacy prepared the study medication. On the morning of the experiment, the study team received a 50-ml syringe with the study drug that was marked with just the subject randomization number. All study personnel and data analysts remained blinded until the data collection was complete and an independent monitor indicated that all data entries were correct and complete.

In 20 subjects, isohypercapnic ventilation was measured from 5 to 10 min before and for 70 min after the start of the 10-min drug infusion. To maintain isohypercapnia and collect ventilation data, we used the dynamic end-tidal forcing technique, which is described in detail elsewhere.^15,16  In brief, subjects breathed through a facemask that was attached to a pneumotachograph and pressure transducer system (#4813; Hans Rudolph Inc., USA) and to a set of mass flow controllers (Bronkhorst High Tech, The Netherlands) for the delivery of oxygen, carbon dioxide, and nitrogen. The mass flow controllers were controlled by a computer running RESREG/ACQ software (Leiden University Medical Center, The Netherlands) that steers end-tidal gas concentrations by varying the inspired concentrations and captures ventilatory data. The inspired and expired oxygen and carbon dioxide concentrations were measured at the mouth using a capnograph (Datex Capnomac, Finland); arterial oxygen saturation was measured by pulse oximetry (Masimo Corporation, USA). In the pharmacokinetic–pharmacodynamic analysis, we used 1-min ventilation averages.

In 20 subjects, the response to activation of cutaneous nociceptors was measured.^11,17  Using a computer-controlled constant current stimulator (Leiden University Medical Center), a 0.1-ms stimulus at 20 Hz was applied to the skin over the tibial bone. The stimulus train increased by 0.5 mA per second from 0 mA to a maximum of 128 mA. The subject was instructed to press a button on a control box at the first sensation of pain (i.e., pain threshold) and end the stimulus train by pressing another button when pain reached the level of tolerance (i.e., pain tolerance). Four pain threshold and tolerance values were obtained in the 30 min prior to drug infusion. These values were averaged and served as baseline value. Further measurements were obtained at t = 10, 15, 30, 45, 60, 75, 90, 105, 120, 150, 180, 210, 240, 300, 360, 420, and 480 min after the start of drug infusion (t = 0 is the start of drug infusion; t = 10 min is end of infusion).

Before (baseline sample), during, and after drug infusion (drug infusion from 0 to 10 min), 2 to 3 ml blood were drawn from the arterial line for measurement of R-dihydroetorphine concentrations in plasma. In the first hour, samples were obtained at t = 2, 5, 8, 10, 12, 15, 18, 20 25, 30, 40, and 60 min, followed in the subsequent hours by samples obtained at t = 1.5, 2, 3, 4, 5, 12, and 24 h. Plasma was separated within 30 min of blood collection and stored at −20°C until analysis. Concentrations of R-dihydroetorphine were determined in human plasma samples using a validated analytical methodology by LGC Ltd. (Fordham, United Kingdom) by liquid chromatography or tandem mass spectrometric method, with stable isotope labeled internal standard incorporating deuterium atoms to distinguish the standard by mass. Chromatographic separation was by ultrahigh performance liquid chromatography using a reverse phase analytical column (100 mm × 2.1 mm, 1.7 µm internal diameter, C18) with a gradient elution using 0.1% (volume/volume %) acetic acid in acetonitrile and water. Samples were analyzed, following precipitation with acetonitrile, using a Sciex API 5000 triple quadrupole mass spectrometer (Sciex, USA) using electrospray ionization. Single specific multiple reaction transitions were monitored for both R-dihydroetorphine and internal standard. The validated analytical range of this assay was 5.00 to 2500 pg/ml with 5.00 pg/ml being the lowest limit of quantification. The interassay precision, as measured by coefficient of variation, at 5.00 pg/ml, 15.0 pg/mL, 100 pg/ml, and 2000 pg/ml was 10.5%, 5.2%, 5.3%, and 6.1%, respectively. The intra-assay accuracy, as measured by relative error, was less than 2.4% and the interassay accuracy was less than 8.8%.

The pharmacokinetic–pharmacodynamic data were analyzed with the mixed-effects modeling software package NONMEM VII (Nonlinear Mixed Effects Modeling; ICON Development Solutions, USA) using a population approach. A two-stage analysis was chosen. From the first stage (pharmacokinetic analysis), individual empirical Bayesian estimates of the pharmacokinetic parameters were obtained and were applied in the second stage (pharmacodynamic analysis; analgesic and respiratory data were analyzed separately). To eliminate the hysteresis between the estimated R-dihydroetorphine concentrations and effect, an effect compartment was postulated. This effect compartment equilibrates with the plasma compartment with rate constant k_e0. To obtain the “best” pharmacokinetic model, two and three compartment models were fitted to the R-dihydroetorphine pharmacokinetic data. The number of compartments in the model was determined by the magnitude of the decrease in the minimum objective function value (χ²-test, P < 0.01 considered significant). Weight and body mass index were considered as covariates.

The 1-min ventilation averages were modeled as follows:^11,12 

where V_B is (predrug) baseline ventilation, V_MIN minimum ventilation, Ce(t) R-dihydroetorphine effect-site concentration, and C₅₀ the effect-site or steady-state concentration causing 50% of the effect between V_B and minimum ventilation (an estimated value of zero for minimum ventilation indicates that at high drug concentrations apnea will occur). γ is a shape parameter and TR a linear trend term. The trend term describes a possible increase in ventilation over time. Such an increase has been observed previously by several research groups and is related to the presence of isohypercapnia.^15,18,19  For example, Dahan et al.¹⁵  observed a trend of 30 to 100 ml/min²  during 15 to 20 min of isohypercapnia. This effect is most prominent during the first hour of isohypercapnia after which no further increase in ventilation is observed.¹⁸  The rise in ventilation is possibly related to slow central neuronal dynamics in response to central chemoreceptor activation.¹⁹  Most μ-opioid receptor agonists tend to inactive central neuronal dynamics with no need for incorporating a trend term when modeling their effect on ventilation.^5,11  However, in the current study we did observe a trend in the ventilation data, and indication that R-dihydroetorphine does not impair central neuronal dynamics.

The threshold and pain responses were analyzed simultaneously as follows:^11,12 

where Pain Response (t) is the stimulus intensity at which the subjects pressed the control button to indicate his pain threshold or tolerance, Baseline is the predrug stimulus intensity at which pain threshold and tolerance were reported, Ce(t) the R-dihydroetorphine effect-site concentration, C₅₀ the effect-site or steady-state concentration at which a 50% increase in stimulus intensity results in a response (pain threshold or tolerance), and γ is a shape parameter.

Model parameters were assumed to be log-normally distributed, except V_B, minimum ventilation and C, which were assumed to be normally distributed. Residual error was assumed to have both an additive and a relative error for concentrations and only an additive error for all effect parameters. Covariance between random effects (η’s) for the three pharmacodynamic end-points were explored using $OMEGA BLOCKs. P values less than 0.01 were considered significant.

To allow a visual predictive check of the final pharmacokinetic or pharmacodynamic models, we estimated the normalized prediction discrepancies in NONMEM.^20,21  To that end, we performed 300 Monte Carlo simulations that were based on the final model considering the distributions of the fixed and random effects. Next, we counted the number of times an observation is greater than the model prediction. The normalized prediction discrepancies are the counts divided by 300, transformed via the inverse normal distribution. Under the null hypothesis that the model is correct, the normalized prediction discrepancies should have a normal distribution. It was visually checked that the normalized prediction discrepancies versus time showed no trends, heteroscedasticity, or both.

For a detailed explanation of the step-by-step construction of the utility surfaces, see Boom et al.¹¹and Roozekrans et al.¹² In brief, we developed utility surfaces that give information on four possible conditions: (1) probability of adequate analgesia without serious respiratory depression, (2) probability of serious respiratory depression without analgesia, (3) probability of absence of respiratory depression and absence of adequate analgesia, and (4) probability of adequate analgesia with serious respiratory depression. The threshold values for adequate analgesia was P(A) > 0.25 and P(A) > 0.5, or the probability of a 25% (A+) or 50% (A++) increase in tolerated electrical current, and for serious respiratory depression P(R) > 0.5 or a reduction in minute ventilation by more than 50% (R+). Additionally, we calculated the probability of less than 25% analgesia (A−, or no analgesia) and less than 50% respiratory depression (R−, no respiratory depression). To obtain these surfaces, we simulated 10,000 pharmacodynamic profiles as function of time (for doses 12.5, 75, 125, and 150 ng/kg) and steady-state or effect-site concentration (from 0 to 100 pg/ml), according to the typical model parameter estimates and the interindividual variances (ω²), as derived from the final pharmacokinetic–pharmacodynamic analyses. Probabilities were calculated from the distribution of occurrence of specific conditions (1 to 4). This process yields iso-utility lines that describe the conditions 1 to 4. By giving specific colors to these four conditions (for example, dark red for the least desirable condition [condition 2] and green for most desirable condition [condition 1]), transitions in-between conditions are depicted by transitions between colors and result in a continuum of probabilities.

The 40 subjects were 23 ± 2 (mean ± SD; range, 18 to 44) yr old, had a body weight of 78 ± 8 (65 to 98) kg, and a body mass index of 23 ± 2 (19 to 23) kg/m². All subjects completed the protocol without serious or unexpected adverse effects. All reported side effects are given in table 1.

We observed a dose-dependent increase in (mean ± SD) plasma R-dihydroetorphine concentrations with peak concentrations (occurring at the end of the 10-min infusion) of 44.8 ± 5.7, 214.2 ± 20.1, 371.1 ± 30.7, and 454.4 ± 42.1 pg/ml at doses 12.5, 75, 125, and 150 ng/kg, respectively (fig. 2). After the termination of the infusion the plasma concentrations dropped rapidly, with a more than 50% decrease in plasma concentrations within 2 min. The final pharmacokinetic model consisted of a conventional three-compartment model with one central (V₁) and two peripheral compartments (V₂ and V₃) that was superior to a two-compartment model with just one peripheral compartment (minimum objective function value = 3,120.368 vs. 3,314.316, P < 0.001). No effect of weight or body mass index was observed on any of the model parameters, most probably due to the fact that we had a homogenous sample of healthy young males weighing 78 ± 8 kg with a body mass index of 23 ± 2 kg/m². Best, median and worst fits are given in figure 3 and goodness of fit plots in figure 4. Inspection of the data fits and all three diagnostic plots (individual predicted vs. measured data, individual weighted residuals vs. time, and normalized prediction discrepancies) indicate that the three-compartment pharmacokinetic model adequately described the data. The pharmacokinetic parameter estimates, given in table 2, show the relatively small value of V₁ (7.7 l), the large clearance from compartment 1 (68 l/h) and equally large intercompartmental clearances. There were significant covariances between clearances 1 and 2 (ω² ± standard error of the estimate = 0.05 ± 0.3), 1 and 3 (0.09 ± 0.03), and 2 and 3 (0.14 ± 0.06).

The ventilation and pain responses were adequately fitted by the pharmacodynamic model (see figs. 3 and 4 for best, medium and worst fits and three goodness-of-fit plots, and table 2 for model parameter estimates). Over the dose range tested, an apparent maximum in respiratory depression greater than zero was identified (minimum ventilation = 6.5 l/min or 33% of baseline ventilation). The pharmacodynamic model with an apparent maximum in ventilation was statistically superior to the model with maximum effect at apnea (minimum ventilation = 0 l/min) with minimum objective function values of 2,243.965 versus 2,354.407 (P < 0.001), respectively.

In contrast, a maximum in antinociception was not reached at the maximum R-dihydroetorphine dose tested. However, the pharmacodynamic model with and without an apparent maximum did not differ in terms of minimum objective function values (values of model without an apparent maximum 1,969.526 and with an apparent maximum 1,967.408, P = 0.150), indicative that we cannot exclude that at higher doses than tested by us an apparent maximum in the analgesic response may occur.

The delay between plasma concentration and effect (t_½k_e0) differed between the two end-points by a factor of 2 (ventilation 0.95 ± 0.20 h vs. pain relief 2.19 ± 0.49 h). The ventilation potency parameter (C_50,V) or the effect-site or steady-state concentration causing 50% of the effect between baseline ventilation and minimum ventilation was 17 ± 4 pg/ml (at this R-dihydroetorphine concentration the ventilation level was 67% of baseline ventilation; the R-dihydroetorphine concentration causing 50% decrease of baseline ventilation was 40% higher, i.e., 27 pg/ml). The analgesia potency parameter (C_50,A) was 34 ± 11 pg/ml; at this concentration the pain threshold and tolerance responses occurred at a 50% increase relative to pre-drug baseline values. We observed a significant covariance between pain threshold and tolerance with ω² = 0.08 ± 0.04. The steady-state relationships between R-dihydroetorphine plasma concentration and effects are given in figure 5 (ventilation red line; pain relief blue line).

The constructed utility surfaces or the continuum of probabilities of R-dihydroetorphine analgesia in the presence or absence of respiratory depression are shown in figures 6 and 7. In figure 6A the probabilities are plotted as function of R-dihydroetorphine effect-site concentration (with the iso-utility lines in fig. 6B). The different conditions that may co-exist are depicted by colors that correspond with analgesia and respiratory thresholds. The colors range from deep green (A++/R− or at least 50% analgesia/no respiratory depression) to green (A+/R− or at least 25% analgesia/no respiratory depression), to yellow (A−/R− or no analgesia/no respiratory depression), and from deep red (A−/R+ or no analgesia/serious respiratory depression) to dark orange (A+/R+ or at least 25% analgesia/serious respiratory depression) and to orange (A++/R+ or at least 50% analgesia/serious respiratory depression). At an R-dihydroetorphine effect-site concentration of 20 pg/ml, the probability of analgesia was 60%, while the probability analgesia without respiratory depression was 45%. At increasing R-dihydroetorphine effect-site concentrations the probability of analgesia increased toward 95% at 100 pg/ml, but the probability of analgesia without respiratory depression was reduced to 20%. The probabilities as function of time for the four doses administered in the study (12.5, 75, 125, and 150 ng/kg; simulated infusion duration is 90 s, enabling comparison with earlier studies)^11,12  are given in figure 6. At all doses, the probability of respiratory depression without analgesia was 5 to 10% (dark red surfaces). The ratio between green and orange surfaces decreased at increasing R-dihydroetorphine doses: 12.5 ng/kg 469%, 75 ng/kg 205%, 125 ng/kg 151%, and 150 ng/kg 133%.

Dihydroetorphine is a six-ring semisynthetic opioid alkaloid (fig. 1), first synthesized in 1967 by Reckitt and Sons Ltd. in England.²²  It is a derivative of thebaine like morphine, hydromorphone, and oxycodone (all five-ring opioid molecules), but like buprenorphine it is a six-ring opioid due to a 6,14-endo-ethano-bridge. In etorphine, another six-ring opioid, the 6,14-endo-ethano-bridge is oxidized to a 6,14-endo-etheno-bridge. Six-ring opioids are characterized by high affinity to opioid receptors (K_i in the nanomolar range).²³  Importantly, the dihydroetorphine molecule has a chiral center at C19 and consequently exists in R− and S-configurations.²³  Studies in the rabbit show that R-dihydroetorphine is a potent analgesic, about 6,000 to 12,000 times more potent than morphine when administered parenterally and with an improved respiratory depression/analgesia ratio compared to morphine.²⁴  Also, human studies indicate that R-dihydroetorphine has high analgesic potency with only mild side effects.¹⁰  Since 1992, R-dihydroetorphine is registered in mainland China for the treatment of severe pain, including labor pain, and anesthesia induction,^10,25  but it is not used clinically in the Western world. To better understand its respiratory effects in relation to its analgesic effects, we studied the effect of four R-dihydroetorphine doses on isohypercapnic ventilation and antinociception in healthy volunteers.

Over the dose range tested (12.5 to 150 ng/kg), we observed an apparent plateau in respiratory depression but not in analgesia. This suggests that R-dihydroetorphine has advantages over other opioids that do not display any plateau in respiratory depression and produce respiratory instability or apnea at high dose (e.g., morphine, fentanyl). However, this is a first and relatively small (n = 40) study on R-dihydroetorphine performed in healthy volunteers under highly-controlled experimental conditions. Whether the apparent plateau is also sustained at higher doses (greater than 150 ng/kg), is present in specific patient populations with a high risk of opioid-induced respiratory depression, or occurs at more intense clinical pain requires further testing. The plateau effect observed at the current R-dihydroetorphine doses is similar to that of other opioids, such as buprenorphine, that shares certain pharmacologic characteristics with R-dihydroetorphine (see two paragraphs below).^4,5 

In order to determine the respiratory effect of R-dihydroetorphine relative to its analgesic effect, we constructed utility surfaces (figs. 6 and 7) rather than calculating the therapeutic ratio (which assumes a similar dose-response relationship with just differences in potency between end-points).^11–13,26  Since our analysis shows that the two end-points differ in their concentration-effect curves and C₅₀s (fig. 5), our approach is more suitable to integrate wanted and unwanted end-points into one function.^26,27  The utility surfaces are based on utility functions that describe the opioids effect in terms of probability of benefit (analgesia) and probability of harm (respiratory depression). For R-dihydroetorphine, we showed a larger probability of analgesia without respiratory depression (green surfaces) than the phenylpiperidines that we tested previously using the same experimental paradigm.^11–13  Based on the utility surfaces that we constructed, R-dihydroetorphine seems an opioid analgesic with a greater benefit than harm, although this has to be considered in light of the current experimental paradigm and requires replication in clinical studies. Still, we argue that it is important to obtain a library of utility surfaces for all clinically available opioids and determine whether such characterizations correlate with clinical observations of opioid-induced respiratory depression and opioid-related fatalities. Our current approach of constructing utility functions is complex as it requires population pharmacokinetic and pharmacodynamic analyses. To simplify matters, we recently proposed a more pragmatic method that enables the construction of utility surfaces without the need for pharmacokinetic data.¹³ 

Our protocol was not designed to clarify the mechanism of the apparent respiratory plateau. However, we believe that this is an important issue that deserves some discussion, and the receptor profile of R-dihydroetorphine, in comparison to that of other opioids (e.g., buprenorphine), may help us identify possible mechanisms. For buprenorphine, the apparent plateau in ventilatory depression is considered related to partial agonism at the μ-opioid receptor (restricting respiratory effect despite full receptors occupancy) and/or full agonism at the nociception/orphanin FQ receptor (with nociception/orphanin FQ receptor activity reducing the respiratory effect from μ-opioid receptor activation).^4–6,28  Since R-dihydroetorphine is a full agonist at the μ-opioid receptor, the first mechanism seems highly improbable. The nociception/orphanin FQ receptor activation may be a possible mechanism for the reduced R-dihydroetorphine respiratory effect at high dose (see for example, Dahan et al. and Linz et al. on the combined μ-opioid and nociception/orphanin FQ receptor agonist cebranopadol).^29,30  However, although R-dihydroetorphine has some affinity for the nociception/orphanin FQ receptor, its affinity is much lower than for the μ-opioid receptor (Ki 0.1 nM vs. 120 nM). Whether such low nociception/orphanin FQ receptor affinity is sufficient to cause profound respiratory protection is questionable. Another possible mechanism may be related to the relatively high affinity of R-dihydroetorphine for the κ-opioid and δ-opioid receptors. Notably, when assessing functional activity of μ-opioid , κ-opioid and δ-opioid receptors, following R-dihydroetorphine receptor activation, μ-opioid and κ-opioid receptors have essentially similar IC₅₀-values for inhibition of cyclic adenosine monophosphate formation, while the IC₅₀-value for δ-opioid receptors is one tenth of that for the μ-opioid receptor (Mundipharma Research Ltd., unpublished observation). There is evidence that κ-opioid and δ-opioid receptor agonists may selectively antagonize μ-opioid receptor agonistic effects, including respiratory depression.^7–9  Finally, other proposed mechanisms involve the intracellular regulatory protein β-arrestin or off-target activity, such as activation of toll-like receptor 4 or the σ₁-receptor.^31–34  Opioid receptors belong to the 7-transmembrane G-protein-coupled receptors that, upon activation, bind to intracellular G-proteins and β-arrestin proteins.^31,32  Based on prolonged morphine-induced analgesia in β-arrestin-2 knockout mice, it was suggested that analgesic efficacy is driven by G-protein activation and side effects such as constipation and tolerance are mediated by β-arrestin.^35,36  However, the β-arrestin effect was limited to morphine and was not observed for fentanyl.³⁶  Further, R-dihydroetorphine is biased toward the β-arrestin pathway; this contradicts the notion that β-arrestin bias results in increased adverse effects such as respiratory depression (Mundipharma Research Ltd., unpublished observation). As mentioned, some opioid side effects could be mediated by off-target effects such as toll-like receptor 4 activation, which has been described for several opioids, including morphine, fentanyl, buprenorphine, and oxycodone.³³  For example, remifentanil-induced hyperalgesia depends on toll-like receptor 4 in mice, and toll-like receptor 4 antagonist TAK-242 attenuated morphine-induced suppression of colonic motility in mice.^37,38  In summary, the mechanisms involved in compound-related differences of tolerability and (respiratory) safety still remain elusive, but appear to be mediated by several factors such as β-arrestin bias, opioid receptor profile and off-target activity. Identifying compounds that show improved safety and tolerability profile will require careful testing of different end-points (such as gastrointestinal motility, immune cells activation and respiratory depression) and assess a favorable utility based on analgesic activity.

We included a trend term in the pharmacodynamic model to account for the observation that at 12.5 ng/kg, R-dihydroetorphine some respiratory stimulation during the 70 min measurement period occurred. As stated in the Methods section, sustained hypercapnia may activate central neuronal dynamics causing some respiratory stimulation.^15,18,19  Due to the inclusion of the trend term, we may have underestimated the respiratory protective effects of R-dihydroetorphine somewhat. Analysis without parameter C is given in figure 6 (yellow line) and yielded a C_50,V of 35 pg/ml and minimum ventilation of 8.9 l/min (44% of baseline), albeit at a significantly higher objective function than the model with trend term. κ-opioid receptor agonists have been associated with dysphoria.³⁹  Interestingly, in our study, during a 24-h observation period, no significant dysphoric effects were observed from R-dihydroetorphine (table 1). We tested only men in the current study. Since men and women quantitatively differ in their opioid analgesic and respiratory depressive effects,^40–43  future studies should compare the R-dihydroetorphine effects in women and men.

The final question that remains is whether our experimental observations can be extrapolated to the clinical setting, and whether the use of R-dihydroetorphine would lead to less respiratory events in treated patients compared to other commonly used opioids. This is a highly relevant topic as there has been a recent increase in the number of fatalities from misuse or abuse of legally prescribed opioids. While a reduced respiratory effect (especially at high dose) and favorable utility surface are certainly advantages over other opioids that lack such a profile, it is important to realize that in real life comedication, underlying respiratory, cardiac and/or renal disease, genetics, overdosing, sex, and age, among others, play an important role as well.⁴⁴  Still, our observation that R-dihydroetorphine (greater than the dose-range tested) has favorable pharmacodynamics and utility surface gives this analgesic an advantage over other commonly used analgesics such as fentanyl. For example, when comparing the utility surfaces of R-dihydroetorphine (fig. 6A) and fentanyl (fig. 8),¹²  it is obvious that R-dihydroetorphine has a much greater green surface, indicative of a high probability of analgesia without respiratory depression, even at high effect-site concentrations. In future studies, we will quantify opioid respiratory and antinociceptive effects in individuals with identified risk factors for respiratory depression. Given the above, it is imperative not to infer clinical utility from our results. The utility function is an experimental tool that is developed to compare opioids under artificial study conditions, with preset definitions of pain and respiratory depression. In future studies, we plan to validate the clinical use of the utility functions.

In conclusion, over the dose-range tested, R-dihydroetorphine exhibited an apparent plateau in respiratory depression but not in analgesia.

This trial was sponsored by Mundipharma Research Ltd. (Cambridge, United Kingdom) and supported by departmental and institutional funds.

At the time the study was conducted, Mr. Baily and Dr. Smith were employees of Mundipharma Research Ltd., Cambridge, United Kingdom. Dr. Oksche is an employee of Mundipharma Research Ltd., Cambridge, United Kingdom. The other authors declare no competing interests.

Full protocol available at: a.dahan@lumc.nl. Raw data available at: a.dahan@lumc.nl.